dri-devel May 2013

dri-devel@lists.freedesktop.org

116 participants
705 discussions

[Bug 33715] New: demo arbocclude regression since r600g: Implement asyncronous query results.
by bugzilla-daemon＠freedesktop.org 13 Apr '14

13 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=33715 Summary: demo arbocclude regression since r600g: Implement asyncronous query results. Product: Mesa Version: git Platform: x86 (IA32) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: lists(a)andyfurniss.entadsl.com The bottom square and number in the mesa demo arbocclude are rendering red. It seems r600c has been doing this for some time, bith swrasts are OK and before this commit r600g got it right also. e7ec53273547335863b2377bea9d35aa9c114c6b is the first bad commit commit e7ec53273547335863b2377bea9d35aa9c114c6b Author: Mathias FrÃ¶hlich <Mathias.Froehlich(a)web.de> Date: Sun Jan 23 22:16:56 2011 +0100 r600g: Implement asyncronous query results. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 9

[Bug 32687] New: XBMC software video renderer is faster than OpenGL video renderer
by bugzilla-daemon＠freedesktop.org 13 Apr '14

13 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=32687 Summary: XBMC software video renderer is faster than OpenGL video renderer Product: Mesa Version: unspecified Platform: x86-64 (AMD64) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: nguyenthetoan(a)gmail.com I have an integrated AMD760G graphic card. I use the latest r600g driver from xorg-edgers package repository for ubuntu Lucid. My most important OpenGL software is XBMC. It has three video renderers: Software, ARB or GLSL. I notice (using the debug information of XBMC) that the Software renderer is actually the fastest one, then ARB and finally GLSL. I guess something is not well optimized in r600g with regard to video rendering. This is especially true for 1080p video (sometime I get only 12-16fps using OpenGL renderers while CPU usage is still only 50%. Software renderer could do 24fps with 100% CPU.). FYI, I also have another similar desktop, but with ATI X1650 PRO card. On this machine, using the r300g driver, the GLSL renderer is fastest (also similar to ARB renderer) and is faster than the Software renderer. So, r300g is well optimized with regard to video rendering. I file this bug hoping the developers could spare sometime optimize r600g for video rendering. Thanks a lot -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 3

[Bug 32557] New: [r600g] z/stencil buffer too small error in dmesg.
by bugzilla-daemon＠freedesktop.org 13 Apr '14

13 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=32557 Summary: [r600g] z/stencil buffer too small error in dmesg. Product: Mesa Version: git Platform: x86-64 (AMD64) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: kdekorte(a)yahoo.com Install wine 1.3.9 Install Flying Model Simulator 2.0 beta 7 under wine download from: http://n.ethz.ch/student/mmoeller/fms/down.html locate and download d3drm.dll and put in the FMS program directory start FMS in dmesg the following errors will occur multiple times [250821.374684] radeon 0000:01:00.0: z/stencil buffer too small (0x00B5FC7F 11648 1 4 -> 2981888 have 2957312) [250821.374688] radeon 0000:01:00.0: r600_packet3_check:1331 invalid cmd stream 507 [250821.374690] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream ! [250821.379713] radeon 0000:01:00.0: z/stencil buffer too small (0x00B5FC7F 11648 1 4 -> 2981888 have 2957312) [250821.379719] radeon 0000:01:00.0: r600_packet3_check:1331 invalid cmd stream 572 [250821.379723] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream ! The application will run, but there will be no display on the screen. Unsure if this is a wine or a r600 error as many other OpenGL applications work. As does Diablo II in Direct3d mode under the same wine. The application works fine under Windows 7 however. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 4

[Bug 28049] New: Failure observed while running the Glean test
by bugzilla-daemon＠freedesktop.org 13 Apr '14

13 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=28049 Summary: Failure observed while running the Glean test Product: Mesa Version: git Platform: x86 (IA32) OS/Version: other Status: NEW Severity: normal Priority: medium Component: Drivers/DRI/R600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: hysvats(a)gmail.com Created an attachment (id=35542) --> (https://bugs.freedesktop.org/attachment.cgi?id=35542) glean-logs KMS Driver Stack details : ========================= 1) Kernel-2.6.33 2) Libdrm-2.4.19 3) Mesa-7.8-rc2 4) Xorg-server-1.7.4 5) ddx: xf86-video-ati-6.13.0 System Environment : =================== O.S. - Ubuntu-9.10(32bit) Asic - M97 CPU - Intel(R) Pentium(R)@ 3.60 GHz Steps to Reproduce: =================== glean -r result -t pbo Observation : ============= 8 tests passed, 2 tests failed (logs attached) -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 4

