From 3980bd3b406addb327d858aebd19e229ea340b9a Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Sun, 31 Jul 2016 23:38:36 -0400 Subject: ext4: reserve xattr index for the Hurd The Hurd is using inode fields which restricts it from using more advanced ext4 file system features, due to design choices made over a decade ago. By giving the Hurd an extended attribute index field we allow it to move the translator and author fields out of the core inode fields, and hopefully we can get rid of ugly hacks such as EXT4_OS_HURD and EXT4_MOUNT2_HURD_COMPAT somday. For more information please see: https://summerofcode.withgoogle.com/projects/#5869799859027968 Signed-off-by: Theodore Ts'o --- fs/ext4/xattr.h | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/ext4/xattr.h b/fs/ext4/xattr.h index 69dd3e6566e0..a92e783fa057 100644 --- a/fs/ext4/xattr.h +++ b/fs/ext4/xattr.h @@ -24,6 +24,7 @@ #define EXT4_XATTR_INDEX_SYSTEM 7 #define EXT4_XATTR_INDEX_RICHACL 8 #define EXT4_XATTR_INDEX_ENCRYPTION 9 +#define EXT4_XATTR_INDEX_HURD 10 /* Reserved for Hurd */ struct ext4_xattr_header { __le32 h_magic; /* magic number for identification */ -- cgit v1.2.3-70-g09d2 From 829fa70dddadf9dd041d62b82cd7cea63943899d Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Mon, 1 Aug 2016 00:51:02 -0400 Subject: ext4: validate that metadata blocks do not overlap superblock A number of fuzzing failures seem to be caused by allocation bitmaps or other metadata blocks being pointed at the superblock. This can cause kernel BUG or WARNings once the superblock is overwritten, so validate the group descriptor blocks to make sure this doesn't happen. Cc: stable@vger.kernel.org Signed-off-by: Theodore Ts'o --- fs/ext4/super.c | 18 +++++++++++++++++- 1 file changed, 17 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index c13a4e464738..da3146e8d49a 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -2211,6 +2211,7 @@ void ext4_group_desc_csum_set(struct super_block *sb, __u32 block_group, /* Called at mount-time, super-block is locked */ static int ext4_check_descriptors(struct super_block *sb, + ext4_fsblk_t sb_block, ext4_group_t *first_not_zeroed) { struct ext4_sb_info *sbi = EXT4_SB(sb); @@ -2241,6 +2242,11 @@ static int ext4_check_descriptors(struct super_block *sb, grp = i; block_bitmap = ext4_block_bitmap(sb, gdp); + if (block_bitmap == sb_block) { + ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " + "Block bitmap for group %u overlaps " + "superblock", i); + } if (block_bitmap < first_block || block_bitmap > last_block) { ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " "Block bitmap for group %u not in group " @@ -2248,6 +2254,11 @@ static int ext4_check_descriptors(struct super_block *sb, return 0; } inode_bitmap = ext4_inode_bitmap(sb, gdp); + if (inode_bitmap == sb_block) { + ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " + "Inode bitmap for group %u overlaps " + "superblock", i); + } if (inode_bitmap < first_block || inode_bitmap > last_block) { ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " "Inode bitmap for group %u not in group " @@ -2255,6 +2266,11 @@ static int ext4_check_descriptors(struct super_block *sb, return 0; } inode_table = ext4_inode_table(sb, gdp); + if (inode_table == sb_block) { + ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " + "Inode table for group %u overlaps " + "superblock", i); + } if (inode_table < first_block || inode_table + sbi->s_itb_per_group - 1 > last_block) { ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " @@ -3757,7 +3773,7 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) goto failed_mount2; } } - if (!ext4_check_descriptors(sb, &first_not_zeroed)) { + if (!ext4_check_descriptors(sb, logical_sb_block, &first_not_zeroed)) { ext4_msg(sb, KERN_ERR, "group descriptors corrupted!"); ret = -EFSCORRUPTED; goto failed_mount2; -- cgit v1.2.3-70-g09d2 From 372ee16386bbf6dc5eeb0387e1ede963debba82a Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 3 Aug 2016 14:11:40 +0100 Subject: rxrpc: Fix races between skb free, ACK generation and replying Inside the kafs filesystem it is possible to occasionally have a call processed and terminated before we've had a chance to check whether we need to clean up the rx queue for that call because afs_send_simple_reply() ends the call when it is done, but this is done in a workqueue item that might happen to run to completion before afs_deliver_to_call() completes. Further, it is possible for rxrpc_kernel_send_data() to be called to send a reply before the last request-phase data skb is released. The rxrpc skb destructor is where the ACK processing is done and the call state is advanced upon release of the last skb. ACK generation is also deferred to a work item because it's possible that the skb destructor is not called in a context where kernel_sendmsg() can be invoked. To this end, the following changes are made: (1) kernel_rxrpc_data_consumed() is added. This should be called whenever an skb is emptied so as to crank the ACK and call states. This does not release the skb, however. kernel_rxrpc_free_skb() must now be called to achieve that. These together replace rxrpc_kernel_data_delivered(). (2) kernel_rxrpc_data_consumed() is wrapped by afs_data_consumed(). This makes afs_deliver_to_call() easier to work as the skb can simply be discarded unconditionally here without trying to work out what the return value of the ->deliver() function means. The ->deliver() functions can, via afs_data_complete(), afs_transfer_reply() and afs_extract_data() mark that an skb has been consumed (thereby cranking the state) without the need to conditionally free the skb to make sure the state is correct on an incoming call for when the call processor tries to send the reply. (3) rxrpc_recvmsg() now has to call kernel_rxrpc_data_consumed() when it has finished with a packet and MSG_PEEK isn't set. (4) rxrpc_packet_destructor() no longer calls rxrpc_hard_ACK_data(). Because of this, we no longer need to clear the destructor and put the call before we free the skb in cases where we don't want the ACK/call state to be cranked. (5) The ->deliver() call-type callbacks are made to return -EAGAIN rather than 0 if they expect more data (afs_extract_data() returns -EAGAIN to the delivery function already), and the caller is now responsible for producing an abort if that was the last packet. (6) There are many bits of unmarshalling code where: ret = afs_extract_data(call, skb, last, ...); switch (ret) { case 0: break; case -EAGAIN: return 0; default: return ret; } is to be found. As -EAGAIN can now be passed back to the caller, we now just return if ret < 0: ret = afs_extract_data(call, skb, last, ...); if (ret < 0) return ret; (7) Checks for trailing data and empty final data packets has been consolidated as afs_data_complete(). So: if (skb->len > 0) return -EBADMSG; if (!last) return 0; becomes: ret = afs_data_complete(call, skb, last); if (ret < 0) return ret; (8) afs_transfer_reply() now checks the amount of data it has against the amount of data desired and the amount of data in the skb and returns an error to induce an abort if we don't get exactly what we want. Without these changes, the following oops can occasionally be observed, particularly if some printks are inserted into the delivery path: general protection fault: 0000 [#1] SMP Modules linked in: kafs(E) af_rxrpc(E) [last unloaded: af_rxrpc] CPU: 0 PID: 1305 Comm: kworker/u8:3 Tainted: G E 4.7.0-fsdevel+ #1303 Hardware name: ASUS All Series/H97-PLUS, BIOS 2306 10/09/2014 Workqueue: kafsd afs_async_workfn [kafs] task: ffff88040be041c0 ti: ffff88040c070000 task.ti: ffff88040c070000 RIP: 0010:[] [] __lock_acquire+0xcf/0x15a1 RSP: 0018:ffff88040c073bc0 EFLAGS: 00010002 RAX: 6b6b6b6b6b6b6b6b RBX: 0000000000000000 RCX: ffff88040d29a710 RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff88040d29a710 RBP: ffff88040c073c70 R08: 0000000000000001 R09: 0000000000000001 R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000 R13: 0000000000000000 R14: ffff88040be041c0 R15: ffffffff814c928f FS: 0000000000000000(0000) GS:ffff88041fa00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fa4595f4750 CR3: 0000000001c14000 CR4: 00000000001406f0 Stack: 0000000000000006 000000000be04930 0000000000000000 ffff880400000000 ffff880400000000 ffffffff8108f847 ffff88040be041c0 ffffffff81050446 ffff8803fc08a920 ffff8803fc08a958 ffff88040be041c0 ffff88040c073c38 Call Trace: [] ? mark_held_locks+0x5e/0x74 [] ? __local_bh_enable_ip+0x9b/0xa1 [] ? trace_hardirqs_on_caller+0x16d/0x189 [] lock_acquire+0x122/0x1b6 [] ? lock_acquire+0x122/0x1b6 [] ? skb_dequeue+0x18/0x61 [] _raw_spin_lock_irqsave+0x35/0x49 [] ? skb_dequeue+0x18/0x61 [] skb_dequeue+0x18/0x61 [] afs_deliver_to_call+0x344/0x39d [kafs] [] afs_process_async_call+0x4c/0xd5 [kafs] [] afs_async_workfn+0xe/0x10 [kafs] [] process_one_work+0x29d/0x57c [] worker_thread+0x24a/0x385 [] ? rescuer_thread+0x2d0/0x2d0 [] kthread+0xf3/0xfb [] ret_from_fork+0x1f/0x40 [] ? kthread_create_on_node+0x1cf/0x1cf Signed-off-by: David Howells Signed-off-by: David S. Miller --- Documentation/networking/rxrpc.txt | 21 ++-- fs/afs/cmservice.c | 78 ++++++------- fs/afs/fsclient.c | 221 ++++++++++++------------------------- fs/afs/internal.h | 14 ++- fs/afs/rxrpc.c | 73 +++++++----- fs/afs/vlclient.c | 11 +- include/net/af_rxrpc.h | 2 +- net/rxrpc/ar-internal.h | 1 + net/rxrpc/call_accept.c | 1 + net/rxrpc/call_event.c | 3 + net/rxrpc/call_object.c | 8 +- net/rxrpc/input.c | 12 +- net/rxrpc/recvmsg.c | 25 +---- net/rxrpc/skbuff.c | 41 +++++-- 14 files changed, 225 insertions(+), 286 deletions(-) (limited to 'fs') diff --git a/Documentation/networking/rxrpc.txt b/Documentation/networking/rxrpc.txt index 16a924c486bf..70c926ae212d 100644 --- a/Documentation/networking/rxrpc.txt +++ b/Documentation/networking/rxrpc.txt @@ -790,13 +790,12 @@ The kernel interface functions are as follows: Data messages can have their contents extracted with the usual bunch of socket buffer manipulation functions. A data message can be determined to be the last one in a sequence with rxrpc_kernel_is_data_last(). When a - data message has been used up, rxrpc_kernel_data_delivered() should be - called on it.. + data message has been used up, rxrpc_kernel_data_consumed() should be + called on it. - Non-data messages should be handled to rxrpc_kernel_free_skb() to dispose - of. It is possible to get extra refs on all types of message for later - freeing, but this may pin the state of a call until the message is finally - freed. + Messages should be handled to rxrpc_kernel_free_skb() to dispose of. It + is possible to get extra refs on all types of message for later freeing, + but this may pin the state of a call until the message is finally freed. (*) Accept an incoming call. @@ -821,12 +820,14 @@ The kernel interface functions are as follows: Other errors may be returned if the call had been aborted (-ECONNABORTED) or had timed out (-ETIME). - (*) Record the delivery of a data message and free it. + (*) Record the delivery of a data message. - void rxrpc_kernel_data_delivered(struct sk_buff *skb); + void rxrpc_kernel_data_consumed(struct rxrpc_call *call, + struct sk_buff *skb); - This is used to record a data message as having been delivered and to - update the ACK state for the call. The socket buffer will be freed. + This is used to record a data message as having been consumed and to + update the ACK state for the call. The message must still be passed to + rxrpc_kernel_free_skb() for disposal by the caller. (*) Free a message. diff --git a/fs/afs/cmservice.c b/fs/afs/cmservice.c index 4b0eff6da674..85737e96ab8b 100644 --- a/fs/afs/cmservice.c +++ b/fs/afs/cmservice.c @@ -189,11 +189,8 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, case 1: _debug("extract FID count"); ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("FID count: %u", call->count); @@ -210,11 +207,8 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, _debug("extract FID array"); ret = afs_extract_data(call, skb, last, call->buffer, call->count * 3 * 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; _debug("unmarshall FID array"); call->request = kcalloc(call->count, @@ -239,11 +233,8 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, case 3: _debug("extract CB count"); ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; tmp = ntohl(call->tmp); _debug("CB count: %u", tmp); @@ -258,11 +249,8 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, _debug("extract CB array"); ret = afs_extract_data(call, skb, last, call->request, call->count * 3 * 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; _debug("unmarshall CB array"); cb = call->request; @@ -278,9 +266,9 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, call->unmarshall++; case 5: - _debug("trailer"); - if (skb->len != 0) - return -EBADMSG; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; /* Record that the message was unmarshalled successfully so * that the call destructor can know do the callback breaking @@ -294,8 +282,6 @@ static int afs_deliver_cb_callback(struct afs_call *call, struct sk_buff *skb, break; } - if (!last) - return 0; call->state = AFS_CALL_REPLYING; @@ -335,13 +321,13 @@ static int afs_deliver_cb_init_call_back_state(struct afs_call *call, { struct afs_server *server; struct in_addr addr; + int ret; _enter(",{%u},%d", skb->len, last); - if (skb->len > 0) - return -EBADMSG; - if (!last) - return 0; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; /* no unmarshalling required */ call->state = AFS_CALL_REPLYING; @@ -371,8 +357,10 @@ static int afs_deliver_cb_init_call_back_state3(struct afs_call *call, _enter(",{%u},%d", skb->len, last); + /* There are some arguments that we ignore */ + afs_data_consumed(call, skb); if (!last) - return 0; + return -EAGAIN; /* no unmarshalling required */ call->state = AFS_CALL_REPLYING; @@ -408,12 +396,13 @@ static void SRXAFSCB_Probe(struct work_struct *work) static int afs_deliver_cb_probe(struct afs_call *call, struct sk_buff *skb, bool last) { + int ret; + _enter(",{%u},%d", skb->len, last); - if (skb->len > 0) - return -EBADMSG; - if (!last) - return 0; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; /* no unmarshalling required */ call->state = AFS_CALL_REPLYING; @@ -460,10 +449,9 @@ static int afs_deliver_cb_probe_uuid(struct afs_call *call, struct sk_buff *skb, _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - if (skb->len > 0) - return -EBADMSG; - if (!last) - return 0; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; switch (call->unmarshall) { case 0: @@ -509,8 +497,9 @@ static int afs_deliver_cb_probe_uuid(struct afs_call *call, struct sk_buff *skb, break; } - if (!last) - return 0; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; call->state = AFS_CALL_REPLYING; @@ -588,12 +577,13 @@ static void SRXAFSCB_TellMeAboutYourself(struct work_struct *work) static int afs_deliver_cb_tell_me_about_yourself(struct afs_call *call, struct sk_buff *skb, bool last) { + int ret; + _enter(",{%u},%d", skb->len, last); - if (skb->len > 0) - return -EBADMSG; - if (!last) - return 0; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; /* no unmarshalling required */ call->state = AFS_CALL_REPLYING; diff --git a/fs/afs/fsclient.c b/fs/afs/fsclient.c index c2e930ec2888..9312b92e54be 100644 --- a/fs/afs/fsclient.c +++ b/fs/afs/fsclient.c @@ -240,15 +240,13 @@ static int afs_deliver_fs_fetch_status(struct afs_call *call, { struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter(",,%u", last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -335,11 +333,8 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call, case 1: _debug("extract data length (MSW)"); ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("DATA length MSW: %u", call->count); @@ -353,11 +348,8 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call, case 2: _debug("extract data length"); ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("DATA length: %u", call->count); @@ -375,11 +367,8 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call, ret = afs_extract_data(call, skb, last, buffer, call->count); kunmap_atomic(buffer); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; } call->offset = 0; @@ -389,11 +378,8 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call, case 4: ret = afs_extract_data(call, skb, last, call->buffer, (21 + 3 + 6) * 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; bp = call->buffer; xdr_decode_AFSFetchStatus(&bp, &vnode->status, vnode, NULL); @@ -405,15 +391,12 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call, call->unmarshall++; case 5: - _debug("trailer"); - if (skb->len != 0) - return -EBADMSG; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; break; } - if (!last) - return 0; - if (call->count < PAGE_SIZE) { _debug("clear"); page = call->reply3; @@ -537,9 +520,8 @@ static int afs_deliver_fs_give_up_callbacks(struct afs_call *call, { _enter(",{%u},%d", skb->len, last); - if (skb->len > 0) - return -EBADMSG; /* shouldn't be any reply data */ - return 0; + /* shouldn't be any reply data */ + return afs_data_complete(call, skb, last); } /* @@ -622,15 +604,13 @@ static int afs_deliver_fs_create_vnode(struct afs_call *call, { struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -721,15 +701,13 @@ static int afs_deliver_fs_remove(struct afs_call *call, { struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -804,15 +782,13 @@ static int afs_deliver_fs_link(struct afs_call *call, { struct afs_vnode *dvnode = call->reply, *vnode = call->reply2; const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -892,15 +868,13 @@ static int afs_deliver_fs_symlink(struct afs_call *call, { struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -999,15 +973,13 @@ static int afs_deliver_fs_rename(struct afs_call *call, { struct afs_vnode *orig_dvnode = call->reply, *new_dvnode = call->reply2; const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -1105,20 +1077,13 @@ static int afs_deliver_fs_store_data(struct afs_call *call, { struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter(",,%u", last); - afs_transfer_reply(call, skb); - if (!last) { - _leave(" = 0 [more]"); - return 0; - } - - if (call->reply_size != call->reply_max) { - _leave(" = -EBADMSG [%u != %u]", - call->reply_size, call->reply_max); - return -EBADMSG; - } + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; @@ -1292,20 +1257,13 @@ static int afs_deliver_fs_store_status(struct afs_call *call, afs_dataversion_t *store_version; struct afs_vnode *vnode = call->reply; const __be32 *bp; + int ret; _enter(",,%u", last); - afs_transfer_reply(call, skb); - if (!last) { - _leave(" = 0 [more]"); - return 0; - } - - if (call->reply_size != call->reply_max) { - _leave(" = -EBADMSG [%u != %u]", - call->reply_size, call->reply_max); - return -EBADMSG; - } + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ store_version = NULL; @@ -1504,11 +1462,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, _debug("extract status"); ret = afs_extract_data(call, skb, last, call->buffer, 12 * 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; bp = call->buffer; xdr_decode_AFSFetchVolumeStatus(&bp, call->reply2); @@ -1518,11 +1473,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, /* extract the volume name length */ case 2: ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("volname length: %u", call->count); @@ -1537,11 +1489,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, if (call->count > 0) { ret = afs_extract_data(call, skb, last, call->reply3, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; } p = call->reply3; @@ -1561,11 +1510,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, case 4: ret = afs_extract_data(call, skb, last, call->buffer, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->offset = 0; call->unmarshall++; @@ -1574,11 +1520,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, /* extract the offline message length */ case 5: ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("offline msg length: %u", call->count); @@ -1593,11 +1536,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, if (call->count > 0) { ret = afs_extract_data(call, skb, last, call->reply3, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; } p = call->reply3; @@ -1617,11 +1557,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, case 7: ret = afs_extract_data(call, skb, last, call->buffer, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->offset = 0; call->unmarshall++; @@ -1630,11 +1567,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, /* extract the message of the day length */ case 8: ret = afs_extract_data(call, skb, last, &call->tmp, 4); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->count = ntohl(call->tmp); _debug("motd length: %u", call->count); @@ -1649,11 +1583,8 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, if (call->count > 0) { ret = afs_extract_data(call, skb, last, call->reply3, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; } p = call->reply3; @@ -1673,26 +1604,20 @@ static int afs_deliver_fs_get_volume_status(struct afs_call *call, case 10: ret = afs_extract_data(call, skb, last, call->buffer, call->count); - switch (ret) { - case 0: break; - case -EAGAIN: return 0; - default: return ret; - } + if (ret < 0) + return ret; call->offset = 0; call->unmarshall++; no_motd_padding: case 11: - _debug("trailer %d", skb->len); - if (skb->len != 0) - return -EBADMSG; + ret = afs_data_complete(call, skb, last); + if (ret < 0) + return ret; break; } - if (!last) - return 0; - _leave(" = 0 [done]"); return 0; } @@ -1764,15 +1689,13 @@ static int afs_deliver_fs_xxxx_lock(struct afs_call *call, struct sk_buff *skb, bool last) { const __be32 *bp; + int ret; _enter("{%u},{%u},%d", call->unmarshall, skb->len, last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ bp = call->buffer; diff --git a/fs/afs/internal.h b/fs/afs/internal.h index 71d5982312f3..df976b2a7f40 100644 --- a/fs/afs/internal.h +++ b/fs/afs/internal.h @@ -609,17 +609,29 @@ extern void afs_proc_cell_remove(struct afs_cell *); */ extern int afs_open_socket(void); extern void afs_close_socket(void); +extern void afs_data_consumed(struct afs_call *, struct sk_buff *); extern int afs_make_call(struct in_addr *, struct afs_call *, gfp_t, const struct afs_wait_mode *); extern struct afs_call *afs_alloc_flat_call(const struct afs_call_type *, size_t, size_t); extern void afs_flat_call_destructor(struct afs_call *); -extern void afs_transfer_reply(struct afs_call *, struct sk_buff *); +extern int afs_transfer_reply(struct afs_call *, struct sk_buff *, bool); extern void afs_send_empty_reply(struct afs_call *); extern void afs_send_simple_reply(struct afs_call *, const void *, size_t); extern int afs_extract_data(struct afs_call *, struct sk_buff *, bool, void *, size_t); +static inline int afs_data_complete(struct afs_call *call, struct sk_buff *skb, + bool last) +{ + if (skb->len > 0) + return -EBADMSG; + afs_data_consumed(call, skb); + if (!last) + return -EAGAIN; + return 0; +} + /* * security.c */ diff --git a/fs/afs/rxrpc.c b/fs/afs/rxrpc.c index 4832de84d52c..14d04c848465 100644 --- a/fs/afs/rxrpc.c +++ b/fs/afs/rxrpc.c @@ -150,10 +150,9 @@ void afs_close_socket(void) } /* - * note that the data in a socket buffer is now delivered and that the buffer - * should be freed + * Note that the data in a socket buffer is now consumed. */ -static void afs_data_delivered(struct sk_buff *skb) +void afs_data_consumed(struct afs_call *call, struct sk_buff *skb) { if (!skb) { _debug("DLVR NULL [%d]", atomic_read(&afs_outstanding_skbs)); @@ -161,9 +160,7 @@ static void afs_data_delivered(struct sk_buff *skb) } else { _debug("DLVR %p{%u} [%d]", skb, skb->mark, atomic_read(&afs_outstanding_skbs)); - if (atomic_dec_return(&afs_outstanding_skbs) == -1) - BUG(); - rxrpc_kernel_data_delivered(skb); + rxrpc_kernel_data_consumed(call->rxcall, skb); } } @@ -489,9 +486,15 @@ static void afs_deliver_to_call(struct afs_call *call) last = rxrpc_kernel_is_data_last(skb); ret = call->type->deliver(call, skb, last); switch (ret) { + case -EAGAIN: + if (last) { + _debug("short data"); + goto unmarshal_error; + } + break; case 0: - if (last && - call->state == AFS_CALL_AWAIT_REPLY) + ASSERT(last); + if (call->state == AFS_CALL_AWAIT_REPLY) call->state = AFS_CALL_COMPLETE; break; case -ENOTCONN: @@ -501,6 +504,7 @@ static void afs_deliver_to_call(struct afs_call *call) abort_code = RX_INVALID_OPERATION; goto do_abort; default: + unmarshal_error: abort_code = RXGEN_CC_UNMARSHAL; if (call->state != AFS_CALL_AWAIT_REPLY) abort_code = RXGEN_SS_UNMARSHAL; @@ -511,9 +515,7 @@ static void afs_deliver_to_call(struct afs_call *call) call->state = AFS_CALL_ERROR; break; } - afs_data_delivered(skb); - skb = NULL; - continue; + break; case RXRPC_SKB_MARK_FINAL_ACK: _debug("Rcv ACK"); call->state = AFS_CALL_COMPLETE; @@ -685,15 +687,35 @@ static void afs_process_async_call(struct afs_call *call) } /* - * empty a socket buffer into a flat reply buffer + * Empty a socket buffer into a flat reply buffer. */ -void afs_transfer_reply(struct afs_call *call, struct sk_buff *skb) +int afs_transfer_reply(struct afs_call *call, struct sk_buff *skb, bool last) { size_t len = skb->len; - if (skb_copy_bits(skb, 0, call->buffer + call->reply_size, len) < 0) - BUG(); - call->reply_size += len; + if (len > call->reply_max - call->reply_size) { + _leave(" = -EBADMSG [%zu > %u]", + len, call->reply_max - call->reply_size); + return -EBADMSG; + } + + if (len > 0) { + if (skb_copy_bits(skb, 0, call->buffer + call->reply_size, + len) < 0) + BUG(); + call->reply_size += len; + } + + afs_data_consumed(call, skb); + if (!last) + return -EAGAIN; + + if (call->reply_size != call->reply_max) { + _leave(" = -EBADMSG [%u != %u]", + call->reply_size, call->reply_max); + return -EBADMSG; + } + return 0; } /* @@ -745,7 +767,8 @@ static void afs_collect_incoming_call(struct work_struct *work) } /* - * grab the operation ID from an incoming cache manager call + * Grab the operation ID from an incoming cache manager call. The socket + * buffer is discarded on error or if we don't yet have sufficient data. */ static int afs_deliver_cm_op_id(struct afs_call *call, struct sk_buff *skb, bool last) @@ -766,12 +789,9 @@ static int afs_deliver_cm_op_id(struct afs_call *call, struct sk_buff *skb, call->offset += len; if (call->offset < 4) { - if (last) { - _leave(" = -EBADMSG [op ID short]"); - return -EBADMSG; - } - _leave(" = 0 [incomplete]"); - return 0; + afs_data_consumed(call, skb); + _leave(" = -EAGAIN"); + return -EAGAIN; } call->state = AFS_CALL_AWAIT_REQUEST; @@ -855,7 +875,7 @@ void afs_send_simple_reply(struct afs_call *call, const void *buf, size_t len) } /* - * extract a piece of data from the received data socket buffers + * Extract a piece of data from the received data socket buffers. */ int afs_extract_data(struct afs_call *call, struct sk_buff *skb, bool last, void *buf, size_t count) @@ -873,10 +893,7 @@ int afs_extract_data(struct afs_call *call, struct sk_buff *skb, call->offset += len; if (call->offset < count) { - if (last) { - _leave(" = -EBADMSG [%d < %zu]", call->offset, count); - return -EBADMSG; - } + afs_data_consumed(call, skb); _leave(" = -EAGAIN"); return -EAGAIN; } diff --git a/fs/afs/vlclient.c b/fs/afs/vlclient.c index 340afd0cd182..f94d1abdc3eb 100644 --- a/fs/afs/vlclient.c +++ b/fs/afs/vlclient.c @@ -64,16 +64,13 @@ static int afs_deliver_vl_get_entry_by_xxx(struct afs_call *call, struct afs_cache_vlocation *entry; __be32 *bp; u32 tmp; - int loop; + int loop, ret; _enter(",,%u", last); - afs_transfer_reply(call, skb); - if (!last) - return 0; - - if (call->reply_size != call->reply_max) - return -EBADMSG; + ret = afs_transfer_reply(call, skb, last); + if (ret < 0) + return ret; /* unmarshall the reply once we've received all of it */ entry = call->reply; diff --git a/include/net/af_rxrpc.h b/include/net/af_rxrpc.h index ac1bc3c49fbd..7b0f88699b25 100644 --- a/include/net/af_rxrpc.h +++ b/include/net/af_rxrpc.h @@ -40,12 +40,12 @@ struct rxrpc_call *rxrpc_kernel_begin_call(struct socket *, unsigned long, gfp_t); int rxrpc_kernel_send_data(struct rxrpc_call *, struct msghdr *, size_t); +void rxrpc_kernel_data_consumed(struct rxrpc_call *, struct sk_buff *); void rxrpc_kernel_abort_call(struct rxrpc_call *, u32); void rxrpc_kernel_end_call(struct rxrpc_call *); bool rxrpc_kernel_is_data_last(struct sk_buff *); u32 rxrpc_kernel_get_abort_code(struct sk_buff *); int rxrpc_kernel_get_error_number(struct sk_buff *); -void rxrpc_kernel_data_delivered(struct sk_buff *); void rxrpc_kernel_free_skb(struct sk_buff *); struct rxrpc_call *rxrpc_kernel_accept_call(struct socket *, unsigned long); int rxrpc_kernel_reject_call(struct socket *); diff --git a/net/rxrpc/ar-internal.h b/net/rxrpc/ar-internal.h index 1bb9e7ac9e14..ff83fb1ddd47 100644 --- a/net/rxrpc/ar-internal.h +++ b/net/rxrpc/ar-internal.h @@ -425,6 +425,7 @@ struct rxrpc_call { spinlock_t lock; rwlock_t state_lock; /* lock for state transition */ atomic_t usage; + atomic_t skb_count; /* Outstanding packets on this call */ atomic_t sequence; /* Tx data packet sequence counter */ u32 local_abort; /* local abort code */ u32 remote_abort; /* remote abort code */ diff --git a/net/rxrpc/call_accept.c b/net/rxrpc/call_accept.c index 0b2832141bd0..9bae21e66d65 100644 --- a/net/rxrpc/call_accept.c +++ b/net/rxrpc/call_accept.c @@ -130,6 +130,7 @@ static int rxrpc_accept_incoming_call(struct rxrpc_local *local, call->state = RXRPC_CALL_SERVER_ACCEPTING; list_add_tail(&call->accept_link, &rx->acceptq); rxrpc_get_call(call); + atomic_inc(&call->skb_count); nsp = rxrpc_skb(notification); nsp->call = call; diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index fc32aa5764a2..f5e99163a09e 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -460,6 +460,7 @@ static void rxrpc_insert_oos_packet(struct rxrpc_call *call, ASSERTCMP(sp->call, ==, NULL); sp->call = call; rxrpc_get_call(call); + atomic_inc(&call->skb_count); /* insert into the buffer in sequence order */ spin_lock_bh(&call->lock); @@ -734,6 +735,7 @@ all_acked: skb->mark = RXRPC_SKB_MARK_FINAL_ACK; sp->call = call; rxrpc_get_call(call); + atomic_inc(&call->skb_count); spin_lock_bh(&call->lock); if (rxrpc_queue_rcv_skb(call, skb, true, true) < 0) BUG(); @@ -793,6 +795,7 @@ static int rxrpc_post_message(struct rxrpc_call *call, u32 mark, u32 error, sp->error = error; sp->call = call; rxrpc_get_call(call); + atomic_inc(&call->skb_count); spin_lock_bh(&call->lock); ret = rxrpc_queue_rcv_skb(call, skb, true, fatal); diff --git a/net/rxrpc/call_object.c b/net/rxrpc/call_object.c index 91287c9d01bb..c47f14fc5e88 100644 --- a/net/rxrpc/call_object.c +++ b/net/rxrpc/call_object.c @@ -491,13 +491,6 @@ void rxrpc_release_call(struct rxrpc_call *call) spin_lock_bh(&call->lock); while ((skb = skb_dequeue(&call->rx_queue)) || (skb = skb_dequeue(&call->rx_oos_queue))) { - sp = rxrpc_skb(skb); - if (sp->call) { - ASSERTCMP(sp->call, ==, call); - rxrpc_put_call(call); - sp->call = NULL; - } - skb->destructor = NULL; spin_unlock_bh(&call->lock); _debug("- zap %s %%%u #%u", @@ -605,6 +598,7 @@ void __rxrpc_put_call(struct rxrpc_call *call) if (atomic_dec_and_test(&call->usage)) { _debug("call %d dead", call->debug_id); + WARN_ON(atomic_read(&call->skb_count) != 0); ASSERTCMP(call->state, ==, RXRPC_CALL_DEAD); rxrpc_queue_work(&call->destroyer); } diff --git a/net/rxrpc/input.c b/net/rxrpc/input.c index 991a20d25093..9e0f58edcd01 100644 --- a/net/rxrpc/input.c +++ b/net/rxrpc/input.c @@ -55,9 +55,6 @@ int rxrpc_queue_rcv_skb(struct rxrpc_call *call, struct sk_buff *skb, if (test_bit(RXRPC_CALL_TERMINAL_MSG, &call->flags)) { _debug("already terminated"); ASSERTCMP(call->state, >=, RXRPC_CALL_COMPLETE); - skb->destructor = NULL; - sp->call = NULL; - rxrpc_put_call(call); rxrpc_free_skb(skb); return 0; } @@ -111,13 +108,7 @@ int rxrpc_queue_rcv_skb(struct rxrpc_call *call, struct sk_buff *skb, ret = 0; out: - /* release the socket buffer */ - if (skb) { - skb->destructor = NULL; - sp->call = NULL; - rxrpc_put_call(call); - rxrpc_free_skb(skb); - } + rxrpc_free_skb(skb); _leave(" = %d", ret); return ret; @@ -200,6 +191,7 @@ static int rxrpc_fast_process_data(struct rxrpc_call *call, sp->call = call; rxrpc_get_call(call); + atomic_inc(&call->skb_count); terminal = ((sp->hdr.flags & RXRPC_LAST_PACKET) && !(sp->hdr.flags & RXRPC_CLIENT_INITIATED)); ret = rxrpc_queue_rcv_skb(call, skb, false, terminal); diff --git a/net/rxrpc/recvmsg.c b/net/rxrpc/recvmsg.c index a3fa2ed85d63..9ed66d533002 100644 --- a/net/rxrpc/recvmsg.c +++ b/net/rxrpc/recvmsg.c @@ -203,6 +203,9 @@ int rxrpc_recvmsg(struct socket *sock, struct msghdr *msg, size_t len, } /* we transferred the whole data packet */ + if (!