diff options
author | Dave Airlie <airlied@redhat.com> | 2021-05-20 13:30:27 +1000 |
---|---|---|
committer | Dave Airlie <airlied@redhat.com> | 2021-05-20 13:31:12 +1000 |
commit | ae25ec2fc6c5a9e5767bf1922cd648501d0f914c (patch) | |
tree | cb07d9b3b578723cdc657c85760758db7bd28c15 /drivers/gpu/drm/amd/amdgpu | |
parent | 41ab70e06e13f81f0da76c5e0734c9bd32b5a4d9 (diff) | |
parent | 30039405ac25665119ff7bc944d33b136ef1c8a9 (diff) |
Merge tag 'drm-misc-next-2021-05-17' of git://anongit.freedesktop.org/drm/drm-misc into drm-next
drm-misc-next for 5.14:
UAPI Changes:
Cross-subsystem Changes:
Core Changes:
* aperture: Fix unlocking on errors
* legacy: Fix some doc comments
Driver Changes:
* drm/amdgpu: Free resource on fence usage query; Fix fence calculation;
* drm/bridge: Lt9611: Add missing MODULE_DEVICE_TABLE
* drm/i915: Print formats with %p4cc
* drm/ingenic: IPU planes are now always of type OVERLAY
* drm/nouveau: Remove left-over reference to struct drm_device.pdev
* drm/panfrost: Disable devfreq if num_supplies > 1; Add Mediatek MT8183 +
DT bindings; Cleanups
* drm/simpledrm: Print resources with %pr; Fix use-after-free errors;
Fix NULL deref; Fix MAINTAINERS entry
* drm/vmwgfx: Fix memory allocation and leak in FIFO allocation; Fix
return value in PCI resource setup
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/YKJs2IfwSYvuGPU7@linux-uq9g.fritz.box
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
-rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 27 | ||||
-rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h | 3 |
2 files changed, 23 insertions, 7 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index 01fe60fedcbe..fc83445fbc40 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -652,12 +652,14 @@ void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr) mutex_destroy(&mgr->lock); } -void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, struct amdgpu_ctx_entity *centity, - ktime_t *total, ktime_t *max) +static void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, + struct amdgpu_ctx_entity *centity, ktime_t *total, ktime_t *max) { ktime_t now, t1; uint32_t i; + *total = *max = 0; + now = ktime_get(); for (i = 0; i < amdgpu_sched_jobs; i++) { struct dma_fence *fence; @@ -669,11 +671,15 @@ void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, struct amdgpu_ctx_entity *cen if (!fence) continue; s_fence = to_drm_sched_fence(fence); - if (!dma_fence_is_signaled(&s_fence->scheduled)) + if (!dma_fence_is_signaled(&s_fence->scheduled)) { + dma_fence_put(fence); continue; + } t1 = s_fence->scheduled.timestamp; - if (t1 >= now) + if (!ktime_before(t1, now)) { + dma_fence_put(fence); continue; + } if (dma_fence_is_signaled(&s_fence->finished) && s_fence->finished.timestamp < now) *total += ktime_sub(s_fence->finished.timestamp, t1); @@ -699,11 +705,22 @@ ktime_t amdgpu_ctx_mgr_fence_usage(struct amdgpu_ctx_mgr *mgr, uint32_t hwip, idp = &mgr->ctx_handles; mutex_lock(&mgr->lock); idr_for_each_entry(idp, ctx, id) { + ktime_t ttotal, tmax; + if (!ctx->entities[hwip][idx]) continue; centity = ctx->entities[hwip][idx]; - amdgpu_ctx_fence_time(ctx, centity, &total, &max); + amdgpu_ctx_fence_time(ctx, centity, &ttotal, &tmax); + + /* Harmonic mean approximation diverges for very small + * values. If ratio < 0.01% ignore + */ + if (AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(tmax, ttotal)) + continue; + + total = ktime_add(total, ttotal); + max = ktime_after(tmax, max) ? tmax : max; } mutex_unlock(&mgr->lock); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h index 10dcf59a5c6b..14db16bc3322 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h @@ -30,6 +30,7 @@ struct drm_file; struct amdgpu_fpriv; #define AMDGPU_MAX_ENTITY_NUM 4 +#define AMDGPU_CTX_FENCE_USAGE_MIN_RATIO(max, total) ((max) > 16384ULL*(total)) struct amdgpu_ctx_entity { uint64_t sequence; @@ -89,6 +90,4 @@ long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout); void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr); ktime_t amdgpu_ctx_mgr_fence_usage(struct amdgpu_ctx_mgr *mgr, uint32_t hwip, uint32_t idx, uint64_t *elapsed); -void amdgpu_ctx_fence_time(struct amdgpu_ctx *ctx, struct amdgpu_ctx_entity *centity, - ktime_t *total, ktime_t *max); #endif |