diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2019-12-03 13:58:22 -0800 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2019-12-03 13:58:22 -0800 |
commit | c3bed3b20e40ab44b98ac5f0471a5bd92a802f5a (patch) | |
tree | 73813bd3b2a851917f6f6a4dcbed22c305bef3b4 /drivers/pci/controller/pci-hyperv.c | |
parent | 4672c2f2026dbc46b20647803426561fe2d460ea (diff) | |
parent | 7e124c40517218e079e580909de2652bddb60ff5 (diff) |
Merge tag 'pci-v5.5-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull PCI updates from Bjorn Helgaas:
"Enumeration:
- Warn if a host bridge has no NUMA info (Yunsheng Lin)
- Add PCI_STD_NUM_BARS for the number of standard BARs (Denis
Efremov)
Resource management:
- Fix boot-time Embedded Controller GPE storm caused by incorrect
resource assignment after ACPI Bus Check Notification (Mika
Westerberg)
- Protect pci_reassign_bridge_resources() against concurrent
addition/removal (Benjamin Herrenschmidt)
- Fix bridge dma_ranges resource list cleanup (Rob Herring)
- Add "pci=hpmmiosize" and "pci=hpmmioprefsize" parameters to control
the MMIO and prefetchable MMIO window sizes of hotplug bridges
independently (Nicholas Johnson)
- Fix MMIO/MMIO_PREF window assignment that assigned more space than
desired (Nicholas Johnson)
- Only enforce bus numbers from bridge EA if the bridge has EA
devices downstream (Subbaraya Sundeep)
- Consolidate DT "dma-ranges" parsing and convert all host drivers to
use shared parsing (Rob Herring)
Error reporting:
- Restore AER capability after resume (Mayurkumar Patel)
- Add PoisonTLPBlocked AER counter (Rajat Jain)
- Use for_each_set_bit() to simplify AER code (Andy Shevchenko)
- Fix AER kernel-doc (Andy Shevchenko)
- Add "pcie_ports=dpc-native" parameter to allow native use of DPC
even if platform didn't grant control over AER (Olof Johansson)
Hotplug:
- Avoid returning prematurely from sysfs requests to enable or
disable a PCIe hotplug slot (Lukas Wunner)
- Don't disable interrupts twice when suspending hotplug ports (Mika
Westerberg)
- Fix deadlocks when PCIe ports are hot-removed while suspended (Mika
Westerberg)
Power management:
- Remove unnecessary ASPM locking (Bjorn Helgaas)
- Add support for disabling L1 PM Substates (Heiner Kallweit)
- Allow re-enabling Clock PM after it has been disabled (Heiner
Kallweit)
- Add sysfs attributes for controlling ASPM link states (Heiner
Kallweit)
- Remove CONFIG_PCIEASPM_DEBUG, including "link_state" and "clk_ctl"
sysfs files (Heiner Kallweit)
- Avoid AMD FCH XHCI USB PME# from D0 defect that prevents wakeup on
USB 2.0 or 1.1 connect events (Kai-Heng Feng)
- Move power state check out of pci_msi_supported() (Bjorn Helgaas)
- Fix incorrect MSI-X masking on resume and revert related nvme quirk
for Kingston NVME SSD running FW E8FK11.T (Jian-Hong Pan)
- Always return devices to D0 when thawing to fix hibernation with
drivers like mlx4 that used legacy power management (previously we
only did it for drivers with new power management ops) (Dexuan Cui)
- Clear PCIe PME Status even for legacy power management (Bjorn
Helgaas)
- Fix PCI PM documentation errors (Bjorn Helgaas)
- Use dev_printk() for more power management messages (Bjorn Helgaas)
- Apply D2 delay as milliseconds, not microseconds (Bjorn Helgaas)
- Convert xen-platform from legacy to generic power management (Bjorn
Helgaas)
- Removed unused .resume_early() and .suspend_late() legacy power
management hooks (Bjorn Helgaas)
- Rearrange power management code for clarity (Rafael J. Wysocki)