(flags & MSG_PEEK)) + rxrpc_kernel_data_consumed(call, skb); + if (sp->hdr.flags & RXRPC_LAST_PACKET) { _debug("last"); if (rxrpc_conn_is_client(call->conn)) { @@ -359,28 +362,6 @@ wait_error: } -/** - * rxrpc_kernel_data_delivered - Record delivery of data message - * @skb: Message holding data - * - * Record the delivery of a data message. This permits RxRPC to keep its - * tracking correct. The socket buffer will be deleted. - */ -void rxrpc_kernel_data_delivered(struct sk_buff *skb) -{ - struct rxrpc_skb_priv *sp = rxrpc_skb(skb); - struct rxrpc_call *call = sp->call; - - ASSERTCMP(sp->hdr.seq, >=, call->rx_data_recv); - ASSERTCMP(sp->hdr.seq, <=, call->rx_data_recv + 1); - call->rx_data_recv = sp->hdr.seq; - - ASSERTCMP(sp->hdr.seq, >, call->rx_data_eaten); - rxrpc_free_skb(skb); -} - -EXPORT_SYMBOL(rxrpc_kernel_data_delivered); - /** * rxrpc_kernel_is_data_last - Determine if data message is last one * @skb: Message holding data diff --git a/net/rxrpc/skbuff.c b/net/rxrpc/skbuff.c index eee0cfd9ac8c..06c51d4b622d 100644 --- a/net/rxrpc/skbuff.c +++ b/net/rxrpc/skbuff.c @@ -98,11 +98,39 @@ static void rxrpc_hard_ACK_data(struct rxrpc_call *call, spin_unlock_bh(&call->lock); } +/** + * rxrpc_kernel_data_consumed - Record consumption of data message + * @call: The call to which the message pertains. + * @skb: Message holding data + * + * Record the consumption of a data message and generate an ACK if appropriate. + * The call state is shifted if this was the final packet. The caller must be + * in process context with no spinlocks held. + * + * TODO: Actually generate the ACK here rather than punting this to the + * workqueue. + */ +void rxrpc_kernel_data_consumed(struct rxrpc_call *call, struct sk_buff *skb) +{ + struct rxrpc_skb_priv *sp = rxrpc_skb(skb); + + _enter("%d,%p{%u}", call->debug_id, skb, sp->hdr.seq); + + ASSERTCMP(sp->call, ==, call); + ASSERTCMP(sp->hdr.type, ==, RXRPC_PACKET_TYPE_DATA); + + /* TODO: Fix the sequence number tracking */ + ASSERTCMP(sp->hdr.seq, >=, call->rx_data_recv); + ASSERTCMP(sp->hdr.seq, <=, call->rx_data_recv + 1); + ASSERTCMP(sp->hdr.seq, >, call->rx_data_eaten); + + call->rx_data_recv = sp->hdr.seq; + rxrpc_hard_ACK_data(call, sp); +} +EXPORT_SYMBOL(rxrpc_kernel_data_consumed); + /* - * destroy a packet that has an RxRPC control buffer - * - advance the hard-ACK state of the parent call (done here in case something - * in the kernel bypasses recvmsg() and steals the packet directly off of the - * socket receive queue) + * Destroy a packet that has an RxRPC control buffer */ void rxrpc_packet_destructor(struct sk_buff *skb) { @@ -112,9 +140,8 @@ void rxrpc_packet_destructor(struct sk_buff *skb) _enter("%p{%p}", skb, call); if (call) { - /* send the final ACK on a client call */ - if (sp->hdr.type == RXRPC_PACKET_TYPE_DATA) - rxrpc_hard_ACK_data(call, sp); + if (atomic_dec_return(&call->skb_count) < 0) + BUG(); rxrpc_put_call(call); sp->call = NULL; } -- cgit v1.2.3-70-g09d2 From 0956254a2d5b9e2141385514553aeef694dfe3b5 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Mon, 8 Aug 2016 15:08:49 +0200 Subject: ovl: don't copy up opaqueness When a copy up of a directory occurs which has the opaque xattr set, the xattr remains in the upper directory. The immediate behavior with overlayfs is that the upper directory is not treated as opaque, however after a remount the opaque flag is used and upper directory is treated as opaque. This causes files created in the lower layer to be hidden when using multiple lower directories. Fix by not copying up the opaque flag. To reproduce: ----8<---------8<---------8<---------8<---------8<---------8<---- mkdir -p l/d/s u v w mnt mount -t overlay overlay -olowerdir=l,upperdir=u,workdir=w mnt rm -rf mnt/d/ mkdir -p mnt/d/n umount mnt mount -t overlay overlay -olowerdir=u:l,upperdir=v,workdir=w mnt touch mnt/d/foo umount mnt mount -t overlay overlay -olowerdir=u:l,upperdir=v,workdir=w mnt ls mnt/d ----8<---------8<---------8<---------8<---------8<---------8<---- output should be: "foo n" Reported-by: Derek McGowan Link: https://bugzilla.kernel.org/show_bug.cgi?id=151291 Signed-off-by: Miklos Szeredi Cc: --- fs/overlayfs/copy_up.c | 2 ++ fs/overlayfs/inode.c | 2 +- fs/overlayfs/overlayfs.h | 1 + 3 files changed, 4 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/overlayfs/copy_up.c b/fs/overlayfs/copy_up.c index 54e5d6681786..43fdc2765aea 100644 --- a/fs/overlayfs/copy_up.c +++ b/fs/overlayfs/copy_up.c @@ -80,6 +80,8 @@ int ovl_copy_xattr(struct dentry *old, struct dentry *new) } for (name = buf; name < (buf + list_size); name += strlen(name) + 1) { + if (ovl_is_private_xattr(name)) + continue; retry: size = vfs_getxattr(old, name, value, value_size); if (size == -ERANGE) diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 1b885c156028..024352f1d405 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -191,7 +191,7 @@ static int ovl_readlink(struct dentry *dentry, char __user *buf, int bufsiz) return err; } -static bool ovl_is_private_xattr(const char *name) +bool ovl_is_private_xattr(const char *name) { #define OVL_XATTR_PRE_NAME OVL_XATTR_PREFIX "." return strncmp(name, OVL_XATTR_PRE_NAME, diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index e4f5c9536bfe..34839bd2b6b8 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -193,6 +193,7 @@ int ovl_removexattr(struct dentry *dentry, const char *name); struct posix_acl *ovl_get_acl(struct inode *inode, int type); int ovl_open_maybe_copy_up(struct dentry *dentry, unsigned int file_flags); int ovl_update_time(struct inode *inode, struct timespec *ts, int flags); +bool ovl_is_private_xattr(const char *name); struct inode *ovl_new_inode(struct super_block *sb, umode_t mode); struct inode *ovl_get_inode(struct super_block *sb, struct inode *realinode); -- cgit v1.2.3-70-g09d2 From d0141191a20289f8955c1e03dad08e42e6f71ca9 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Thu, 11 Aug 2016 11:50:30 -0400 Subject: ext4: fix xattr shifting when expanding inodes The code in ext4_expand_extra_isize_ea() treated new_extra_isize argument sometimes as the desired target i_extra_isize and sometimes as the amount by which we need to grow current i_extra_isize. These happen to coincide when i_extra_isize is 0 which used to be the common case and so nobody noticed this until recently when we added i_projid to the inode and so i_extra_isize now needs to grow from 28 to 32 bytes. The result of these bugs was that we sometimes unnecessarily decided to move xattrs out of inode even if there was enough space and we often ended up corrupting in-inode xattrs because arguments to ext4_xattr_shift_entries() were just wrong. This could demonstrate itself as BUG_ON in ext4_xattr_shift_entries() triggering. Fix the problem by introducing new isize_diff variable and use it where appropriate. CC: stable@vger.kernel.org # 4.4.x Reported-by: Dave Chinner Signed-off-by: Jan Kara Signed-off-by: Theodore Ts'o --- fs/ext4/xattr.c | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index 39e9cfb1b371..cb1d7b4482de 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1353,11 +1353,13 @@ int ext4_expand_extra_isize_ea(struct inode *inode, int new_extra_isize, size_t min_offs, free; int total_ino; void *base, *start, *end; - int extra_isize = 0, error = 0, tried_min_extra_isize = 0; + int error = 0, tried_min_extra_isize = 0; int s_min_extra_isize = le16_to_cpu(EXT4_SB(inode->i_sb)->s_es->s_min_extra_isize); + int isize_diff; /* How much do we need to grow i_extra_isize */ down_write(&EXT4_I(inode)->xattr_sem); retry: + isize_diff = new_extra_isize - EXT4_I(inode)->i_extra_isize; if (EXT4_I(inode)->i_extra_isize >= new_extra_isize) { up_write(&EXT4_I(inode)->xattr_sem); return 0; @@ -1382,7 +1384,7 @@ retry: goto cleanup; free = ext4_xattr_free_space(last, &min_offs, base, &total_ino); - if (free >= new_extra_isize) { + if (free >= isize_diff) { entry = IFIRST(header); ext4_xattr_shift_entries(entry, EXT4_I(inode)->i_extra_isize - new_extra_isize, (void *)raw_inode + @@ -1414,7 +1416,7 @@ retry: end = bh->b_data + bh->b_size; min_offs = end - base; free = ext4_xattr_free_space(first, &min_offs, base, NULL); - if (free < new_extra_isize) { + if (free < isize_diff) { if (!tried_min_extra_isize && s_min_extra_isize) { tried_min_extra_isize++; new_extra_isize = s_min_extra_isize; @@ -1428,7 +1430,7 @@ retry: free = inode->i_sb->s_blocksize; } - while (new_extra_isize > 0) { + while (isize_diff > 0) { size_t offs, size, entry_size; struct ext4_xattr_entry *small_entry = NULL; struct ext4_xattr_info i = { @@ -1459,7 +1461,7 @@ retry: EXT4_XATTR_SIZE(le32_to_cpu(last->e_value_size)) + EXT4_XATTR_LEN(last->e_name_len); if (total_size <= free && total_size < min_total_size) { - if (total_size < new_extra_isize) { + if (total_size < isize_diff) { small_entry = last; } else { entry = last; @@ -1516,20 +1518,19 @@ retry: goto cleanup; entry = IFIRST(header); - if (entry_size + EXT4_XATTR_SIZE(size) >= new_extra_isize) - shift_bytes = new_extra_isize; + if (entry_size + EXT4_XATTR_SIZE(size) >= isize_diff) + shift_bytes = isize_diff; else shift_bytes = entry_size + size; /* Adjust the offsets and shift the remaining entries ahead */ - ext4_xattr_shift_entries(entry, EXT4_I(inode)->i_extra_isize - - shift_bytes, (void *)raw_inode + - EXT4_GOOD_OLD_INODE_SIZE + extra_isize + shift_bytes, + ext4_xattr_shift_entries(entry, -shift_bytes, + (void *)raw_inode + EXT4_GOOD_OLD_INODE_SIZE + + EXT4_I(inode)->i_extra_isize + shift_bytes, (void *)header, total_ino - entry_size, inode->i_sb->s_blocksize); - extra_isize += shift_bytes; - new_extra_isize -= shift_bytes; - EXT4_I(inode)->i_extra_isize = extra_isize; + isize_diff -= shift_bytes; + EXT4_I(inode)->i_extra_isize += shift_bytes; i.name = b_entry_name; i.value = buffer; -- cgit v1.2.3-70-g09d2 From 418c12d08dc64a45107c467ec1ba29b5e69b0715 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Thu, 11 Aug 2016 11:58:32 -0400 Subject: ext4: fix xattr shifting when expanding inodes part 2 When multiple xattrs need to be moved out of inode, we did not properly recompute total size of xattr headers in the inode and the new header position. Thus when moving the second and further xattr we asked ext4_xattr_shift_entries() to move too much and from the wrong place, resulting in possible xattr value corruption or general memory corruption. CC: stable@vger.kernel.org # 4.4.x Signed-off-by: Jan Kara Signed-off-by: Theodore Ts'o --- fs/ext4/xattr.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index cb1d7b4482de..b18b1ff7cc27 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1516,6 +1516,7 @@ retry: error = ext4_xattr_ibody_set(handle, inode, &i, is); if (error) goto cleanup; + total_ino -= entry_size; entry = IFIRST(header); if (entry_size + EXT4_XATTR_SIZE(size) >= isize_diff) @@ -1526,11 +1527,11 @@ retry: ext4_xattr_shift_entries(entry, -shift_bytes, (void *)raw_inode + EXT4_GOOD_OLD_INODE_SIZE + EXT4_I(inode)->i_extra_isize + shift_bytes, - (void *)header, total_ino - entry_size, - inode->i_sb->s_blocksize); + (void *)header, total_ino, inode->i_sb->s_blocksize); isize_diff -= shift_bytes; EXT4_I(inode)->i_extra_isize += shift_bytes; + header = IHDR(inode, raw_inode); i.name = b_entry_name; i.value = buffer; -- cgit v1.2.3-70-g09d2 From 443a8c41cd49de66a3fda45b32b9860ea0292b84 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Thu, 11 Aug 2016 12:00:01 -0400 Subject: ext4: properly align shifted xattrs when expanding inodes We did not count with the padding of xattr value when computing desired shift of xattrs in the inode when expanding i_extra_isize. As a result we could create unaligned start of inline xattrs. Account for alignment properly. CC: stable@vger.kernel.org # 4.4.x- Signed-off-by: Jan Kara --- fs/ext4/xattr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index b18b1ff7cc27..c893f00b6dc0 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1522,7 +1522,7 @@ retry: if (entry_size + EXT4_XATTR_SIZE(size) >= isize_diff) shift_bytes = isize_diff; else - shift_bytes = entry_size + size; + shift_bytes = entry_size + EXT4_XATTR_SIZE(size); /* Adjust the offsets and shift the remaining entries ahead */ ext4_xattr_shift_entries(entry, -shift_bytes, (void *)raw_inode + EXT4_GOOD_OLD_INODE_SIZE + -- cgit v1.2.3-70-g09d2 From 2e81a4eeedcaa66e35f58b81e0755b87057ce392 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Thu, 11 Aug 2016 12:38:55 -0400 Subject: ext4: avoid deadlock when expanding inode size When we need to move xattrs into external xattr block, we call ext4_xattr_block_set() from ext4_expand_extra_isize_ea(). That may end up calling ext4_mark_inode_dirty() again which will recurse back into the inode expansion code leading to deadlocks. Protect from recursion using EXT4_STATE_NO_EXPAND inode flag and move its management into ext4_expand_extra_isize_ea() since its manipulation is safe there (due to xattr_sem) from possible races with ext4_xattr_set_handle() which plays with it as well. CC: stable@vger.kernel.org # 4.4.x Signed-off-by: Jan Kara Signed-off-by: Theodore Ts'o --- fs/ext4/inode.c | 2 -- fs/ext4/xattr.c | 19 +++++++++++++------ 2 files changed, 13 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 5a6277d80f7c..13c95b290263 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -5466,8 +5466,6 @@ int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode) sbi->s_want_extra_isize, iloc, handle); if (ret) { - ext4_set_inode_state(inode, - EXT4_STATE_NO_EXPAND); if (mnt_count != le16_to_cpu(sbi->s_es->s_mnt_count)) { ext4_warning(inode->i_sb, diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index c893f00b6dc0..2eb935ca5d9e 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1358,12 +1358,14 @@ int ext4_expand_extra_isize_ea(struct inode *inode, int new_extra_isize, int isize_diff; /* How much do we need to grow i_extra_isize */ down_write(&EXT4_I(inode)->xattr_sem); + /* + * Set EXT4_STATE_NO_EXPAND to avoid recursion when marking inode dirty + */ + ext4_set_inode_state(inode, EXT4_STATE_NO_EXPAND); retry: isize_diff = new_extra_isize - EXT4_I(inode)->i_extra_isize; - if (EXT4_I(inode)->i_extra_isize >= new_extra_isize) { - up_write(&EXT4_I(inode)->xattr_sem); - return 0; - } + if (EXT4_I(inode)->i_extra_isize >= new_extra_isize) + goto out; header = IHDR(inode, raw_inode); entry = IFIRST(header); @@ -1392,8 +1394,7 @@ retry: (void *)header, total_ino, inode->i_sb->s_blocksize); EXT4_I(inode)->i_extra_isize = new_extra_isize; - error = 0; - goto cleanup; + goto out; } /* @@ -1553,6 +1554,8 @@ retry: kfree(bs); } brelse(bh); +out: + ext4_clear_inode_state(inode, EXT4_STATE_NO_EXPAND); up_write(&EXT4_I(inode)->xattr_sem); return 0; @@ -1564,6 +1567,10 @@ cleanup: kfree(is); kfree(bs); brelse(bh); + /* + * We deliberately leave EXT4_STATE_NO_EXPAND set here since inode + * size expansion failed. + */ up_write(&EXT4_I(inode)->xattr_sem); return error; } -- cgit v1.2.3-70-g09d2 From 1c8d477a77e2d1d3504419e7f2e02e6422becf9a Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 14 Aug 2016 12:47:49 -0400 Subject: pNFS/flexfiles: Fix layoutstat periodic reporting Putting the periodicity timer in the mirror instances is causing non-scalable reporting behaviour and missed reporting intervals. When you recall layouts and/or implement client side mirroring, it leads to consecutive reports with only a few ms between RPC calls. Signed-off-by: Trond Myklebust Fixes: d0379a5d066a9 ("pNFS/flexfiles: Support server-supplied...") --- fs/nfs/flexfilelayout/flexfilelayout.c | 8 ++++---- fs/nfs/flexfilelayout/flexfilelayout.h | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c index e6206eaf2bdf..ee1c94c7614c 100644 --- a/fs/nfs/flexfilelayout/flexfilelayout.c +++ b/fs/nfs/flexfilelayout/flexfilelayout.c @@ -37,6 +37,7 @@ ff_layout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags) if (ffl) { INIT_LIST_HEAD(&ffl->error_list); INIT_LIST_HEAD(&ffl->mirrors); + ffl->last_report_time = ktime_get(); return &ffl->generic_hdr; } else return NULL; @@ -640,19 +641,18 @@ nfs4_ff_layoutstat_start_io(struct nfs4_ff_layout_mirror *mirror, { static const ktime_t notime = {0}; s64 report_interval = FF_LAYOUTSTATS_REPORT_INTERVAL; + struct nfs4_flexfile_layout *ffl = FF_LAYOUT_FROM_HDR(mirror->layout); nfs4_ff_start_busy_timer(&layoutstat->busy_timer, now); if (ktime_equal(mirror->start_time, notime)) mirror->start_time = now; - if (ktime_equal(mirror->last_report_time, notime)) - mirror->last_report_time = now; if (mirror->report_interval != 0) report_interval = (s64)mirror->report_interval * 1000LL; else if (layoutstats_timer != 0) report_interval = (s64)layoutstats_timer * 1000LL; - if (ktime_to_ms(ktime_sub(now, mirror->last_report_time)) >= + if (ktime_to_ms(ktime_sub(now, ffl->last_report_time)) >= report_interval) { - mirror->last_report_time = now; + ffl->last_report_time = now; return true; } diff --git a/fs/nfs/flexfilelayout/flexfilelayout.h b/fs/nfs/flexfilelayout/flexfilelayout.h index 1bcdb15d0c41..3ee0c9fcea76 100644 --- a/fs/nfs/flexfilelayout/flexfilelayout.h +++ b/fs/nfs/flexfilelayout/flexfilelayout.h @@ -84,7 +84,6 @@ struct nfs4_ff_layout_mirror { struct nfs4_ff_layoutstat read_stat; struct nfs4_ff_layoutstat write_stat; ktime_t start_time; - ktime_t last_report_time; u32 report_interval; }; @@ -101,6 +100,7 @@ struct nfs4_flexfile_layout { struct pnfs_ds_commit_info commit_info; struct list_head mirrors; struct list_head error_list; /* nfs4_ff_layout_ds_err */ + ktime_t last_report_time; /* Layoutstat report times */ }; static inline struct nfs4_flexfile_layout * -- cgit v1.2.3-70-g09d2 From a956beda19a6b39fbc19d0aaf21947acdc18cf74 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 16 Aug 2016 10:26:47 -0400 Subject: NFS: Allow the mount option retrans=0 We should allow retrans=0 as just meaning that every timeout is a major timeout, and that there is no increment in the timeout value. For instance, this means that we would allow TCP users to specify a flat timeout value of 60s, by specifying "timeo=600,retrans=0" in their mount option string. Siged-off-by: Trond Myklebust --- fs/nfs/client.c | 10 +++++----- fs/nfs/internal.h | 5 ++++- fs/nfs/super.c | 19 +++++++++++++++++-- 3 files changed, 26 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/nfs/client.c b/fs/nfs/client.c index 003ebce4bbc4..1e106780a237 100644 --- a/fs/nfs/client.c +++ b/fs/nfs/client.c @@ -426,7 +426,7 @@ EXPORT_SYMBOL_GPL(nfs_mark_client_ready); * Initialise the timeout values for a connection */ void nfs_init_timeout_values(struct rpc_timeout *to, int proto, - unsigned int timeo, unsigned int retrans) + int timeo, int retrans) { to->to_initval = timeo * HZ / 10; to->to_retries = retrans; @@ -434,9 +434,9 @@ void nfs_init_timeout_values(struct rpc_timeout *to, int proto, switch (proto) { case XPRT_TRANSPORT_TCP: case XPRT_TRANSPORT_RDMA: - if (to->to_retries == 0) + if (retrans == NFS_UNSPEC_RETRANS) to->to_retries = NFS_DEF_TCP_RETRANS; - if (to->to_initval == 0) + if (timeo == NFS_UNSPEC_TIMEO || to->to_retries == 0) to->to_initval = NFS_DEF_TCP_TIMEO * HZ / 10; if (to->to_initval > NFS_MAX_TCP_TIMEOUT) to->to_initval = NFS_MAX_TCP_TIMEOUT; @@ -449,9 +449,9 @@ void nfs_init_timeout_values(struct rpc_timeout *to, int proto, to->to_exponential = 0; break; case XPRT_TRANSPORT_UDP: - if (to->to_retries == 0) + if (retrans == NFS_UNSPEC_RETRANS) to->to_retries = NFS_DEF_UDP_RETRANS; - if (!to->to_initval) + if (timeo == NFS_UNSPEC_TIMEO || to->to_initval == 0) to->to_initval = NFS_DEF_UDP_TIMEO * HZ / 10; if (to->to_initval > NFS_MAX_UDP_TIMEOUT) to->to_initval = NFS_MAX_UDP_TIMEOUT; diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h index 7ce5e023c3c3..74935a19e4bf 100644 --- a/fs/nfs/internal.h +++ b/fs/nfs/internal.h @@ -58,6 +58,9 @@ struct nfs_clone_mount { */ #define NFS_UNSPEC_PORT (-1) +#define NFS_UNSPEC_RETRANS (UINT_MAX) +#define NFS_UNSPEC_TIMEO (UINT_MAX) + /* * Maximum number of pages that readdir can use for creating * a vmapped array of pages. @@ -156,7 +159,7 @@ struct nfs_client *nfs_get_client(const struct nfs_client_initdata *, int nfs_probe_fsinfo(struct nfs_server *server, struct nfs_fh *, struct nfs_fattr *); void nfs_server_insert_lists(struct nfs_server *); void nfs_server_remove_lists(struct nfs_server *); -void nfs_init_timeout_values(struct rpc_timeout *, int, unsigned int, unsigned int); +void nfs_init_timeout_values(struct rpc_timeout *to, int proto, int timeo, int retrans); int nfs_init_server_rpcclient(struct nfs_server *, const struct rpc_timeout *t, rpc_authflavor_t); struct nfs_server *nfs_alloc_server(void); diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 18d446e1a82b..d39601381adf 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -923,6 +923,8 @@ static struct nfs_parsed_mount_data *nfs_alloc_parsed_mount_data(void) data = kzalloc(sizeof(*data), GFP_KERNEL); if (data) { + data->timeo = NFS_UNSPEC_TIMEO; + data->retrans = NFS_UNSPEC_RETRANS; data->acregmin = NFS_DEF_ACREGMIN; data->acregmax = NFS_DEF_ACREGMAX; data->acdirmin = NFS_DEF_ACDIRMIN; @@ -1189,6 +1191,19 @@ static int nfs_get_option_ul(substring_t args[], unsigned long *option) return rc; } +static int nfs_get_option_ul_bound(substring_t args[], unsigned long *option, + unsigned long l_bound, unsigned long u_bound) +{ + int ret; + + ret = nfs_get_option_ul(args, option); + if (ret != 0) + return ret; + if (*option < l_bound || *option > u_bound) + return -ERANGE; + return 0; +} + /* * Error-check and convert a string of mount options from user space into * a data structure. The whole mount string is processed; bad options are @@ -1352,12 +1367,12 @@ static int nfs_parse_mount_options(char *raw, mnt->bsize = option; break; case Opt_timeo: - if (nfs_get_option_ul(args, &option) || option == 0) + if (nfs_get_option_ul_bound(args, &option, 1, INT_MAX)) goto out_invalid_value; mnt->timeo = option; break; case Opt_retrans: - if (nfs_get_option_ul(args, &option) || option == 0) + if (nfs_get_option_ul_bound(args, &option, 0, INT_MAX)) goto out_invalid_value; mnt->retrans = option; break; -- cgit v1.2.3-70-g09d2 From 15d03055cf39fe61714aeda8d0a722b3137531ed Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 16 Aug 2016 11:08:22 -0400 Subject: pNFS/flexfiles: Set reasonable default retrans values for the data channel Prior to this patch, the retrans value was set at 5, meaning that we could see a maximum retransmission timeout value of more than 6 minutes. That's a tad high for NFSv3 where the protocol does allow the server to drop requests at any time. Since this is a data channel, let's just set retrans to 0, and the default timeout to 60s. The user can continue to adjust these defaults using the dataserver_retrans and dataserver_timeo module parameters. Signed-off-by: Trond Myklebust --- fs/nfs/flexfilelayout/flexfilelayoutdev.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c index 0aa36be71fce..970efba05ae1 100644 --- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c +++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c @@ -17,8 +17,8 @@ #define NFSDBG_FACILITY NFSDBG_PNFS_LD -static unsigned int dataserver_timeo = NFS4_DEF_DS_TIMEO; -static unsigned int dataserver_retrans = NFS4_DEF_DS_RETRANS; +static unsigned int dataserver_timeo = NFS_DEF_TCP_RETRANS; +static unsigned int dataserver_retrans; void nfs4_ff_layout_put_deviceid(struct nfs4_ff_layout_ds *mirror_ds) { -- cgit v1.2.3-70-g09d2 From 4dd3fd7197303739094183b139bae3142a3d55e6 Mon Sep 17 00:00:00 2001 From: Brian Foster Date: Wed, 17 Aug 2016 08:30:28 +1000 Subject: xfs: don't assert fail on non-async buffers on ioacct decrement The buffer I/O accounting mechanism tracks async buffers under I/O. As an optimization, the buffer I/O count is incremented only once on the first async I/O for a given hold cycle of a buffer and decremented once the buffer is released to the LRU (or freed). xfs_buf_ioacct_dec() has an ASSERT() check for an XBF_ASYNC buffer, but we have one or two corner cases where a buffer can be submitted for I/O multiple times via different methods in a single hold cycle. If an async I/O occurs first, the I/O count is incremented. If a sync I/O occurs before the hold count drops, XBF_ASYNC is cleared by the time the I/O count is decremented. Remove the async assert check from xfs_buf_ioacct_dec() as this is a perfectly valid scenario. For the purposes of I/O accounting, we really only care about the buffer async state at I/O submission time. Discovered-and-analyzed-by: Dave Chinner Signed-off-by: Brian Foster Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/xfs_buf.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index 47a318ce82e0..607cc29bba21 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -115,7 +115,6 @@ xfs_buf_ioacct_dec( if (!(bp->b_flags & _XBF_IN_FLIGHT)) return; - ASSERT(bp->b_flags & XBF_ASYNC); bp->b_flags &= ~_XBF_IN_FLIGHT; percpu_counter_dec(&bp->b_target->bt_io_count); } -- cgit v1.2.3-70-g09d2 From 0af32fb468b4a4434dd759d68611763658650b59 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:30:28 +1000 Subject: xfs: fix bogus space reservation in xfs_iomap_write_allocate The space reservations was without an explaination in commit "Add error reporting calls in error paths that return EFSCORRUPTED" back in 2003. There is no reason to reserve disk blocks in the transaction when allocating blocks for delalloc space as we already reserved the space when creating the delalloc extent. With this fix we stop running out of the reserved pool in generic/229, which has happened for long time with small blocksize file systems, and has increased in severity with the new buffered write path. [ dchinner: we still need to pass the block reservation into xfs_bmapi_write() to ensure we don't deadlock during AG selection. See commit dbd5c8c ("xfs: pass total block res. as total xfs_bmapi_write() parameter") for more details on why this is necessary. ] Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/xfs_iomap.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c index 2114d53df433..4398932a7d1b 100644 --- a/fs/xfs/xfs_iomap.c +++ b/fs/xfs/xfs_iomap.c @@ -715,12 +715,16 @@ xfs_iomap_write_allocate( * is in the delayed allocation extent on which we sit * but before our buffer starts. */ - nimaps = 0; while (nimaps == 0) { nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK); - - error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, nres, + /* + * We have already reserved space for the extent and any + * indirect blocks when creating the delalloc extent, + * there is no need to reserve space in this transaction + * again. + */ + error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, 0, 0, XFS_TRANS_RESERVE, &tp); if (error) return error; -- cgit v1.2.3-70-g09d2 From 8b2180b3bf0338625cab07da6543acb436df9c40 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 17 Aug 2016 08:31:33 +1000 Subject: xfs: don't invalidate whole file on DAX read/write When we do DAX IO, we try to invalidate the entire page cache held on the file. This is incorrect as it will trash the entire mapping tree that now tracks dirty state in exceptional entries in the radix tree slots. What we are trying to do is remove cached pages (e.g from reads into holes) that sit in the radix tree over the range we are about to write to. Hence we should just limit the invalidation to the range we are about to overwrite. Reported-by: Jan Kara Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/xfs_file.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c index ed95e5bb04e6..e612a0233710 100644 --- a/fs/xfs/xfs_file.c +++ b/fs/xfs/xfs_file.c @@ -741,9 +741,20 @@ xfs_file_dax_write( * page is inserted into the pagecache when we have to serve a write * fault on a hole. It should never be dirtied and can simply be * dropped from the pagecache once we get real data for the page. + * + * XXX: This is racy against mmap, and there's nothing we can do about + * it. dax_do_io() should really do this invalidation internally as + * it will know if we've allocated over a holei for this specific IO and + * if so it needs to update the mapping tree and invalidate existing + * PTEs over the newly allocated range. Remove this invalidation when + * dax_do_io() is fixed up. */ if (mapping->nrpages) { - ret = invalidate_inode_pages2(mapping); + loff_t end = iocb->ki_pos + iov_iter_count(from) - 1; + + ret = invalidate_inode_pages2_range(mapping, + iocb->ki_pos >> PAGE_SHIFT, + end >> PAGE_SHIFT); WARN_ON_ONCE(ret); } -- cgit v1.2.3-70-g09d2 From f32866fdc9ed077c3bc7160f4f53d6a479201d46 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Wed, 17 Aug 2016 08:31:49 +1000 Subject: xfs: store rmapbt block count in the AGF Track the number of blocks used for the rmapbt in the AGF. When we get to the AG reservation code we need this counter to quickly make our reservation during mount. Signed-off-by: Darrick J. Wong Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_alloc.c | 1 + fs/xfs/libxfs/xfs_format.h | 11 ++++++++--- fs/xfs/libxfs/xfs_rmap_btree.c | 6 ++++++ fs/xfs/xfs_fsops.c | 1 + 4 files changed, 16 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c index 776ae2f325d1..af1a72efa64c 100644 --- a/fs/xfs/libxfs/xfs_alloc.c +++ b/fs/xfs/libxfs/xfs_alloc.c @@ -2264,6 +2264,7 @@ xfs_alloc_log_agf( offsetof(xfs_agf_t, agf_longest), offsetof(xfs_agf_t, agf_btreeblks), offsetof(xfs_agf_t, agf_uuid), + offsetof(xfs_agf_t, agf_rmap_blocks), sizeof(xfs_agf_t) }; diff --git a/fs/xfs/libxfs/xfs_format.h b/fs/xfs/libxfs/xfs_format.h index f814d42c73b2..e6a8bea0f7ba 100644 --- a/fs/xfs/libxfs/xfs_format.h +++ b/fs/xfs/libxfs/xfs_format.h @@ -640,12 +640,15 @@ typedef struct xfs_agf { __be32 agf_btreeblks; /* # of blocks held in AGF btrees */ uuid_t agf_uuid; /* uuid of filesystem */ + __be32 agf_rmap_blocks; /* rmapbt blocks used */ + __be32 agf_padding; /* padding */ + /* * reserve some contiguous space for future logged fields before we add * the unlogged fields. This makes the range logging via flags and * structure offsets much simpler. */ - __be64 agf_spare64[16]; + __be64 agf_spare64[15]; /* unlogged fields, written during buffer writeback. */ __be64 agf_lsn; /* last write sequence */ @@ -670,7 +673,8 @@ typedef struct xfs_agf { #define XFS_AGF_LONGEST 0x00000400 #define XFS_AGF_BTREEBLKS 0x00000800 #define XFS_AGF_UUID 0x00001000 -#define XFS_AGF_NUM_BITS 13 +#define XFS_AGF_RMAP_BLOCKS 0x00002000 +#define XFS_AGF_NUM_BITS 14 #define XFS_AGF_ALL_BITS ((1 << XFS_AGF_NUM_BITS) - 1) #define XFS_AGF_FLAGS \ @@ -686,7 +690,8 @@ typedef struct xfs_agf { { XFS_AGF_FREEBLKS, "FREEBLKS" }, \ { XFS_AGF_LONGEST, "LONGEST" }, \ { XFS_AGF_BTREEBLKS, "BTREEBLKS" }, \ - { XFS_AGF_UUID, "UUID" } + { XFS_AGF_UUID, "UUID" }, \ + { XFS_AGF_RMAP_BLOCKS, "RMAP_BLOCKS" } /* disk block (xfs_daddr_t) in the AG */ #define XFS_AGF_DADDR(mp) ((xfs_daddr_t)(1 << (mp)->m_sectbb_log)) diff --git a/fs/xfs/libxfs/xfs_rmap_btree.c b/fs/xfs/libxfs/xfs_rmap_btree.c index bc1faebc84ec..17b8eeb34ac8 100644 --- a/fs/xfs/libxfs/xfs_rmap_btree.c +++ b/fs/xfs/libxfs/xfs_rmap_btree.c @@ -98,6 +98,8 @@ xfs_rmapbt_alloc_block( union xfs_btree_ptr *new, int *stat) { + struct xfs_buf *agbp = cur->bc_private.a.agbp; + struct xfs_agf *agf = XFS_BUF_TO_AGF(agbp); int error; xfs_agblock_t bno; @@ -124,6 +126,8 @@ xfs_rmapbt_alloc_block( xfs_trans_agbtree_delta(cur->bc_tp, 1); new->s = cpu_to_be32(bno); + be32_add_cpu(&agf->agf_rmap_blocks, 1); + xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS); XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT); *stat = 1; @@ -143,6 +147,8 @@ xfs_rmapbt_free_block( bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp)); trace_xfs_rmapbt_free_block(cur->bc_mp, cur->bc_private.a.agno, bno, 1); + be32_add_cpu(&agf->agf_rmap_blocks, -1); + xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS); error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1); if (error) return error; diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c index 0f96847b90e1..0b7f986745c1 100644 --- a/fs/xfs/xfs_fsops.c +++ b/fs/xfs/xfs_fsops.c @@ -248,6 +248,7 @@ xfs_growfs_data_private( agf->agf_roots[XFS_BTNUM_RMAPi] = cpu_to_be32(XFS_RMAP_BLOCK(mp)); agf->agf_levels[XFS_BTNUM_RMAPi] = cpu_to_be32(1); + agf->agf_rmap_blocks = cpu_to_be32(1); } agf->agf_flfirst = cpu_to_be32(1); -- cgit v1.2.3-70-g09d2 From 97dd8c9ee60c98f09be86a2e7e61b18f8d8ed4e9 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:39:47 +1000 Subject: iomap: remove superflous mark_page_accessed from iomap_write_actor This catches up with commit 2457ae ("mm: non-atomically mark page accessed during page cache allocation where possible"), which moved the initial access marking into the pagecache allocator. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/iomap.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index 48141b8eff5f..f39c3181fa10 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -199,7 +199,6 @@ again: pagefault_enable(); flush_dcache_page(page); - mark_page_accessed(page); status = iomap_write_end(inode, pos, bytes, copied, page); if (unlikely(status < 0)) -- cgit v1.2.3-70-g09d2 From 274c887494cb248eb05f8180bda8298942d98625 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:40:18 +1000 Subject: iomap: remove superflous pagefault_disable from iomap_write_actor iov_iter_copy_from_user_atomic disables page faults internally, no need to do it around the call. This also brings the iomap code in line with the original filemap version. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/iomap.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index f39c3181fa10..74712e25184b 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -194,9 +194,7 @@ again: if (mapping_writably_mapped(inode->i_mapping)) flush_dcache_page(page); - pagefault_disable(); copied = iov_iter_copy_from_user_atomic(page, i, offset, bytes); - pagefault_enable(); flush_dcache_page(page); -- cgit v1.2.3-70-g09d2 From 8896b8f60951ff0a8b7092f495a445714a1bddc4 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 17 Aug 2016 08:41:10 +1000 Subject: iomap: fiemap should honor the FIEMAP_FLAG_SYNC flag The flag is checked as supported, but then we do an unconditional sync of the file, regardless of whether the flag is set or not. Make the sync conditional on having the FIEMAP_FLAG_SYNC flag set. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/iomap.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index 74712e25184b..56c19e617a26 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -467,9 +467,11 @@ int iomap_fiemap(struct inode *inode, struct fiemap_extent_info *fi, if (ret) return ret; - ret = filemap_write_and_wait(inode->i_mapping); - if (ret) - return ret; + if (fi->fi_flags & FIEMAP_FLAG_SYNC) { + ret = filemap_write_and_wait(inode->i_mapping); + if (ret) + return ret; + } while (len > 0) { ret = iomap_apply(inode, start, len, 0, ops, &ctx, -- cgit v1.2.3-70-g09d2 From ac2dc058bce81b83631ad5accb90b8f4abe613b7 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 17 Aug 2016 08:41:34 +1000 Subject: iomap: prepare iomap_fiemap for attribute mappings By bassing through an -ENOENT, similar to the old XFS implementation of FIEMAP_FLAG_XATTR. Signed-off-by: Dave Chinner [hch: split from a larger patch] Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/iomap.