[Bug 46711] New: vmwgfx does not work in qemu/kvm
by bugzilla-daemon＠bugzilla.kernel.org 09 Apr '14

09 Apr '14

https://bugzilla.kernel.org/show_bug.cgi?id=46711 URL: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1 037606 Summary: vmwgfx does not work in qemu/kvm Product: Drivers Version: 2.5 Kernel Version: 3.6-rc3 Platform: All OS/Version: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: Video(DRI - non Intel) AssignedTo: drivers_video-dri(a)kernel-bugs.osdl.org ReportedBy: hramrach(a)gmail.com Regression: No vmwgfx fails in kvm (running something like kvm -m 2048 -vga vmware -hda ubuntu.img): [ 29.856068] [drm] Initialized drm 1.1.0 20060810 [ 29.905192] [drm] Capabilities: [ 29.905196] [drm] Rect copy. [ 29.905197] [drm] Cursor. [ 29.905198] [drm] Cursor bypass. [ 29.905198] [drm] Cursor bypass 2. [ 29.905200] [drm] VRAM at 0xfd000000 size is 16384 kiB [ 29.905201] [drm] MMIO at 0xfe000000 size is 64 kiB [ 29.905203] [drm] global init. [ 29.907091] [TTM] Zone kernel: Available graphics memory: 1025792 kiB [ 29.907093] [TTM] Initializing pool allocator [ 29.907097] [TTM] Initializing DMA pool allocator [ 29.907432] [drm:vmw_driver_load] *ERROR* Hardware has no pitchlock [ 29.907475] [TTM] Finalizing pool allocator [ 29.907478] [TTM] Finalizing DMA pool allocator [ 29.907494] [TTM] Zone kernel: Used memory at exit: 0 kiB [ 29.907676] vmwgfx: probe of 0000:00:02.0 failed with error -38 -- Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching the assignee of the bug.

1 6

[Bug 41086] New: [r600] Screen update problems when scrolling to the right in Java applications
by bugzilla-daemon＠freedesktop.org 09 Apr '14

09 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=41086 Summary: [r600] Screen update problems when scrolling to the right in Java applications Product: DRI Version: XOrg CVS Platform: x86-64 (AMD64) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: DRM/Radeon AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: h.judt(a)gmx.at Created an attachment (id=51459) --> (https://bugs.freedesktop.org/attachment.cgi?id=51459) Screenshot showing garbage area after scrolling to the right. Example application for testing: TV Browser (see screenshot) When scrolling to the right, the contents of the scrolled area become garbled. Moving the mouse pointer over elements in the garbled area apparently causes a redraw of that element, fixing the issue, but the other elements are not redrawn. This does not occur when scrolling to the left. If the area is garbled and one scrolls to the left, everything is redrawn and appears correctly again. This may be confined to java applications, and I only noticed this in tvbrowser (http://tvbrowser.org). Unfortunately I don't have any other java applications available to test. Reproducible on the following hardware: ATI Technologies Inc Mobility Radeon HD 3400 Series ATI Technologies Inc Device 6719 (HD 6950 Cayman) -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 18

[Bug 43829] New: Resuming my AMD A4-300 based laptop leaves the screen black
by bugzilla-daemon＠freedesktop.org 03 Apr '14