- Decode power states more clearly ("4" or "D4" really refers to
"D3cold") (Bjorn Helgaas)
- Notice when reading PM Control register returns an error (~0)
instead of interpreting it as being in D3hot (Bjorn Helgaas)
- Add missing link delays required by the PCIe spec (Mika Westerberg)
Virtualization:
- Move pci_prg_resp_pasid_required() to CONFIG_PCI_PRI (Bjorn
Helgaas)
- Allow VFs to use PRI (the PF PRI is shared by the VFs, but the code
previously didn't recognize that) (Kuppuswamy Sathyanarayanan)
- Allow VFs to use PASID (the PF PASID capability is shared by the
VFs, but the code previously didn't recognize that) (Kuppuswamy
Sathyanarayanan)
- Disconnect PF and VF ATS enablement, since ATS in PFs and
associated VFs can be enabled independently (Kuppuswamy
Sathyanarayanan)
- Cache PRI and PASID capability offsets (Kuppuswamy Sathyanarayanan)
- Cache the PRI PRG Response PASID Required bit (Bjorn Helgaas)
- Consolidate ATS declarations in linux/pci-ats.h (Krzysztof
Wilczynski)
- Remove unused PRI and PASID stubs (Bjorn Helgaas)
- Removed unnecessary EXPORT_SYMBOL_GPL() from ATS, PRI, and PASID
interfaces that are only used by built-in IOMMU drivers (Bjorn
Helgaas)
- Hide PRI and PASID state restoration functions used only inside the
PCI core (Bjorn Helgaas)
- Add a DMA alias quirk for the Intel VCA NTB (Slawomir Pawlowski)
- Serialize sysfs sriov_numvfs reads vs writes (Pierre Crégut)
- Update Cavium ACS quirk for ThunderX2 and ThunderX3 (George
Cherian)
- Fix the UPDCR register address in the Intel ACS quirk (Steffen
Liebergeld)
- Unify ACS quirk implementations (Bjorn Helgaas)
Amlogic Meson host bridge driver:
- Fix meson PERST# GPIO polarity problem (Remi Pommarel)
- Add DT bindings for Amlogic Meson G12A (Neil Armstrong)
- Fix meson clock names to match DT bindings (Neil Armstrong)
- Add meson support for Amlogic G12A SoC with separate shared PHY
(Neil Armstrong)
- Add meson extended PCIe PHY functions for Amlogic G12A USB3+PCIe
combo PHY (Neil Armstrong)
- Add arm64 DT for Amlogic G12A PCIe controller node (Neil Armstrong)
- Add commented-out description of VIM3 USB3/PCIe mux in arm64 DT
(Neil Armstrong)
Broadcom iProc host bridge driver:
- Invalidate iProc PAXB address mapping before programming it
(Abhishek Shah)
- Fix iproc-msi and mvebu __iomem annotations (Ben Dooks)
Cadence host bridge driver:
- Refactor Cadence PCIe host controller to use as a library for both
host and endpoint (Tom Joseph)
Freescale Layerscape host bridge driver:
- Add layerscape LS1028a support (Xiaowei Bao)
Intel VMD host bridge driver:
- Add VMD bus 224-255 restriction decode (Jon Derrick)
- Add VMD 8086:9A0B device ID (Jon Derrick)
- Remove Keith from VMD maintainer list (Keith Busch)
Marvell ARMADA 3700 / Aardvark host bridge driver:
- Use LTSSM state to build link training flag since Aardvark doesn't
implement the Link Training bit (Remi Pommarel)
- Delay before training Aardvark link in case PERST# was asserted
before the driver probe (Remi Pommarel)
- Fix Aardvark issues with Root Control reads and writes (Remi
Pommarel)
- Don't rely on jiffies in Aardvark config access path since
interrupts may be disabled (Remi Pommarel)
- Fix Aardvark big-endian support (Grzegorz Jaszczyk)
Marvell ARMADA 370 / XP host bridge driver:
- Make mvebu_pci_bridge_emul_ops static (Ben Dooks)
Microsoft Hyper-V host bridge driver:
- Add hibernation support for Hyper-V virtual PCI devices (Dexuan
Cui)