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index 56c19e617a26..d9d1f50c36aa 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -476,6 +476,9 @@ int iomap_fiemap(struct inode *inode, struct fiemap_extent_info *fi, while (len > 0) { ret = iomap_apply(inode, start, len, 0, ops, &ctx, iomap_fiemap_actor); + /* inode with no (attribute) mapping will give ENOENT */ + if (ret == -ENOENT) + break; if (ret < 0) return ret; if (ret == 0) -- cgit v1.2.3-70-g09d2 From f20ac7ab17fcf7046bd9bc4166faf9580f713afd Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:42:34 +1000 Subject: iomap: mark ->iomap_end as optional No need to implement it for read-only mappings. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/iomap.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index d9d1f50c36aa..0342254646e3 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -84,8 +84,11 @@ iomap_apply(struct inode *inode, loff_t pos, loff_t length, unsigned flags, * Now the data has been copied, commit the range we've copied. This * should not fail unless the filesystem has had a fatal error. */ - ret = ops->iomap_end(inode, pos, length, written > 0 ? written : 0, - flags, &iomap); + if (ops->iomap_end) { + ret = ops->iomap_end(inode, pos, length, + written > 0 ? written : 0, + flags, &iomap); + } return written ? written : ret; } -- cgit v1.2.3-70-g09d2 From b95a21271b30544a9fb992269d79ed1e1978e023 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:44:52 +1000 Subject: xfs: simplify xfs_file_iomap_begin We'll never get nimap == 0 for a successful return from xfs_bmapi_read, so don't try to handle it. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/xfs_iomap.c | 14 ++++---------- fs/xfs/xfs_trace.h | 1 - 2 files changed, 4 insertions(+), 11 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c index 2114d53df433..1cce760399a2 100644 --- a/fs/xfs/xfs_iomap.c +++ b/fs/xfs/xfs_iomap.c @@ -1037,20 +1037,14 @@ xfs_file_iomap_begin( return error; trace_xfs_iomap_alloc(ip, offset, length, 0, &imap); - xfs_bmbt_to_iomap(ip, iomap, &imap); - } else if (nimaps) { - xfs_iunlock(ip, XFS_ILOCK_EXCL); - trace_xfs_iomap_found(ip, offset, length, 0, &imap); - xfs_bmbt_to_iomap(ip, iomap, &imap); } else { + ASSERT(nimaps); + xfs_iunlock(ip, XFS_ILOCK_EXCL); - trace_xfs_iomap_not_found(ip, offset, length, 0, &imap); - iomap->blkno = IOMAP_NULL_BLOCK; - iomap->type = IOMAP_HOLE; - iomap->offset = offset; - iomap->length = length; + trace_xfs_iomap_found(ip, offset, length, 0, &imap); } + xfs_bmbt_to_iomap(ip, iomap, &imap); return 0; } diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index 551b7e26980c..7e88bec3f359 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -1298,7 +1298,6 @@ DEFINE_IOMAP_EVENT(xfs_get_blocks_alloc); DEFINE_IOMAP_EVENT(xfs_get_blocks_map_direct); DEFINE_IOMAP_EVENT(xfs_iomap_alloc); DEFINE_IOMAP_EVENT(xfs_iomap_found); -DEFINE_IOMAP_EVENT(xfs_iomap_not_found); DECLARE_EVENT_CLASS(xfs_simple_io_class, TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count), -- cgit v1.2.3-70-g09d2 From 1d4795e7bde075588c90df2175349bb2251802d5 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Aug 2016 08:45:30 +1000 Subject: xfs: (re-)implement FIEMAP_FLAG_XATTR Use a special read-only iomap_ops implementation to support fiemap on the attr fork. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/xfs_iomap.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ fs/xfs/xfs_iomap.h | 1 + fs/xfs/xfs_iops.c | 9 ++++++++- 3 files changed, 54 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c index 1cce760399a2..697c8fd4200f 100644 --- a/fs/xfs/xfs_iomap.c +++ b/fs/xfs/xfs_iomap.c @@ -1106,3 +1106,48 @@ struct iomap_ops xfs_iomap_ops = { .iomap_begin = xfs_file_iomap_begin, .iomap_end = xfs_file_iomap_end, }; + +static int +xfs_xattr_iomap_begin( + struct inode *inode, + loff_t offset, + loff_t length, + unsigned flags, + struct iomap *iomap) +{ + struct xfs_inode *ip = XFS_I(inode); + struct xfs_mount *mp = ip->i_mount; + xfs_fileoff_t offset_fsb = XFS_B_TO_FSBT(mp, offset); + xfs_fileoff_t end_fsb = XFS_B_TO_FSB(mp, offset + length); + struct xfs_bmbt_irec imap; + int nimaps = 1, error = 0; + unsigned lockmode; + + if (XFS_FORCED_SHUTDOWN(mp)) + return -EIO; + + lockmode = xfs_ilock_data_map_shared(ip); + + /* if there are no attribute fork or extents, return ENOENT */ + if (XFS_IFORK_Q(ip) || !ip->i_d.di_anextents) { + error = -ENOENT; + goto out_unlock; + } + + ASSERT(ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL); + error = xfs_bmapi_read(ip, offset_fsb, end_fsb - offset_fsb, &imap, + &nimaps, XFS_BMAPI_ENTIRE | XFS_BMAPI_ATTRFORK); +out_unlock: + xfs_iunlock(ip, lockmode); + + if (!error) { + ASSERT(nimaps); + xfs_bmbt_to_iomap(ip, iomap, &imap); + } + + return error; +} + +struct iomap_ops xfs_xattr_iomap_ops = { + .iomap_begin = xfs_xattr_iomap_begin, +}; diff --git a/fs/xfs/xfs_iomap.h b/fs/xfs/xfs_iomap.h index e066d045e2ff..fb8aca3d69ab 100644 --- a/fs/xfs/xfs_iomap.h +++ b/fs/xfs/xfs_iomap.h @@ -35,5 +35,6 @@ void xfs_bmbt_to_iomap(struct xfs_inode *, struct iomap *, struct xfs_bmbt_irec *); extern struct iomap_ops xfs_iomap_ops; +extern struct iomap_ops xfs_xattr_iomap_ops; #endif /* __XFS_IOMAP_H__*/ diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c index ab820f84ed50..b24c3102fa93 100644 --- a/fs/xfs/xfs_iops.c +++ b/fs/xfs/xfs_iops.c @@ -1009,7 +1009,14 @@ xfs_vn_fiemap( int error; xfs_ilock(XFS_I(inode), XFS_IOLOCK_SHARED); - error = iomap_fiemap(inode, fieinfo, start, length, &xfs_iomap_ops); + if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) { + fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR; + error = iomap_fiemap(inode, fieinfo, start, length, + &xfs_xattr_iomap_ops); + } else { + error = iomap_fiemap(inode, fieinfo, start, length, + &xfs_iomap_ops); + } xfs_iunlock(XFS_I(inode), XFS_IOLOCK_SHARED); return error; -- cgit v1.2.3-70-g09d2 From a03f1a6633144300ef4a3a33e95dfa11866f1299 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Wed, 17 Aug 2016 11:12:57 +1000 Subject: xfs: remove OWN_AG rmap when allocating a block from the AGFL When we're really tight on space, xfs_alloc_ag_vextent_small() can allocate a block from the AGFL and give it to the caller. Since the caller is never the AGFL-fixing method, we must remove the OWN_AG reverse mapping because it will clash with whatever rmap the caller wants to set up. This bug was discovered by running generic/299 repeatedly. Signed-off-by: Darrick J. Wong Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_alloc.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c index af1a72efa64c..3dd8f1d54498 100644 --- a/fs/xfs/libxfs/xfs_alloc.c +++ b/fs/xfs/libxfs/xfs_alloc.c @@ -1582,6 +1582,7 @@ xfs_alloc_ag_vextent_small( xfs_extlen_t *flenp, /* result length */ int *stat) /* status: 0-freelist, 1-normal/none */ { + struct xfs_owner_info oinfo; int error; xfs_agblock_t fbno; xfs_extlen_t flen; @@ -1624,6 +1625,18 @@ xfs_alloc_ag_vextent_small( error0); args->wasfromfl = 1; trace_xfs_alloc_small_freelist(args); + + /* + * If we're feeding an AGFL block to something that + * doesn't live in the free space, we need to clear + * out the OWN_AG rmap. + */ + xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG); + error = xfs_rmap_free(args->tp, args->agbp, args->agno, + fbno, 1, &oinfo); + if (error) + goto error0; + *stat = 0; return 0; } -- cgit v1.2.3-70-g09d2 From b873b798af6386f3c7ca1636d4989e9b8f9d1794 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 4 Aug 2016 11:38:25 -0700 Subject: Revert "f2fs: use percpu_rw_semaphore" LKP reported -36.3% regression of fsmark.files_per_sec due to this patch. I've confirmed that fxmark [1] has also slight regression for DWAL. [1] https://github.com/sslab-gatech/fxmark This reverts commit ec795418c41850056feb956534edf059dc1155d4. --- fs/f2fs/f2fs.h | 12 ++++++------ fs/f2fs/node.c | 47 +++++++++++++++++++++++------------------------ fs/f2fs/super.c | 6 +----- 3 files changed, 30 insertions(+), 35 deletions(-) (limited to 'fs') diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index 675fa79d86f6..14f5fe2b841e 100644 --- a/fs/f2fs/f2fs.h +++ b/fs/f2fs/f2fs.h @@ -538,7 +538,7 @@ struct f2fs_nm_info { /* NAT cache management */ struct radix_tree_root nat_root;/* root of the nat entry cache */ struct radix_tree_root nat_set_root;/* root of the nat set cache */ - struct percpu_rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ + struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ struct list_head nat_entries; /* cached nat entry list (clean) */ unsigned int nat_cnt; /* the # of cached nat entries */ unsigned int dirty_nat_cnt; /* total num of nat entries in set */ @@ -787,7 +787,7 @@ struct f2fs_sb_info { struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ struct inode *meta_inode; /* cache meta blocks */ struct mutex cp_mutex; /* checkpoint procedure lock */ - struct percpu_rw_semaphore cp_rwsem; /* blocking FS operations */ + struct rw_semaphore cp_rwsem; /* blocking FS operations */ struct rw_semaphore node_write; /* locking node writes */ wait_queue_head_t cp_wait; unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ @@ -1074,22 +1074,22 @@ static inline void clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) { - percpu_down_read(&sbi->cp_rwsem); + down_read(&sbi->cp_rwsem); } static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) { - percpu_up_read(&sbi->cp_rwsem); + up_read(&sbi->cp_rwsem); } static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) { - percpu_down_write(&sbi->cp_rwsem); + down_write(&sbi->cp_rwsem); } static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) { - percpu_up_write(&sbi->cp_rwsem); + up_write(&sbi->cp_rwsem); } static inline int __get_cp_reason(struct f2fs_sb_info *sbi) diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index b2fa4b615925..f75d197d5beb 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -206,14 +206,14 @@ int need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid) struct nat_entry *e; bool need = false; - percpu_down_read(&nm_i->nat_tree_lock); + down_read(&nm_i->nat_tree_lock); e = __lookup_nat_cache(nm_i, nid); if (e) { if (!get_nat_flag(e, IS_CHECKPOINTED) && !get_nat_flag(e, HAS_FSYNCED_INODE)) need = true; } - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); return need; } @@ -223,11 +223,11 @@ bool is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid) struct nat_entry *e; bool is_cp = true; - percpu_down_read(&nm_i->nat_tree_lock); + down_read(&nm_i->nat_tree_lock); e = __lookup_nat_cache(nm_i, nid); if (e && !get_nat_flag(e, IS_CHECKPOINTED)) is_cp = false; - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); return is_cp; } @@ -237,13 +237,13 @@ bool need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino) struct nat_entry *e; bool need_update = true; - percpu_down_read(&nm_i->nat_tree_lock); + down_read(&nm_i->nat_tree_lock); e = __lookup_nat_cache(nm_i, ino); if (e && get_nat_flag(e, HAS_LAST_FSYNC) && (get_nat_flag(e, IS_CHECKPOINTED) || get_nat_flag(e, HAS_FSYNCED_INODE))) need_update = false; - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); return need_update; } @@ -284,7 +284,7 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni, struct f2fs_nm_info *nm_i = NM_I(sbi); struct nat_entry *e; - percpu_down_write(&nm_i->nat_tree_lock); + down_write(&nm_i->nat_tree_lock); e = __lookup_nat_cache(nm_i, ni->nid); if (!e) { e = grab_nat_entry(nm_i, ni->nid); @@ -334,7 +334,7 @@ static void set_node_addr(struct f2fs_sb_info *sbi, struct node_info *ni, set_nat_flag(e, HAS_FSYNCED_INODE, true); set_nat_flag(e, HAS_LAST_FSYNC, fsync_done); } - percpu_up_write(&nm_i->nat_tree_lock); + up_write(&nm_i->nat_tree_lock); } int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink) @@ -342,7 +342,8 @@ int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink) struct f2fs_nm_info *nm_i = NM_I(sbi); int nr = nr_shrink; - percpu_down_write(&nm_i->nat_tree_lock); + if (!down_write_trylock(&nm_i->nat_tree_lock)) + return 0; while (nr_shrink && !list_empty(&nm_i->nat_entries)) { struct nat_entry *ne; @@ -351,7 +352,7 @@ int try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink) __del_from_nat_cache(nm_i, ne); nr_shrink--; } - percpu_up_write(&nm_i->nat_tree_lock); + up_write(&nm_i->nat_tree_lock); return nr - nr_shrink; } @@ -373,13 +374,13 @@ void get_node_info(struct f2fs_sb_info *sbi, nid_t nid, struct node_info *ni) ni->nid = nid; /* Check nat cache */ - percpu_down_read(&nm_i->nat_tree_lock); + down_read(&nm_i->nat_tree_lock); e = __lookup_nat_cache(nm_i, nid); if (e) { ni->ino = nat_get_ino(e); ni->blk_addr = nat_get_blkaddr(e); ni->version = nat_get_version(e); - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); return; } @@ -403,11 +404,11 @@ void get_node_info(struct f2fs_sb_info *sbi, nid_t nid, struct node_info *ni) node_info_from_raw_nat(ni, &ne); f2fs_put_page(page, 1); cache: - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); /* cache nat entry */ - percpu_down_write(&nm_i->nat_tree_lock); + down_write(&nm_i->nat_tree_lock); cache_nat_entry(sbi, nid, &ne); - percpu_up_write(&nm_i->nat_tree_lock); + up_write(&nm_i->nat_tree_lock); } /* @@ -1788,7 +1789,7 @@ void build_free_nids(struct f2fs_sb_info *sbi) ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), FREE_NID_PAGES, META_NAT, true); - percpu_down_read(&nm_i->nat_tree_lock); + down_read(&nm_i->nat_tree_lock); while (1) { struct page *page = get_current_nat_page(sbi, nid); @@ -1820,7 +1821,7 @@ void build_free_nids(struct f2fs_sb_info *sbi) remove_free_nid(nm_i, nid); } up_read(&curseg->journal_rwsem); - percpu_up_read(&nm_i->nat_tree_lock); + up_read(&nm_i->nat_tree_lock); ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nm_i->next_scan_nid), nm_i->ra_nid_pages, META_NAT, false); @@ -2209,7 +2210,7 @@ void flush_nat_entries(struct f2fs_sb_info *sbi) if (!nm_i->dirty_nat_cnt) return; - percpu_down_write(&nm_i->nat_tree_lock); + down_write(&nm_i->nat_tree_lock); /* * if there are no enough space in journal to store dirty nat @@ -2232,7 +2233,7 @@ void flush_nat_entries(struct f2fs_sb_info *sbi) list_for_each_entry_safe(set, tmp, &sets, set_list) __flush_nat_entry_set(sbi, set); - percpu_up_write(&nm_i->nat_tree_lock); + up_write(&nm_i->nat_tree_lock); f2fs_bug_on(sbi, nm_i->dirty_nat_cnt); } @@ -2268,8 +2269,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi) mutex_init(&nm_i->build_lock); spin_lock_init(&nm_i->free_nid_list_lock); - if (percpu_init_rwsem(&nm_i->nat_tree_lock)) - return -ENOMEM; + init_rwsem(&nm_i->nat_tree_lock); nm_i->next_scan_nid = le32_to_cpu(sbi->ckpt->next_free_nid); nm_i->bitmap_size = __bitmap_size(sbi, NAT_BITMAP); @@ -2326,7 +2326,7 @@ void destroy_node_manager(struct f2fs_sb_info *sbi) spin_unlock(&nm_i->free_nid_list_lock); /* destroy nat cache */ - percpu_down_write(&nm_i->nat_tree_lock); + down_write(&nm_i->nat_tree_lock); while ((found = __gang_lookup_nat_cache(nm_i, nid, NATVEC_SIZE, natvec))) { unsigned idx; @@ -2351,9 +2351,8 @@ void destroy_node_manager(struct f2fs_sb_info *sbi) kmem_cache_free(nat_entry_set_slab, setvec[idx]); } } - percpu_up_write(&nm_i->nat_tree_lock); + up_write(&nm_i->nat_tree_lock); - percpu_free_rwsem(&nm_i->nat_tree_lock); kfree(nm_i->nat_bitmap); sbi->nm_info = NULL; kfree(nm_i); diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c index 1b86d3f638ef..7f863a645ab1 100644 --- a/fs/f2fs/super.c +++ b/fs/f2fs/super.c @@ -706,8 +706,6 @@ static void destroy_percpu_info(struct f2fs_sb_info *sbi) percpu_counter_destroy(&sbi->nr_pages[i]); percpu_counter_destroy(&sbi->alloc_valid_block_count); percpu_counter_destroy(&sbi->total_valid_inode_count); - - percpu_free_rwsem(&sbi->cp_rwsem); } static void f2fs_put_super(struct super_block *sb) @@ -1483,9 +1481,6 @@ static int init_percpu_info(struct f2fs_sb_info *sbi) { int i, err; - if (percpu_init_rwsem(&sbi->cp_rwsem)) - return -ENOMEM; - for (i = 0; i < NR_COUNT_TYPE; i++) { err = percpu_counter_init(&sbi->nr_pages[i], 0, GFP_KERNEL); if (err) @@ -1686,6 +1681,7 @@ try_onemore: sbi->write_io[i].bio = NULL; } + init_rwsem(&sbi->cp_rwsem); init_waitqueue_head(&sbi->cp_wait); init_sb_info(sbi); -- cgit v1.2.3-70-g09d2 From 3024c9a1fefb3ac0d1b0b078a2e3f2f69478daab Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Sat, 6 Aug 2016 21:09:41 +0800 Subject: Revert "f2fs: move i_size_write in f2fs_write_end" This reverts commit a2ee0a300344a6da76186129b078113354fe13d2. When testing with generic/032 of xfstest suit, failure message will be reported as below: generic/032 8s ... [failed, exit status 1] - output mismatch (see results/generic/032.out.bad) --- tests/generic/032.out 2015-01-11 16:52:27.643681072 +0800 +++ results/generic/032.out.bad 2016-08-06 13:44:43.861330500 +0800 @@ -1,5 +1,5 @@ QA output created by 032 -100 iterations -0000000 cdcd cdcd cdcd cdcd cdcd cdcd cdcd cdcd -* -0100000 +1: [768..775]: unwritten +Unwritten extents found! ... (Run 'diff -u tests/generic/032.out results/generic/032.out.bad' to see the entire diff) Ran: generic/032 Failures: generic/032 Failed 1 of 1 tests In write_end(), we should update i_size of inode before unlock page, otherwise, we will lose newly updated data in following race condition. Thread A Thread B - write_end - unlock page - writepages - lock_page - writepage if page is out-of-range of file size, we will skip writting the page. - update i_size Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/data.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index d64d2a515cb2..ccb401eebc11 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -1699,11 +1699,11 @@ static int f2fs_write_end(struct file *file, trace_f2fs_write_end(inode, pos, len, copied); set_page_dirty(page); - f2fs_put_page(page, 1); if (pos + copied > i_size_read(inode)) f2fs_i_size_write(inode, pos + copied); + f2fs_put_page(page, 1); f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); return copied; } -- cgit v1.2.3-70-g09d2 From fe8494bfc8c2914fca821d4ae994aef039be5cf1 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Thu, 4 Aug 2016 20:13:02 +0800 Subject: f2fs: allow copying file range only in between regular files Only if two input files are regular files, we allow copying data in range of them, otherwise, deny it. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 0e493f63ea41..685e629f768b 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2086,8 +2086,8 @@ static int f2fs_move_file_range(struct file *file_in, loff_t pos_in, if (unlikely(f2fs_readonly(src->i_sb))) return -EROFS; - if (S_ISDIR(src->i_mode) || S_ISDIR(dst->i_mode)) - return -EISDIR; + if (!S_ISREG(src->i_mode) || !S_ISREG(dst->i_mode)) + return -EINVAL; if (f2fs_encrypted_inode(src) || f2fs_encrypted_inode(dst)) return -EOPNOTSUPP; -- cgit v1.2.3-70-g09d2 From 20a3d61d46e1fb45efa6eb4637c0dcd3f00a14e9 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Thu, 4 Aug 2016 20:13:03 +0800 Subject: f2fs: avoid potential deadlock in f2fs_move_file_range Thread A Thread B - inode_lock fileA - inode_lock fileB - inode_lock fileA - inode_lock fileB We may encounter above potential deadlock during moving file range in concurrent scenario. This patch fixes the issue by using inode_trylock instead. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 685e629f768b..47abb96098e4 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2093,8 +2093,12 @@ static int f2fs_move_file_range(struct file *file_in, loff_t pos_in, return -EOPNOTSUPP; inode_lock(src); - if (src != dst) - inode_lock(dst); + if (src != dst) { + if (!inode_trylock(dst)) { + ret = -EBUSY; + goto out; + } + } ret = -EINVAL; if (pos_in + len > src->i_size || pos_in + len < pos_in) @@ -2152,6 +2156,7 @@ static int f2fs_move_file_range(struct file *file_in, loff_t pos_in, out_unlock: if (src != dst) inode_unlock(dst); +out: inode_unlock(src); return ret; } -- cgit v1.2.3-70-g09d2 From 9a0fe86745b8e95f7ea39933a956f5771332c430 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 19 Aug 2016 15:33:12 -0400 Subject: pNFS: Handle NFS4ERR_OLD_STATEID correctly in LAYOUTSTAT calls We normally want to update the stateid and then retry, Signed-off-by: Trond Myklebust --- fs/nfs/nfs42proc.c | 34 +++++++++++++++++++++++++++++----- fs/nfs/pnfs.c | 1 - 2 files changed, 29 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c index 6f4752734804..64b43b4ad9dd 100644 --- a/fs/nfs/nfs42proc.c +++ b/fs/nfs/nfs42proc.c @@ -318,10 +318,22 @@ static void nfs42_layoutstat_prepare(struct rpc_task *task, void *calldata) { struct nfs42_layoutstat_data *data = calldata; - struct nfs_server *server = NFS_SERVER(data->args.inode); + struct inode *inode = data->inode; + struct nfs_server *server = NFS_SERVER(inode); + struct pnfs_layout_hdr *lo; + spin_lock(&inode->i_lock); + lo = NFS_I(inode)->layout; + if (!pnfs_layout_is_valid(lo)) { + spin_unlock(&inode->i_lock); + rpc_exit(task, 0); + return; + } + nfs4_stateid_copy(&data->args.stateid, &lo->plh_stateid); + spin_unlock(&inode->i_lock); nfs41_setup_sequence(nfs4_get_session(server), &data->args.seq_args, &data->res.seq_res, task); + } static void @@ -341,11 +353,11 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata) case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_DELEG_REVOKED: case -NFS4ERR_STALE_STATEID: - case -NFS4ERR_OLD_STATEID: case -NFS4ERR_BAD_STATEID: spin_lock(&inode->i_lock); lo = NFS_I(inode)->layout; - if (lo && nfs4_stateid_match(&data->args.stateid, + if (pnfs_layout_is_valid(lo) && + nfs4_stateid_match(&data->args.stateid, &lo->plh_stateid)) { LIST_HEAD(head); @@ -359,11 +371,23 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata) } else spin_unlock(&inode->i_lock); break; + case -NFS4ERR_OLD_STATEID: + spin_lock(&inode->i_lock); + lo = NFS_I(inode)->layout; + if (pnfs_layout_is_valid(lo) && + nfs4_stateid_match_other(&data->args.stateid, + &lo->plh_stateid)) { + /* Do we need to delay before resending? */ + if (!nfs4_stateid_is_newer(&lo->plh_stateid, + &data->args.stateid)) + rpc_delay(task, HZ); + rpc_restart_call_prepare(task); + } + spin_unlock(&inode->i_lock); + break; case -ENOTSUPP: case -EOPNOTSUPP: NFS_SERVER(inode)->caps &= ~NFS_CAP_LAYOUTSTATS; - default: - break; } dprintk("%s server returns %d\n", __func__, task->tk_status); diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 70806cae0d36..bf98f1b2595f 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -2510,7 +2510,6 @@ pnfs_report_layoutstat(struct inode *inode, gfp_t gfp_flags) data->args.fh = NFS_FH(inode); data->args.inode = inode; - nfs4_stateid_copy(&data->args.stateid, &hdr->plh_stateid); status = ld->prepare_layoutstats(&data->args); if (status) goto out_free; -- cgit v1.2.3-70-g09d2 From e9e5e3fae8da7e237049e00e0bfc9e32fd808fe8 Mon Sep 17 00:00:00 2001 From: Vegard Nossum Date: Mon, 22 Aug 2016 12:47:43 +0200 Subject: bdev: fix NULL pointer dereference I got this: kasan: GPF could be caused by NULL-ptr deref or user memory access general protection fault: 0000 [#1] PREEMPT SMP KASAN Dumping ftrace buffer: (ftrace buffer empty) CPU: 0 PID: 5505 Comm: syz-executor Not tainted 4.8.0-rc2+ #161 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.3-0-ge2fc41e-prebuilt.qemu-project.org 04/01/2014 task: ffff880113415940 task.stack: ffff880118350000 RIP: 0010:[] [] bd_mount+0x52/0xa0 RSP: 0018:ffff880118357ca0 EFLAGS: 00010207 RAX: dffffc0000000000 RBX: ffffffffffffffff RCX: ffffc90000bb6000 RDX: 0000000000000018 RSI: ffffffff846d6b20 RDI: 00000000000000c7 RBP: ffff880118357cb0 R08: ffff880115967c68 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffff8801188211e8 R13: ffffffff847baa20 R14: ffff8801139cb000 R15: 0000000000000080 FS: 00007fa3ff6c0700(0000) GS:ffff88011aa00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fc1d8cc7e78 CR3: 0000000109f20000 CR4: 00000000000006f0 DR0: 000000000000001e DR1: 000000000000001e DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600 Stack: ffff880112cfd6c0 ffff8801188211e8 ffff880118357cf0 ffffffff8167f207 ffffffff816d7a1e ffff880112a413c0 ffffffff847baa20 ffff8801188211e8 0000000000000080 ffff880112cfd6c0 ffff880118357d38 ffffffff816dce0a Call Trace: [] mount_fs+0x97/0x2e0 [] ? alloc_vfsmnt+0x55e/0x760 [] vfs_kern_mount+0x7a/0x300 [] ? _raw_read_unlock+0x2c/0x50 [] do_mount+0x3d7/0x2730 [] ? trace_do_page_fault+0x1f4/0x3a0 [] ? copy_mount_string+0x40/0x40 [] ? memset+0x31/0x40 [] ? copy_mount_options+0x1ee/0x320 [] SyS_mount+0xb2/0x120 [] ? copy_mnt_ns+0x970/0x970 [] do_syscall_64+0x1c4/0x4e0 [] entry_SYSCALL64_slow_path+0x25/0x25 Code: 83 e8 63 1b fc ff 48 85 c0 48 89 c3 74 4c e8 56 35 d1 ff 48 8d bb c8 00 00 00 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <80> 3c 02 00 75 36 4c 8b a3 c8 00 00 00 48 b8 00 00 00 00 00 fc RIP [] bd_mount+0x52/0xa0 RSP ---[ end trace 13690ad962168b98 ]--- mount_pseudo() returns ERR_PTR(), not NULL, on error. Fixes: 3684aa7099e0 ("block-dev: enable writeback cgroup support") Cc: Shaohua Li Cc: Tejun Heo Cc: Jens Axboe Cc: stable@vger.kernel.org Signed-off-by: Vegard Nossum Signed-off-by: Jens Axboe --- fs/block_dev.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/block_dev.c b/fs/block_dev.c index c3cdde87cc8c..e17bdbdfe9b1 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -646,7 +646,7 @@ static struct dentry *bd_mount(struct file_system_type *fs_type, { struct dentry *dent; dent = mount_pseudo(fs_type, "bdev:", &bdev_sops, NULL, BDEVFS_MAGIC); - if (dent) + if (!IS_ERR(dent)) dent->d_sb->s_iflags |= SB_I_CGROUPWB; return dent; } -- cgit v1.2.3-70-g09d2 From b88fa69eaa8649f11828158c7b65c4bcd886ebd5 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 23 Aug 2016 11:19:33 -0400 Subject: pNFS: The client must not do I/O to the DS if it's lease has expired Ensure that the client conforms to the normative behaviour described in RFC5661 Section 12.7.2: "If a client believes its lease has expired, it MUST NOT send I/O to the storage device until it has validated its lease." So ensure that we wait for the lease to be validated before using the layout. Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v3.20+ --- fs/nfs/pnfs.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index bf98f1b2595f..6daf034645c8 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -1555,6 +1555,7 @@ pnfs_update_layout(struct inode *ino, } lookup_again: + nfs4_client_recover_expired_lease(clp); first = false; spin_lock(&ino->i_lock); lo = pnfs_find_alloc_layout(ino, ctx, gfp_flags); -- cgit v1.2.3-70-g09d2 From 41963c10c47a35185e68cb9049f7a3493c94d2d7 Mon Sep 17 00:00:00 2001 From: Benjamin Coddington Date: Mon, 22 Aug 2016 14:11:16 -0400 Subject: pnfs/blocklayout: update last_write_offset atomically with extents Block/SCSI layout write completion may add committable extents to the extent tree before updating the layout's last-written byte under the inode lock. If a sync happens before this value is updated, then prepare_layoutcommit may find and encode these extents which would produce a LAYOUTCOMMIT request whose encoded extents are larger than the request's loca_length. Fix this by using a last-written byte value that is updated atomically with the extent tree so that commitable extents always match. Signed-off-by: Benjamin Coddington Signed-off-by: Trond Myklebust --- fs/nfs/blocklayout/blocklayout.c | 2 +- fs/nfs/blocklayout/blocklayout.h | 3 ++- fs/nfs/blocklayout/extent_tree.c | 10 +++++++--- 3 files changed, 10 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfs/blocklayout/blocklayout.c b/fs/nfs/blocklayout/blocklayout.c index f55a4e756047..217847679f0e 100644 --- a/fs/nfs/blocklayout/blocklayout.c +++ b/fs/nfs/blocklayout/blocklayout.c @@ -346,7 +346,7 @@ static void bl_write_cleanup(struct work_struct *work) PAGE_SIZE - 1) & (loff_t)PAGE_MASK; ext_tree_mark_written(bl, start >> SECTOR_SHIFT, - (end - start) >> SECTOR_SHIFT); + (end - start) >> SECTOR_SHIFT, end); } pnfs_ld_write_done(hdr); diff --git a/fs/nfs/blocklayout/blocklayout.h b/fs/nfs/blocklayout/blocklayout.h index 18e6fd0b9506..efc007f00742 100644 --- a/fs/nfs/blocklayout/blocklayout.h +++ b/fs/nfs/blocklayout/blocklayout.h @@ -141,6 +141,7 @@ struct pnfs_block_layout { struct rb_root bl_ext_ro; spinlock_t bl_ext_lock; /* Protects list manipulation */ bool bl_scsi_layout; + u64 bl_lwb; }; static inline struct pnfs_block_layout * @@ -182,7 +183,7 @@ int ext_tree_insert(struct pnfs_block_layout *bl, int ext_tree_remove(struct pnfs_block_layout *bl, bool rw, sector_t start, sector_t end); int ext_tree_mark_written(struct pnfs_block_layout *bl, sector_t start, - sector_t len); + sector_t len, u64 lwb); bool ext_tree_lookup(struct pnfs_block_layout *bl, sector_t isect, struct pnfs_block_extent *ret, bool rw); int ext_tree_prepare_commit(struct nfs4_layoutcommit_args *arg); diff --git a/fs/nfs/blocklayout/extent_tree.c b/fs/nfs/blocklayout/extent_tree.c index 992bcb19c11e..c85fbfd2d0d9 100644 --- a/fs/nfs/blocklayout/extent_tree.c +++ b/fs/nfs/blocklayout/extent_tree.c @@ -402,7 +402,7 @@ ext_tree_split(struct rb_root *root, struct pnfs_block_extent *be, int ext_tree_mark_written(struct pnfs_block_layout *bl, sector_t start, - sector_t len) + sector_t len, u64 lwb) { struct rb_root *root = &bl->bl_ext_rw; sector_t end = start + len; @@ -471,6 +471,8 @@ ext_tree_mark_written(struct pnfs_block_layout *bl, sector_t start, } } out: + if (bl->bl_lwb < lwb) + bl->bl_lwb = lwb; spin_unlock(&bl->bl_ext_lock); __ext_put_deviceids(&tmp); @@ -518,7 +520,7 @@ static __be32 *encode_scsi_range(struct pnfs_block_extent *be, __be32 *p) } static int ext_tree_encode_commit(struct pnfs_block_layout *bl, __be32 *p, - size_t buffer_size, size_t *count) + size_t buffer_size, size_t *count, __u64 *lastbyte) { struct pnfs_block_extent *be; int ret = 0; @@ -542,6 +544,8 @@ static int ext_tree_encode_commit(struct pnfs_block_layout *bl, __be32 *p, p = encode_block_extent(be, p); be->be_tag = EXTENT_COMMITTING; } + *lastbyte = bl->bl_lwb - 1; + bl->bl_lwb = 0; spin_unlock(&bl->bl_ext_lock); return ret; @@ -564,7 +568,7 @@ ext_tree_prepare_commit(struct nfs4_layoutcommit_args *arg) arg->layoutupdate_pages = &arg->layoutupdate_page; retry: - ret = ext_tree_encode_commit(bl, start_p + 1, buffer_size, &count); + ret = ext_tree_encode_commit(bl, start_p + 1, buffer_size, &count, &arg->lastbytewritten); if (unlikely(ret)) { ext_tree_free_commitdata(arg, buffer_size); -- cgit v1.2.3-70-g09d2 From c0082e985fdf77b02fc9e0dac3b58504dcf11b7a Mon Sep 17 00:00:00 2001 From: Vincent Stehlé Date: Fri, 12 Aug 2016 15:26:30 +0200 Subject: ubifs: Fix assertion in layout_in_gaps() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit An assertion in layout_in_gaps() verifies that the gap_lebs pointer is below the maximum bound. When computing this maximum bound the idx_lebs count is multiplied by sizeof(int), while C pointers arithmetic does take into account the size of the pointed elements implicitly already. Remove the multiplication to fix the assertion. Fixes: 1e51764a3c2ac05a ("UBIFS: add new flash file system") Cc: Signed-off-by: Vincent Stehlé Cc: Artem Bityutskiy Signed-off-by: Artem Bityutskiy Signed-off-by: Richard Weinberger --- fs/ubifs/tnc_commit.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ubifs/tnc_commit.c b/fs/ubifs/tnc_commit.c index b45345d701e7..51157da3f76e 100644 --- a/fs/ubifs/tnc_commit.c +++ b/fs/ubifs/tnc_commit.c @@ -370,7 +370,7 @@ static int layout_in_gaps(struct ubifs_info *c, int cnt) p = c->gap_lebs; do { - ubifs_assert(p < c->gap_lebs + sizeof(int) * c->lst.idx_lebs); + ubifs_assert(p < c->gap_lebs + c->lst.idx_lebs); written = layout_leb_in_gaps(c, p); if (written < 0) { err = written; -- cgit v1.2.3-70-g09d2 From 17ce1eb0b64eb27d4f9180daae7495fa022c7b0d Mon Sep 17 00:00:00 2001 From: Richard Weinberger Date: Sun, 31 Jul 2016 21:42:23 +0200 Subject: ubifs: Fix xattr generic handler usage UBIFS uses full names to work with xattrs, therefore we have to use xattr_full_name() to obtain the xattr prefix as string. Cc: Cc: Andreas Gruenbacher Fixes: 2b88fc21ca ("ubifs: Switch to generic xattr handlers") Signed-off-by: Richard Weinberger Reviewed-by: Andreas Gruenbacher Tested-by: Dongsheng Yang --- fs/ubifs/xattr.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ubifs/xattr.c b/fs/ubifs/xattr.c index e237811f09ce..11a004114eba 100644 --- a/fs/ubifs/xattr.c +++ b/fs/ubifs/xattr.c @@ -575,7 +575,8 @@ static int ubifs_xattr_get(const struct xattr_handler *handler, dbg_gen("xattr '%s', ino %lu ('%pd'), buf size %zd", name, inode->i_ino, dentry, size); - return __ubifs_getxattr(inode, name, buffer, size); + name = xattr_full_name(handler, name); + return __ubifs_getxattr(inode, name, buffer, size); } static int ubifs_xattr_set(const struct xattr_handler *handler, @@ -586,6 +587,8 @@ static int ubifs_xattr_set(const struct xattr_handler *handler, dbg_gen("xattr '%s', host ino %lu ('%pd'), size %zd", name, inode->i_ino, dentry, size); + name = xattr_full_name(handler, name); + if (value) return __ubifs_setxattr(inode, name, value, size, flags); else -- cgit v1.2.3-70-g09d2 From 1c1ea4f781db9f754842b9c31d1eff400d17cddc Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 19 Jul 2016 15:36:05 -0700 Subject: Btrfs: fix memory leak of reloc_root When some critical errors occur and FS would be flipped into RO, if we have an on-going balance, we can end up with a memory leak of root->reloc_root since btrfs_drop_snapshots() bails out without freeing reloc_root at the very early start. However, we're not able to free reloc_root in btrfs_drop_snapshots() because its caller, merge_reloc_roots(), still needs to access it to cleanup reloc_root's rbtree. This makes us free reloc_root when we're going to free fs/file roots. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index ff2362dca91a..4a34c9fcc1a7 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3744,8 +3744,15 @@ void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info, if (btrfs_root_refs(&root->root_item) == 0) synchronize_srcu(&fs_info->subvol_srcu); - if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { btrfs_free_log(NULL, root); + if (root->reloc_root) { + free_extent_buffer(root->reloc_root->node); + free_extent_buffer(root->reloc_root->commit_root); + btrfs_put_fs_root(root->reloc_root); + root->reloc_root = NULL; + } + } if (root->free_ino_pinned) __btrfs_remove_free_space_cache(root->free_ino_pinned); -- cgit v1.2.3-70-g09d2 From d8422ba334f9df16e071bc77707e55fd7f8446ae Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Wed, 20 Jul 2016 15:04:18 +0800 Subject: btrfs: backref: Fix soft lockup in __merge_refs function When over 1000 file extents refers to one extent, find_parent_nodes() will be obviously slow, due to the O(n^2)~O(n^3) loops inside __merge_refs(). The following ftrace shows the cubic growth of execution time: 256 refs 5) + 91.768 us | __add_keyed_refs.isra.12 [btrfs](); 5) 1.447 us | __add_missing_keys.isra.13 [btrfs](); 5) ! 114.544 us | __merge_refs [btrfs](); 5) ! 136.399 us | __merge_refs [btrfs](); 512 refs 6) ! 279.859 us | __add_keyed_refs.isra.12 [btrfs](); 6) 3.164 us | __add_missing_keys.isra.13 [btrfs](); 6) ! 442.498 us | __merge_refs [btrfs](); 6) # 2091.073 us | __merge_refs [btrfs](); and 1024 refs 7) ! 368.683 us | __add_keyed_refs.isra.12 [btrfs](); 7) 4.810 us | __add_missing_keys.isra.13 [btrfs](); 7) # 2043.428 us | __merge_refs [btrfs](); 7) * 18964.23 us | __merge_refs [btrfs](); And sort them into the following char: (Unit: us) ------------------------------------------------------------------------ Trace function | 256 ref | 512 refs | 1024 refs | ------------------------------------------------------------------------ __add_keyed_refs | 91 | 249 | 368 | __add_missing_keys | 1 | 3 | 4 | __merge_refs 1st call | 114 | 442 | 2043 | __merge_refs 2nd call | 136 | 2091 | 18964 | ------------------------------------------------------------------------ We can see the that __add_keyed_refs() grows almost in linear behavior. And __add_missing_keys() in this case doesn't change much or takes much time. While for the 1st __merge_refs() it's square growth for the 2nd __merge_refs() call it's cubic growth. It's no doubt that merge_refs() will take a long long time to execute if the number of refs continues its grows. So add a cond_resced() into the loop of __merge_refs(). Although this will solve the problem of soft lockup, we need to use the new rb_tree based structure introduced by Lu Fengqi to really solve the long execution time. Signed-off-by: Qu Wenruo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 2b88439c2ee8..455a6b2fd539 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -589,6 +589,7 @@ static void __merge_refs(struct list_head *head, int mode) list_del(&ref2->list); kmem_cache_free(btrfs_prelim_ref_cache, ref2); + cond_resched(); } } -- cgit v1.2.3-70-g09d2 From f3bca8028bd934e96257b8bd1143e6474fe98465 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Wed, 20 Jul 2016 17:33:44 -0700 Subject: Btrfs: add ASSERT for block group's memory leak This adds several ASSERT()' s to report memory leak of block group cache. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index c2b81b0d3fe0..e962b0e25324 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -9942,6 +9942,7 @@ void btrfs_put_block_group_cache(struct btrfs_fs_info *info) block_group->iref = 0; block_group->inode = NULL; spin_unlock(&block_group->lock); + ASSERT(block_group->io_ctl.inode == NULL); iput(inode); last = block_group->key.objectid + block_group->key.offset; btrfs_put_block_group(block_group); @@ -9999,6 +10000,10 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info) free_excluded_extents(info->extent_root, block_group); btrfs_remove_free_space_cache(block_group); + ASSERT(list_empty(&block_group->dirty_list)); + ASSERT(list_empty(&block_group->io_list)); + ASSERT(list_empty(&block_group->bg_list)); + ASSERT(atomic_read(&block_group->count) == 1); btrfs_put_block_group(block_group); spin_lock(&info->block_group_cache_lock); -- cgit v1.2.3-70-g09d2 From eecba891d38051ebf7f4af6394d188a5fd151a6a Mon Sep 17 00:00:00 2001 From: Alex Lyakas Date: Sun, 6 Dec 2015 12:32:31 +0200 Subject: btrfs: flush_space: treat return value of do_chunk_alloc properly do_chunk_alloc returns 1 when it succeeds to allocate a new chunk. But flush_space will not convert this to 0, and will also return 1. As a result, reserve_metadata_bytes will think that flush_space failed, and may potentially return this value "1" to the caller (depends how reserve_metadata_bytes was called). The caller will also treat this as an error. For example, btrfs_block_rsv_refill does: int ret = -ENOSPC; ... ret = reserve_metadata_bytes(root, block_rsv, num_bytes, flush); if (!ret) { block_rsv_add_bytes(block_rsv, num_bytes, 0); return 0; } return ret; So it will return -ENOSPC. Signed-off-by: Alex Lyakas Reviewed-by: Josef Bacik Reviewed-by: Liu Bo Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index e962b0e25324..d2cc16ff3e21 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4882,7 +4882,7 @@ static int flush_space(struct btrfs_root *root, btrfs_get_alloc_profile(root, 0), CHUNK_ALLOC_NO_FORCE); btrfs_end_transaction(trans, root); - if (ret == -ENOSPC) + if (ret > 0 || ret == -ENOSPC) ret = 0; break; case COMMIT_TRANS: -- cgit v1.2.3-70-g09d2 From d2c609b834d62f1e91f1635a27dca29f7806d3d6 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Mon, 15 Aug 2016 12:10:33 -0400 Subject: btrfs: properly track when rescan worker is running The qgroup_flags field is overloaded such that it reflects the on-disk status of qgroups and the runtime state. The BTRFS_QGROUP_STATUS_FLAG_RESCAN flag is used to indicate that a rescan operation is in progress, but if the file system is unmounted while a rescan is running, the rescan operation is paused. If the file system is then mounted read-only, the flag will still be present but the rescan operation will not have been resumed. When we go to umount, btrfs_qgroup_wait_for_completion will see the flag and interpret it to mean that the rescan worker is still running and will wait for a completion that will never come. This patch uses a separate flag to indicate when the worker is running. The locking and state surrounding the qgroup rescan worker needs a lot of attention beyond this patch but this is enough to avoid a hung umount. Cc: # v4.4+ Signed-off-by; Jeff Mahoney Reviewed-by: Qu Wenruo Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 1 + fs/btrfs/disk-io.c | 1 + fs/btrfs/qgroup.c | 9 ++++++++- 3 files changed, 10 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index f66a0ba9a2a9..acc6850a118f 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1028,6 +1028,7 @@ struct btrfs_fs_info { struct btrfs_workqueue *qgroup_rescan_workers; struct completion qgroup_rescan_completion; struct btrfs_work qgroup_rescan_work; + bool qgroup_rescan_running; /* protected by qgroup_rescan_lock */ /* filesystem state */ unsigned long fs_state; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4a34c9fcc1a7..f5b2a7fb4575 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2304,6 +2304,7 @@ static void btrfs_init_qgroup(struct btrfs_fs_info *fs_info) fs_info->quota_enabled = 0; fs_info->pending_quota_state = 0; fs_info->qgroup_ulist = NULL; + fs_info->qgroup_rescan_running = false; mutex_init(&fs_info->qgroup_rescan_lock); } diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 93ee1c18ef9d..3fe295e2c84a 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -2303,6 +2303,10 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work) int err = -ENOMEM; int ret = 0; + mutex_lock(&fs_info->qgroup_rescan_lock); + fs_info->qgroup_rescan_running = true; + mutex_unlock(&fs_info->qgroup_rescan_lock); + path = btrfs_alloc_path(); if (!path) goto out; @@ -2369,6 +2373,9 @@ out: } done: + mutex_lock(&fs_info->qgroup_rescan_lock); + fs_info->qgroup_rescan_running = false; + mutex_unlock(&fs_info->qgroup_rescan_lock); complete_all(&fs_info->qgroup_rescan_completion); } @@ -2494,7 +2501,7 @@ int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info) mutex_lock(&fs_info->qgroup_rescan_lock); spin_lock(&fs_info->qgroup_lock); - running = fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN; + running = fs_info->qgroup_rescan_running; spin_unlock(&fs_info->qgroup_lock); mutex_unlock(&fs_info->qgroup_rescan_lock); -- cgit v1.2.3-70-g09d2 From d06f23d6a947c9abae41dc46be69a56baf36f436 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Mon, 8 Aug 2016 22:08:06 -0400 Subject: btrfs: waiting on qgroup rescan should not always be interruptible We wait on qgroup rescan completion in three places: file system shutdown, the quota disable ioctl, and the rescan wait ioctl. If the user sends a signal while we're waiting, we continue happily along. This is expected behavior for the rescan wait ioctl. It's racy in the shutdown path but mostly works due to other unrelated synchronization points. In the quota disable path, it Oopses the kernel pretty much immediately. Cc: # v4.4+ Signed-off-by: Jeff Mahoney Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 2 +- fs/btrfs/ioctl.c | 2 +- fs/btrfs/qgroup.c | 12 +++++++++--- fs/btrfs/qgroup.h | 3 ++- 4 files changed, 13 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index f5b2a7fb4575..7857f64e1cae 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -3864,7 +3864,7 @@ void close_ctree(struct btrfs_root *root) smp_mb(); /* wait for the qgroup rescan worker to stop */ - btrfs_qgroup_wait_for_completion(fs_info); + btrfs_qgroup_wait_for_completion(fs_info, false); /* wait for the uuid_scan task to finish */ down(&fs_info->uuid_tree_rescan_sem); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 14ed1e9e6bc8..b2a2da5893af 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -5084,7 +5084,7 @@ static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg) if (!capable(CAP_SYS_ADMIN)) return -EPERM; - return btrfs_qgroup_wait_for_completion(root->fs_info); + return btrfs_qgroup_wait_for_completion(root->fs_info, true); } static long _btrfs_ioctl_set_received_subvol(struct file *file, diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 3fe295e2c84a..13eb6a7a4db1 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -995,7 +995,7 @@ int btrfs_quota_disable(struct btrfs_trans_handle *trans, goto out; fs_info->quota_enabled = 0; fs_info->pending_quota_state = 0; - btrfs_qgroup_wait_for_completion(fs_info); + btrfs_qgroup_wait_for_completion(fs_info, false); spin_lock(&fs_info->qgroup_lock); quota_root = fs_info->quota_root; fs_info->quota_root = NULL; @@ -2494,7 +2494,8 @@ btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info) return 0; } -int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info) +int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info, + bool interruptible) { int running; int ret = 0; @@ -2505,9 +2506,14 @@ int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info) spin_unlock(&fs_info->qgroup_lock); mutex_unlock(&fs_info->qgroup_rescan_lock); - if (running) + if (!running) + return 0; + + if (interruptible) ret = wait_for_completion_interruptible( &fs_info->qgroup_rescan_completion); + else + wait_for_completion(&fs_info->qgroup_rescan_completion); return ret; } diff --git a/fs/btrfs/qgroup.h b/fs/btrfs/qgroup.h index 710887c06aaf..af3e5578cad7 100644 --- a/fs/btrfs/qgroup.h +++ b/fs/btrfs/qgroup.h @@ -46,7 +46,8 @@ int btrfs_quota_disable(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info); int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info); void btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info); -int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info); +int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info, + bool interruptible); int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info, u64 src, u64 dst); int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans, -- cgit v1.2.3-70-g09d2 From cb93b52cc005ba0e470845b519c662e661d5113c Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Mon, 15 Aug 2016 10:36:50 +0800 Subject: btrfs: qgroup: Refactor btrfs_qgroup_insert_dirty_extent() Refactor btrfs_qgroup_insert_dirty_extent() function, to two functions: 1. btrfs_qgroup_insert_dirty_extent_nolock() Almost the same with original code. For delayed_ref usage, which has delayed refs locked. Change the return value type to int, since caller never needs the pointer, but only needs to know if they need to free the allocated memory. 2. btrfs_qgroup_insert_dirty_extent() The more encapsulated version. Will do the delayed_refs lock, memory allocation, quota enabled check and other things. The original design is to keep exported functions to minimal, but since more btrfs hacks exposed, like replacing path in balance, we need to record dirty extents manually, so we have to add such functions. Also, add comment for both functions, to info developers how to keep qgroup correct when doing hacks. Cc: Mark Fasheh Signed-off-by: Qu Wenruo Reviewed-and-Tested-by: Goldwyn Rodrigues Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/delayed-ref.c | 7 ++----- fs/btrfs/extent-tree.c | 37 +++++-------------------------------- fs/btrfs/qgroup.c | 41 +++++++++++++++++++++++++++++++++++------ fs/btrfs/qgroup.h | 33 +++++++++++++++++++++++++++++---- 4 files changed, 71 insertions(+), 47 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c index d9ddcfc18c91..ac02e041464b 100644 --- a/fs/btrfs/delayed-ref.c +++ b/fs/btrfs/delayed-ref.c @@ -541,7 +541,6 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info, struct btrfs_delayed_ref_head *existing; struct btrfs_delayed_ref_head *head_ref = NULL; struct btrfs_delayed_ref_root *delayed_refs; - struct btrfs_qgroup_extent_record *qexisting; int count_mod = 1; int must_insert_reserved = 0; @@ -606,10 +605,8 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info, qrecord->num_bytes = num_bytes; qrecord->old_roots = NULL; - qexisting = btrfs_qgroup_insert_dirty_extent(fs_info, - delayed_refs, - qrecord); - if (qexisting) + if(btrfs_qgroup_insert_dirty_extent_nolock(fs_info, + delayed_refs, qrecord)) kfree(qrecord); } diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d2cc16ff3e21..d10872748c43 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -8521,35 +8521,6 @@ reada: wc->reada_slot = slot; } -/* - * These may not be seen by the usual inc/dec ref code so we have to - * add them here. - */ -static int record_one_subtree_extent(struct btrfs_trans_handle *trans, - struct btrfs_root *root, u64 bytenr, - u64 num_bytes) -{ - struct btrfs_qgroup_extent_record *qrecord; - struct btrfs_delayed_ref_root *delayed_refs; - - qrecord = kmalloc(sizeof(*qrecord), GFP_NOFS); - if (!qrecord) - return -ENOMEM; - - qrecord->bytenr = bytenr; - qrecord->num_bytes = num_bytes; - qrecord->old_roots = NULL; - - delayed_refs = &trans->transaction->delayed_refs; - spin_lock(&delayed_refs->lock); - if (btrfs_qgroup_insert_dirty_extent(trans->fs_info, - delayed_refs, qrecord)) - kfree(qrecord); - spin_unlock(&delayed_refs->lock); - - return 0; -} - static int account_leaf_items(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *eb) @@ -8583,7 +8554,8 @@ static int account_leaf_items(struct btrfs_trans_handle *trans, num_bytes = btrfs_file_extent_disk_num_bytes(eb, fi); - ret = record_one_subtree_extent(trans, root, bytenr, num_bytes); + ret = btrfs_qgroup_insert_dirty_extent(trans, root->fs_info, + bytenr, num_bytes, GFP_NOFS); if (ret) return ret; } @@ -8732,8 +8704,9 @@ walk_down: btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK); path->locks[level] = BTRFS_READ_LOCK_BLOCKING; - ret = record_one_subtree_extent(trans, root, child_bytenr, - root->nodesize); + ret = btrfs_qgroup_insert_dirty_extent(trans, + root->fs_info, child_bytenr, + root->nodesize, GFP_NOFS); if (ret) goto out; } diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index 13eb6a7a4db1..8db2e29fdcf4 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -1453,10 +1453,9 @@ int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans, return ret; } -struct btrfs_qgroup_extent_record * -btrfs_qgroup_insert_dirty_extent(struct btrfs_fs_info *fs_info, - struct btrfs_delayed_ref_root *delayed_refs, - struct btrfs_qgroup_extent_record *record) +int btrfs_qgroup_insert_dirty_extent_nolock(struct btrfs_fs_info *fs_info, + struct btrfs_delayed_ref_root *delayed_refs, + struct btrfs_qgroup_extent_record *record) { struct rb_node **p = &delayed_refs->dirty_extent_root.rb_node; struct rb_node *parent_node = NULL; @@ -1475,12 +1474,42 @@ btrfs_qgroup_insert_dirty_extent(struct btrfs_fs_info *fs_info, else if (bytenr > entry->bytenr) p = &(*p)->rb_right; else - return entry; + return 1; } rb_link_node(&record->node, parent_node, p); rb_insert_color(&record->node, &delayed_refs->dirty_extent_root); - return NULL; + return 0; +} + +int btrfs_qgroup_insert_dirty_extent(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 bytenr, u64 num_bytes, + gfp_t gfp_flag) +{ + struct btrfs_qgroup_extent_record *record; + struct btrfs_delayed_ref_root *delayed_refs; + int ret; + + if (!fs_info->quota_enabled || bytenr == 0 || num_bytes == 0) + return 0; + if (WARN_ON(trans == NULL)) + return -EINVAL; + record = kmalloc(sizeof(*record), gfp_flag); + if (!record) + return -ENOMEM; + + delayed_refs = &trans->transaction->delayed_refs; + record->bytenr = bytenr; + record->num_bytes = num_bytes; + record->old_roots = NULL; + + spin_lock(&delayed_refs->lock); + ret = btrfs_qgroup_insert_dirty_extent_nolock(fs_info, delayed_refs, + record); + spin_unlock(&delayed_refs->lock); + if (ret > 0) + kfree(record); + return 0; } #define UPDATE_NEW 0 diff --git a/fs/btrfs/qgroup.h b/fs/btrfs/qgroup.h index af3e5578cad7..1bc64c864b62 100644 --- a/fs/btrfs/qgroup.h +++ b/fs/btrfs/qgroup.h @@ -64,10 +64,35 @@ void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info); struct btrfs_delayed_extent_op; int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info); -struct btrfs_qgroup_extent_record * -btrfs_qgroup_insert_dirty_extent(struct btrfs_fs_info *fs_info, - struct btrfs_delayed_ref_root *delayed_refs, - struct btrfs_qgroup_extent_record *record); +/* + * Insert one dirty extent record into @delayed_refs, informing qgroup to + * account that extent at commit trans time. + * + * No lock version, caller must acquire delayed ref lock and allocate memory. + * + * Return 0 for success insert + * Return >0 for existing record, caller can free @record safely. + * Error is not possible + */ +int btrfs_qgroup_insert_dirty_extent_nolock( + struct btrfs_fs_info *fs_info, + struct btrfs_delayed_ref_root *delayed_refs, + struct btrfs_qgroup_extent_record *record); + +/* + * Insert one dirty extent record into @delayed_refs, informing qgroup to + * account that extent at commit trans time. + * + * Better encapsulated version. + * + * Return 0 if the operation is done. + * Return <0 for error, like memory allocation failure or invalid parameter + * (NULL trans) + */ +int btrfs_qgroup_insert_dirty_extent(struct btrfs_trans_handle *trans, + struct btrfs_fs_info *fs_info, u64 bytenr, u64 num_bytes, + gfp_t gfp_flag); + int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info, -- cgit v1.2.3-70-g09d2 From 62b99540a1d91e46422f0e04de50fc723812c421 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Mon, 15 Aug 2016 10:36:51 +0800 Subject: btrfs: relocation: Fix leaking qgroups numbers on data extents This patch fixes a REGRESSION introduced in 4.2, caused by the big quota rework. When balancing data extents, qgroup will leak all its numbers for relocated data extents. The relocation is done in the following steps for data extents: 1) Create data reloc tree and inode 2) Copy all data extents to data reloc tree And commit transaction 3) Create tree reloc tree(special snapshot) for any related subvolumes 4) Replace file extent in tree reloc tree with new extents in data reloc tree And commit transaction 5) Merge tree reloc tree with original fs, by swapping tree blocks For 1)~4), since tree reloc tree and data reloc tree doesn't count to qgroup, everything is OK. But for 5), the swapping of tree blocks will only info qgroup to track metadata extents. If metadata extents contain file extents, qgroup number for file extents will get lost, leading to corrupted qgroup accounting. The fix is, before commit transaction of step 5), manually info qgroup to track all file extents in data reloc tree. Since at commit transaction time, the tree swapping is done, and qgroup will account these data extents correctly. Cc: Mark Fasheh Reported-by: Mark Fasheh Reported-by: Filipe Manana Signed-off-by: Qu Wenruo Tested-by: Goldwyn Rodrigues Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/relocation.c | 109 +++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 103 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index b26a5aea41b4..27480ef9813c 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -31,6 +31,7 @@ #include "async-thread.h" #include "free-space-cache.h" #include "inode-map.h" +#include "qgroup.h" /* * backref_node, mapping_node and tree_block start with this @@ -3916,6 +3917,90 @@ int prepare_to_relocate(struct reloc_control *rc) return 0; } +/* + * Qgroup fixer for data chunk relocation. + * The data relocation is done in the following steps + * 1) Copy data extents into data reloc tree + * 2) Create tree reloc tree(special snapshot) for related subvolumes + * 3) Modify file extents in tree reloc tree + * 4) Merge tree reloc tree with original fs tree, by swapping tree blocks + * + * The problem is, data and tree reloc tree are not accounted to qgroup, + * and 4) will only info qgroup to track tree blocks change, not file extents + * in the tree blocks. + * + * The good news is, related data extents are all in data reloc tree, so we + * only need to info qgroup to track all file extents in data reloc tree + * before commit trans. + */ +static int qgroup_fix_relocated_data_extents(struct btrfs_trans_handle *trans, + struct reloc_control *rc) +{ + struct btrfs_fs_info *fs_info = rc->extent_root->fs_info; + struct inode *inode = rc->data_inode; + struct btrfs_root *data_reloc_root = BTRFS_I(inode)->root; + struct btrfs_path *path; + struct btrfs_key key; + int ret = 0; + + if (!fs_info->quota_enabled) + return 0; + + /* + * Only for stage where we update data pointers the qgroup fix is + * valid. + * For MOVING_DATA stage, we will miss the timing of swapping tree + * blocks, and won't fix it. + */ + if (!(rc->stage == UPDATE_DATA_PTRS && rc->extents_found)) + return 0; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + key.objectid = btrfs_ino(inode); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = 0; + + ret = btrfs_search_slot(NULL, data_reloc_root, &key, path, 0, 0); + if (ret < 0) + goto out; + + lock_extent(&BTRFS_I(inode)->io_tree, 0, (u64)-1); + while (1) { + struct btrfs_file_extent_item *fi; + + btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); + if (key.objectid > btrfs_ino(inode)) + break; + if (key.type != BTRFS_EXTENT_DATA_KEY) + goto next; + fi = btrfs_item_ptr(path->nodes[0], path->slots[0], + struct btrfs_file_extent_item); + if (btrfs_file_extent_type(path->nodes[0], fi) != + BTRFS_FILE_EXTENT_REG) + goto next; + ret = btrfs_qgroup_insert_dirty_extent(trans, fs_info, + btrfs_file_extent_disk_bytenr(path->nodes[0], fi), + btrfs_file_extent_disk_num_bytes(path->nodes[0], fi), + GFP_NOFS); + if (ret < 0) + break; +next: + ret = btrfs_next_item(data_reloc_root, path); + if (ret < 0) + break; + if (ret > 0) { + ret = 0; + break; + } + } + unlock_extent(&BTRFS_I(inode)->io_tree, 0 , (u64)-1); +out: + btrfs_free_path(path); + return ret; +} + static noinline_for_stack int relocate_block_group(struct reloc_control *rc) { struct rb_root blocks = RB_ROOT; @@ -4102,10 +4187,16 @@ restart: /* get rid of pinned extents */ trans = btrfs_join_transaction(rc->extent_root); - if (IS_ERR(trans)) + if (IS_ERR(trans)) { err = PTR_ERR(trans); - else - btrfs_commit_transaction(trans, rc->extent_root); + goto out_free; + } + err = qgroup_fix_relocated_data_extents(trans, rc); + if (err < 0) { + btrfs_abort_transaction(trans, err); + goto out_free; + } + btrfs_commit_transaction(trans, rc->extent_root); out_free: btrfs_free_block_rsv(rc->extent_root, rc->block_rsv); btrfs_free_path(path); @@ -4468,10 +4559,16 @@ int btrfs_recover_relocation(struct btrfs_root *root) unset_reloc_control(rc); trans = btrfs_join_transaction(rc->extent_root); - if (IS_ERR(trans)) + if (IS_ERR(trans)) { err = PTR_ERR(trans); - else - err = btrfs_commit_transaction(trans, rc->extent_root); + goto out_free; + } + err = qgroup_fix_relocated_data_extents(trans, rc); + if (err < 0) { + btrfs_abort_transaction(trans, err); + goto out_free; + } + err = btrfs_commit_transaction(trans, rc->extent_root); out_free: kfree(rc); out: -- cgit v1.2.3-70-g09d2 From df2c95f33e0a28b22509e4ee85365eedf32a1056 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Mon, 15 Aug 2016 10:36:52 +0800 Subject: btrfs: qgroup: Fix qgroup incorrectness caused by log replay When doing log replay at mount time(after power loss), qgroup will leak numbers of replayed data extents. The cause is almost the same of balance. So fix it by manually informing qgroup for owner changed extents. The bug can be detected by btrfs/119 test case. Cc: Mark Fasheh Signed-off-by: Qu Wenruo Reviewed-and-Tested-by: Goldwyn Rodrigues Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/tree-log.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index fff3f3efa436..ffe92da81b8a 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -27,6 +27,7 @@ #include "backref.h" #include "hash.h" #include "compression.h" +#include "qgroup.h" /* magic values for the inode_only field in btrfs_log_inode: * @@ -680,6 +681,21 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans, ins.type = BTRFS_EXTENT_ITEM_KEY; offset = key->offset - btrfs_file_extent_offset(eb, item); + /* + * Manually record dirty extent, as here we did a shallow + * file extent item copy and skip normal backref update, + * but modifying extent tree all by ourselves. + * So need to manually record dirty extent for qgroup, + * as the owner of the file extent changed from log tree + * (doesn't affect qgroup) to fs/file tree(affects qgroup) + */ + ret = btrfs_qgroup_insert_dirty_extent(trans, root->fs_info, + btrfs_file_extent_disk_bytenr(eb, item), + btrfs_file_extent_disk_num_bytes(eb, item), + GFP_NOFS); + if (ret < 0) + goto out; + if (ins.objectid > 0) { u64 csum_start; u64 csum_end; -- cgit v1.2.3-70-g09d2 From dcb40c196fc85c6dfb28456480e5a882e26f567d Mon Sep 17 00:00:00 2001 From: Wang Xiaoguang Date: Mon, 25 Jul 2016 15:51:38 +0800 Subject: btrfs: use correct offset for reloc_inode in prealloc_file_extent_cluster() In prealloc_file_extent_cluster(), btrfs_check_data_free_space() uses wrong file offset for reloc_inode, it uses cluster->start and cluster->end, which indeed are extent's bytenr. The correct value should be cluster->[start|end] minus block group's start bytenr. start bytenr cluster->start | | extent | extent | ...| extent | |----------------------------------------------------------------| | block group reloc_inode | Signed-off-by: Wang Xiaoguang Reviewed-by: Josef Bacik Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/relocation.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 27480ef9813c..71b4b70f56b9 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -3038,12 +3038,14 @@ int prealloc_file_extent_cluster(struct inode *inode, u64 num_bytes; int nr = 0; int ret = 0; + u64 prealloc_start = cluster->start - offset; + u64 prealloc_end = cluster->end - offset; BUG_ON(cluster->start != cluster->boundary[0]); inode_lock(inode); - ret = btrfs_check_data_free_space(inode, cluster->start, - cluster->end + 1 - cluster->start); + ret = btrfs_check_data_free_space(inode, prealloc_start, + prealloc_end + 1 - prealloc_start); if (ret) goto out; @@ -3064,8 +3066,8 @@ int prealloc_file_extent_cluster(struct inode *inode, break; nr++; } - btrfs_free_reserved_data_space(inode, cluster->start, - cluster->end + 1 - cluster->start); + btrfs_free_reserved_data_space(inode, prealloc_start, + prealloc_end + 1 - prealloc_start); out: inode_unlock(inode); return ret; -- cgit v1.2.3-70-g09d2 From 4824f1f412f75e9f84b9cecbde828e8f4699f82d Mon Sep 17 00:00:00 2001 From: Wang Xiaoguang Date: Mon, 25 Jul 2016 15:51:39 +0800 Subject: btrfs: divide btrfs_update_reserved_bytes() into two functions MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This patch divides btrfs_update_reserved_bytes() into btrfs_add_reserved_bytes() and btrfs_free_reserved_bytes(), and next patch will extend btrfs_add_reserved_bytes（）to fix some false ENOSPC error, please see later patch for detailed info. Signed-off-by: Wang Xiaoguang Reviewed-by: Josef Bacik Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 97 +++++++++++++++++++++++++++++--------------------- 1 file changed, 57 insertions(+), 40 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d10872748c43..f1121db65878 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -104,9 +104,10 @@ static int find_next_key(struct btrfs_path *path, int level, struct btrfs_key *key); static void dump_space_info(struct btrfs_space_info *info, u64 bytes, int dump_block_groups); -static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, - int delalloc); +static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache, + u64 num_bytes, int reserve, int delalloc); +static int btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache, + u64 num_bytes, int delalloc); static int block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv, u64 num_bytes); int btrfs_pin_extent(struct btrfs_root *root, @@ -6497,19 +6498,14 @@ void btrfs_wait_block_group_reservations(struct btrfs_block_group_cache *bg) } /** - * btrfs_update_reserved_bytes - update the block_group and space info counters + * btrfs_add_reserved_bytes - update the block_group and space info counters * @cache: The cache we are manipulating * @num_bytes: The number of bytes in question * @reserve: One of the reservation enums * @delalloc: The blocks are allocated for the delalloc write * - * This is called by the allocator when it reserves space, or by somebody who is - * freeing space that was never actually used on disk. For example if you - * reserve some space for a new leaf in transaction A and before transaction A - * commits you free that leaf, you call this with reserve set to 0 in order to - * clear the reservation. - * - * Metadata reservations should be called with RESERVE_ALLOC so we do the proper + * This is called by the allocator when it reserves space. Metadata + * reservations should be called with RESERVE_ALLOC so we do the proper * ENOSPC accounting. For data we handle the reservation through clearing the * delalloc bits in the io_tree. We have to do this since we could end up * allocating less disk space for the amount of data we have reserved in the @@ -6519,44 +6515,65 @@ void btrfs_wait_block_group_reservations(struct btrfs_block_group_cache *bg) * make the reservation and return -EAGAIN, otherwise this function always * succeeds. */ -static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, int delalloc) +static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache, + u64 num_bytes, int reserve, int delalloc) { struct btrfs_space_info *space_info = cache->space_info; int ret = 0; spin_lock(&space_info->lock); spin_lock(&cache->lock); - if (reserve != RESERVE_FREE) { - if (cache->ro) { - ret = -EAGAIN; - } else { - cache->reserved += num_bytes; - space_info->bytes_reserved += num_bytes; - if (reserve == RESERVE_ALLOC) { - trace_btrfs_space_reservation(cache->fs_info, - "space_info", space_info->flags, - num_bytes, 0); - space_info->bytes_may_use -= num_bytes; - } - - if (delalloc) - cache->delalloc_bytes += num_bytes; - } + if (cache->ro) { + ret = -EAGAIN; } else { - if (cache->ro) - space_info->bytes_readonly += num_bytes; - cache->reserved -= num_bytes; - space_info->bytes_reserved -= num_bytes; + cache->reserved += num_bytes; + space_info->bytes_reserved += num_bytes; + if (reserve == RESERVE_ALLOC) { + trace_btrfs_space_reservation(cache->fs_info, + "space_info", space_info->flags, + num_bytes, 0); + space_info->bytes_may_use -= num_bytes; + } if (delalloc) - cache->delalloc_bytes -= num_bytes; + cache->delalloc_bytes += num_bytes; } spin_unlock(&cache->lock); spin_unlock(&space_info->lock); return ret; } +/** + * btrfs_free_reserved_bytes - update the block_group and space info counters + * @cache: The cache we are manipulating + * @num_bytes: The number of bytes in question + * @delalloc: The blocks are allocated for the delalloc write + * + * This is called by somebody who is freeing space that was never actually used + * on disk. For example if you reserve some space for a new leaf in transaction + * A and before transaction A commits you free that leaf, you call this with + * reserve set to 0 in order to clear the reservation. + */ + +static int btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache, + u64 num_bytes, int delalloc) +{ + struct btrfs_space_info *space_info = cache->space_info; + int ret = 0; + + spin_lock(&space_info->lock); + spin_lock(&cache->lock); + if (cache->ro) + space_info->bytes_readonly += num_bytes; + cache->reserved -= num_bytes; + space_info->bytes_reserved -= num_bytes; + + if (delalloc) + cache->delalloc_bytes -= num_bytes; + spin_unlock(&cache->lock); + spin_unlock(&space_info->lock); + return ret; +} void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans, struct btrfs_root *root) { @@ -7191,7 +7208,7 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans, WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)); btrfs_add_free_space(cache, buf->start, buf->len); - btrfs_update_reserved_bytes(cache, buf->len, RESERVE_FREE, 0); + btrfs_free_reserved_bytes(cache, buf->len, 0); btrfs_put_block_group(cache); trace_btrfs_reserved_extent_free(root, buf->start, buf->len); pin = 0; @@ -7763,8 +7780,8 @@ checks: search_start - offset); BUG_ON(offset > search_start); - ret = btrfs_update_reserved_bytes(block_group, num_bytes, - alloc_type, delalloc); + ret = btrfs_add_reserved_bytes(block_group, num_bytes, + alloc_type, delalloc); if (ret == -EAGAIN) { btrfs_add_free_space(block_group, offset, num_bytes); goto loop; @@ -7995,7 +8012,7 @@ static int __btrfs_free_reserved_extent(struct btrfs_root *root, if (btrfs_test_opt(root->fs_info, DISCARD)) ret = btrfs_discard_extent(root, start, len, NULL); btrfs_add_free_space(cache, start, len); - btrfs_update_reserved_bytes(cache, len, RESERVE_FREE, delalloc); + btrfs_free_reserved_bytes(cache, len, delalloc); trace_btrfs_reserved_extent_free(root, start, len); } @@ -8223,8 +8240,8 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, if (!block_group) return -EINVAL; - ret = btrfs_update_reserved_bytes(block_group, ins->offset, - RESERVE_ALLOC_NO_ACCOUNT, 0); + ret = btrfs_add_reserved_bytes(block_group, ins->offset, + RESERVE_ALLOC_NO_ACCOUNT, 0); BUG_ON(ret); /* logic error */ ret = alloc_reserved_file_extent(trans, root, 0, root_objectid, 0, owner, offset, ins, 1); -- cgit v1.2.3-70-g09d2 From 18513091af9483ba84328d42092bd4d42a3c958f Mon Sep 17 00:00:00 2001 From: Wang Xiaoguang Date: Mon, 25 Jul 2016 15:51:40 +0800 Subject: btrfs: update btrfs_space_info's bytes_may_use timely This patch can fix some false ENOSPC errors, below test script can reproduce one false ENOSPC error: #!