03 Apr '14

https://bugs.freedesktop.org/show_bug.cgi?id=43829 Bug #: 43829 Summary: Resuming my AMD A4-300 based laptop leaves the screen black Classification: Unclassified Product: DRI Version: XOrg CVS Platform: Other OS/Version: All Status: NEW Severity: normal Priority: medium Component: DRM/Radeon AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: zboszor(a)pr.hu Created attachment 54422 --> https://bugs.freedesktop.org/attachment.cgi?id=54422 dmesg showing suspend/resume I have installed Fedora 16 on my new ASUS K53TA laptop that has these devices: # lspci 00:00.0 Host bridge: Advanced Micro Devices [AMD] Family 12h Processor Root Complex 00:01.0 VGA compatible controller: ATI Technologies Inc Device 9648 00:01.1 Audio device: ATI Technologies Inc Device 1714 00:02.0 PCI bridge: Advanced Micro Devices [AMD] Family 12h Processor Root Port 00:04.0 PCI bridge: Advanced Micro Devices [AMD] Family 12h Processor Root Port 00:05.0 PCI bridge: Advanced Micro Devices [AMD] Family 12h Processor Root Port 00:10.0 USB controller: Advanced Micro Devices [AMD] Hudson USB XHCI Controller (rev 03) 00:11.0 SATA controller: Advanced Micro Devices [AMD] Hudson SATA Controller [AHCI mode] (rev 40) 00:12.0 USB controller: Advanced Micro Devices [AMD] Hudson USB OHCI Controller (rev 11) 00:12.2 USB controller: Advanced Micro Devices [AMD] Hudson USB EHCI Controller (rev 11) 00:14.0 SMBus: Advanced Micro Devices [AMD] Hudson SMBus Controller (rev 13) 00:14.2 Audio device: Advanced Micro Devices [AMD] Hudson Azalia Controller (rev 01) 00:14.3 ISA bridge: Advanced Micro Devices [AMD] Hudson LPC Bridge (rev 11) 00:14.4 PCI bridge: Advanced Micro Devices [AMD] Hudson PCI Bridge (rev 40) 00:14.7 SD Host controller: Advanced Micro Devices [AMD] Hudson SD Flash Controller 00:18.0 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 0 (rev 43) 00:18.1 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 1 00:18.2 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 2 00:18.3 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 3 00:18.4 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 4 00:18.5 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 6 00:18.6 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 5 00:18.7 Host bridge: Advanced Micro Devices [AMD] Family 12h/14h Processor Function 7 01:00.0 VGA compatible controller: ATI Technologies Inc Whistler [AMD Radeon HD 6600M Series] (rev ff) 02:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 06) 03:00.0 Network controller: Atheros Communications Inc. AR9285 Wireless Network Adapter (PCI-Express) (rev 01) I had to install kernel-3.2.0-0.rc5.git1.1.fc17.x86_64 from Rawhide to make KMS work on the integrated video. I tried suspend/resume by closing the lid, suspend worked. Resume woke up the machine but left the screen black. dmesg for the suspend/resume is attached. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 61