- Track Hyper-V pci_protocol_version per-hbus, not globally (Dexuan
Cui)
- Avoid kmemleak false positive on hv hbus buffer (Dexuan Cui)
Mobiveil host bridge driver:
- Change mobiveil csr_read()/write() function names that conflict
with riscv arch functions (Kefeng Wang)
NVIDIA Tegra host bridge driver:
- Fix Tegra CLKREQ dependency programming (Vidya Sagar)
Renesas R-Car host bridge driver:
- Remove unnecessary header include from rcar (Andrew Murray)
- Tighten register index checking for rcar inbound range programming
(Marek Vasut)
- Fix rcar inbound range alignment calculation to improve packing of
multiple entries (Marek Vasut)
- Update rcar MACCTLR setting to match documentation (Yoshihiro
Shimoda)
- Clear bit 0 of MACCTLR before PCIETCTLR.CFINIT per manual
(Yoshihiro Shimoda)
- Add Marek Vasut and Yoshihiro Shimoda as R-Car maintainers (Simon
Horman)
Rockchip host bridge driver:
- Make rockchip 0V9 and 1V8 power regulators non-optional (Robin
Murphy)
Socionext UniPhier host bridge driver:
- Set uniphier to host (RC) mode always (Kunihiko Hayashi)
Endpoint drivers:
- Fix endpoint driver sign extension problem when shifting page
number to phys_addr_t (Alan Mikhak)
Misc:
- Add NumaChip SPDX header (Krzysztof Wilczynski)
- Replace EXTRA_CFLAGS with ccflags-y (Krzysztof Wilczynski)
- Remove unused includes (Krzysztof Wilczynski)
- Removed unused sysfs attribute groups (Ben Dooks)
- Remove PTM and ASPM dependencies on PCIEPORTBUS (Bjorn Helgaas)
- Add PCIe Link Control 2 register field definitions to replace magic
numbers in AMDGPU and Radeon CIK/SI (Bjorn Helgaas)
- Fix incorrect Link Control 2 Transmit Margin usage in AMDGPU and
Radeon CIK/SI PCIe Gen3 link training (Bjorn Helgaas)
- Use pcie_capability_read_word() instead of pci_read_config_word()
in AMDGPU and Radeon CIK/SI (Frederick Lawler)
- Remove unused pci_irq_get_node() Greg Kroah-Hartman)
- Make asm/msi.h mandatory and simplify PCI_MSI_IRQ_DOMAIN Kconfig
(Palmer Dabbelt, Michal Simek)
- Read all 64 bits of Switchtec part_event_bitmap (Logan Gunthorpe)
- Fix erroneous intel-iommu dependency on CONFIG_AMD_IOMMU (Bjorn
Helgaas)
- Fix bridge emulation big-endian support (Grzegorz Jaszczyk)
- Fix dwc find_next_bit() usage (Niklas Cassel)
- Fix pcitest.c fd leak (Hewenliang)
- Fix typos and comments (Bjorn Helgaas)
- Fix Kconfig whitespace errors (Krzysztof Kozlowski)"
* tag 'pci-v5.5-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: (160 commits)
PCI: Remove PCI_MSI_IRQ_DOMAIN architecture whitelist
asm-generic: Make msi.h a mandatory include/asm header
Revert "nvme: Add quirk for Kingston NVME SSD running FW E8FK11.T"
PCI/MSI: Fix incorrect MSI-X masking on resume
PCI/MSI: Move power state check out of pci_msi_supported()
PCI/MSI: Remove unused pci_irq_get_node()
PCI: hv: Avoid a kmemleak false positive caused by the hbus buffer
PCI: hv: Change pci_protocol_version to per-hbus
PCI: hv: Add hibernation support
PCI: hv: Reorganize the code in preparation of hibernation
MAINTAINERS: Remove Keith from VMD maintainer
PCI/ASPM: Remove PCIEASPM_DEBUG Kconfig option and related code
PCI/ASPM: Add sysfs attributes for controlling ASPM link states
PCI: Fix indentation
drm/radeon: Prefer pcie_capability_read_word()
drm/radeon: Replace numbers with PCI_EXP_LNKCTL2 definitions
drm/radeon: Correct Transmit Margin masks
drm/amdgpu: Prefer pcie_capability_read_word()
PCI: uniphier: Set mode register to host mode
drm/amdgpu: Replace numbers with PCI_EXP_LNKCTL2 definitions
...