/bin/bash dd if=/dev/zero of=fs.img bs=$((1024*1024)) count=128 dev=$(losetup --show -f fs.img) mkfs.btrfs -f -M $dev mkdir /tmp/mntpoint mount $dev /tmp/mntpoint cd /tmp/mntpoint xfs_io -f -c "falloc 0 $((64*1024*1024))" testfile Above script will fail for ENOSPC reason, but indeed fs still has free space to satisfy this request. Please see call graph: btrfs_fallocate() |-> btrfs_alloc_data_chunk_ondemand() | bytes_may_use += 64M |-> btrfs_prealloc_file_range() |-> btrfs_reserve_extent() |-> btrfs_add_reserved_bytes() | alloc_type is RESERVE_ALLOC_NO_ACCOUNT, so it does not | change bytes_may_use, and bytes_reserved += 64M. Now | bytes_may_use + bytes_reserved == 128M, which is greater | than btrfs_space_info's total_bytes, false enospc occurs. | Note, the bytes_may_use decrease operation will be done in | end of btrfs_fallocate(), which is too late. Here is another simple case for buffered write: CPU 1 | CPU 2 | |-> cow_file_range() |-> __btrfs_buffered_write() |-> btrfs_reserve_extent() | | | | | | | | | ..... | |-> btrfs_check_data_free_space() | | | | |-> extent_clear_unlock_delalloc() | In CPU 1, btrfs_reserve_extent()->find_free_extent()-> btrfs_add_reserved_bytes() do not decrease bytes_may_use, the decrease operation will be delayed to be done in extent_clear_unlock_delalloc(). Assume in this case, btrfs_reserve_extent() reserved 128MB data, CPU2's btrfs_check_data_free_space() tries to reserve 100MB data space. If 100MB > data_sinfo->total_bytes - data_sinfo->bytes_used - data_sinfo->bytes_reserved - data_sinfo->bytes_pinned - data_sinfo->bytes_readonly - data_sinfo->bytes_may_use btrfs_check_data_free_space() will try to allcate new data chunk or call btrfs_start_delalloc_roots(), or commit current transaction in order to reserve some free space, obviously a lot of work. But indeed it's not necessary as long as decreasing bytes_may_use timely, we still have free space, decreasing 128M from bytes_may_use. To fix this issue, this patch chooses to update bytes_may_use for both data and metadata in btrfs_add_reserved_bytes(). For compress path, real extent length may not be equal to file content length, so introduce a ram_bytes argument for btrfs_reserve_extent(), find_free_extent() and btrfs_add_reserved_bytes(), it's becasue bytes_may_use is increased by file content length. Then compress path can update bytes_may_use correctly. Also now we can discard RESERVE_ALLOC_NO_ACCOUNT, RESERVE_ALLOC and RESERVE_FREE. As we know, usually EXTENT_DO_ACCOUNTING is used for error path. In run_delalloc_nocow(), for inode marked as NODATACOW or extent marked as PREALLOC, we also need to update bytes_may_use, but can not pass EXTENT_DO_ACCOUNTING, because it also clears metadata reservation, so here we introduce EXTENT_CLEAR_DATA_RESV flag to indicate btrfs_clear_bit_hook() to update btrfs_space_info's bytes_may_use. Meanwhile __btrfs_prealloc_file_range() will call btrfs_free_reserved_data_space() internally for both sucessful and failed path, btrfs_prealloc_file_range()'s callers does not need to call btrfs_free_reserved_data_space() any more. Signed-off-by: Wang Xiaoguang Reviewed-by: Josef Bacik Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 2 +- fs/btrfs/extent-tree.c | 56 +++++++++++++++++--------------------------------- fs/btrfs/extent_io.h | 1 + fs/btrfs/file.c | 26 +++++++++++++---------- fs/btrfs/inode-map.c | 3 +-- fs/btrfs/inode.c | 37 ++++++++++++++++++++++++--------- fs/btrfs/relocation.c | 11 ++++++++-- 7 files changed, 73 insertions(+), 63 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index acc6850a118f..09cdff0d58e8 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2579,7 +2579,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 root_objectid, u64 owner, u64 offset, struct btrfs_key *ins); -int btrfs_reserve_extent(struct btrfs_root *root, u64 num_bytes, +int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes, u64 num_bytes, u64 min_alloc_size, u64 empty_size, u64 hint_byte, struct btrfs_key *ins, int is_data, int delalloc); int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root, diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index f1121db65878..133c93b0a0f1 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -60,21 +60,6 @@ enum { CHUNK_ALLOC_FORCE = 2, }; -/* - * Control how reservations are dealt with. - * - * RESERVE_FREE - freeing a reservation. - * RESERVE_ALLOC - allocating space and we need to update bytes_may_use for - * ENOSPC accounting - * RESERVE_ALLOC_NO_ACCOUNT - allocating space and we should not update - * bytes_may_use as the ENOSPC accounting is done elsewhere - */ -enum { - RESERVE_FREE = 0, - RESERVE_ALLOC = 1, - RESERVE_ALLOC_NO_ACCOUNT = 2, -}; - static int update_block_group(struct btrfs_trans_handle *trans, struct btrfs_root *root, u64 bytenr, u64 num_bytes, int alloc); @@ -105,7 +90,7 @@ static int find_next_key(struct btrfs_path *path, int level, static void dump_space_info(struct btrfs_space_info *info, u64 bytes, int dump_block_groups); static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, int delalloc); + u64 ram_bytes, u64 num_bytes, int delalloc); static int btrfs_free_reserved_bytes(struct btrfs_block_group_cache *cache, u64 num_bytes, int delalloc); static int block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv, @@ -3502,7 +3487,6 @@ again: dcs = BTRFS_DC_SETUP; else if (ret == -ENOSPC) set_bit(BTRFS_TRANS_CACHE_ENOSPC, &trans->transaction->flags); - btrfs_free_reserved_data_space(inode, 0, num_pages); out_put: iput(inode); @@ -6500,8 +6484,9 @@ void btrfs_wait_block_group_reservations(struct btrfs_block_group_cache *bg) /** * btrfs_add_reserved_bytes - update the block_group and space info counters * @cache: The cache we are manipulating + * @ram_bytes: The number of bytes of file content, and will be same to + * @num_bytes except for the compress path. * @num_bytes: The number of bytes in question - * @reserve: One of the reservation enums * @delalloc: The blocks are allocated for the delalloc write * * This is called by the allocator when it reserves space. Metadata @@ -6516,7 +6501,7 @@ void btrfs_wait_block_group_reservations(struct btrfs_block_group_cache *bg) * succeeds. */ static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache, - u64 num_bytes, int reserve, int delalloc) + u64 ram_bytes, u64 num_bytes, int delalloc) { struct btrfs_space_info *space_info = cache->space_info; int ret = 0; @@ -6528,13 +6513,11 @@ static int btrfs_add_reserved_bytes(struct btrfs_block_group_cache *cache, } else { cache->reserved += num_bytes; space_info->bytes_reserved += num_bytes; - if (reserve == RESERVE_ALLOC) { - trace_btrfs_space_reservation(cache->fs_info, - "space_info", space_info->flags, - num_bytes, 0); - space_info->bytes_may_use -= num_bytes; - } + trace_btrfs_space_reservation(cache->fs_info, + "space_info", space_info->flags, + ram_bytes, 0); + space_info->bytes_may_use -= ram_bytes; if (delalloc) cache->delalloc_bytes += num_bytes; } @@ -7433,9 +7416,9 @@ btrfs_release_block_group(struct btrfs_block_group_cache *cache, * the free space extent currently. */ static noinline int find_free_extent(struct btrfs_root *orig_root, - u64 num_bytes, u64 empty_size, - u64 hint_byte, struct btrfs_key *ins, - u64 flags, int delalloc) + u64 ram_bytes, u64 num_bytes, u64 empty_size, + u64 hint_byte, struct btrfs_key *ins, + u64 flags, int delalloc) { int ret = 0; struct btrfs_root *root = orig_root->fs_info->extent_root; @@ -7447,8 +7430,6 @@ static noinline int find_free_extent(struct btrfs_root *orig_root, struct btrfs_space_info *space_info; int loop = 0; int index = __get_raid_index(flags); - int alloc_type = (flags & BTRFS_BLOCK_GROUP_DATA) ? - RESERVE_ALLOC_NO_ACCOUNT : RESERVE_ALLOC; bool failed_cluster_refill = false; bool failed_alloc = false; bool use_cluster = true; @@ -7780,8 +7761,8 @@ checks: search_start - offset); BUG_ON(offset > search_start); - ret = btrfs_add_reserved_bytes(block_group, num_bytes, - alloc_type, delalloc); + ret = btrfs_add_reserved_bytes(block_group, ram_bytes, + num_bytes, delalloc); if (ret == -EAGAIN) { btrfs_add_free_space(block_group, offset, num_bytes); goto loop; @@ -7953,7 +7934,7 @@ again: up_read(&info->groups_sem); } -int btrfs_reserve_extent(struct btrfs_root *root, +int btrfs_reserve_extent(struct btrfs_root *root, u64 ram_bytes, u64 num_bytes, u64 min_alloc_size, u64 empty_size, u64 hint_byte, struct btrfs_key *ins, int is_data, int delalloc) @@ -7965,8 +7946,8 @@ int btrfs_reserve_extent(struct btrfs_root *root, flags = btrfs_get_alloc_profile(root, is_data); again: WARN_ON(num_bytes < root->sectorsize); - ret = find_free_extent(root, num_bytes, empty_size, hint_byte, ins, - flags, delalloc); + ret = find_free_extent(root, ram_bytes, num_bytes, empty_size, + hint_byte, ins, flags, delalloc); if (!ret && !is_data) { btrfs_dec_block_group_reservations(root->fs_info, ins->objectid); @@ -7975,6 +7956,7 @@ again: num_bytes = min(num_bytes >> 1, ins->offset); num_bytes = round_down(num_bytes, root->sectorsize); num_bytes = max(num_bytes, min_alloc_size); + ram_bytes = num_bytes; if (num_bytes == min_alloc_size) final_tried = true; goto again; @@ -8241,7 +8223,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans, return -EINVAL; ret = btrfs_add_reserved_bytes(block_group, ins->offset, - RESERVE_ALLOC_NO_ACCOUNT, 0); + ins->offset, 0); BUG_ON(ret); /* logic error */ ret = alloc_reserved_file_extent(trans, root, 0, root_objectid, 0, owner, offset, ins, 1); @@ -8385,7 +8367,7 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans, if (IS_ERR(block_rsv)) return ERR_CAST(block_rsv); - ret = btrfs_reserve_extent(root, blocksize, blocksize, + ret = btrfs_reserve_extent(root, blocksize, blocksize, blocksize, empty_size, hint, &ins, 0, 0); if (ret) goto out_unuse; diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index c0c1c4fef6ce..b52ca5db01cb 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -20,6 +20,7 @@ #define EXTENT_DAMAGED (1U << 14) #define EXTENT_NORESERVE (1U << 15) #define EXTENT_QGROUP_RESERVED (1U << 16) +#define EXTENT_CLEAR_DATA_RESV (1U << 17) #define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK) #define EXTENT_CTLBITS (EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC) diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 5842423f8f47..3391f2adf0c8 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2675,6 +2675,7 @@ static long btrfs_fallocate(struct file *file, int mode, alloc_start = round_down(offset, blocksize); alloc_end = round_up(offset + len, blocksize); + cur_offset = alloc_start; /* Make sure we aren't being give some crap mode */ if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) @@ -2767,7 +2768,6 @@ static long btrfs_fallocate(struct file *file, int mode, /* First, check if we exceed the qgroup limit */ INIT_LIST_HEAD(&reserve_list); - cur_offset = alloc_start; while (1) { em = btrfs_get_extent(inode, NULL, 0, cur_offset, alloc_end - cur_offset, 0); @@ -2794,6 +2794,14 @@ static long btrfs_fallocate(struct file *file, int mode, last_byte - cur_offset); if (ret < 0) break; + } else { + /* + * Do not need to reserve unwritten extent for this + * range, free reserved data space first, otherwise + * it'll result in false ENOSPC error. + */ + btrfs_free_reserved_data_space(inode, cur_offset, + last_byte - cur_offset); } free_extent_map(em); cur_offset = last_byte; @@ -2811,6 +2819,9 @@ static long btrfs_fallocate(struct file *file, int mode, range->start, range->len, 1 << inode->i_blkbits, offset + len, &alloc_hint); + else + btrfs_free_reserved_data_space(inode, range->start, + range->len); list_del(&range->list); kfree(range); } @@ -2845,18 +2856,11 @@ out_unlock: unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, &cached_state, GFP_KERNEL); out: - /* - * As we waited the extent range, the data_rsv_map must be empty - * in the range, as written data range will be released from it. - * And for prealloacted extent, it will also be released when - * its metadata is written. - * So this is completely used as cleanup. - */ - btrfs_qgroup_free_data(inode, alloc_start, alloc_end - alloc_start); inode_unlock(inode); /* Let go of our reservation. */ - btrfs_free_reserved_data_space(inode, alloc_start, - alloc_end - alloc_start); + if (ret != 0) + btrfs_free_reserved_data_space(inode, alloc_start, + alloc_end - cur_offset); return ret; } diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c index aa6fabaee72e..359ee861b5a4 100644 --- a/fs/btrfs/inode-map.c +++ b/fs/btrfs/inode-map.c @@ -495,10 +495,9 @@ again: ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, prealloc, prealloc, prealloc, &alloc_hint); if (ret) { - btrfs_delalloc_release_space(inode, 0, prealloc); + btrfs_delalloc_release_metadata(inode, prealloc); goto out_put; } - btrfs_free_reserved_data_space(inode, 0, prealloc); ret = btrfs_write_out_ino_cache(root, trans, path, inode); out_put: diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 29636624a427..bcea20015d9a 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -566,6 +566,8 @@ cont: PAGE_SET_WRITEBACK | page_error_op | PAGE_END_WRITEBACK); + btrfs_free_reserved_data_space_noquota(inode, start, + end - start + 1); goto free_pages_out; } } @@ -742,7 +744,7 @@ retry: lock_extent(io_tree, async_extent->start, async_extent->start + async_extent->ram_size - 1); - ret = btrfs_reserve_extent(root, + ret = btrfs_reserve_extent(root, async_extent->ram_size, async_extent->compressed_size, async_extent->compressed_size, 0, alloc_hint, &ins, 1, 1); @@ -969,7 +971,8 @@ static noinline int cow_file_range(struct inode *inode, EXTENT_DEFRAG, PAGE_UNLOCK | PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK); - + btrfs_free_reserved_data_space_noquota(inode, start, + end - start + 1); *nr_written = *nr_written + (end - start + PAGE_SIZE) / PAGE_SIZE; *page_started = 1; @@ -989,7 +992,7 @@ static noinline int cow_file_range(struct inode *inode, unsigned long op; cur_alloc_size = disk_num_bytes; - ret = btrfs_reserve_extent(root, cur_alloc_size, + ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size, root->sectorsize, 0, alloc_hint, &ins, 1, 1); if (ret < 0) @@ -1489,8 +1492,10 @@ out_check: extent_clear_unlock_delalloc(inode, cur_offset, cur_offset + num_bytes - 1, locked_page, EXTENT_LOCKED | - EXTENT_DELALLOC, PAGE_UNLOCK | - PAGE_SET_PRIVATE2); + EXTENT_DELALLOC | + EXTENT_CLEAR_DATA_RESV, + PAGE_UNLOCK | PAGE_SET_PRIVATE2); + if (!nolock && nocow) btrfs_end_write_no_snapshoting(root); cur_offset = extent_end; @@ -1807,7 +1812,9 @@ static void btrfs_clear_bit_hook(struct inode *inode, return; if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID - && do_list && !(state->state & EXTENT_NORESERVE)) + && do_list && !(state->state & EXTENT_NORESERVE) + && (*bits & (EXTENT_DO_ACCOUNTING | + EXTENT_CLEAR_DATA_RESV))) btrfs_free_reserved_data_space_noquota(inode, state->start, len); @@ -7252,7 +7259,7 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode, int ret; alloc_hint = get_extent_allocation_hint(inode, start, len); - ret = btrfs_reserve_extent(root, len, root->sectorsize, 0, + ret = btrfs_reserve_extent(root, len, len, root->sectorsize, 0, alloc_hint, &ins, 1, 1); if (ret) return ERR_PTR(ret); @@ -7752,6 +7759,13 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock, ret = PTR_ERR(em2); goto unlock_err; } + /* + * For inode marked NODATACOW or extent marked PREALLOC, + * use the existing or preallocated extent, so does not + * need to adjust btrfs_space_info's bytes_may_use. + */ + btrfs_free_reserved_data_space_noquota(inode, + start, len); goto unlock; } } @@ -7786,7 +7800,6 @@ unlock: i_size_write(inode, start + len); adjust_dio_outstanding_extents(inode, dio_data, len); - btrfs_free_reserved_data_space(inode, start, len); WARN_ON(dio_data->reserve < len); dio_data->reserve -= len; dio_data->unsubmitted_oe_range_end = start + len; @@ -10306,6 +10319,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, u64 last_alloc = (u64)-1; int ret = 0; bool own_trans = true; + u64 end = start + num_bytes - 1; if (trans) own_trans = false; @@ -10327,8 +10341,8 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode, * sized chunks. */ cur_bytes = min(cur_bytes, last_alloc); - ret = btrfs_reserve_extent(root, cur_bytes, min_size, 0, - *alloc_hint, &ins, 1, 0); + ret = btrfs_reserve_extent(root, cur_bytes, cur_bytes, + min_size, 0, *alloc_hint, &ins, 1, 0); if (ret) { if (own_trans) btrfs_end_transaction(trans, root); @@ -10414,6 +10428,9 @@ next: if (own_trans) btrfs_end_transaction(trans, root); } + if (cur_offset < end) + btrfs_free_reserved_data_space(inode, cur_offset, + end - cur_offset + 1); return ret; } diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 71b4b70f56b9..8a2c2a07987b 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -3040,6 +3040,7 @@ int prealloc_file_extent_cluster(struct inode *inode, int ret = 0; u64 prealloc_start = cluster->start - offset; u64 prealloc_end = cluster->end - offset; + u64 cur_offset; BUG_ON(cluster->start != cluster->boundary[0]); inode_lock(inode); @@ -3049,6 +3050,7 @@ int prealloc_file_extent_cluster(struct inode *inode, if (ret) goto out; + cur_offset = prealloc_start; while (nr < cluster->nr) { start = cluster->boundary[nr] - offset; if (nr + 1 < cluster->nr) @@ -3058,16 +3060,21 @@ int prealloc_file_extent_cluster(struct inode *inode, lock_extent(&BTRFS_I(inode)->io_tree, start, end); num_bytes = end + 1 - start; + if (cur_offset < start) + btrfs_free_reserved_data_space(inode, cur_offset, + start - cur_offset); ret = btrfs_prealloc_file_range(inode, 0, start, num_bytes, num_bytes, end + 1, &alloc_hint); + cur_offset = end + 1; unlock_extent(&BTRFS_I(inode)->io_tree, start, end); if (ret) break; nr++; } - btrfs_free_reserved_data_space(inode, prealloc_start, - prealloc_end + 1 - prealloc_start); + if (cur_offset < prealloc_end) + btrfs_free_reserved_data_space(inode, cur_offset, + prealloc_end + 1 - cur_offset); out: inode_unlock(inode); return ret; -- cgit v1.2.3-70-g09d2 From 9e7cc91a6d18a4973c6d2cc104871439c9e94f3d Mon Sep 17 00:00:00 2001 From: Wang Xiaoguang Date: Mon, 1 Aug 2016 13:28:08 +0800 Subject: btrfs: fix fsfreeze hang caused by delayed iputs deal When running fstests generic/068, sometimes we got below deadlock: xfs_io D ffff8800331dbb20 0 6697 6693 0x00000080 ffff8800331dbb20 ffff88007acfc140 ffff880034d895c0 ffff8800331dc000 ffff880032d243e8 fffffffeffffffff ffff880032d24400 0000000000000001 ffff8800331dbb38 ffffffff816a9045 ffff880034d895c0 ffff8800331dbba8 Call Trace: [] schedule+0x35/0x80 [] rwsem_down_read_failed+0xf2/0x140 [] ? __filemap_fdatawrite_range+0xd1/0x100 [] call_rwsem_down_read_failed+0x18/0x30 [] ? btrfs_alloc_block_rsv+0x2c/0xb0 [btrfs] [] percpu_down_read+0x35/0x50 [] __sb_start_write+0x2c/0x40 [] start_transaction+0x2a5/0x4d0 [btrfs] [] btrfs_join_transaction+0x17/0x20 [btrfs] [] btrfs_evict_inode+0x3c4/0x5d0 [btrfs] [] evict+0xba/0x1a0 [] iput+0x196/0x200 [] btrfs_run_delayed_iputs+0x70/0xc0 [btrfs] [] btrfs_commit_transaction+0x928/0xa80 [btrfs] [] btrfs_freeze+0x30/0x40 [btrfs] [] freeze_super+0xf0/0x190 [] do_vfs_ioctl+0x4a5/0x5c0 [] ? do_audit_syscall_entry+0x66/0x70 [] ? syscall_trace_enter_phase1+0x11f/0x140 [] SyS_ioctl+0x79/0x90 [] do_syscall_64+0x62/0x110 [] entry_SYSCALL64_slow_path+0x25/0x25 >From this warning, freeze_super() already holds SB_FREEZE_FS, but btrfs_freeze() will call btrfs_commit_transaction() again, if btrfs_commit_transaction() finds that it has delayed iputs to handle, it'll start_transaction(), which will try to get SB_FREEZE_FS lock again, then deadlock occurs. The root cause is that in btrfs, sync_filesystem(sb) does not make sure all metadata is updated. There still maybe some codes adding delayed iputs, see below sample race window: CPU1 | CPU2 |-> freeze_super() | |-> sync_filesystem(sb); | | |-> cleaner_kthread() | | |-> btrfs_delete_unused_bgs() | | |-> btrfs_remove_chunk() | | |-> btrfs_remove_block_group() | | |-> btrfs_add_delayed_iput() | | |-> sb->s_writers.frozen = SB_FREEZE_FS; | |-> sb_wait_write(sb, SB_FREEZE_FS); | | acquire SB_FREEZE_FS lock. | | | |-> btrfs_freeze() | |-> btrfs_commit_transaction() | |-> btrfs_run_delayed_iputs() | | will handle delayed iputs, | | that means start_transaction() | | will be called, which will try | | to get SB_FREEZE_FS lock. | To fix this issue, introduce a "int fs_frozen" to record internally whether fs has been frozen. If fs has been frozen, we can not handle delayed iputs. Signed-off-by: Wang Xiaoguang Reviewed-by: David Sterba [ add comment to btrfs_freeze ] Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 2 ++ fs/btrfs/disk-io.c | 1 + fs/btrfs/super.c | 16 ++++++++++++++++ fs/btrfs/transaction.c | 7 ++++++- 4 files changed, 25 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 09cdff0d58e8..ec4154faab61 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1080,6 +1080,8 @@ struct btrfs_fs_info { struct list_head pinned_chunks; int creating_free_space_tree; + /* Used to record internally whether fs has been frozen */ + int fs_frozen; }; struct btrfs_subvolume_writers { diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 7857f64e1cae..17062223fac3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2631,6 +2631,7 @@ int open_ctree(struct super_block *sb, atomic_set(&fs_info->qgroup_op_seq, 0); atomic_set(&fs_info->reada_works_cnt, 0); atomic64_set(&fs_info->tree_mod_seq, 0); + fs_info->fs_frozen = 0; fs_info->sb = sb; fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE; fs_info->metadata_ratio = 0; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 864ce334f696..4071fe2bd098 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -2241,6 +2241,13 @@ static int btrfs_freeze(struct super_block *sb) struct btrfs_trans_handle *trans; struct btrfs_root *root = btrfs_sb(sb)->tree_root; + root->fs_info->fs_frozen = 1; + /* + * We don't need a barrier here, we'll wait for any transaction that + * could be in progress on other threads (and do delayed iputs that + * we want to avoid on a frozen filesystem), or do the commit + * ourselves. + */ trans = btrfs_attach_transaction_barrier(root); if (IS_ERR(trans)) { /* no transaction, don't bother */ @@ -2251,6 +2258,14 @@ static int btrfs_freeze(struct super_block *sb) return btrfs_commit_transaction(trans, root); } +static int btrfs_unfreeze(struct super_block *sb) +{ + struct btrfs_root *root = btrfs_sb(sb)->tree_root; + + root->fs_info->fs_frozen = 0; + return 0; +} + static int btrfs_show_devname(struct seq_file *m, struct dentry *root) { struct btrfs_fs_info *fs_info = btrfs_sb(root->d_sb); @@ -2299,6 +2314,7 @@ static const struct super_operations btrfs_super_ops = { .statfs = btrfs_statfs, .remount_fs = btrfs_remount, .freeze_fs = btrfs_freeze, + .unfreeze_fs = btrfs_unfreeze, }; static const struct file_operations btrfs_ctl_fops = { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 9cca0a721961..95d41919d034 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -2278,8 +2278,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, kmem_cache_free(btrfs_trans_handle_cachep, trans); + /* + * If fs has been frozen, we can not handle delayed iputs, otherwise + * it'll result in deadlock about SB_FREEZE_FS. + */ if (current != root->fs_info->transaction_kthread && - current != root->fs_info->cleaner_kthread) + current != root->fs_info->cleaner_kthread && + !root->fs_info->fs_frozen) btrfs_run_delayed_iputs(root); return ret; -- cgit v1.2.3-70-g09d2 From 28b737f6ede3661fe610937706c4a6f50e9ab769 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Fri, 29 Jul 2016 11:09:50 -0700 Subject: Btrfs: clarify do_chunk_alloc()'s return value Function start_transaction() can return ERR_PTR(1) when flush is BTRFS_RESERVE_FLUSH_LIMIT, so the call graph is start_transaction (return ERR_PTR(1)) -> btrfs_block_rsv_add (return 1) -> reserve_metadata_bytes (return 1) -> flush_space (return 1) -> do_chunk_alloc (return 1) With BTRFS_RESERVE_FLUSH_LIMIT, if flush_space is already on the flush_state of ALLOC_CHUNK and it successfully allocates a new chunk, then instead of trying to reserve space again, reserve_metadata_bytes returns 1 immediately. Eventually the callers who call start_transaction() usually just do the IS_ERR() check which ERR_PTR(1) can pass, then it'll get a panic when dereferencing a pointer which is ERR_PTR(1). The following patch fixes the above problem. "btrfs: flush_space: treat return value of do_chunk_alloc properly" https://patchwork.kernel.org/patch/7778651/ This add comments to clarify do_chunk_alloc()'s return value. Signed-off-by: Liu Bo Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 133c93b0a0f1..d5d3cfb1f66f 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4457,6 +4457,15 @@ void check_system_chunk(struct btrfs_trans_handle *trans, } } +/* + * If force is CHUNK_ALLOC_FORCE: + * - return 1 if it successfully allocates a chunk, + * - return errors including -ENOSPC otherwise. + * If force is NOT CHUNK_ALLOC_FORCE: + * - return 0 if it doesn't need to allocate a new chunk, + * - return 1 if it successfully allocates a chunk, + * - return errors including -ENOSPC otherwise. + */ static int do_chunk_alloc(struct btrfs_trans_handle *trans, struct btrfs_root *extent_root, u64 flags, int force) { -- cgit v1.2.3-70-g09d2 From 142388194191a3edc9ba01cfcfd8b691e0971fb2 Mon Sep 17 00:00:00 2001 From: Anand Jain Date: Fri, 22 Jul 2016 06:04:53 +0800 Subject: btrfs: do not background blkdev_put() At the end of unmount/dev-delete, if the device exclusive open is not actually closed, then there might be a race with another program in the userland who is trying to open the device in exclusive mode and it may fail for eg: unmount /btrfs; fsck /dev/x btrfs dev del /dev/x /btrfs; fsck /dev/x so here background blkdev_put() is not a choice Signed-off-by: Anand Jain Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index e217035d63df..9a6f0dcb3e9c 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -834,10 +834,6 @@ static void __free_device(struct work_struct *work) struct btrfs_device *device; device = container_of(work, struct btrfs_device, rcu_work); - - if (device->bdev) - blkdev_put(device->bdev, device->mode); - rcu_string_free(device->name); kfree(device); } @@ -852,6 +848,17 @@ static void free_device(struct rcu_head *head) schedule_work(&device->rcu_work); } +static void btrfs_close_bdev(struct btrfs_device *device) +{ + if (device->bdev && device->writeable) { + sync_blockdev(device->bdev); + invalidate_bdev(device->bdev); + } + + if (device->bdev) + blkdev_put(device->bdev, device->mode); +} + static void btrfs_close_one_device(struct btrfs_device *device) { struct btrfs_fs_devices *fs_devices = device->fs_devices; @@ -870,10 +877,7 @@ static void btrfs_close_one_device(struct btrfs_device *device) if (device->missing) fs_devices->missing_devices--; - if (device->bdev && device->writeable) { - sync_blockdev(device->bdev); - invalidate_bdev(device->bdev); - } + btrfs_close_bdev(device); new_device = btrfs_alloc_device(NULL, &device->devid, device->uuid); @@ -1932,6 +1936,8 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path, u64 devid) btrfs_sysfs_rm_device_link(root->fs_info->fs_devices, device); } + btrfs_close_bdev(device); + call_rcu(&device->rcu, free_device); num_devices = btrfs_super_num_devices(root->fs_info->super_copy) - 1; @@ -2025,6 +2031,9 @@ void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info, /* zero out the old super if it is writable */ btrfs_scratch_superblocks(srcdev->bdev, srcdev->name->str); } + + btrfs_close_bdev(srcdev); + call_rcu(&srcdev->rcu, free_device); /* @@ -2080,6 +2089,8 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info, * the device_list_mutex lock. */ btrfs_scratch_superblocks(tgtdev->bdev, tgtdev->name->str); + + btrfs_close_bdev(tgtdev); call_rcu(&tgtdev->rcu, free_device); } -- cgit v1.2.3-70-g09d2 From 187ee58c62c1d0d238d3dc4835869d33e1869906 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 18 Aug 2016 15:30:06 -0400 Subject: Btrfs: fix em leak in find_first_block_group We need to call free_extent_map() on the em we look up. Signed-off-by: Josef Bacik Reviewed-by: Omar Sandoval Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d5d3cfb1f66f..60d4ae7ce974 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -9887,6 +9887,7 @@ static int find_first_block_group(struct btrfs_root *root, } else { ret = 0; } + free_extent_map(em); goto out; } path->slots[0]++; -- cgit v1.2.3-70-g09d2 From 35bbb97fc898aeb874cb7c8b746f091caa359994 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Wed, 17 Aug 2016 21:58:33 -0400 Subject: btrfs: don't create or leak aliased root while cleaning up orphans commit 909c3a22da3 (Btrfs: fix loading of orphan roots leading to BUG_ON) avoids the BUG_ON but can add an aliased root to the dead_roots list or leak the root. Since we've already been loading roots into the radix tree, we should use it before looking the root up on disk. Cc: # 4.5 Signed-off-by: Jeff Mahoney Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 4 ++-- fs/btrfs/disk-io.h | 2 ++ fs/btrfs/root-tree.c | 27 ++++++++++++++++++--------- 3 files changed, 22 insertions(+), 11 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 17062223fac3..edda47162752 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1624,8 +1624,8 @@ fail: return ret; } -static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info, - u64 root_id) +struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info, + u64 root_id) { struct btrfs_root *root; diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index c9d42c92da2b..e0efe45e5282 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -68,6 +68,8 @@ struct extent_buffer *btrfs_find_tree_block(struct btrfs_fs_info *fs_info, struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root, struct btrfs_key *location); int btrfs_init_fs_root(struct btrfs_root *root); +struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info, + u64 root_id); int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root); void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info); diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c index 7fd7e1830cfe..091296062456 100644 --- a/fs/btrfs/root-tree.c +++ b/fs/btrfs/root-tree.c @@ -272,6 +272,23 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root) root_key.objectid = key.offset; key.offset++; + /* + * The root might have been inserted already, as before we look + * for orphan roots, log replay might have happened, which + * triggers a transaction commit and qgroup accounting, which + * in turn reads and inserts fs roots while doing backref + * walking. + */ + root = btrfs_lookup_fs_root(tree_root->fs_info, + root_key.objectid); + if (root) { + WARN_ON(!test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, + &root->state)); + if (btrfs_root_refs(&root->root_item) == 0) + btrfs_add_dead_root(root); + continue; + } + root = btrfs_read_fs_root(tree_root, &root_key); err = PTR_ERR_OR_ZERO(root); if (err && err != -ENOENT) { @@ -310,16 +327,8 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root) set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state); err = btrfs_insert_fs_root(root->fs_info, root); - /* - * The root might have been inserted already, as before we look - * for orphan roots, log replay might have happened, which - * triggers a transaction commit and qgroup accounting, which - * in turn reads and inserts fs roots while doing backref - * walking. - */ - if (err == -EEXIST) - err = 0; if (err) { + BUG_ON(err == -EEXIST); btrfs_free_fs_root(root); break; } -- cgit v1.2.3-70-g09d2 From 053ab70f0604224c7893b43f9d9d5efa283580d6 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 23 Aug 2016 17:37:45 -0700 Subject: Btrfs: check btree node's nritems When btree node (level = 1) has nritems which equals to zero, we can end up with panic due to insert_ptr()'s BUG_ON(slot > nritems); where slot is 1 and nritems is 0, as copy_for_split() calls insert_ptr(.., path->slots[1] + 1, ...); A invalid value results in the whole mess, this adds the check for btree's node nritems so that we stop reading block when when something is wrong. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index edda47162752..474209f50844 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -613,6 +613,19 @@ static noinline int check_leaf(struct btrfs_root *root, return 0; } +static int check_node(struct btrfs_root *root, struct extent_buffer *node) +{ + unsigned long nr = btrfs_header_nritems(node); + + if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root)) { + btrfs_crit(root->fs_info, + "corrupt node: block %llu root %llu nritems %lu", + node->start, root->objectid, nr); + return -EIO; + } + return 0; +} + static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio, u64 phy_offset, struct page *page, u64 start, u64 end, int mirror) @@ -683,6 +696,9 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio, ret = -EIO; } + if (found_level > 0 && check_node(root, eb)) + ret = -EIO; + if (!ret) set_extent_buffer_uptodate(eb); err: -- cgit v1.2.3-70-g09d2 From 1ba98d086fe3a14d6a31f2f66dbab70c45d00f63 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Tue, 23 Aug 2016 15:22:58 -0700 Subject: Btrfs: detect corruption when non-root leaf has zero item Right now we treat leaf which has zero item as a valid one because we could have an empty tree, that is, a root that is also a leaf without any item, however, in the same case but when the leaf is not a root, we can end up with hitting the BUG_ON(1) in btrfs_extend_item() called by setup_inline_extent_backref(). This makes us check the situation as a corruption if leaf is not its own root. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 23 ++++++++++++++++++++++- 1 file changed, 22 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 474209f50844..70e76ad11fbf 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -560,8 +560,29 @@ static noinline int check_leaf(struct btrfs_root *root, u32 nritems = btrfs_header_nritems(leaf); int slot; - if (nritems == 0) + if (nritems == 0) { + struct btrfs_root *check_root; + + key.objectid = btrfs_header_owner(leaf); + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + + check_root = btrfs_get_fs_root(root->fs_info, &key, false); + /* + * The only reason we also check NULL here is that during + * open_ctree() some roots has not yet been set up. + */ + if (!IS_ERR_OR_NULL(check_root)) { + /* if leaf is the root, then it's fine */ + if (leaf->start != + btrfs_root_bytenr(&check_root->root_item)) { + CORRUPT("non-root leaf's nritems is 0", + leaf, root, 0); + return -EIO; + } + } return 0; + } /* Check the 0 item */ if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) != -- cgit v1.2.3-70-g09d2 From 28a235931b56d4e7bdd51f6733daf95f2b269da8 Mon Sep 17 00:00:00 2001 From: Filipe Manana Date: Tue, 23 Aug 2016 21:13:51 +0100 Subject: Btrfs: fix lockdep warning on deadlock against an inode's log mutex Commit 44f714dae50a ("Btrfs: improve performance on fsync against new inode after rename/unlink"), which landed in 4.8-rc2, introduced a possibility for a deadlock due to double locking of an inode's log mutex by the same task, which lockdep reports with: [23045.433975] ============================================= [23045.434748] [ INFO: possible recursive locking detected ] [23045.435426] 4.7.0-rc6-btrfs-next-34+ #1 Not tainted [23045.436044] --------------------------------------------- [23045.436044] xfs_io/3688 is trying to acquire lock: [23045.436044] (&ei->log_mutex){+.+...}, at: [] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] but task is already holding lock: [23045.436044] (&ei->log_mutex){+.+...}, at: [] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] other info that might help us debug this: [23045.436044] Possible unsafe locking scenario: [23045.436044] CPU0 [23045.436044] ---- [23045.436044] lock(&ei->log_mutex); [23045.436044] lock(&ei->log_mutex); [23045.436044] *** DEADLOCK *** [23045.436044] May be due to missing lock nesting notation [23045.436044] 3 locks held by xfs_io/3688: [23045.436044] #0: (&sb->s_type->i_mutex_key#15){+.+...}, at: [] btrfs_sync_file+0x14e/0x425 [btrfs] [23045.436044] #1: (sb_internal#2){.+.+.+}, at: [] __sb_start_write+0x5f/0xb0 [23045.436044] #2: (&ei->log_mutex){+.+...}, at: [] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] stack backtrace: [23045.436044] CPU: 4 PID: 3688 Comm: xfs_io Not tainted 4.7.0-rc6-btrfs-next-34+ #1 [23045.436044] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.1-0-gb3ef39f-prebuilt.qemu-project.org 04/01/2014 [23045.436044] 0000000000000000 ffff88022f5f7860 ffffffff8127074d ffffffff82a54b70 [23045.436044] ffffffff82a54b70 ffff88022f5f7920 ffffffff81092897 ffff880228015d68 [23045.436044] 0000000000000000 ffffffff82a54b70 ffffffff829c3f00 ffff880228015d68 [23045.436044] Call Trace: [23045.436044] [] dump_stack+0x67/0x90 [23045.436044] [] __lock_acquire+0xcbb/0xe4e [23045.436044] [] ? mark_lock+0x24/0x201 [23045.436044] [] ? mark_held_locks+0x5e/0x74 [23045.436044] [] lock_acquire+0x12f/0x1c3 [23045.436044] [] ? lock_acquire+0x12f/0x1c3 [23045.436044] [] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [] mutex_lock_nested+0x77/0x3a7 [23045.436044] [] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [] ? btrfs_release_delayed_node+0xb/0xd [btrfs] [23045.436044] [] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [] ? vprintk_emit+0x453/0x465 [23045.436044] [] btrfs_log_inode+0x66e/0xc95 [btrfs] [23045.436044] [] log_new_dir_dentries+0x26c/0x359 [btrfs] [23045.436044] [] btrfs_log_inode_parent+0x4a6/0x628 [btrfs] [23045.436044] [] btrfs_log_dentry_safe+0x5a/0x75 [btrfs] [23045.436044] [] btrfs_sync_file+0x304/0x425 [btrfs] [23045.436044] [] vfs_fsync_range+0x8c/0x9e [23045.436044] [] vfs_fsync+0x1c/0x1e [23045.436044] [] do_fsync+0x31/0x4a [23045.436044] [] SyS_fsync+0x10/0x14 [23045.436044] [] entry_SYSCALL_64_fastpath+0x18/0xa8 [23045.436044] [] ? trace_hardirqs_off_caller+0x3f/0xaa An example reproducer for this is: $ mkfs.btrfs -f /dev/sdb $ mount /dev/sdb /mnt $ mkdir /mnt/dir $ touch /mnt/dir/foo $ sync $ mv /mnt/dir/foo /mnt/dir/bar $ touch /mnt/dir/foo $ xfs_io -c "fsync" /mnt/dir/bar This is because while logging the inode of file bar we end up logging its parent directory (since its inode has an unlink_trans field matching the current transaction id due to the rename operation), which in turn logs the inodes for all its new dentries, so that the new inode for the new file named foo gets logged which in turn triggered another logging attempt for the inode we are fsync'ing, since that inode had an old name that corresponds to the name of the new inode. So fix this by ensuring that when logging the inode for a new dentry that has a name matching an old name of some other inode, we don't log again the original inode that we are fsync'ing. Fixes: 44f714dae50a ("Btrfs: improve performance on fsync against new inode after rename/unlink") Signed-off-by: Filipe Manana Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/file.c | 2 +- fs/btrfs/tree-log.c | 5 +++-- fs/btrfs/tree-log.h | 5 ++++- 3 files changed, 8 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 3391f2adf0c8..fea31a4a6e36 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -2070,7 +2070,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) } trans->sync = true; - btrfs_init_log_ctx(&ctx); + btrfs_init_log_ctx(&ctx, inode); ret = btrfs_log_dentry_safe(trans, root, dentry, start, end, &ctx); if (ret < 0) { diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index ffe92da81b8a..e935035ac034 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -2823,7 +2823,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, */ mutex_unlock(&root->log_mutex); - btrfs_init_log_ctx(&root_log_ctx); + btrfs_init_log_ctx(&root_log_ctx, NULL); mutex_lock(&log_root_tree->log_mutex); atomic_inc(&log_root_tree->log_batch); @@ -4757,7 +4757,8 @@ again: if (ret < 0) { err = ret; goto out_unlock; - } else if (ret > 0) { + } else if (ret > 0 && ctx && + other_ino != btrfs_ino(ctx->inode)) { struct btrfs_key inode_key; struct inode *other_inode; diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h index a9f1b75d080d..ab858e31ccbc 100644 --- a/fs/btrfs/tree-log.h +++ b/fs/btrfs/tree-log.h @@ -30,15 +30,18 @@ struct btrfs_log_ctx { int log_transid; int io_err; bool log_new_dentries; + struct inode *inode; struct list_head list; }; -static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx) +static inline void btrfs_init_log_ctx(struct btrfs_log_ctx *ctx, + struct inode *inode) { ctx->log_ret = 0; ctx->log_transid = 0; ctx->io_err = 0; ctx->log_new_dentries = false; + ctx->inode = inode; INIT_LIST_HEAD(&ctx->list); } -- cgit v1.2.3-70-g09d2 From 5bb53c0fb8e0fc2e34287d5d0fcadc784de913e1 Mon Sep 17 00:00:00 2001 From: Andrey Ryabinin Date: Tue, 23 Aug 2016 18:55:31 +0300 Subject: fs/block_dev: fix potential NULL ptr deref in freeze_bdev() Calling freeze_bdev() twice on the same block device without mounted filesystem get_super() will return NULL, which will lead to NULL-ptr dereference later in drop_super(). Check get_super() result to fix that. Note, that this is a purely theoretical issue. We have only 3 freeze_bdev() callers. 2 of them are in filesystem code and used on a device with mounted fs. The third one in lock_fs() has protection in upper-layer code against freezing block device the second time without thawing it first. Signed-off-by: Andrey Ryabinin Reviewed-by: Christoph Hellwig Signed-off-by: Jens Axboe --- fs/block_dev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/block_dev.c b/fs/block_dev.c index e17bdbdfe9b1..08ae99343d92 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -249,7 +249,8 @@ struct super_block *freeze_bdev(struct block_device *bdev) * thaw_bdev drops it. */ sb = get_super(bdev); - drop_super(sb); + if (sb) + drop_super(sb); mutex_unlock(&bdev->bd_fsfreeze_mutex); return sb; } -- cgit v1.2.3-70-g09d2 From ed150e1a5cf20c04cf0b2d2c34e498fc1d6519be Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Fri, 26 Aug 2016 15:58:40 +1000 Subject: xfs: don't perform lookups on zero-height btrees If the caller passes in a cursor to a zero-height btree (which is impossible), we never set block to anything but NULL, which causes the later dereference of it to crash. Instead, just return -EFSCORRUPTED. Signed-off-by: Darrick J. Wong Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_btree.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c index b5c213a051cd..33f14067c320 100644 --- a/fs/xfs/libxfs/xfs_btree.c +++ b/fs/xfs/libxfs/xfs_btree.c @@ -1814,6 +1814,10 @@ xfs_btree_lookup( XFS_BTREE_STATS_INC(cur, lookup); + /* No such thing as a zero-level tree. */ + if (cur->bc_nlevels == 0) + return -EFSCORRUPTED; + block = NULL; keyno = 0; -- cgit v1.2.3-70-g09d2 From 738f57c16a2bb527c705641f0fc1c68ff8cba72a Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Fri, 26 Aug 2016 15:59:19 +1000 Subject: xfs: disallow mounting of realtime + rmap filesystems Since the kernel doesn't currently support the realtime rmapbt, don't allow such filesystems to be mounted. Support will appear in a future release. Signed-off-by: Darrick J. Wong Reviewed-by: Carlos Maiolino Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/xfs_super.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c index 24ef83ef04de..fd6be45b3a1e 100644 --- a/fs/xfs/xfs_super.c +++ b/fs/xfs/xfs_super.c @@ -1574,9 +1574,16 @@ xfs_fs_fill_super( } } - if (xfs_sb_version_hasrmapbt(&mp->m_sb)) + if (xfs_sb_version_hasrmapbt(&mp->m_sb)) { + if (mp->m_sb.sb_rblocks) { + xfs_alert(mp, + "EXPERIMENTAL reverse mapping btree not compatible with realtime device!"); + error = -EINVAL; + goto out_filestream_unmount; + } xfs_alert(mp, "EXPERIMENTAL reverse mapping btree feature enabled. Use at your own risk!"); + } error = xfs_mountfs(mp); if (error) -- cgit v1.2.3-70-g09d2 From da1f039d6947b1a49f13b39a6de0df2a3e9e1ed1 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Fri, 26 Aug 2016 15:59:31 +1000 Subject: xfs: don't log the entire end of the AGF When we're logging the last non-spare field in the AGF, we don't need to log the spare fields, so plumb in a new AGF logging flag to help us avoid that. Signed-off-by: Darrick J. Wong Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_alloc.c | 2 ++ fs/xfs/libxfs/xfs_format.h | 6 ++++-- 2 files changed, 6 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c index 3dd8f1d54498..05b5243d89f6 100644 --- a/fs/xfs/libxfs/xfs_alloc.c +++ b/fs/xfs/libxfs/xfs_alloc.c @@ -2278,6 +2278,8 @@ xfs_alloc_log_agf( offsetof(xfs_agf_t, agf_btreeblks), offsetof(xfs_agf_t, agf_uuid), offsetof(xfs_agf_t, agf_rmap_blocks), + /* needed so that we don't log the whole rest of the structure: */ + offsetof(xfs_agf_t, agf_spare64), sizeof(xfs_agf_t) }; diff --git a/fs/xfs/libxfs/xfs_format.h b/fs/xfs/libxfs/xfs_format.h index e6a8bea0f7ba..270fb5cf4fa1 100644 --- a/fs/xfs/libxfs/xfs_format.h +++ b/fs/xfs/libxfs/xfs_format.h @@ -674,7 +674,8 @@ typedef struct xfs_agf { #define XFS_AGF_BTREEBLKS 0x00000800 #define XFS_AGF_UUID 0x00001000 #define XFS_AGF_RMAP_BLOCKS 0x00002000 -#define XFS_AGF_NUM_BITS 14 +#define XFS_AGF_SPARE64 0x00004000 +#define XFS_AGF_NUM_BITS 15 #define XFS_AGF_ALL_BITS ((1 << XFS_AGF_NUM_BITS) - 1) #define XFS_AGF_FLAGS \ @@ -691,7 +692,8 @@ typedef struct xfs_agf { { XFS_AGF_LONGEST, "LONGEST" }, \ { XFS_AGF_BTREEBLKS, "BTREEBLKS" }, \ { XFS_AGF_UUID, "UUID" }, \ - { XFS_AGF_RMAP_BLOCKS, "RMAP_BLOCKS" } + { XFS_AGF_RMAP_BLOCKS, "RMAP_BLOCKS" }, \ + { XFS_AGF_SPARE64, "SPARE64" } /* disk block (xfs_daddr_t) in the AG */ #define XFS_AGF_DADDR(mp) ((xfs_daddr_t)(1 << (mp)->m_sectbb_log)) -- cgit v1.2.3-70-g09d2 From 722278997bc964349e23e7061d541f8df3133a04 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Fri, 26 Aug 2016 15:59:50 +1000 Subject: xfs: fix some key handling problems in _btree_simple_query_range We only need the record's high key for the first record that we look at; for all records, we /definitely/ need the regular record key. Therefore, fix how the simple range query function gets its keys. Signed-off-by: Darrick J. Wong Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_btree.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c index 33f14067c320..b70d9f918156 100644 --- a/fs/xfs/libxfs/xfs_btree.c +++ b/fs/xfs/libxfs/xfs_btree.c @@ -4563,10 +4563,10 @@ xfs_btree_simple_query_range( error = xfs_btree_get_rec(cur, &recp, &stat); if (error || !stat) break; - cur->bc_ops->init_high_key_from_rec(&rec_key, recp); /* Skip if high_key(rec) < low_key. */ if (firstrec) { + cur->bc_ops->init_high_key_from_rec(&rec_key, recp); firstrec = false; diff = cur->bc_ops->diff_two_keys(cur, low_key, &rec_key); @@ -4575,6 +4575,7 @@ xfs_btree_simple_query_range( } /* Stop if high_key < low_key(rec). */ + cur->bc_ops->init_key_from_rec(&rec_key, recp); diff = cur->bc_ops->diff_two_keys(cur, &rec_key, high_key); if (diff > 0) break; -- cgit v1.2.3-70-g09d2 From 5b5c2dbd3c9bcfa89fba9709c12ecc0a445c6e40 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Fri, 26 Aug 2016 16:00:10 +1000 Subject: xfs: simple btree query range should look right if LE lookup fails If the initial LOOKUP_LE in the simple query range fails to find anything, we should attempt to increment the btree cursor to see if there actually /are/ records for what we're trying to find. Without this patch, a bnobt range query of (0, $agsize) returns no results because the leftmost record never has a startblock of zero. Signed-off-by: Darrick J. Wong Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_btree.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c index b70d9f918156..08569792fe20 100644 --- a/fs/xfs/libxfs/xfs_btree.c +++ b/fs/xfs/libxfs/xfs_btree.c @@ -4558,6 +4558,13 @@ xfs_btree_simple_query_range( if (error) goto out; + /* Nothing? See if there's anything to the right. */ + if (!stat) { + error = xfs_btree_increment(cur, 0, &stat); + if (error) + goto out; + } + while (stat) { /* Find the record. */ error = xfs_btree_get_rec(cur, &recp, &stat); -- cgit v1.2.3-70-g09d2 From f3d7ebdeb2c297bd26272384e955033493ca291c Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Fri, 26 Aug 2016 16:01:30 +1000 Subject: xfs: fix superblock inprogress check From inspection, the superblock sb_inprogress check is done in the verifier and triggered only for the primary superblock via a "bp->b_bn == XFS_SB_DADDR" check. Unfortunately, the primary superblock is an uncached buffer, and hence it is configured by xfs_buf_read_uncached() with: bp->b_bn = XFS_BUF_DADDR_NULL; /* always null for uncached buffers */ And so this check never triggers. Fix it. cc: Signed-off-by: Dave Chinner Reviewed-by: Brian Foster Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_sb.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c index 0e3d4f5ec33c..4aecc5fefe96 100644 --- a/fs/xfs/libxfs/xfs_sb.c +++ b/fs/xfs/libxfs/xfs_sb.c @@ -583,7 +583,8 @@ xfs_sb_verify( * Only check the in progress field for the primary superblock as * mkfs.xfs doesn't clear it from secondary superblocks. */ - return xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR, + return xfs_mount_validate_sb(mp, &sb, + bp->b_maps[0].bm_bn == XFS_SB_DADDR, check_version); } -- cgit v1.2.3-70-g09d2 From 800b2694f890cc35a1bda63501fc71c94389d517 Mon Sep 17 00:00:00 2001 From: Brian Foster Date: Fri, 26 Aug 2016 16:01:59 +1000 Subject: xfs: prevent dropping ioend completions during buftarg wait xfs_wait_buftarg() waits for all pending I/O, drains the ioend completion workqueue and walks the LRU until all buffers in the cache have been released. This is traditionally an unmount operation` but the mechanism is also reused during filesystem freeze. xfs_wait_buftarg() invokes drain_workqueue() as part of the quiesce, which is intended more for a shutdown sequence in that it indicates to the queue that new operations are not expected once the drain has begun. New work jobs after this point result in a WARN_ON_ONCE() and are otherwise dropped. With filesystem freeze, however, read operations are allowed and can proceed during or after the workqueue drain. If such a read occurs during the drain sequence, the workqueue infrastructure complains about the queued ioend completion work item and drops it on the floor. As a result, the buffer remains on the LRU and the freeze never completes. Despite the fact that the overall buffer cache cleanup is not necessary during freeze, fix up this operation such that it is safe to invoke during non-unmount quiesce operations. Replace the drain_workqueue() call with flush_workqueue(), which runs a similar serialization on pending workqueue jobs without causing new jobs to be dropped. This is safe for unmount as unmount independently locks out new operations by the time xfs_wait_buftarg() is invoked. cc: Signed-off-by: Brian Foster Reviewed-by: Christoph Hellwig Signed-off-by: Dave Chinner --- fs/xfs/xfs_buf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c index 607cc29bba21..b5b9bffe3520 100644 --- a/fs/xfs/xfs_buf.c +++ b/fs/xfs/xfs_buf.c @@ -1611,7 +1611,7 @@ xfs_wait_buftarg( */ while (percpu_counter_sum(&btp->bt_io_count)) delay(100); - drain_workqueue(btp->bt_mount->m_buf_workqueue); + flush_workqueue(btp->bt_mount->m_buf_workqueue); /* loop until there is nothing left on the lru list. */ while (list_lru_count(&btp->bt_lru)) { -- cgit v1.2.3-70-g09d2 From 079d37df3397d48aab0f014986c1b0a1ca6256aa Mon Sep 17 00:00:00 2001 From: Eric Ren Date: Thu, 25 Aug 2016 17:20:59 +0800 Subject: dlm: fix malfunction of dlm_tool caused by debugfs changes MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit With the current kernel, `dlm_tool lockdebug` fails as below: "dlm_tool lockdebug ED0BD86DCE724393918A1AE8FDBF1EE3 can't open /sys/kernel/debug/dlm/ED0BD86DCE724393918A1AE8FDBF1EE3: Operation not permitted" This is because table_open() depends on file->f_op to tell which seq_file ops should be passed down. But, the original file ops in file->f_op is replaced by "debugfs_full_proxy_file_operations" with commit 49d200deaa68 ("debugfs: prevent access to removed files' private data"). Currently, I can think up 2 solutions: 1st, replace debugfs_create_file() with debugfs_create_file_unsafe(); 2nd, make different table_open#() accordingly. The 1st one is neat, but I don't thoroughly understand its risk. Maybe someone has a better one. Signed-off-by: Eric Ren Signed-off-by: David Teigland --- fs/dlm/debug_fs.c | 62 ++++++++++++++++++++++++++++++++++++++++++------------- 1 file changed, 48 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/dlm/debug_fs.c b/fs/dlm/debug_fs.c index eea64912c9c0..466f7d60edc2 100644 --- a/fs/dlm/debug_fs.c +++ b/fs/dlm/debug_fs.c @@ -607,20 +607,54 @@ static const struct file_operations format2_fops; static const struct file_operations format3_fops; static const struct file_operations format4_fops; -static int table_open(struct inode *inode, struct file *file) +static int table_open1(struct inode *inode, struct file *file) { struct seq_file *seq; - int ret = -1; + int ret; - if (file->f_op == &format1_fops) - ret = seq_open(file, &format1_seq_ops); - else if (file->f_op == &format2_fops) - ret = seq_open(file, &format2_seq_ops); - else if (file->f_op == &format3_fops) - ret = seq_open(file, &format3_seq_ops); - else if (file->f_op == &format4_fops) - ret = seq_open(file, &format4_seq_ops); + ret = seq_open(file, &format1_seq_ops); + if (ret) + return ret; + + seq = file->private_data; + seq->private = inode->i_private; /* the dlm_ls */ + return 0; +} + +static int table_open2(struct inode *inode, struct file *file) +{ + struct seq_file *seq; + int ret; + + ret = seq_open(file, &format2_seq_ops); + if (ret) + return ret; + + seq = file->private_data; + seq->private = inode->i_private; /* the dlm_ls */ + return 0; +} + +static int table_open3(struct inode *inode, struct file *file) +{ + struct seq_file *seq; + int ret; + + ret = seq_open(file, &format3_seq_ops); + if (ret) + return ret; + + seq = file->private_data; + seq->private = inode->i_private; /* the dlm_ls */ + return 0; +} + +static int table_open4(struct inode *inode, struct file *file) +{ + struct seq_file *seq; + int ret; + ret = seq_open(file, &format4_seq_ops); if (ret) return ret; @@ -631,7 +665,7 @@ static int table_open(struct inode *inode, struct file *file) static const struct file_operations format1_fops = { .owner = THIS_MODULE, - .open = table_open, + .open = table_open1, .read = seq_read, .llseek = seq_lseek, .release = seq_release @@ -639,7 +673,7 @@ static const struct file_operations format1_fops = { static const struct file_operations format2_fops = { .owner = THIS_MODULE, - .open = table_open, + .open = table_open2, .read = seq_read, .llseek = seq_lseek, .release = seq_release @@ -647,7 +681,7 @@ static const struct file_operations format2_fops = { static const struct file_operations format3_fops = { .owner = THIS_MODULE, - .open = table_open, + .open = table_open3, .read = seq_read, .llseek = seq_lseek, .release = seq_release @@ -655,7 +689,7 @@ static const struct file_operations format3_fops = { static const struct file_operations format4_fops = { .owner = THIS_MODULE, - .open = table_open, + .open = table_open4, .read = seq_read, .llseek = seq_lseek, .release = seq_release -- cgit v1.2.3-70-g09d2 From 088bf2ff5d12e2e32ee52a4024fec26e582f44d3 Mon Sep 17 00:00:00 2001 From: Vegard Nossum Date: Thu, 25 Aug 2016 15:17:11 -0700 Subject: fs/seq_file: fix out-of-bounds read seq_read() is a nasty piece of work, not to mention buggy. It has (I think) an old bug which allows unprivileged userspace to read beyond the end of m->buf. I was getting these: BUG: KASAN: slab-out-of-bounds in seq_read+0xcd2/0x1480 at addr ffff880116889880 Read of size 2713 by task trinity-c2/1329 CPU: 2 PID: 1329 Comm: trinity-c2 Not tainted 4.8.0-rc1+ #96 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.3-0-ge2fc41e-prebuilt.qemu-project.org 04/01/2014 Call Trace: kasan_object_err+0x1c/0x80 kasan_report_error+0x2cb/0x7e0 kasan_report+0x4e/0x80 check_memory_region+0x13e/0x1a0 kasan_check_read+0x11/0x20 seq_read+0xcd2/0x1480 proc_reg_read+0x10b/0x260 do_loop_readv_writev.part.5+0x140/0x2c0 do_readv_writev+0x589/0x860 vfs_readv+0x7b/0xd0 do_readv+0xd8/0x2c0 SyS_readv+0xb/0x10 do_syscall_64+0x1b3/0x4b0 entry_SYSCALL64_slow_path+0x25/0x25 Object at ffff880116889100, in cache kmalloc-4096 size: 4096 Allocated: PID = 1329 save_stack_trace+0x26/0x80 save_stack+0x46/0xd0 kasan_kmalloc+0xad/0xe0 __kmalloc+0x1aa/0x4a0 seq_buf_alloc+0x35/0x40 seq_read+0x7d8/0x1480 proc_reg_read+0x10b/0x260 do_loop_readv_writev.part.5+0x140/0x2c0 do_readv_writev+0x589/0x860 vfs_readv+0x7b/0xd0 do_readv+0xd8/0x2c0 SyS_readv+0xb/0x10 do_syscall_64+0x1b3/0x4b0 return_from_SYSCALL_64+0x0/0x6a Freed: PID = 0 (stack is not available) Memory state around the buggy address: ffff88011688a000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ffff88011688a080: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 >ffff88011688a100: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ^ ffff88011688a180: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff88011688a200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ================================================================== Disabling lock debugging due to kernel taint This seems to be the same thing that Dave Jones was seeing here: https://lkml.org/lkml/2016/8/12/334 There are multiple issues here: 1) If we enter the function with a non-empty buffer, there is an attempt to flush it. But it was not clearing m->from after doing so, which means that if we try to do this flush twice in a row without any call to traverse() in between, we are going to be reading from the wrong place -- the splat above, fixed by this patch. 2) If there's a short write to userspace because of page faults, the buffer may already contain multiple lines (i.e. pos has advanced by more than 1), but we don't save the progress that was made so the next call will output what we've already returned previously. Since that is a much less serious issue (and I have a headache after staring at seq_read() for the past 8 hours), I'll leave that for now. Link: http://lkml.kernel.org/r/1471447270-32093-1-git-send-email-vegard.nossum@oracle.com Signed-off-by: Vegard Nossum Reported-by: Dave Jones Cc: Al Viro Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/seq_file.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/seq_file.c b/fs/seq_file.c index 19f532e7d35e..6dc4296eed62 100644 --- a/fs/seq_file.c +++ b/fs/seq_file.c @@ -223,8 +223,10 @@ ssize_t seq_read(struct file *file, char __user *buf, size_t size, loff_t *ppos) size -= n; buf += n; copied += n; - if (!m->count) + if (!m->count) { + m->from = 0; m->index++; + } if (!size) goto Done; } -- cgit v1.2.3-70-g09d2 From e09c978aae5bedfdb379be80363b024b7d82638b Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 27 Aug 2016 23:44:04 -0400 Subject: NFSv4.1: Fix Oopsable condition in server callback races The slot table hasn't been an array since v3.7. Ensure that we use nfs4_lookup_slot() to access the slot correctly. Fixes: 87dda67e7386 ("NFSv4.1: Allow SEQUENCE to resize the slot table...") Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v3.8+ --- fs/nfs/callback_proc.c | 5 +---- fs/nfs/nfs4session.c | 33 +++++++++++++++++++++++++++++++++ fs/nfs/nfs4session.h | 1 + 3 files changed, 35 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index c92a75e066a6..a4cf6d2c14a4 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -454,11 +454,8 @@ static bool referring_call_exists(struct nfs_client *clp, ((u32 *)&rclist->rcl_sessionid.data)[3], ref->rc_sequenceid, ref->rc_slotid); - spin_lock(&tbl->slot_tbl_lock); - status = (test_bit(ref->rc_slotid, tbl->used_slots) && - tbl->slots[ref->rc_slotid].seq_nr == + status = nfs4_slot_seqid_in_use(tbl, ref->rc_slotid, ref->rc_sequenceid); - spin_unlock(&tbl->slot_tbl_lock); if (status) goto out; } diff --git a/fs/nfs/nfs4session.c b/fs/nfs/nfs4session.c index 332d06e64fa9..c1f4c208f38a 100644 --- a/fs/nfs/nfs4session.c +++ b/fs/nfs/nfs4session.c @@ -172,6 +172,39 @@ struct nfs4_slot *nfs4_lookup_slot(struct nfs4_slot_table *tbl, u32 slotid) return ERR_PTR(-E2BIG); } +static int nfs4_slot_get_seqid(struct nfs4_slot_table *tbl, u32 slotid, + u32 *seq_nr) + __must_hold(&tbl->slot_tbl_lock) +{ + struct nfs4_slot *slot; + + slot = nfs4_lookup_slot(tbl, slotid); + if (IS_ERR(slot)) + return PTR_ERR(slot); + *seq_nr = slot->seq_nr; + return 0; +} + +/* + * nfs4_slot_seqid_in_use - test if a slot sequence id is still in use + * + * Given a slot table, slot id and sequence number, determine if the + * RPC call in question is still in flight. This function is mainly + * intended for use by the callback channel. + */ +bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, u32 slotid, u32 seq_nr) +{ + u32 cur_seq; + bool ret = false; + + spin_lock(&tbl->slot_tbl_lock); + if (nfs4_slot_get_seqid(tbl, slotid, &cur_seq) == 0 && + cur_seq == seq_nr && test_bit(slotid, tbl->used_slots)) + ret = true; + spin_unlock(&tbl->slot_tbl_lock); + return ret; +} + /* * nfs4_alloc_slot - efficiently look for a free slot * diff --git a/fs/nfs/nfs4session.h b/fs/nfs/nfs4session.h index 5b51298d1d03..33cace62b50b 100644 --- a/fs/nfs/nfs4session.h +++ b/fs/nfs/nfs4session.h @@ -78,6 +78,7 @@ extern int nfs4_setup_slot_table(struct nfs4_slot_table *tbl, extern void nfs4_shutdown_slot_table(struct nfs4_slot_table *tbl); extern struct nfs4_slot *nfs4_alloc_slot(struct nfs4_slot_table *tbl); extern struct nfs4_slot *nfs4_lookup_slot(struct nfs4_slot_table *tbl, u32 slotid); +extern bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, u32 slotid, u32 seq_nr); extern bool nfs4_try_to_lock_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *slot); extern void nfs4_free_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *slot); extern void nfs4_slot_tbl_drain_complete(struct nfs4_slot_table *tbl); -- cgit v1.2.3-70-g09d2 From 045d2a6d076a2ecd7043ea543ea198af943f8b16 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 28 Aug 2016 13:25:43 -0400 Subject: NFSv4.1: Delay callback processing when there are referring triples If CB_SEQUENCE tells us that the processing of this request depends on the completion of one or more referring triples (see RFC 5661 Section 2.10.6.3), delay the callback processing until after the RPC requests being referred to have completed. If we end up delaying for more than 1/2 second, then fall back to returning NFS4ERR_DELAY in reply to the callback. Signed-off-by: Trond Myklebust --- fs/nfs/callback_proc.c | 4 ++-- fs/nfs/nfs4proc.c | 2 ++ fs/nfs/nfs4session.c | 22 +++++++++++++++++++++- fs/nfs/nfs4session.h | 5 ++++- 4 files changed, 29 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index a4cf6d2c14a4..c35932967722 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -454,8 +454,8 @@ static bool referring_call_exists(struct nfs_client *clp, ((u32 *)&rclist->rcl_sessionid.data)[3], ref->rc_sequenceid, ref->rc_slotid); - status = nfs4_slot_seqid_in_use(tbl, ref->rc_slotid, - ref->rc_sequenceid); + status = nfs4_slot_wait_on_seqid(tbl, ref->rc_slotid, + ref->rc_sequenceid, HZ >> 1) < 0; if (status) goto out; } diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 1949bbd806eb..0cc0c319cfdd 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -686,6 +686,8 @@ out_unlock: res->sr_slot = NULL; if (send_new_highest_used_slotid) nfs41_notify_server(session->clp); + if (waitqueue_active(&tbl->slot_waitq)) + wake_up_all(&tbl->slot_waitq); } int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) diff --git a/fs/nfs/nfs4session.c b/fs/nfs/nfs4session.c index c1f4c208f38a..b62973045a3e 100644 --- a/fs/nfs/nfs4session.c +++ b/fs/nfs/nfs4session.c @@ -28,6 +28,7 @@ static void nfs4_init_slot_table(struct nfs4_slot_table *tbl, const char *queue) tbl->highest_used_slotid = NFS4_NO_SLOT; spin_lock_init(&tbl->slot_tbl_lock); rpc_init_priority_wait_queue(&tbl->slot_tbl_waitq, queue); + init_waitqueue_head(&tbl->slot_waitq); init_completion(&tbl->complete); } @@ -192,7 +193,8 @@ static int nfs4_slot_get_seqid(struct nfs4_slot_table *tbl, u32 slotid, * RPC call in question is still in flight. This function is mainly * intended for use by the callback channel. */ -bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, u32 slotid, u32 seq_nr) +static bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, + u32 slotid, u32 seq_nr) { u32 cur_seq; bool ret = false; @@ -205,6 +207,24 @@ bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, u32 slotid, u32 seq_nr) return ret; } +/* + * nfs4_slot_wait_on_seqid - wait until a slot sequence id is complete + * + * Given a slot table, slot id and sequence number, wait until the + * corresponding RPC call completes. This function is mainly + * intended for use by the callback channel. + */ +int nfs4_slot_wait_on_seqid(struct nfs4_slot_table *tbl, + u32 slotid, u32 seq_nr, + unsigned long timeout) +{ + if (wait_event_timeout(tbl->slot_waitq, + !nfs4_slot_seqid_in_use(tbl, slotid, seq_nr), + timeout) == 0) + return -ETIMEDOUT; + return 0; +} + /* * nfs4_alloc_slot - efficiently look for a free slot * diff --git a/fs/nfs/nfs4session.h b/fs/nfs/nfs4session.h index 33cace62b50b..fa75d7db3db3 100644 --- a/fs/nfs/nfs4session.h +++ b/fs/nfs/nfs4session.h @@ -36,6 +36,7 @@ struct nfs4_slot_table { unsigned long used_slots[SLOT_TABLE_SZ]; /* used/unused bitmap */ spinlock_t slot_tbl_lock; struct rpc_wait_queue slot_tbl_waitq; /* allocators may wait here */ + wait_queue_head_t slot_waitq; /* Completion wait on slot */ u32 max_slots; /* # slots in table */ u32 max_slotid; /* Max allowed slotid value */ u32 highest_used_slotid; /* sent to server on each SEQ. @@ -78,7 +79,9 @@ extern int nfs4_setup_slot_table(struct nfs4_slot_table *tbl, extern void nfs4_shutdown_slot_table(struct nfs4_slot_table *tbl); extern struct nfs4_slot *nfs4_alloc_slot(struct nfs4_slot_table *tbl); extern struct nfs4_slot *nfs4_lookup_slot(struct nfs4_slot_table *tbl, u32 slotid); -extern bool nfs4_slot_seqid_in_use(struct nfs4_slot_table *tbl, u32 slotid, u32 seq_nr); +extern int nfs4_slot_wait_on_seqid(struct nfs4_slot_table *tbl, + u32 slotid, u32 seq_nr, + unsigned long timeout); extern bool nfs4_try_to_lock_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *slot); extern void nfs4_free_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *slot); extern void nfs4_slot_tbl_drain_complete(struct nfs4_slot_table *tbl); -- cgit v1.2.3-70-g09d2 From 07e8dcbda71ef87e9cbdc42b5bb16a44c1ab839b Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 28 Aug 2016 10:28:25 -0400 Subject: NFSv4.1: Defer bumping the slot sequence number until we free the slot For operations like OPEN or LAYOUTGET, which return recallable state (i.e. delegations and layouts) we want to enable the mechanism for resolving recall races in RFC5661 Section 2.10.6.3. To do so, we will want to defer bumping the slot's sequence number until we have finished processing the RPC results. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 9 +++++++-- fs/nfs/nfs4session.h | 3 ++- 2 files changed, 9 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 0cc0c319cfdd..de4a89d3d740 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -666,6 +666,11 @@ static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res) tbl = slot->table; session = tbl->session; + /* Bump the slot sequence number */ + if (slot->seq_done) + slot->seq_nr++; + slot->seq_done = 0; + spin_lock(&tbl->slot_tbl_lock); /* Be nice to the server: try to ensure that the last transmitted * value for highest_user_slotid <= target_highest_slotid @@ -716,7 +721,7 @@ int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) switch (res->sr_status) { case 0: /* Update the slot's sequence and clientid lease timer */ - ++slot->seq_nr; + slot->seq_done = 1; clp = session->clp; do_renew_lease(clp, res->sr_timestamp); /* Check sequence flags */ @@ -771,7 +776,7 @@ int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) goto retry_nowait; default: /* Just update the slot sequence no. */ - ++slot->seq_nr; + slot->seq_done = 1; } out: /* The session may be reset by one of the error handlers. */ diff --git a/fs/nfs/nfs4session.h b/fs/nfs/nfs4session.h index fa75d7db3db3..f703b755351b 100644 --- a/fs/nfs/nfs4session.h +++ b/fs/nfs/nfs4session.h @@ -21,7 +21,8 @@ struct nfs4_slot { unsigned long generation; u32 slot_nr; u32 seq_nr; - unsigned int interrupted : 1; + unsigned int interrupted : 1, + seq_done : 1; }; /* Sessions */ -- cgit v1.2.3-70-g09d2 From 2e80dbe7ac51a911e8a828407b1a48c5ba938cd2 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 28 Aug 2016 11:50:26 -0400 Subject: NFSv4.1: Close callback races for OPEN, LAYOUTGET and LAYOUTRETURN Defer freeing the slot until after we have processed the results from OPEN and LAYOUTGET. This means that the server can rely on the mechanism in RFC5661 Section 2.10.6.3 to ensure that replies to an OPEN or LAYOUTGET/RETURN RPC call don't race with the callbacks that apply to them. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 78 +++++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 65 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index de4a89d3d740..f5aecaabcb7c 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -634,15 +634,11 @@ out_sleep: } EXPORT_SYMBOL_GPL(nfs40_setup_sequence); -static int nfs40_sequence_done(struct rpc_task *task, - struct nfs4_sequence_res *res) +static void nfs40_sequence_free_slot(struct nfs4_sequence_res *res) { struct nfs4_slot *slot = res->sr_slot; struct nfs4_slot_table *tbl; - if (slot == NULL) - goto out; - tbl = slot->table; spin_lock(&tbl->slot_tbl_lock); if (!nfs41_wake_and_assign_slot(tbl, slot)) @@ -650,7 +646,13 @@ static int nfs40_sequence_done(struct rpc_task *task, spin_unlock(&tbl->slot_tbl_lock); res->sr_slot = NULL; -out: +} + +static int nfs40_sequence_done(struct rpc_task *task, + struct nfs4_sequence_res *res) +{ + if (res->sr_slot != NULL) + nfs40_sequence_free_slot(res); return 1; } @@ -695,7 +697,8 @@ out_unlock: wake_up_all(&tbl->slot_waitq); } -int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) +static int nfs41_sequence_process(struct rpc_task *task, + struct nfs4_sequence_res *res) { struct nfs4_session *session; struct nfs4_slot *slot = res->sr_slot; @@ -781,11 +784,11 @@ int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) out: /* The session may be reset by one of the error handlers. */ dprintk("%s: Error %d free the slot \n", __func__, res->sr_status); - nfs41_sequence_free_slot(res); out_noaction: return ret; retry_nowait: if (rpc_restart_call_prepare(task)) { + nfs41_sequence_free_slot(res); task->tk_status = 0; ret = 0; } @@ -796,8 +799,37 @@ out_retry: rpc_delay(task, NFS4_POLL_RETRY_MAX); return 0; } + +int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) +{ + if (!nfs41_sequence_process(task, res)) + return 0; + if (res->sr_slot != NULL) + nfs41_sequence_free_slot(res); + return 1; + +} EXPORT_SYMBOL_GPL(nfs41_sequence_done); +static int nfs4_sequence_process(struct rpc_task *task, struct nfs4_sequence_res *res) +{ + if (res->sr_slot == NULL) + return 1; + if (res->sr_slot->table->session != NULL) + return nfs41_sequence_process(task, res); + return nfs40_sequence_done(task, res); +} + +static void nfs4_sequence_free_slot(struct nfs4_sequence_res *res) +{ + if (res->sr_slot != NULL) { + if (res->sr_slot->table->session != NULL) + nfs41_sequence_free_slot(res); + else + nfs40_sequence_free_slot(res); + } +} + int nfs4_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) { if (res->sr_slot == NULL) @@ -927,6 +959,17 @@ static int nfs4_setup_sequence(const struct nfs_server *server, args, res, task); } +static int nfs4_sequence_process(struct rpc_task *task, struct nfs4_sequence_res *res) +{ + return nfs40_sequence_done(task, res); +} + +static void nfs4_sequence_free_slot(struct nfs4_sequence_res *res) +{ + if (res->sr_slot != NULL) + nfs40_sequence_free_slot(res); +} + int nfs4_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res) { @@ -1204,6 +1247,7 @@ static void nfs4_opendata_free(struct kref *kref) struct super_block *sb = p->dentry->d_sb; nfs_free_seqid(p->o_arg.seqid); + nfs4_sequence_free_slot(&p->o_res.seq_res); if (p->state != NULL) nfs4_put_open_state(p->state); nfs4_put_state_owner(p->owner); @@ -1663,9 +1707,14 @@ err: static struct nfs4_state * nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data) { + struct nfs4_state *ret; + if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) - return _nfs4_opendata_reclaim_to_nfs4_state(data); - return _nfs4_opendata_to_nfs4_state(data); + ret =_nfs4_opendata_reclaim_to_nfs4_state(data); + else + ret = _nfs4_opendata_to_nfs4_state(data); + nfs4_sequence_free_slot(&data->o_res.seq_res); + return ret; } static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state *state) @@ -2063,7 +2112,7 @@ static void nfs4_open_done(struct rpc_task *task, void *calldata) data->rpc_status = task->tk_status; - if (!nfs4_sequence_done(task, &data->o_res.seq_res)) + if (!nfs4_sequence_process(task, &data->o_res.seq_res)) return; if (task->tk_status == 0) { @@ -7871,7 +7920,7 @@ static void nfs4_layoutget_done(struct rpc_task *task, void *calldata) struct nfs4_layoutget *lgp = calldata; dprintk("--> %s\n", __func__); - nfs41_sequence_done(task, &lgp->res.seq_res); + nfs41_sequence_process(task, &lgp->res.seq_res); dprintk("<-- %s\n", __func__); } @@ -8087,6 +8136,7 @@ nfs4_proc_layoutget(struct nfs4_layoutget *lgp, long *timeout, gfp_t gfp_flags) /* if layoutp->len is 0, nfs4_layoutget_prepare called rpc_exit */ if (status == 0 && lgp->res.layoutp->len) lseg = pnfs_layout_process(lgp); + nfs4_sequence_free_slot(&lgp->res.seq_res); rpc_put_task(task); dprintk("<-- %s status=%d\n", __func__, status); if (status) @@ -8113,7 +8163,7 @@ static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata) dprintk("--> %s\n", __func__); - if (!nfs41_sequence_done(task, &lrp->res.seq_res)) + if (!nfs41_sequence_process(task, &lrp->res.seq_res)) return; server = NFS_SERVER(lrp->args.inode); @@ -8125,6 +8175,7 @@ static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata) case -NFS4ERR_DELAY: if (nfs4_async_handle_error(task, server, NULL, NULL) != -EAGAIN) break; + nfs4_sequence_free_slot(&lrp->res.seq_res); rpc_restart_call_prepare(task); return; } @@ -8145,6 +8196,7 @@ static void nfs4_layoutreturn_release(void *calldata) pnfs_set_layout_stateid(lo, &lrp->res.stateid, true); pnfs_clear_layoutreturn_waitbit(lo); spin_unlock(&lo->plh_inode->i_lock); + nfs4_sequence_free_slot(&lrp->res.seq_res); pnfs_free_lseg_list(&freeme); pnfs_put_layout_hdr(lrp->args.layout); nfs_iput_and_deactive(lrp->inode); -- cgit v1.2.3-70-g09d2 From d138027a8256a3e9d7657c8d0dae84c08ef2cfe1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 28 Aug 2016 12:19:04 -0400 Subject: NFSv4.1: Remove obsolete and incorrrect assignment in nfs4_callback_sequence Signed-off-by: Trond Myklebust --- fs/nfs/callback_proc.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index c35932967722..f953ef6b2f2e 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -484,7 +484,6 @@ __be32 nfs4_callback_sequence(struct cb_sequenceargs *args, goto out; tbl = &clp->cl_session->bc_slot_table; - slot = tbl->slots + args->csa_slotid; /* Set up res before grabbing the spinlock */ memcpy(&res->csr_sessionid, &args->csa_sessionid, -- cgit v1.2.3-70-g09d2 From 17de0a9ff3df8f54f2f47746d118112d4e61d973 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 29 Aug 2016 11:33:58 +1000 Subject: iomap: don't set FIEMAP_EXTENT_MERGED for extent based filesystems Filesystems like XFS that use extents should not set the FIEMAP_EXTENT_MERGED flag in the fiemap extent structures. To allow for both behaviors for the upcoming gfs2 usage split the iomap type field into type and flags, and only set FIEMAP_EXTENT_MERGED if the IOMAP_F_MERGED flag is set. The flags field will also come in handy for future features such as shared extents on reflink-enabled file systems. Reported-by: Andreas Gruenbacher Signed-off-by: Christoph Hellwig Acked-by: Darrick J. Wong Signed-off-by: Dave Chinner --- fs/iomap.c | 5 ++++- include/linux/iomap.h | 8 +++++++- 2 files changed, 11 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/iomap.c b/fs/iomap.c index 0342254646e3..706270f21b35 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -428,9 +428,12 @@ static int iomap_to_fiemap(struct fiemap_extent_info *fi, break; } + if (iomap->flags & IOMAP_F_MERGED) + flags |= FIEMAP_EXTENT_MERGED; + return fiemap_fill_next_extent(fi, iomap->offset, iomap->blkno != IOMAP_NULL_BLOCK ? iomap->blkno << 9: 0, - iomap->length, flags | FIEMAP_EXTENT_MERGED); + iomap->length, flags); } diff --git a/include/linux/iomap.h b/include/linux/iomap.h index 3267df461012..3d70ece10313 100644 --- a/include/linux/iomap.h +++ b/include/linux/iomap.h @@ -18,6 +18,11 @@ struct vm_fault; #define IOMAP_MAPPED 0x03 /* blocks allocated @blkno */ #define IOMAP_UNWRITTEN 0x04 /* blocks allocated @blkno in unwritten state */ +/* + * Flags for iomap mappings: + */ +#define IOMAP_F_MERGED 0x01 /* contains multiple blocks/extents */ + /* * Magic value for blkno: */ @@ -27,7 +32,8 @@ struct iomap { sector_t blkno; /* 1st sector of mapping, 512b units */ loff_t offset; /* file offset of mapping, bytes */ u64 length; /* length of mapping, bytes */ - int type; /* type of mapping */ + u16 type; /* type of mapping */ + u16 flags; /* flags for mapping */ struct block_device *bdev; /* block device for I/O */ }; -- cgit v1.2.3-70-g09d2 From 3dc147359e3dcdf0648f1e2c11f62cfae3160df0 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 29 Aug 2016 15:12:54 -0400 Subject: pNFS/flexfiles: Fix an Oopsable condition when connection to the DS fails If the attempt to connect to a DS fails inside ff_layout_pg_init_read or ff_layout_pg_init_write, then we currently end up clearing the layout segment carried by the struct nfs_pageio_descriptor, causing an Oops when we later call into ff_layout_read_pagelist/ff_layout_write_pagelist. The fix is to ensure we return the layout and then retry. Fixes: 446ca2195303 ("pNFS/flexfiles: When initing reads or writes, we...") Cc: stable@vger.kernel.org # v4.7+ Signed-off-by: Trond Myklebust --- fs/nfs/flexfilelayout/flexfilelayout.c | 37 +++++++++++++++---------------- fs/nfs/flexfilelayout/flexfilelayoutdev.c | 19 ++++++++-------- 2 files changed, 28 insertions(+), 28 deletions(-) (limited to 'fs') diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c index ee1c94c7614c..51b51369704c 100644 --- a/fs/nfs/flexfilelayout/flexfilelayout.c +++ b/fs/nfs/flexfilelayout/flexfilelayout.c @@ -806,11 +806,14 @@ ff_layout_choose_best_ds_for_read(struct pnfs_layout_segment *lseg, { struct nfs4_ff_layout_segment *fls = FF_LAYOUT_LSEG(lseg); struct nfs4_pnfs_ds *ds; + bool fail_return = false; int idx; /* mirrors are sorted by efficiency */ for (idx = start_idx; idx < fls->mirror_array_cnt; idx++) { - ds = nfs4_ff_layout_prepare_ds(lseg, idx, false); + if (idx+1 == fls->mirror_array_cnt) + fail_return = true; + ds = nfs4_ff_layout_prepare_ds(lseg, idx, fail_return); if (ds) { *best_idx = idx; return ds; @@ -859,6 +862,7 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio, struct nfs4_pnfs_ds *ds; int ds_idx; +retry: /* Use full layout for now */ if (!pgio->pg_lseg) ff_layout_pg_get_read(pgio, req, false); @@ -871,10 +875,13 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio, ds = ff_layout_choose_best_ds_for_read(pgio->pg_lseg, 0, &ds_idx); if (!ds) { - if (ff_layout_no_fallback_to_mds(pgio->pg_lseg)) - goto out_pnfs; - else + if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg)) goto out_mds; + pnfs_put_lseg(pgio->pg_lseg); + pgio->pg_lseg = NULL; + /* Sleep for 1 second before retrying */ + ssleep(1); + goto retry; } mirror = FF_LAYOUT_COMP(pgio->pg_lseg, ds_idx); @@ -890,12 +897,6 @@ out_mds: pnfs_put_lseg(pgio->pg_lseg); pgio->pg_lseg = NULL; nfs_pageio_reset_read_mds(pgio); - return; - -out_pnfs: - pnfs_set_lo_fail(pgio->pg_lseg); - pnfs_put_lseg(pgio->pg_lseg); - pgio->pg_lseg = NULL; } static void @@ -909,6 +910,7 @@ ff_layout_pg_init_write(struct nfs_pageio_descriptor *pgio, int i; int status; +retry: if (!pgio->pg_lseg) { pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode, req->wb_context, @@ -940,10 +942,13 @@ ff_layout_pg_init_write(struct nfs_pageio_descriptor *pgio, for (i = 0; i < pgio->pg_mirror_count; i++) { ds = nfs4_ff_layout_prepare_ds(pgio->pg_lseg, i, true); if (!ds) { - if (ff_layout_no_fallback_to_mds(pgio->pg_lseg)) - goto out_pnfs; - else + if (!ff_layout_no_fallback_to_mds(pgio->pg_lseg)) goto out_mds; + pnfs_put_lseg(pgio->pg_lseg); + pgio->pg_lseg = NULL; + /* Sleep for 1 second before retrying */ + ssleep(1); + goto retry; } pgm = &pgio->pg_mirrors[i]; mirror = FF_LAYOUT_COMP(pgio->pg_lseg, i); @@ -956,12 +961,6 @@ out_mds: pnfs_put_lseg(pgio->pg_lseg); pgio->pg_lseg = NULL; nfs_pageio_reset_write_mds(pgio); - return; - -out_pnfs: - pnfs_set_lo_fail(pgio->pg_lseg); - pnfs_put_lseg(pgio->pg_lseg); - pgio->pg_lseg = NULL; } static unsigned int diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c index 970efba05ae1..f7a3f6b05369 100644 --- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c +++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c @@ -379,7 +379,7 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx, devid = &mirror->mirror_ds->id_node; if (ff_layout_test_devid_unavailable(devid)) - goto out; + goto out_fail; ds = mirror->mirror_ds->ds; /* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */ @@ -405,15 +405,16 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx, mirror->mirror_ds->ds_versions[0].rsize = max_payload; if (mirror->mirror_ds->ds_versions[0].wsize > max_payload) mirror->mirror_ds->ds_versions[0].wsize = max_payload; - } else { - ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout), - mirror, lseg->pls_range.offset, - lseg->pls_range.length, NFS4ERR_NXIO, - OP_ILLEGAL, GFP_NOIO); - if (fail_return || !ff_layout_has_available_ds(lseg)) - pnfs_error_mark_layout_for_return(ino, lseg); - ds = NULL; + goto out; } + ff_layout_track_ds_error(FF_LAYOUT_FROM_HDR(lseg->pls_layout), + mirror, lseg->pls_range.offset, + lseg->pls_range.length, NFS4ERR_NXIO, + OP_ILLEGAL, GFP_NOIO); +out_fail: + if (fail_return || !ff_layout_has_available_ds(lseg)) + pnfs_error_mark_layout_for_return(ino, lseg); + ds = NULL; out: return ds; } -- cgit v1.2.3-70-g09d2 From ea78d80866ce375defb2fdd1c8a3aafec95e0f85 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Tue, 30 Aug 2016 13:51:39 +1000 Subject: xfs: track log done items directly in the deferred pending work item Christoph reports slab corruption when a deferred refcount update aborts during _defer_finish(). The cause of this was broken log item state tracking in xfs_defer_pending -- upon an abort, _defer_trans_abort() will call abort_intent on all intent items, including the ones that have already had a done item attached. This is incorrect because each intent item has 2 refcount: the first is released when the intent item is committed to the log; and the second is released when the _done_ item is committed to the log, or by the intent creator if there is no done item. In other words, once we log the done item, responsibility for releasing the intent item's second refcount is transferred to the done item and /must not/ be performed by anything else. The dfp_committed flag should have been tracking whether or not we had a done item so that _defer_trans_abort could decide if it needs to abort the intent item, but due to a thinko this was not the case. Rip it out and track the done item directly so that we do the right thing w.r.t. intent item freeing. Signed-off-by: Darrick J. Wong Reported-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_defer.c | 17 ++++------------- fs/xfs/libxfs/xfs_defer.h | 2 +- fs/xfs/xfs_trace.h | 2 +- 3 files changed, 6 insertions(+), 15 deletions(-) (limited to 'fs') diff --git a/fs/xfs/libxfs/xfs_defer.c b/fs/xfs/libxfs/xfs_defer.c index 054a2032fdb3..c221d0ecd52e 100644 --- a/fs/xfs/libxfs/xfs_defer.c +++ b/fs/xfs/libxfs/xfs_defer.c @@ -194,7 +194,7 @@ xfs_defer_trans_abort( /* Abort intent items. */ list_for_each_entry(dfp, &dop->dop_pending, dfp_list) { trace_xfs_defer_pending_abort(tp->t_mountp, dfp); - if (dfp->dfp_committed) + if (!dfp->dfp_done) dfp->dfp_type->abort_intent(dfp->dfp_intent); } @@ -290,7 +290,6 @@ xfs_defer_finish( struct xfs_defer_pending *dfp; struct list_head *li; struct list_head *n; - void *done_item = NULL; void *state; int error = 0; void (*cleanup_fn)(struct xfs_trans *, void *, int); @@ -309,19 +308,11 @@ xfs_defer_finish( if (error) goto out; - /* Mark all pending intents as committed. */ - list_for_each_entry_reverse(dfp, &dop->dop_pending, dfp_list) { - if (dfp->dfp_committed) - break; - trace_xfs_defer_pending_commit((*tp)->t_mountp, dfp); - dfp->dfp_committed = true; - } - /* Log an intent-done item for the first pending item. */ dfp = list_first_entry(&dop->dop_pending, struct xfs_defer_pending, dfp_list); trace_xfs_defer_pending_finish((*tp)->t_mountp, dfp); - done_item = dfp->dfp_type->create_done(*tp, dfp->dfp_intent, + dfp->dfp_done = dfp->dfp_type->create_done(*tp, dfp->dfp_intent, dfp->dfp_count); cleanup_fn = dfp->dfp_type->finish_cleanup; @@ -331,7 +322,7 @@ xfs_defer_finish( list_del(li); dfp->dfp_count--; error = dfp->dfp_type->finish_item(*tp, dop, li, - done_item, &state); + dfp->dfp_done, &state); if (error) { /* * Clean up after ourselves and jump out. @@ -428,8 +419,8 @@ xfs_defer_add( dfp = kmem_alloc(sizeof(struct xfs_defer_pending), KM_SLEEP | KM_NOFS); dfp->dfp_type = defer_op_types[type]; - dfp->dfp_committed = false; dfp->dfp_intent = NULL; + dfp->dfp_done = NULL; dfp->dfp_count = 0; INIT_LIST_HEAD(&dfp->dfp_work); list_add_tail(&dfp->dfp_list, &dop->dop_intake); diff --git a/fs/xfs/libxfs/xfs_defer.h b/fs/xfs/libxfs/xfs_defer.h index cc3981c48296..e96533d178cf 100644 --- a/fs/xfs/libxfs/xfs_defer.h +++ b/fs/xfs/libxfs/xfs_defer.h @@ -30,8 +30,8 @@ struct xfs_defer_op_type; struct xfs_defer_pending { const struct xfs_defer_op_type *dfp_type; /* function pointers */ struct list_head dfp_list; /* pending items */ - bool dfp_committed; /* committed trans? */ void *dfp_intent; /* log intent item */ + void *dfp_done; /* log done item */ struct list_head dfp_work; /* work items */ unsigned int dfp_count; /* # extent items */ }; diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index 7e88bec3f359..d303a665dba9 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -2295,7 +2295,7 @@ DECLARE_EVENT_CLASS(xfs_defer_pending_class, __entry->dev = mp ? mp->m_super->s_dev : 0; __entry->type = dfp->dfp_type->type; __entry->intent = dfp->dfp_intent; - __entry->committed = dfp->dfp_committed; + __entry->committed = dfp->dfp_done != NULL; __entry->nr = dfp->dfp_count; ), TP_printk("dev %d:%d optype %d intent %p committed %d nr %d\n", -- cgit v1.2.3-70-g09d2 From 52442f9b11b7e5d4a38d99143011831fd171f8d9 Mon Sep 17 00:00:00 2001 From: Benjamin Coddington Date: Tue, 30 Aug 2016 09:20:32 -0400 Subject: NFS4: Avoid migration loops If a server returns itself as a location while migrating, the client may end up getting stuck attempting to migrate twice to the same server. Catch this by checking if the nfs_client found is the same as the existing client. For the other two callers to nfs4_set_client, the nfs_client will always be ERR_PTR(-EINVAL). Signed-off-by: Benjamin Coddington Signed-off-by: Trond Myklebust --- fs/nfs/nfs4client.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c index 8d7d08d4f95f..cd3b7cfdde16 100644 --- a/fs/nfs/nfs4client.c +++ b/fs/nfs/nfs4client.c @@ -817,6 +817,11 @@ static int nfs4_set_client(struct nfs_server *server, goto error; } + if (server->nfs_client == clp) { + error = -ELOOP; + goto error; + } + /* * Query for the lease time on clientid setup or renewal * -- cgit v1.2.3-70-g09d2 From 98b0f80c2396224bbbed81792b526e6c72ba9efa Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 29 Aug 2016 11:15:36 -0400 Subject: NFSv4.x: Fix a refcount leak in nfs_callback_up_net On error, the callers expect us to return without bumping nn->cb_users[]. Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v3.7+ --- fs/nfs/callback.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c index a7f2e6e33305..52a28311e2a4 100644 --- a/fs/nfs/callback.c +++ b/fs/nfs/callback.c @@ -275,6 +275,7 @@ static int nfs_callback_up_net(int minorversion, struct svc_serv *serv, err_socks: svc_rpcb_cleanup(serv, net); err_bind: + nn->cb_users[minorversion]--; dprintk("NFS: Couldn't create callback socket: err = %d; " "net = %p\n", ret, net); return ret; -- cgit v1.2.3-70-g09d2 From df6a58c5c5aa8ecb1e088ecead3fa33ae70181f1 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Fri, 17 Jun 2016 17:51:17 -0400 Subject: kernfs: don't depend on d_find_any_alias() when generating notifications kernfs_notify_workfn() sends out file modified events for the scheduled kernfs_nodes. Because the modifications aren't from userland, it doesn't have the matching file struct at hand and can't use fsnotify_modify(). Instead, it looked up the inode and then used d_find_any_alias() to find the dentry and used fsnotify_parent() and fsnotify() directly to generate notifications. The assumption was that the relevant dentries would have been pinned if there are listeners, which isn't true as inotify doesn't pin dentries at all and watching the parent doesn't pin the child dentries even for dnotify. This led to, for example, inotify watchers not getting notifications if the system is under memory pressure and the matching dentries got reclaimed. It can also be triggered through /proc/sys/vm/drop_caches or a remount attempt which involves shrinking dcache. fsnotify_parent() only uses the dentry to access the parent inode, which kernfs can do easily. Update kernfs_notify_workfn() so that it uses fsnotify() directly for both the parent and target inodes without going through d_find_any_alias(). While at it, supply the target file name to fsnotify() from kernfs_node->name. Signed-off-by: Tejun Heo Reported-by: Evgeny Vereshchagin Fixes: d911d9874801 ("kernfs: make kernfs_notify() trigger inotify events too") Cc: John McCutchan Cc: Robert Love Cc: Eric Paris Cc: stable@vger.kernel.org # v3.16+ Signed-off-by: Greg Kroah-Hartman --- fs/kernfs/file.c | 28 +++++++++++++++++++++------- 1 file changed, 21 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/kernfs/file.c b/fs/kernfs/file.c index e1574008adc9..2bcb86e6e6ca 100644 --- a/fs/kernfs/file.c +++ b/fs/kernfs/file.c @@ -840,21 +840,35 @@ repeat: mutex_lock(&kernfs_mutex); list_for_each_entry(info, &kernfs_root(kn)->supers, node) { + struct kernfs_node *parent; struct inode *inode; - struct dentry *dentry; + /* + * We want fsnotify_modify() on @kn but as the + * modifications aren't originating from userland don't + * have the matching @file available. Look up the inodes + * and generate the events manually. + */ inode = ilookup(info->sb, kn->ino); if (!inode) continue; - dentry = d_find_any_alias(inode); - if (dentry) { - fsnotify_parent(NULL, dentry, FS_MODIFY); - fsnotify(inode, FS_MODIFY, inode, FSNOTIFY_EVENT_INODE, - NULL, 0); - dput(dentry); + parent = kernfs_get_parent(kn); + if (parent) { + struct inode *p_inode; + + p_inode = ilookup(info->sb, parent->ino); + if (p_inode) { + fsnotify(p_inode, FS_MODIFY | FS_EVENT_ON_CHILD, + inode, FSNOTIFY_EVENT_INODE, kn->name, 0); + iput(p_inode); + } + + kernfs_put(parent); } + fsnotify(inode, FS_MODIFY, inode, FSNOTIFY_EVENT_INODE, + kn->name, 0); iput(inode); } -- cgit v1.2.3-70-g09d2 From 17d0774f80681020eccc9638d925a23f1fc4f671 Mon Sep 17 00:00:00 2001 From: Konstantin Khlebnikov Date: Wed, 22 Jun 2016 21:42:16 +0300 Subject: sysfs: correctly handle read offset on PREALLOC attrs Attributes declared with __ATTR_PREALLOC use sysfs_kf_read() which returns zero bytes for non-zero offset. This breaks script checkarray in mdadm tool in debian where /bin/sh is 'dash' because its builtin 'read' reads only one byte at a time. Script gets 'i' instead of 'idle' when reads current action from /sys/block/$dev/md/sync_action and as a result does nothing. This patch adds trivial implementation of partial read: generate whole string and move required part into buffer head. Signed-off-by: Konstantin Khlebnikov Fixes: 4ef67a8c95f3 ("sysfs/kernfs: make read requests on pre-alloc files use the buffer.") Link: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=787950 Cc: Stable # v3.19+ Acked-by: Tejun Heo Signed-off-by: Greg Kroah-Hartman --- fs/sysfs/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c index f35523d4fa3a..b803213d1307 100644 --- a/fs/sysfs/file.c +++ b/fs/sysfs/file.c @@ -114,9 +114,15 @@ static ssize_t sysfs_kf_read(struct kernfs_open_file *of, char *buf, * If buf != of->prealloc_buf, we don't know how * large it is, so cannot safely pass it to ->show */ - if (pos || WARN_ON_ONCE(buf != of->prealloc_buf)) + if (WARN_ON_ONCE(buf != of->prealloc_buf)) return 0; len = ops->show(kobj, of->kn->priv, buf); + if (pos) { + if (len <= pos) + return 0; + len -= pos; + memmove(buf, buf + pos, len); + } return min(count, len); } -- cgit v1.2.3-70-g09d2 From 9f834ec18defc369d73ccf9e87a2790bfa05bf46 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Mon, 22 Aug 2016 16:41:46 -0700 Subject: binfmt_elf: switch to new creds when switching to new mm MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We used to delay switching to the new credentials until after we had mapped the executable (and possible elf interpreter). That was kind of odd to begin with, since the new executable will actually then _run_ with the new creds, but whatever. The bigger problem was that we also want to make sure that we turn off prof events and tracing before we start mapping the new executable state. So while this is a cleanup, it's also a fix for a possible information leak. Reported-by: Robert Święcki Tested-by: Peter Zijlstra Acked-by: David Howells Acked-by: Oleg Nesterov Acked-by: Andy Lutomirski Acked-by: Eric W. Biederman Cc: Willy Tarreau Cc: Kees Cook Cc: Al Viro Signed-off-by: Linus Torvalds --- fs/binfmt_elf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c index 7f6aff3f72eb..e5495f37c6ed 100644 --- a/fs/binfmt_elf.c +++ b/fs/binfmt_elf.c @@ -853,6 +853,7 @@ static int load_elf_binary(struct linux_binprm *bprm) current->flags |= PF_RANDOMIZE; setup_new_exec(bprm); + install_exec_creds(bprm); /* Do this so that we can load the interpreter, if need be. We will change some of these later */ @@ -1044,7 +1045,6 @@ static int load_elf_binary(struct linux_binprm *bprm) goto out; #endif /* ARCH_HAS_SETUP_ADDITIONAL_PAGES */ - install_exec_creds(bprm); retval = create_elf_tables(bprm, &loc->elf_ex, load_addr, interp_load_addr); if (retval < 0) -- cgit v1.2.3-70-g09d2 From cd81a9170e69e018bbaba547c1fd85a585f5697a Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Tue, 23 Aug 2016 16:20:38 +0200 Subject: mm: introduce get_task_exe_file For more convenient access if one has a pointer to the task. As a minor nit take advantage of the fact that only task lock + rcu are needed to safely grab ->exe_file. This saves mm refcount dance. Use the helper in proc_exe_link. Signed-off-by: Mateusz Guzik Acked-by: Konstantin Khlebnikov Acked-by: Richard Guy Briggs Cc: # 4.3.x Signed-off-by: Paul Moore --- fs/proc/base.c | 7 +------ include/linux/mm.h | 1 + kernel/fork.c | 23 +++++++++++++++++++++++ 3 files changed, 25 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/proc/base.c b/fs/proc/base.c index 0d163a84082d..da8b1943ba04 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -1552,18 +1552,13 @@ static const struct file_operations proc_pid_set_comm_operations = { static int proc_exe_link(struct dentry *dentry, struct path *exe_path) { struct task_struct *task; - struct mm_struct *mm; struct file *exe_file; task = get_proc_task(d_inode(dentry)); if (!task) return -ENOENT; - mm = get_task_mm(task); + exe_file = get_task_exe_file(task); put_task_struct(task); - if (!mm) - return -ENOENT; - exe_file = get_mm_exe_file(mm); - mmput(mm); if (exe_file) { *exe_path = exe_file->f_path; path_get(&exe_file->f_path); diff --git a/include/linux/mm.h b/include/linux/mm.h index 8f468e0d2534..004c73a988b7 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1987,6 +1987,7 @@ extern void mm_drop_all_locks(struct mm_struct *mm); extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); extern struct file *get_mm_exe_file(struct mm_struct *mm); +extern struct file *get_task_exe_file(struct task_struct *task); extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); diff --git a/kernel/fork.c b/kernel/fork.c index d277e83ed3e0..42451aeb245f 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -773,6 +773,29 @@ struct file *get_mm_exe_file(struct mm_struct *mm) } EXPORT_SYMBOL(get_mm_exe_file); +/** + * get_task_exe_file - acquire a reference to the task's executable file + * + * Returns %NULL if task's mm (if any) has no associated executable file or + * this is a kernel thread with borrowed mm (see the comment above get_task_mm). + * User must release file via fput(). + */ +struct file *get_task_exe_file(struct task_struct *task) +{ + struct file *exe_file = NULL; + struct mm_struct *mm; + + task_lock(task); + mm = task->mm; + if (mm) { + if (!(task->flags & PF_KTHREAD)) + exe_file = get_mm_exe_file(mm); + } + task_unlock(task); + return exe_file; +} +EXPORT_SYMBOL(get_task_exe_file); + /** * get_task_mm - acquire a reference to the task's mm * -- cgit v1.2.3-70-g09d2 From 38b256973ea90fc7c2b7e1b734fa0e8b83538d50 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:11:59 +0200 Subject: ovl: handle umask and posix_acl_default correctly on creation Setting MS_POSIXACL in sb->s_flags has the side effect of passing mode to create functions without masking against umask. Another problem when creating over a whiteout is that the default posix acl is not inherited from the parent dir (because the real parent dir at the time of creation is the work directory). Fix these problems by: a) If upper fs does not have MS_POSIXACL, then mask mode with umask. b) If creating over a whiteout, call posix_acl_create() to get the inherited acls. After creation (but before moving to the final destination) set these acls on the created file. posix_acl_create() also updates the file creation mode as appropriate. Fixes: 39a25b2b3762 ("ovl: define ->get_acl() for overlay inodes") Signed-off-by: Miklos Szeredi --- fs/overlayfs/dir.c | 54 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 54 insertions(+) (limited to 'fs') diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c index 12bcd07b9e32..f485dd4288e4 100644 --- a/fs/overlayfs/dir.c +++ b/fs/overlayfs/dir.c @@ -12,6 +12,8 @@ #include #include #include +#include +#include #include "overlayfs.h" void ovl_cleanup(struct inode *wdir, struct dentry *wdentry) @@ -186,6 +188,9 @@ static int ovl_create_upper(struct dentry *dentry, struct inode *inode, struct dentry *newdentry; int err; + if (!hardlink && !IS_POSIXACL(udir)) + stat->mode &= ~current_umask(); + inode_lock_nested(udir, I_MUTEX_PARENT); newdentry = lookup_one_len(dentry->d_name.name, upperdir, dentry->d_name.len); @@ -335,6 +340,32 @@ out_free: return ret; } +static int ovl_set_upper_acl(struct dentry *upperdentry, const char *name, + const struct posix_acl *acl) +{ + void *buffer; + size_t size; + int err; + + if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !acl) + return 0; + + size = posix_acl_to_xattr(NULL, acl, NULL, 0); + buffer = kmalloc(size, GFP_KERNEL); + if (!buffer) + return -ENOMEM; + + size = posix_acl_to_xattr(&init_user_ns, acl, buffer, size); + err = size; + if (err < 0) + goto out_free; + + err = vfs_setxattr(upperdentry, name, buffer, size, XATTR_CREATE); +out_free: + kfree(buffer); + return err; +} + static int ovl_create_over_whiteout(struct dentry *dentry, struct inode *inode, struct kstat *stat, const char *link, struct dentry *hardlink) @@ -346,10 +377,18 @@ static int ovl_create_over_whiteout(struct dentry *dentry, struct inode *inode, struct dentry *upper; struct dentry *newdentry; int err; + struct posix_acl *acl, *default_acl; if (WARN_ON(!workdir)) return -EROFS; + if (!hardlink) { + err = posix_acl_create(dentry->d_parent->d_inode, + &stat->mode, &default_acl, &acl); + if (err) + return err; + } + err = ovl_lock_rename_workdir(workdir, upperdir); if (err) goto out; @@ -384,6 +423,17 @@ static int ovl_create_over_whiteout(struct dentry *dentry, struct inode *inode, if (err) goto out_cleanup; } + if (!hardlink) { + err = ovl_set_upper_acl(newdentry, XATTR_NAME_POSIX_ACL_ACCESS, + acl); + if (err) + goto out_cleanup; + + err = ovl_set_upper_acl(newdentry, XATTR_NAME_POSIX_ACL_DEFAULT, + default_acl); + if (err) + goto out_cleanup; + } if (!hardlink && S_ISDIR(stat->mode)) { err = ovl_set_opaque(newdentry); @@ -410,6 +460,10 @@ out_dput: out_unlock: unlock_rename(workdir, upperdir); out: + if (!hardlink) { + posix_acl_release(acl); + posix_acl_release(default_acl); + } return err; out_cleanup: -- cgit v1.2.3-70-g09d2 From c11b9fdd6a612f376a5e886505f1c54c16d8c380 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:11:59 +0200 Subject: ovl: remove posix_acl_default from workdir Clear out posix acl xattrs on workdir and also reset the mode after creation so that an inherited sgid bit is cleared. Signed-off-by: Miklos Szeredi Cc: --- fs/overlayfs/super.