Diffstat (limited to 'drivers/pci/controller/pci-hyperv.c')
-rw-r--r-- | drivers/pci/controller/pci-hyperv.c | 218 |
1 files changed, 184 insertions, 34 deletions
diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c index f1f300218fab..9977abff92fc 100644 --- a/drivers/pci/controller/pci-hyperv.c +++ b/drivers/pci/controller/pci-hyperv.c @@ -76,11 +76,6 @@ static enum pci_protocol_version_t pci_protocol_versions[] = { PCI_PROTOCOL_VERSION_1_1, }; -/* - * Protocol version negotiated by hv_pci_protocol_negotiation(). - */ -static enum pci_protocol_version_t pci_protocol_version; - #define PCI_CONFIG_MMIO_LENGTH 0x2000 #define CFG_PAGE_OFFSET 0x1000 #define CFG_PAGE_SIZE (PCI_CONFIG_MMIO_LENGTH - CFG_PAGE_OFFSET) @@ -307,7 +302,7 @@ struct pci_bus_relations { struct pci_q_res_req_response { struct vmpacket_descriptor hdr; s32 status; /* negative values are failures */ - u32 probed_bar[6]; + u32 probed_bar[PCI_STD_NUM_BARS]; } __packed; struct pci_set_power { @@ -455,12 +450,15 @@ enum hv_pcibus_state { hv_pcibus_init = 0, hv_pcibus_probed, hv_pcibus_installed, + hv_pcibus_removing, hv_pcibus_removed, hv_pcibus_maximum }; struct hv_pcibus_device { struct pci_sysdata sysdata; + /* Protocol version negotiated with the host */ + enum pci_protocol_version_t protocol_version; enum hv_pcibus_state state; refcount_t remove_lock; struct hv_device *hdev; @@ -539,7 +537,7 @@ struct hv_pci_dev { * What would be observed if one wrote 0xFFFFFFFF to a BAR and then * read it back, for each of the BAR offsets within config space. */ - u32 probed_bar[6]; + u32 probed_bar[PCI_STD_NUM_BARS]; }; struct hv_pci_compl { @@ -1224,7 +1222,7 @@ static void hv_irq_unmask(struct irq_data *data) * negative effect (yet?). */ - if (pci_protocol_version >= PCI_PROTOCOL_VERSION_1_2) { + if (hbus->protocol_version >= PCI_PROTOCOL_VERSION_1_2) { /* * PCI_PROTOCOL_VERSION_1_2 supports the VP_SET version of the * HVCALL_RETARGET_INTERRUPT hypercall, which also coincides @@ -1394,7 +1392,7 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) ctxt.pci_pkt.completion_func = hv_pci_compose_compl; ctxt.pci_pkt.compl_ctxt = ∁ - switch (pci_protocol_version) { + switch (hbus->protocol_version) { case PCI_PROTOCOL_VERSION_1_1: size = hv_compose_msi_req_v1(&ctxt.int_pkts.v1, dest, @@ -1610,7 +1608,7 @@ static void survey_child_resources(struct hv_pcibus_device *hbus) * so it's sufficient to just add them up without tracking alignment. */ list_for_each_entry(hpdev, &hbus->children, list_entry) { - for (i = 0; i < 6; i++) { + for (i = 0; i < PCI_STD_NUM_BARS; i++) { if (hpdev->probed_bar[i] & PCI_BASE_ADDRESS_SPACE_IO) dev_err(&hbus->hdev->device, "There's an I/O BAR in this list!\n"); @@ -1681,10 +1679,27 @@ static void prepopulate_bars(struct hv_pcibus_device *hbus) spin_lock_irqsave(&hbus->device_list_lock, flags); + /* + * Clear the memory enable bit, in case it's already set. This occurs + * in the suspend path of hibernation, where the device is suspended, + * resumed and suspended again: see hibernation_snapshot() and + * hibernation_platform_enter(). + * + * If the memory enable bit is already set, Hyper-V sliently ignores + * the below BAR updates, and the related PCI device driver can not + * work, because reading from the device register(s) always returns + * 0xFFFFFFFF. + */ + list_for_each_entry(hpdev, &hbus->children, list_entry) { + _hv_pcifront_read_config(hpdev, PCI_COMMAND, 2, &command); + command &= ~PCI_COMMAND_MEMORY; + _hv_pcifront_write_config(hpdev, PCI_COMMAND, 2, command); + } + /* Pick addresses for the BARs. */ do { list_for_each_entry(hpdev, &hbus->children, list_entry) { - for (i = 0; i < 6; i++) { + for (i = 0; i < PCI_STD_NUM_BARS; i++) { bar_val = hpdev->probed_bar[i]; if (bar_val == 0) continue; @@ -1841,7 +1856,7 @@ static void q_resource_requirements(void *context, struct pci_response *resp, "query resource requirements failed: %x\n", resp->status); } else { - for (i = 0; i < 6; i++) { + for (i = 0; i < PCI_STD_NUM_BARS; i++) { completion->hpdev->probed_bar[i] = q_res_req->probed_bar[i]; } @@ -2107,6 +2122,12 @@ static void hv_pci_devices_present(struct hv_pcibus_device *hbus, unsigned long flags; bool pending_dr; + if (hbus->state == hv_pcibus_removing) { + dev_info(&hbus->hdev->device, + "PCI VMBus BUS_RELATIONS: ignored\n"); + return; + } + dr_wrk = kzalloc(sizeof(*dr_wrk), GFP_NOWAIT); if (!dr_wrk) return; @@ -2223,11 +2244,19 @@ static void hv_eject_device_work(struct work_struct *work) */ static void hv_pci_eject_device(struct hv_pci_dev *hpdev) { + struct hv_pcibus_device *hbus = hpdev->hbus; + struct hv_device *hdev = hbus->hdev; + + if (hbus->state == hv_pcibus_removing) { + dev_info(&hdev->device, "PCI VMBus EJECT: ignored\n"); + return; + } + hpdev->state = hv_pcichild_ejecting; get_pcichild(hpdev); INIT_WORK(&hpdev->wrk, hv_eject_device_work); - get_hvpcibus(hpdev->hbus); - queue_work(hpdev->hbus->wq, &hpdev->wrk); + get_hvpcibus(hbus); + queue_work(hbus->wq, &hpdev->wrk); } /** @@ -2379,8 +2408,11 @@ static void hv_pci_onchannelcallback(void *context) * failing if the host doesn't support the necessary protocol * level. */ -static int hv_pci_protocol_negotiation(struct hv_device *hdev) +static int hv_pci_protocol_negotiation(struct hv_device *hdev, + enum pci_protocol_version_t version[], + int num_version) { + struct hv_pcibus_device *hbus = hv_get_drvdata(hdev); struct pci_version_request *version_req; struct hv_pci_compl comp_pkt; struct pci_packet *pkt; @@ -2403,8 +2435,8 @@ static int hv_pci_protocol_negotiation(struct hv_device *hdev) version_req = (struct pci_version_request *)&pkt->message; version_req->message_type.type = PCI_QUERY_PROTOCOL_VERSION; - for (i = 0; i < ARRAY_SIZE(pci_protocol_versions); i++) { - version_req->protocol_version = pci_protocol_versions[i]; + for (i = 0; i < num_version; i++) { + version_req->protocol_version = version[i]; ret = vmbus_sendpacket(hdev->channel, version_req, sizeof(struct pci_version_request), (unsigned long)pkt, VM_PKT_DATA_INBAND, @@ -2420,10 +2452,10 @@ static int hv_pci_protocol_negotiation(struct hv_device *hdev) } if (comp_pkt.completion_status >= 0) { - pci_protocol_version = pci_protocol_versions[i]; + hbus->protocol_version = version[i]; dev_info(&hdev->device, "PCI VMBus probing: Using version %#x\n", - pci_protocol_version); + hbus->protocol_version); goto exit; } @@ -2707,7 +2739,7 @@ static int hv_send_resources_allocated(struct hv_device *hdev) u32 wslot; int ret; - size_res = (pci_protocol_version < PCI_PROTOCOL_VERSION_1_2) + size_res = (hbus->protocol_version < PCI_PROTOCOL_VERSION_1_2) ? sizeof(*res_assigned) : sizeof(*res_assigned2); pkt = kmalloc(sizeof(*pkt) + size_res, GFP_KERNEL); @@ -2726,7 +2758,7 @@ static int hv_send_resources_allocated(struct hv_device *hdev) pkt->completion_func = hv_pci_generic_compl; pkt->compl_ctxt = &comp_pkt; - if (pci_protocol_version < PCI_PROTOCOL_VERSION_1_2) { + if (hbus->protocol_version < PCI_PROTOCOL_VERSION_1_2) { res_assigned = (struct pci_resources_assigned *)&pkt->message; res_assigned->message_type.type = @@ -2870,9 +2902,27 @@ static int hv_pci_probe(struct hv_device *hdev, * hv_pcibus_device contains the hypercall arguments for retargeting in * hv_irq_unmask(). Those must not cross a page boundary. */ - BUILD_BUG_ON(sizeof(*hbus) > PAGE_SIZE); + BUILD_BUG_ON(sizeof(*hbus) > HV_HYP_PAGE_SIZE); - hbus = (struct hv_pcibus_device *)get_zeroed_page(GFP_KERNEL); + /* + * With the recent 59bb47985c1d ("mm, sl[aou]b: guarantee natural + * alignment for kmalloc(power-of-two)"), kzalloc() is able to allocate + * a 4KB buffer that is guaranteed to be 4KB-aligned. Here the size and + * alignment of hbus is important because hbus's field + * retarget_msi_interrupt_params must not cross a 4KB page boundary. + * + * Here we prefer kzalloc to get_zeroed_page(), because a buffer + * allocated by the latter is not tracked and scanned by kmemleak, and + * hence kmemleak reports the pointer contained in the hbus buffer + * (i.e. the hpdev struct, which is created in new_pcichild_device() and + * is tracked by hbus->children) as memory leak (false positive). + * + * If the kernel doesn't have 59bb47985c1d, get_zeroed_page() *must* be + * used to allocate the hbus buffer and we can avoid the kmemleak false + * positive by using kmemleak_alloc() and kmemleak_free() to ask + * kmemleak to track and scan the hbus buffer. + */ + hbus = (struct hv_pcibus_device *)kzalloc(HV_HYP_PAGE_SIZE, GFP_KERNEL); if (!hbus) return -ENOMEM; hbus->state = hv_pcibus_init; @@ -2930,7 +2980,8 @@ static int hv_pci_probe(struct hv_device *hdev, hv_set_drvdata(hdev, hbus); - ret = hv_pci_protocol_negotiation(hdev); + ret = hv_pci_protocol_negotiation(hdev, pci_protocol_versions, + ARRAY_SIZE(pci_protocol_versions)); if (ret) goto close; @@ -3011,7 +3062,7 @@ free_bus: return ret; } -static void hv_pci_bus_exit(struct hv_device *hdev) +static int hv_pci_bus_exit(struct hv_device *hdev, bool hibernating) { struct hv_pcibus_device *hbus = hv_get_drvdata(hdev); struct { @@ -3027,16 +3078,20 @@ static void hv_pci_bus_exit(struct hv_device *hdev) * access the per-channel ringbuffer any longer. */ if (hdev->channel->rescind) - return; + return 0; - /* Delete any children which might still exist. */ - memset(&relations, 0, sizeof(relations)); - hv_pci_devices_present(hbus, &relations); + if (!hibernating) { + /* Delete any children which might still exist. */ + memset(&relations, 0, sizeof(relations)); + hv_pci_devices_present(hbus, &relations); + } ret = hv_send_resources_released(hdev); - if (ret) + if (ret) { dev_err(&hdev->device, "Couldn't send resources released packet(s)\n"); + return ret; + } memset(&pkt.teardown_packet, 0, sizeof(pkt.teardown_packet)); init_completion(&comp_pkt.host_event); @@ -3049,8 +3104,13 @@ static void hv_pci_bus_exit(struct hv_device *hdev) (unsigned long)&pkt.teardown_packet, VM_PKT_DATA_INBAND, VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); - if (!ret) - wait_for_completion_timeout(&comp_pkt.host_event, 10 * HZ); + if (ret) + return ret; + + if (wait_for_completion_timeout(&comp_pkt.host_event, 10 * HZ) == 0) + return -ETIMEDOUT; + + return 0; } /** @@ -3062,6 +3122,7 @@ static void