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'fs') diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index 4036132842b5..452fb7130efa 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -814,6 +814,10 @@ retry: struct kstat stat = { .mode = S_IFDIR | 0, }; + struct iattr attr = { + .ia_valid = ATTR_MODE, + .ia_mode = stat.mode, + }; if (work->d_inode) { err = -EEXIST; @@ -829,6 +833,21 @@ retry: err = ovl_create_real(dir, work, &stat, NULL, NULL, true); if (err) goto out_dput; + + err = vfs_removexattr(work, XATTR_NAME_POSIX_ACL_DEFAULT); + if (err && err != -ENODATA) + goto out_dput; + + err = vfs_removexattr(work, XATTR_NAME_POSIX_ACL_ACCESS); + if (err && err != -ENODATA) + goto out_dput; + + /* Clear any inherited mode bits */ + inode_lock(work->d_inode); + err = notify_change(work, &attr, NULL); + inode_unlock(work->d_inode); + if (err) + goto out_dput; } out_unlock: inode_unlock(dir); -- cgit v1.2.3-70-g09d2 From eea2fb4851e9dcbab6b991aaf47e2e024f1f55a0 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:11:59 +0200 Subject: ovl: proper cleanup of workdir When mounting overlayfs it needs a clean "work" directory under the supplied workdir. Previously the mount code removed this directory if it already existed and created a new one. If the removal failed (e.g. directory was not empty) then it fell back to a read-only mount not using the workdir. While this has never been reported, it is possible to get a non-empty "work" dir from a previous mount of overlayfs in case of crash in the middle of an operation using the work directory. In this case the left over state should be discarded and the overlay filesystem will be consistent, guaranteed by the atomicity of operations on moving to/from the workdir to the upper layer. This patch implements cleaning out any files left in workdir. It is implemented using real recursion for simplicity, but the depth is limited to 2, because the worst case is that of a directory containing whiteouts under "work". Signed-off-by: Miklos Szeredi Cc: --- fs/overlayfs/overlayfs.h | 2 ++ fs/overlayfs/readdir.c | 63 +++++++++++++++++++++++++++++++++++++++++++++++- fs/overlayfs/super.c | 2 +- 3 files changed, 65 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index 34839bd2b6b8..9a95e2c5653e 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -179,6 +179,8 @@ int ovl_check_empty_dir(struct dentry *dentry, struct list_head *list); void ovl_cleanup_whiteouts(struct dentry *upper, struct list_head *list); void ovl_cache_free(struct list_head *list); int ovl_check_d_type_supported(struct path *realpath); +void ovl_workdir_cleanup(struct inode *dir, struct vfsmount *mnt, + struct dentry *dentry, int level); /* inode.c */ int ovl_setattr(struct dentry *dentry, struct iattr *attr); diff --git a/fs/overlayfs/readdir.c b/fs/overlayfs/readdir.c index cf37fc76fc9f..f241b4ee3d8a 100644 --- a/fs/overlayfs/readdir.c +++ b/fs/overlayfs/readdir.c @@ -248,7 +248,7 @@ static inline int ovl_dir_read(struct path *realpath, err = rdd->err; } while (!err && rdd->count); - if (!err && rdd->first_maybe_whiteout) + if (!err && rdd->first_maybe_whiteout && rdd->dentry) err = ovl_check_whiteouts(realpath->dentry, rdd); fput(realfile); @@ -606,3 +606,64 @@ int ovl_check_d_type_supported(struct path *realpath) return rdd.d_type_supported; } + +static void ovl_workdir_cleanup_recurse(struct path *path, int level) +{ + int err; + struct inode *dir = path->dentry->d_inode; + LIST_HEAD(list); + struct ovl_cache_entry *p; + struct ovl_readdir_data rdd = { + .ctx.actor = ovl_fill_merge, + .dentry = NULL, + .list = &list, + .root = RB_ROOT, + .is_lowest = false, + }; + + err = ovl_dir_read(path, &rdd); + if (err) + goto out; + + inode_lock_nested(dir, I_MUTEX_PARENT); + list_for_each_entry(p, &list, l_node) { + struct dentry *dentry; + + if (p->name[0] == '.') { + if (p->len == 1) + continue; + if (p->len == 2 && p->name[1] == '.') + continue; + } + dentry = lookup_one_len(p->name, path->dentry, p->len); + if (IS_ERR(dentry)) + continue; + if (dentry->d_inode) + ovl_workdir_cleanup(dir, path->mnt, dentry, level); + dput(dentry); + } + inode_unlock(dir); +out: + ovl_cache_free(&list); +} + +void ovl_workdir_cleanup(struct inode *dir, struct vfsmount *mnt, + struct dentry *dentry, int level) +{ + int err; + + if (!d_is_dir(dentry) || level > 1) { + ovl_cleanup(dir, dentry); + return; + } + + err = ovl_do_rmdir(dir, dentry); + if (err) { + struct path path = { .mnt = mnt, .dentry = dentry }; + + inode_unlock(dir); + ovl_workdir_cleanup_recurse(&path, level + 1); + inode_lock_nested(dir, I_MUTEX_PARENT); + ovl_cleanup(dir, dentry); + } +} diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index 452fb7130efa..219534e5ca0b 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -825,7 +825,7 @@ retry: goto out_dput; retried = true; - ovl_cleanup(dir, work); + ovl_workdir_cleanup(dir, mnt, work, 0); dput(work); goto retry; } -- cgit v1.2.3-70-g09d2 From 5201dc449e4b6b6d7e92f7f974269b11681f98b5 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:11:59 +0200 Subject: ovl: use cached acl on underlying layer Instead of calling ->get_acl() directly, use get_acl() to get the cached value. We will have the acl cached on the underlying inode anyway, because we do permission checking on the both the overlay and the underlying fs. So, since we already have double caching, this improves performance without any cost. Signed-off-by: Miklos Szeredi --- fs/overlayfs/inode.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 024352f1d405..d50d1ead1b6f 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -10,6 +10,7 @@ #include #include #include +#include #include "overlayfs.h" static int ovl_copy_up_truncate(struct dentry *dentry) @@ -314,14 +315,14 @@ struct posix_acl *ovl_get_acl(struct inode *inode, int type) const struct cred *old_cred; struct posix_acl *acl; - if (!IS_POSIXACL(realinode)) + if (!IS_ENABLED(CONFIG_FS_POSIX_ACL) || !IS_POSIXACL(realinode)) return NULL; if (!realinode->i_op->get_acl) return NULL; old_cred = ovl_override_creds(inode->i_sb); - acl = realinode->i_op->get_acl(realinode, type); + acl = get_acl(realinode, type); revert_creds(old_cred); return acl; -- cgit v1.2.3-70-g09d2 From 2a3a2a3f35249412e35fbb48b743348c40373409 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:11:59 +0200 Subject: ovl: don't cache acl on overlay layer Some operations (setxattr/chmod) can make the cached acl stale. We either need to clear overlay's acl cache for the affected inode or prevent acl caching on the overlay altogether. Preventing caching has the following advantages: - no double caching, less memory used - overlay cache doesn't go stale when fs clears it's own cache Possible disadvantage is performance loss. If that becomes a problem get_acl() can be optimized for overlayfs. This patch disables caching by pre setting i_*acl to a value that - has bit 0 set, so is_uncached_acl() will return true - is not equal to ACL_NOT_CACHED, so get_acl() will not overwrite it The constant -3 was chosen for this purpose. Fixes: 39a25b2b3762 ("ovl: define ->get_acl() for overlay inodes") Signed-off-by: Miklos Szeredi --- fs/overlayfs/inode.c | 3 +++ include/linux/fs.h | 1 + 2 files changed, 4 insertions(+) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index d50d1ead1b6f..47a4f33df47b 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -416,6 +416,9 @@ static void ovl_fill_inode(struct inode *inode, umode_t mode) inode->i_ino = get_next_ino(); inode->i_mode = mode; inode->i_flags |= S_NOCMTIME; +#ifdef CONFIG_FS_POSIX_ACL + inode->i_acl = inode->i_default_acl = ACL_DONT_CACHE; +#endif mode &= S_IFMT; switch (mode) { diff --git a/include/linux/fs.h b/include/linux/fs.h index 3523bf62f328..901e25d495cc 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -574,6 +574,7 @@ static inline void mapping_allow_writable(struct address_space *mapping) struct posix_acl; #define ACL_NOT_CACHED ((void *)(-1)) +#define ACL_DONT_CACHE ((void *)(-3)) static inline struct posix_acl * uncached_acl_sentinel(struct task_struct *task) -- cgit v1.2.3-70-g09d2 From fd36570a8805f39b40a0ebde19b08603aa201d17 Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Thu, 18 Aug 2016 16:58:35 +0100 Subject: ovl: fix spelling mistake: "directries" -> "directories" Trivial fix to spelling mistake in pr_err message. Signed-off-by: Colin Ian King Signed-off-by: Miklos Szeredi --- fs/overlayfs/super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index 219534e5ca0b..6aad7d4e2601 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -1151,7 +1151,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent) err = -EINVAL; stacklen = ovl_split_lowerdirs(lowertmp); if (stacklen > OVL_MAX_STACK) { - pr_err("overlayfs: too many lower directries, limit is %d\n", + pr_err("overlayfs: too many lower directories, limit is %d\n", OVL_MAX_STACK); goto out_free_lowertmp; } else if (!ufs->config.upperdir && stacklen == 1) { -- cgit v1.2.3-70-g09d2 From fe2b75952347762a21f67d9df1199137ae5988b2 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Mon, 22 Aug 2016 17:59:22 +0200 Subject: ovl: Fix OVL_XATTR_PREFIX Make sure ovl_own_xattr_handler only matches attribute names starting with "overlay.", not "overlayXXX". Signed-off-by: Andreas Gruenbacher Fixes: d837a49bd57f ("ovl: fix POSIX ACL setting") Signed-off-by: Miklos Szeredi --- fs/overlayfs/inode.c | 5 ++--- fs/overlayfs/overlayfs.h | 4 ++-- 2 files changed, 4 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 47a4f33df47b..f523511b324f 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -194,9 +194,8 @@ static int ovl_readlink(struct dentry *dentry, char __user *buf, int bufsiz) bool ovl_is_private_xattr(const char *name) { -#define OVL_XATTR_PRE_NAME OVL_XATTR_PREFIX "." - return strncmp(name, OVL_XATTR_PRE_NAME, - sizeof(OVL_XATTR_PRE_NAME) - 1) == 0; + return strncmp(name, OVL_XATTR_PREFIX, + sizeof(OVL_XATTR_PREFIX) - 1) == 0; } int ovl_setxattr(struct dentry *dentry, struct inode *inode, diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index 9a95e2c5653e..f50c390683a3 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -24,8 +24,8 @@ enum ovl_path_type { (OVL_TYPE_MERGE(type) || !OVL_TYPE_UPPER(type)) -#define OVL_XATTR_PREFIX XATTR_TRUSTED_PREFIX "overlay" -#define OVL_XATTR_OPAQUE OVL_XATTR_PREFIX ".opaque" +#define OVL_XATTR_PREFIX XATTR_TRUSTED_PREFIX "overlay." +#define OVL_XATTR_OPAQUE OVL_XATTR_PREFIX "opaque" #define OVL_ISUPPER_MASK 1UL -- cgit v1.2.3-70-g09d2 From 0c97be22f928b85110504c4bbb8574facb4bd0c0 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Mon, 22 Aug 2016 16:36:49 +0200 Subject: ovl: Get rid of ovl_xattr_noacl_handlers array Use an ordinary #ifdef to conditionally include the POSIX ACL handlers in ovl_xattr_handlers, like the other filesystems do. Flag the code that is now only used conditionally with __maybe_unused. Signed-off-by: Andreas Gruenbacher Signed-off-by: Miklos Szeredi --- fs/overlayfs/super.c | 28 ++++++++++++---------------- 1 file changed, 12 insertions(+), 16 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index 6aad7d4e2601..c35619195385 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -986,10 +986,11 @@ static unsigned int ovl_split_lowerdirs(char *str) return ctr; } -static int ovl_posix_acl_xattr_set(const struct xattr_handler *handler, - struct dentry *dentry, struct inode *inode, - const char *name, const void *value, - size_t size, int flags) +static int __maybe_unused +ovl_posix_acl_xattr_set(const struct xattr_handler *handler, + struct dentry *dentry, struct inode *inode, + const char *name, const void *value, + size_t size, int flags) { struct dentry *workdir = ovl_workdir(dentry); struct inode *realinode = ovl_inode_real(inode, NULL); @@ -1040,13 +1041,15 @@ static int ovl_own_xattr_set(const struct xattr_handler *handler, return -EPERM; } -static const struct xattr_handler ovl_posix_acl_access_xattr_handler = { +static const struct xattr_handler __maybe_unused +ovl_posix_acl_access_xattr_handler = { .name = XATTR_NAME_POSIX_ACL_ACCESS, .flags = ACL_TYPE_ACCESS, .set = ovl_posix_acl_xattr_set, }; -static const struct xattr_handler ovl_posix_acl_default_xattr_handler = { +static const struct xattr_handler __maybe_unused +ovl_posix_acl_default_xattr_handler = { .name = XATTR_NAME_POSIX_ACL_DEFAULT, .flags = ACL_TYPE_DEFAULT, .set = ovl_posix_acl_xattr_set, @@ -1063,19 +1066,15 @@ static const struct xattr_handler ovl_other_xattr_handler = { }; static const struct xattr_handler *ovl_xattr_handlers[] = { +#ifdef CONFIG_FS_POSIX_ACL &ovl_posix_acl_access_xattr_handler, &ovl_posix_acl_default_xattr_handler, +#endif &ovl_own_xattr_handler, &ovl_other_xattr_handler, NULL }; -static const struct xattr_handler *ovl_xattr_noacl_handlers[] = { - &ovl_own_xattr_handler, - &ovl_other_xattr_handler, - NULL, -}; - static int ovl_fill_super(struct super_block *sb, void *data, int silent) { struct path upperpath = { NULL, NULL }; @@ -1288,10 +1287,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent) sb->s_magic = OVERLAYFS_SUPER_MAGIC; sb->s_op = &ovl_super_operations; - if (IS_ENABLED(CONFIG_FS_POSIX_ACL)) - sb->s_xattr = ovl_xattr_handlers; - else - sb->s_xattr = ovl_xattr_noacl_handlers; + sb->s_xattr = ovl_xattr_handlers; sb->s_root = root_dentry; sb->s_fs_info = ufs; sb->s_flags |= MS_POSIXACL; -- cgit v1.2.3-70-g09d2 From 0e585ccc13b3edbb187fb4f1b7cc9397f17d64a9 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Mon, 22 Aug 2016 17:22:11 +0200 Subject: ovl: Switch to generic_removexattr Commit d837a49bd57f ("ovl: fix POSIX ACL setting") switches from iop->setxattr from ovl_setxattr to generic_setxattr, so switch from ovl_removexattr to generic_removexattr as well. As far as permission checking goes, the same rules should apply in either case. While doing that, rename ovl_setxattr to ovl_xattr_set to indicate that this is not an iop->setxattr implementation and remove the unused inode argument. Move ovl_other_xattr_set above ovl_own_xattr_set so that they match the order of handlers in ovl_xattr_handlers. Signed-off-by: Andreas Gruenbacher Fixes: d837a49bd57f ("ovl: fix POSIX ACL setting") Signed-off-by: Miklos Szeredi --- fs/overlayfs/dir.c | 2 +- fs/overlayfs/inode.c | 65 ++++++++++++++++-------------------------------- fs/overlayfs/overlayfs.h | 6 ++--- fs/overlayfs/super.c | 18 +++++++------- 4 files changed, 33 insertions(+), 58 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c index f485dd4288e4..791c6a209656 100644 --- a/fs/overlayfs/dir.c +++ b/fs/overlayfs/dir.c @@ -1006,7 +1006,7 @@ const struct inode_operations ovl_dir_inode_operations = { .setxattr = generic_setxattr, .getxattr = ovl_getxattr, .listxattr = ovl_listxattr, - .removexattr = ovl_removexattr, + .removexattr = generic_removexattr, .get_acl = ovl_get_acl, .update_time = ovl_update_time, }; diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index f523511b324f..94bca710e6d2 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -198,25 +198,38 @@ bool ovl_is_private_xattr(const char *name) sizeof(OVL_XATTR_PREFIX) - 1) == 0; } -int ovl_setxattr(struct dentry *dentry, struct inode *inode, - const char *name, const void *value, - size_t size, int flags) +int ovl_xattr_set(struct dentry *dentry, const char *name, const void *value, + size_t size, int flags) { int err; - struct dentry *upperdentry; + struct path realpath; + enum ovl_path_type type = ovl_path_real(dentry, &realpath); const struct cred *old_cred; err = ovl_want_write(dentry); if (err) goto out; + if (!value && !OVL_TYPE_UPPER(type)) { + err = vfs_getxattr(realpath.dentry, name, NULL, 0); + if (err < 0) + goto out_drop_write; + } + err = ovl_copy_up(dentry); if (err) goto out_drop_write; - upperdentry = ovl_dentry_upper(dentry); + if (!OVL_TYPE_UPPER(type)) + ovl_path_upper(dentry, &realpath); + old_cred = ovl_override_creds(dentry->d_sb); - err = vfs_setxattr(upperdentry, name, value, size, flags); + if (value) + err = vfs_setxattr(realpath.dentry, name, value, size, flags); + else { + WARN_ON(flags != XATTR_REPLACE); + err = vfs_removexattr(realpath.dentry, name); + } revert_creds(old_cred); out_drop_write: @@ -272,42 +285,6 @@ ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size) return res; } -int ovl_removexattr(struct dentry *dentry, const char *name) -{ - int err; - struct path realpath; - enum ovl_path_type type = ovl_path_real(dentry, &realpath); - const struct cred *old_cred; - - err = ovl_want_write(dentry); - if (err) - goto out; - - err = -ENODATA; - if (ovl_is_private_xattr(name)) - goto out_drop_write; - - if (!OVL_TYPE_UPPER(type)) { - err = vfs_getxattr(realpath.dentry, name, NULL, 0); - if (err < 0) - goto out_drop_write; - - err = ovl_copy_up(dentry); - if (err) - goto out_drop_write; - - ovl_path_upper(dentry, &realpath); - } - - old_cred = ovl_override_creds(dentry->d_sb); - err = vfs_removexattr(realpath.dentry, name); - revert_creds(old_cred); -out_drop_write: - ovl_drop_write(dentry); -out: - return err; -} - struct posix_acl *ovl_get_acl(struct inode *inode, int type) { struct inode *realinode = ovl_inode_real(inode, NULL); @@ -393,7 +370,7 @@ static const struct inode_operations ovl_file_inode_operations = { .setxattr = generic_setxattr, .getxattr = ovl_getxattr, .listxattr = ovl_listxattr, - .removexattr = ovl_removexattr, + .removexattr = generic_removexattr, .get_acl = ovl_get_acl, .update_time = ovl_update_time, }; @@ -406,7 +383,7 @@ static const struct inode_operations ovl_symlink_inode_operations = { .setxattr = generic_setxattr, .getxattr = ovl_getxattr, .listxattr = ovl_listxattr, - .removexattr = ovl_removexattr, + .removexattr = generic_removexattr, .update_time = ovl_update_time, }; diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index f50c390683a3..5769aaf151a3 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -185,13 +185,11 @@ void ovl_workdir_cleanup(struct inode *dir, struct vfsmount *mnt, /* inode.c */ int ovl_setattr(struct dentry *dentry, struct iattr *attr); int ovl_permission(struct inode *inode, int mask); -int ovl_setxattr(struct dentry *dentry, struct inode *inode, - const char *name, const void *value, - size_t size, int flags); +int ovl_xattr_set(struct dentry *dentry, const char *name, const void *value, + size_t size, int flags); ssize_t ovl_getxattr(struct dentry *dentry, struct inode *inode, const char *name, void *value, size_t size); ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size); -int ovl_removexattr(struct dentry *dentry, const char *name); struct posix_acl *ovl_get_acl(struct inode *inode, int type); int ovl_open_maybe_copy_up(struct dentry *dentry, unsigned int file_flags); int ovl_update_time(struct inode *inode, struct timespec *ts, int flags); diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index c35619195385..45a2eb0b4693 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -1018,21 +1018,13 @@ ovl_posix_acl_xattr_set(const struct xattr_handler *handler, posix_acl_release(acl); - return ovl_setxattr(dentry, inode, handler->name, value, size, flags); + return ovl_xattr_set(dentry, handler->name, value, size, flags); out_acl_release: posix_acl_release(acl); return err; } -static int ovl_other_xattr_set(const struct xattr_handler *handler, - struct dentry *dentry, struct inode *inode, - const char *name, const void *value, - size_t size, int flags) -{ - return ovl_setxattr(dentry, inode, name, value, size, flags); -} - static int ovl_own_xattr_set(const struct xattr_handler *handler, struct dentry *dentry, struct inode *inode, const char *name, const void *value, @@ -1041,6 +1033,14 @@ static int ovl_own_xattr_set(const struct xattr_handler *handler, return -EPERM; } +static int ovl_other_xattr_set(const struct xattr_handler *handler, + struct dentry *dentry, struct inode *inode, + const char *name, const void *value, + size_t size, int flags) +{ + return ovl_xattr_set(dentry, name, value, size, flags); +} + static const struct xattr_handler __maybe_unused ovl_posix_acl_access_xattr_handler = { .name = XATTR_NAME_POSIX_ACL_ACCESS, -- cgit v1.2.3-70-g09d2 From ce31513a9114f74fe3e9caa6534d201bdac7238d Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:12:00 +0200 Subject: ovl: copyattr after setting POSIX ACL Setting POSIX acl may also modify the file mode, so need to copy that up to the overlay inode. Reported-by: Eryu Guan Fixes: d837a49bd57f ("ovl: fix POSIX ACL setting") Signed-off-by: Miklos Szeredi --- fs/overlayfs/super.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index 45a2eb0b4693..cba2c9fea98c 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -1018,7 +1018,11 @@ ovl_posix_acl_xattr_set(const struct xattr_handler *handler, posix_acl_release(acl); - return ovl_xattr_set(dentry, handler->name, value, size, flags); + err = ovl_xattr_set(dentry, handler->name, value, size, flags); + if (!err) + ovl_copyattr(ovl_inode_real(inode, NULL), inode); + + return err; out_acl_release: posix_acl_release(acl); -- cgit v1.2.3-70-g09d2 From 0eb45fc3bb7a2cf9c9c93d9e95986a841e5f4625 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Mon, 22 Aug 2016 17:52:55 +0200 Subject: ovl: Switch to generic_getxattr Now that overlayfs has xattr handlers for iop->{set,remove}xattr, use those same handlers for iop->getxattr as well. Signed-off-by: Andreas Gruenbacher Signed-off-by: Miklos Szeredi --- fs/overlayfs/dir.c | 2 +- fs/overlayfs/inode.c | 11 ++++------- fs/overlayfs/overlayfs.h | 4 ++-- fs/overlayfs/super.c | 26 ++++++++++++++++++++++++++ 4 files changed, 33 insertions(+), 10 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c index 791c6a209656..1560fdc09a5f 100644 --- a/fs/overlayfs/dir.c +++ b/fs/overlayfs/dir.c @@ -1004,7 +1004,7 @@ const struct inode_operations ovl_dir_inode_operations = { .permission = ovl_permission, .getattr = ovl_dir_getattr, .setxattr = generic_setxattr, - .getxattr = ovl_getxattr, + .getxattr = generic_getxattr, .listxattr = ovl_listxattr, .removexattr = generic_removexattr, .get_acl = ovl_get_acl, diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 94bca710e6d2..1878591f6a2d 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -238,16 +238,13 @@ out: return err; } -ssize_t ovl_getxattr(struct dentry *dentry, struct inode *inode, - const char *name, void *value, size_t size) +int ovl_xattr_get(struct dentry *dentry, const char *name, + void *value, size_t size) { struct dentry *realdentry = ovl_dentry_real(dentry); ssize_t res; const struct cred *old_cred; - if (ovl_is_private_xattr(name)) - return -ENODATA; - old_cred = ovl_override_creds(dentry->d_sb); res = vfs_getxattr(realdentry, name, value, size); revert_creds(old_cred); @@ -368,7 +365,7 @@ static const struct inode_operations ovl_file_inode_operations = { .permission = ovl_permission, .getattr = ovl_getattr, .setxattr = generic_setxattr, - .getxattr = ovl_getxattr, + .getxattr = generic_getxattr, .listxattr = ovl_listxattr, .removexattr = generic_removexattr, .get_acl = ovl_get_acl, @@ -381,7 +378,7 @@ static const struct inode_operations ovl_symlink_inode_operations = { .readlink = ovl_readlink, .getattr = ovl_getattr, .setxattr = generic_setxattr, - .getxattr = ovl_getxattr, + .getxattr = generic_getxattr, .listxattr = ovl_listxattr, .removexattr = generic_removexattr, .update_time = ovl_update_time, diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index 5769aaf151a3..5813ccff8cd9 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -187,8 +187,8 @@ int ovl_setattr(struct dentry *dentry, struct iattr *attr); int ovl_permission(struct inode *inode, int mask); int ovl_xattr_set(struct dentry *dentry, const char *name, const void *value, size_t size, int flags); -ssize_t ovl_getxattr(struct dentry *dentry, struct inode *inode, - const char *name, void *value, size_t size); +int ovl_xattr_get(struct dentry *dentry, const char *name, + void *value, size_t size); ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size); struct posix_acl *ovl_get_acl(struct inode *inode, int type); int ovl_open_maybe_copy_up(struct dentry *dentry, unsigned int file_flags); diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index cba2c9fea98c..a4585f961bf9 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -986,6 +986,14 @@ static unsigned int ovl_split_lowerdirs(char *str) return ctr; } +static int __maybe_unused +ovl_posix_acl_xattr_get(const struct xattr_handler *handler, + struct dentry *dentry, struct inode *inode, + const char *name, void *buffer, size_t size) +{ + return ovl_xattr_get(dentry, handler->name, buffer, size); +} + static int __maybe_unused ovl_posix_acl_xattr_set(const struct xattr_handler *handler, struct dentry *dentry, struct inode *inode, @@ -1029,6 +1037,13 @@ out_acl_release: return err; } +static int ovl_own_xattr_get(const struct xattr_handler *handler, + struct dentry *dentry, struct inode *inode, + const char *name, void *buffer, size_t size) +{ + return -EPERM; +} + static int ovl_own_xattr_set(const struct xattr_handler *handler, struct dentry *dentry, struct inode *inode, const char *name, const void *value, @@ -1037,6 +1052,13 @@ static int ovl_own_xattr_set(const struct xattr_handler *handler, return -EPERM; } +static int ovl_other_xattr_get(const struct xattr_handler *handler, + struct dentry *dentry, struct inode *inode, + const char *name, void *buffer, size_t size) +{ + return ovl_xattr_get(dentry, name, buffer, size); +} + static int ovl_other_xattr_set(const struct xattr_handler *handler, struct dentry *dentry, struct inode *inode, const char *name, const void *value, @@ -1049,6 +1071,7 @@ static const struct xattr_handler __maybe_unused ovl_posix_acl_access_xattr_handler = { .name = XATTR_NAME_POSIX_ACL_ACCESS, .flags = ACL_TYPE_ACCESS, + .get = ovl_posix_acl_xattr_get, .set = ovl_posix_acl_xattr_set, }; @@ -1056,16 +1079,19 @@ static const struct xattr_handler __maybe_unused ovl_posix_acl_default_xattr_handler = { .name = XATTR_NAME_POSIX_ACL_DEFAULT, .flags = ACL_TYPE_DEFAULT, + .get = ovl_posix_acl_xattr_get, .set = ovl_posix_acl_xattr_set, }; static const struct xattr_handler ovl_own_xattr_handler = { .prefix = OVL_XATTR_PREFIX, + .get = ovl_own_xattr_get, .set = ovl_own_xattr_set, }; static const struct xattr_handler ovl_other_xattr_handler = { .prefix = "", /* catch all */ + .get = ovl_other_xattr_get, .set = ovl_other_xattr_set, }; -- cgit v1.2.3-70-g09d2 From 7cb35119d067191ce9ebc380a599db0b03cbd9d9 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Thu, 1 Sep 2016 11:12:00 +0200 Subject: ovl: listxattr: use strnlen() Be defensive about what underlying fs provides us in the returned xattr list buffer. If it's not properly null terminated, bail out with a warning insead of BUG. Signed-off-by: Miklos Szeredi Cc: --- fs/overlayfs/inode.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 1878591f6a2d..c75625c1efa3 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -255,7 +255,8 @@ ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size) { struct dentry *realdentry = ovl_dentry_real(dentry); ssize_t res; - int off; + size_t len; + char *s; const struct cred *old_cred; old_cred = ovl_override_creds(dentry->d_sb); @@ -265,17 +266,19 @@ ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size) return res; /* filter out private xattrs */ - for (off = 0; off < res;) { - char *s = list + off; - size_t slen = strlen(s) + 1; + for (s = list, len = res; len;) { + size_t slen = strnlen(s, len) + 1; - BUG_ON(off + slen > res); + /* underlying fs providing us with an broken xattr list? */ + if (WARN_ON(slen > len)) + return -EIO; + len -= slen; if (ovl_is_private_xattr(s)) { res -= slen; - memmove(s, s + slen, res - off); + memmove(s, s + slen, len); } else { - off += slen; + s += slen; } } -- cgit v1.2.3-70-g09d2 From 3dc09ec895f098cedd789a620c90ff1bf7f779a1 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 24 Aug 2016 11:57:52 -0400 Subject: Btrfs: kill invalid ASSERT() in process_all_refs() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Suppose you have the following tree in snap1 on a file system mounted with -o inode_cache so that inode numbers are recycled └── [ 258] a └── [ 257] b and then you remove b, rename a to c, and then re-create b in c so you have the following tree └── [ 258] c └── [ 257] b and then you try to do an incremental send you will hit ASSERT(pending_move == 0); in process_all_refs(). This is because we assume that any recycling of inodes will not have a pending change in our path, which isn't the case. This is the case for the DELETE side, since we want to remove the old file using the old path, but on the create side we could have a pending move and need to do the normal pending rename dance. So remove this ASSERT() and put a comment about why we ignore pending_move. Thanks, Signed-off-by: Josef Bacik Signed-off-by: David Sterba --- fs/btrfs/send.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c index efe129fe2678..a87675ffd02b 100644 --- a/fs/btrfs/send.c +++ b/fs/btrfs/send.c @@ -4268,10 +4268,12 @@ static int process_all_refs(struct send_ctx *sctx, } btrfs_release_path(path); + /* + * We don't actually care about pending_move as we are simply + * re-creating this inode and will be rename'ing it into place once we + * rename the parent directory. + */ ret = process_recorded_refs(sctx, &pending_move); - /* Only applicable to an incremental send. */ - ASSERT(pending_move == 0); - out: btrfs_free_path(path); return ret; -- cgit v1.2.3-70-g09d2 From a9b1fc851db054ddec703dc7951ed00620600b26 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Wed, 31 Aug 2016 16:43:33 -0700 Subject: Btrfs: fix endless loop in balancing block groups Qgroup function may overwrite the saved error 'err' with 0 in case quota is not enabled, and this ends up with a endless loop in balance because we keep going back to balance the same block group. It really should use 'ret' instead. Signed-off-by: Liu Bo Reviewed-by: Qu Wenruo Signed-off-by: David Sterba --- fs/btrfs/relocation.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 8a2c2a07987b..c0c13dc6fe12 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -4200,9 +4200,11 @@ restart: err = PTR_ERR(trans); goto out_free; } - err = qgroup_fix_relocated_data_extents(trans, rc); - if (err < 0) { - btrfs_abort_transaction(trans, err); + ret = qgroup_fix_relocated_data_extents(trans, rc); + if (ret < 0) { + btrfs_abort_transaction(trans, ret); + if (!err) + err = ret; goto out_free; } btrfs_commit_transaction(trans, rc->extent_root); -- cgit v1.2.3-70-g09d2 From e0af24849efb0eea572cf22d22bb65d164cb8a6f Mon Sep 17 00:00:00 2001 From: Wang Xiaoguang Date: Wed, 31 Aug 2016 19:46:16 +0800 Subject: btrfs: fix one bug that process may endlessly wait for ticket in wait_reserve_ticket() If can_overcommit() in btrfs_calc_reclaim_metadata_size() returns true, btrfs_async_reclaim_metadata_space() will not reclaim metadata space, just return directly and also forget to wake up process which are waiting for their tickets, so these processes will wait endlessly. Fstests case generic/172 with mount option "-o compress=lzo" have revealed this bug in my test machine. Here if we have tickets to handle, we must handle them first. Signed-off-by: Wang Xiaoguang Reviewed-by: Josef Bacik Signed-off-by: David Sterba --- fs/btrfs/extent-tree.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 60d4ae7ce974..64676a16d32a 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4901,11 +4901,6 @@ btrfs_calc_reclaim_metadata_size(struct btrfs_root *root, u64 expected; u64 to_reclaim = 0; - to_reclaim = min_t(u64, num_online_cpus() * SZ_1M, SZ_16M); - if (can_overcommit(root, space_info, to_reclaim, - BTRFS_RESERVE_FLUSH_ALL)) - return 0; - list_for_each_entry(ticket, &space_info->tickets, list) to_reclaim += ticket->bytes; list_for_each_entry(ticket, &space_info->priority_tickets, list) @@ -4913,6 +4908,11 @@ btrfs_calc_reclaim_metadata_size(struct btrfs_root *root, if (to_reclaim) return to_reclaim; + to_reclaim = min_t(u64, num_online_cpus() * SZ_1M, SZ_16M); + if (can_overcommit(root, space_info, to_reclaim, + BTRFS_RESERVE_FLUSH_ALL)) + return 0; + used = space_info->bytes_used + space_info->bytes_reserved + space_info->bytes_pinned + space_info->bytes_readonly + space_info->bytes_may_use; -- cgit v1.2.3-70-g09d2 From 3e423945ea94412283eaba8bfbe9d6e0a80b434f Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sat, 3 Sep 2016 11:02:50 -0700 Subject: devpts: return NULL pts 'priv' entry for non-devpts nodes In commit 8ead9dd54716 ("devpts: more pty driver interface cleanups") I made devpts_get_priv() just return the dentry->fs_data directly. And because I thought it wouldn't happen, I added a warning if you ever saw a pts node that wasn't on devpts. And no, that warning never triggered under any actual real use, but you can trigger it by creating nonsensical pts nodes by hand. So just revert the warning, and make devpts_get_priv() return NULL for that case like it used to. Reported-by: Dmitry Vyukov Cc: stable@vger.kernel.org # 4.6+ Cc: Eric W Biederman" Signed-off-by: Linus Torvalds --- fs/devpts/inode.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/devpts/inode.c b/fs/devpts/inode.c index d116453b0276..79a5941c2474 100644 --- a/fs/devpts/inode.c +++ b/fs/devpts/inode.c @@ -585,7 +585,8 @@ struct dentry *devpts_pty_new(struct pts_fs_info *fsi, int index, void *priv) */ void *devpts_get_priv(struct dentry *dentry) { - WARN_ON_ONCE(dentry->d_sb->s_magic != DEVPTS_SUPER_MAGIC); + if (dentry->d_sb->s_magic != DEVPTS_SUPER_MAGIC) + return NULL; return dentry->d_fsdata; } -- cgit v1.2.3-70-g09d2