hv_pci_bus_exit(struct hv_device *hdev) static int hv_pci_remove(struct hv_device *hdev) { struct hv_pcibus_device *hbus; + int ret; hbus = hv_get_drvdata(hdev); if (hbus->state == hv_pcibus_installed) { @@ -3074,7 +3135,7 @@ static int hv_pci_remove(struct hv_device *hdev) hbus->state = hv_pcibus_removed; } - hv_pci_bus_exit(hdev); + ret = hv_pci_bus_exit(hdev, false); vmbus_close(hdev->channel); @@ -3090,10 +3151,97 @@ static int hv_pci_remove(struct hv_device *hdev) hv_put_dom_num(hbus->sysdata.domain); - free_page((unsigned long)hbus); + kfree(hbus); + return ret; +} + +static int hv_pci_suspend(struct hv_device *hdev) +{ + struct hv_pcibus_device *hbus = hv_get_drvdata(hdev); + enum hv_pcibus_state old_state; + int ret; + + /* + * hv_pci_suspend() must make sure there are no pending work items + * before calling vmbus_close(), since it runs in a process context + * as a callback in dpm_suspend(). When it starts to run, the channel + * callback hv_pci_onchannelcallback(), which runs in a tasklet + * context, can be still running concurrently and scheduling new work + * items onto hbus->wq in hv_pci_devices_present() and + * hv_pci_eject_device(), and the work item handlers can access the + * vmbus channel, which can be being closed by hv_pci_suspend(), e.g. + * the work item handler pci_devices_present_work() -> + * new_pcichild_device() writes to the vmbus channel. + * + * To eliminate the race, hv_pci_suspend() disables the channel + * callback tasklet, sets hbus->state to hv_pcibus_removing, and + * re-enables the tasklet. This way, when hv_pci_suspend() proceeds, + * it knows that no new work item can be scheduled, and then it flushes + * hbus->wq and safely closes the vmbus channel. + */ + tasklet_disable(&hdev->channel->callback_event); + + /* Change the hbus state to prevent new work items. */ + old_state = hbus->state; + if (hbus->state == hv_pcibus_installed) + hbus->state = hv_pcibus_removing; + + tasklet_enable(&hdev->channel->callback_event); + + if (old_state != hv_pcibus_installed) + return -EINVAL; + + flush_workqueue(hbus->wq); + + ret = hv_pci_bus_exit(hdev, true); + if (ret) + return ret; + + vmbus_close(hdev->channel); + return 0; } +static int hv_pci_resume(struct hv_device *hdev) +{ + struct hv_pcibus_device *hbus = hv_get_drvdata(hdev); + enum pci_protocol_version_t version[1]; + int ret; + + hbus->state = hv_pcibus_init; + + ret = vmbus_open(hdev->channel, pci_ring_size, pci_ring_size, NULL, 0, + hv_pci_onchannelcallback, hbus); + if (ret) + return ret; + + /* Only use the version that was in use before hibernation. */ + version[0] = hbus->protocol_version; + ret = hv_pci_protocol_negotiation(hdev, version, 1); + if (ret) + goto out; + + ret = hv_pci_query_relations(hdev); + if (ret) + goto out; + + ret = hv_pci_enter_d0(hdev); + if (ret) + goto out; + + ret = hv_send_resources_allocated(hdev); + if (ret) + goto out; + + prepopulate_bars(hbus); + + hbus->state = hv_pcibus_installed; + return 0; +out: + vmbus_close(hdev->channel); + return ret; +} + static const struct hv_vmbus_device_id hv_pci_id_table[] = { /* PCI Pass-through Class ID */ /* 44C4F61D-4444-4400-9D52-802E27EDE19F */ @@ -3108,6 +3256,8 @@ static struct hv_driver hv_pci_drv = { .id_table = hv_pci_id_table, .probe = hv_pci_probe, .remove = hv_pci_remove, + .suspend = hv_pci_suspend, + .resume = hv_pci_resume, }; static void __exit exit_hv_pci_drv(void) |