[PATCH] drm/ttm: remove fence_lock
by Maarten Lankhorst 25 Mar '14

25 Mar '14

With the nouveau calls fixed there were only 2 places left that used fence_lock without a reservation, those are fixed in this patch: ttm_bo_cleanup_refs_or_queue is fixed by simply doing things the other way around. ttm_bo_cleanup_refs is fixed by taking a reservation first, then a pointer to the fence object and backs off from the reservation if waiting has to be performed. Signed-off-by: Maarten Lankhorst <maarten.lankhorst(a)canonical.com> --- This patch should be applied only after all the previous patches I sent are applied, since it depends on sync_obj_arg removal (kinda, probably fails to apply otherwise), uses ttm_bo_is_reserved, and depends on nouveau wait behavior being fixed. drivers/gpu/drm/nouveau/nouveau_bo.c | 4 - drivers/gpu/drm/nouveau/nouveau_gem.c | 15 +--- drivers/gpu/drm/radeon/radeon_object.c | 2 - drivers/gpu/drm/ttm/ttm_bo.c | 124 ++++++++++++-------------------- drivers/gpu/drm/ttm/ttm_bo_util.c | 3 - drivers/gpu/drm/ttm/ttm_bo_vm.c | 5 - drivers/gpu/drm/ttm/ttm_execbuf_util.c | 2 - drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 3 - include/drm/ttm/ttm_bo_api.h | 12 +-- include/drm/ttm/ttm_bo_driver.h | 3 - 10 files changed, 52 insertions(+), 121 deletions(-) diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index cee7448..9749c45 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -1449,10 +1449,8 @@ nouveau_bo_fence(struct nouveau_bo *nvbo, struct nouveau_fence *fence) if (likely(fence)) nouveau_fence_ref(fence); - spin_lock(&nvbo->bo.bdev->fence_lock); old_fence = nvbo->bo.sync_obj; nvbo->bo.sync_obj = fence; - spin_unlock(&nvbo->bo.bdev->fence_lock); nouveau_fence_unref(&old_fence); } @@ -1552,9 +1550,7 @@ nouveau_bo_vma_del(struct nouveau_bo *nvbo, struct nouveau_vma *vma) if (vma->node) { if (nvbo->bo.mem.mem_type != TTM_PL_SYSTEM) { ttm_bo_reserve(&nvbo->bo, false, false, false, 0); - spin_lock(&nvbo->bo.bdev->fence_lock); ttm_bo_wait(&nvbo->bo, false, false, false); - spin_unlock(&nvbo->bo.bdev->fence_lock); ttm_bo_unreserve(&nvbo->bo); nouveau_vm_unmap(vma); } diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index 6d8391d..eaa700a 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -391,18 +391,11 @@ retry: static int validate_sync(struct nouveau_channel *chan, struct nouveau_bo *nvbo) { - struct nouveau_fence *fence = NULL; + struct nouveau_fence *fence = nvbo->bo.sync_obj; int ret = 0; - spin_lock(&nvbo->bo.bdev->fence_lock); - if (nvbo->bo.sync_obj) - fence = nouveau_fence_ref(nvbo->bo.sync_obj); - spin_unlock(&nvbo->bo.bdev->fence_lock); - - if (fence) { + if (fence) ret = nouveau_fence_sync(fence, chan); - nouveau_fence_unref(&fence); - } return ret; } @@ -614,9 +607,7 @@ nouveau_gem_pushbuf_reloc_apply(struct drm_device *dev, data |= r->vor; } - spin_lock(&nvbo->bo.bdev->fence_lock); ret = ttm_bo_wait(&nvbo->bo, false, false, false); - spin_unlock(&nvbo->bo.bdev->fence_lock); if (ret) { NV_ERROR(drm, "reloc wait_idle failed: %d\n", ret); break; @@ -848,11 +839,9 @@ nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data, ret = ttm_bo_reserve(&nvbo->bo, true, false, false, 0); if (!ret) { - spin_lock(&nvbo->bo.bdev->fence_lock); ret = ttm_bo_wait(&nvbo->bo, true, true, true); if (!no_wait && ret) fence = nouveau_fence_ref(nvbo->bo.sync_obj); - spin_unlock(&nvbo->bo.bdev->fence_lock); ttm_bo_unreserve(&nvbo->bo); } diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 808c444..df430e4 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -613,12 +613,10 @@ int radeon_bo_wait(struct radeon_bo *bo, u32 *mem_type, bool no_wait) r = ttm_bo_reserve(&bo->tbo, true, no_wait, false, 0); if (unlikely(r != 0)) return r; - spin_lock(&bo->tbo.bdev->fence_lock); if (mem_type) *mem_type = bo->tbo.mem.mem_type; if (bo->tbo.sync_obj) r = ttm_bo_wait(&bo->tbo, true, true, no_wait); - spin_unlock(&bo->tbo.bdev->fence_lock); ttm_bo_unreserve(&bo->tbo); return r; } diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index f6d7026..69fc29b 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -498,28 +498,23 @@ static void ttm_bo_cleanup_refs_or_queue(struct ttm_buffer_object *bo) { struct ttm_bo_device *bdev = bo->bdev; struct ttm_bo_global *glob = bo->glob; - struct ttm_bo_driver *driver; + struct ttm_bo_driver *driver = bdev->driver; void *sync_obj = NULL; int put_count; int ret; - spin_lock(&bdev->fence_lock); - (void) ttm_bo_wait(bo, false, false, true); - if (!bo->sync_obj) { - - spin_lock(&glob->lru_lock); - - /** - * Lock inversion between bo:reserve and bdev::fence_lock here, - * but that's OK, since we're only trylocking. - */ - - ret = ttm_bo_reserve_locked(bo, false, true, false, 0); + spin_lock(&glob->lru_lock); + ret = ttm_bo_reserve_locked(bo, false, true, false, 0); + if (!ret) { + ret = ttm_bo_wait(bo, false, false, true); - if (unlikely(ret == -EBUSY)) + if (unlikely(ret == -EBUSY)) { + sync_obj = driver->sync_obj_ref(bo->sync_obj); + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); goto queue; + } - spin_unlock(&bdev->fence_lock); put_count = ttm_bo_del_from_lru(bo); spin_unlock(&glob->lru_lock); @@ -528,18 +523,11 @@ static void ttm_bo_cleanup_refs_or_queue(struct ttm_buffer_object *bo) ttm_bo_list_ref_sub(bo, put_count, true); return; - } else { - spin_lock(&glob->lru_lock); } queue: - driver = bdev->driver; - if (bo->sync_obj) - sync_obj = driver->sync_obj_ref(bo->sync_obj); - kref_get(&bo->list_kref); list_add_tail(&bo->ddestroy, &bdev->ddestroy); spin_unlock(&glob->lru_lock); - spin_unlock(&bdev->fence_lock); if (sync_obj) { driver->sync_obj_flush(sync_obj); @@ -565,25 +553,15 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool no_wait_gpu) { struct ttm_bo_device *bdev = bo->bdev; + struct ttm_bo_driver *driver = bdev->driver; struct ttm_bo_global *glob = bo->glob; int put_count; int ret = 0; + void *sync_obj; retry: - spin_lock(&bdev->fence_lock); - ret = ttm_bo_wait(bo, false, interruptible, no_wait_gpu); - spin_unlock(&bdev->fence_lock); - - if (unlikely(ret != 0)) - return ret; - spin_lock(&glob->lru_lock); - if (unlikely(list_empty(&bo->ddestroy))) { - spin_unlock(&glob->lru_lock); - return 0; - } - ret = ttm_bo_reserve_locked(bo, interruptible, no_wait_reserve, false, 0); @@ -592,18 +570,37 @@ retry: return ret; } - /** - * We can re-check for sync object without taking - * the bo::lock since setting the sync object requires - * also bo::reserved. A busy object at this point may - * be caused by another thread recently starting an accelerated - * eviction. - */ + if (unlikely(list_empty(&bo->ddestroy))) { + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); + spin_unlock(&glob->lru_lock); + return 0; + } + ret = ttm_bo_wait(bo, false, false, true); + + if (ret) { + if (no_wait_gpu) { + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); + spin_unlock(&glob->lru_lock); + return ret; + } - if (unlikely(bo->sync_obj)) { + /** + * Take a reference to the fence and unreserve, if the wait + * was succesful and no new sync_obj was attached, + * ttm_bo_wait in retry will return ret = 0, and end the loop. + */ + + sync_obj = driver->sync_obj_ref(&bo->sync_obj); atomic_set(&bo->reserved, 0); wake_up_all(&bo->event_queue); spin_unlock(&glob->lru_lock); + + ret = driver->sync_obj_wait(bo->sync_obj, false, interruptible); + driver->sync_obj_unref(&sync_obj); + if (ret) + return ret; goto retry; } @@ -735,9 +732,7 @@ static int ttm_bo_evict(struct ttm_buffer_object *bo, bool interruptible, struct ttm_placement placement; int ret = 0; - spin_lock(&bdev->fence_lock); ret = ttm_bo_wait(bo, false, interruptible, no_wait_gpu); - spin_unlock(&bdev->fence_lock); if (unlikely(ret != 0)) { if (ret != -ERESTARTSYS) { @@ -1054,7 +1049,6 @@ int ttm_bo_move_buffer(struct ttm_buffer_object *bo, { int ret = 0; struct ttm_mem_reg mem; - struct ttm_bo_device *bdev = bo->bdev; BUG_ON(!ttm_bo_is_reserved(bo)); @@ -1063,9 +1057,7 @@ int ttm_bo_move_buffer(struct ttm_buffer_object *bo, * Have the driver move function wait for idle when necessary, * instead of doing it here. */ - spin_lock(&bdev->fence_lock); ret = ttm_bo_wait(bo, false, interruptible, no_wait_gpu); - spin_unlock(&bdev->fence_lock); if (ret) return ret; mem.num_pages = bo->num_pages; @@ -1554,7 +1546,6 @@ int ttm_bo_device_init(struct ttm_bo_device *bdev, bdev->glob = glob; bdev->need_dma32 = need_dma32; bdev->val_seq = 0; - spin_lock_init(&bdev->fence_lock); mutex_lock(&glob->device_list_mutex); list_add_tail(&bdev->device_list, &glob->device_list); mutex_unlock(&glob->device_list_mutex); @@ -1690,52 +1681,33 @@ int ttm_bo_wait(struct ttm_buffer_object *bo, bool lazy, bool interruptible, bool no_wait) { struct ttm_bo_driver *driver = bo->bdev->driver; - struct ttm_bo_device *bdev = bo->bdev; - void *sync_obj; int ret = 0; + WARN_ON_ONCE(!ttm_bo_is_reserved(bo)); + if (likely(bo->sync_obj == NULL)) return 0; - while (bo->sync_obj) { - + if (bo->sync_obj) { if (driver->sync_obj_signaled(bo->sync_obj)) { void *tmp_obj = bo->sync_obj; bo->sync_obj = NULL; clear_bit(TTM_BO_PRIV_FLAG_MOVING, &bo->priv_flags); - spin_unlock(&bdev->fence_lock); driver->sync_obj_unref(&tmp_obj); - spin_lock(&bdev->fence_lock); - continue; + return 0; } if (no_wait) return -EBUSY; - sync_obj = driver->sync_obj_ref(bo->sync_obj); - spin_unlock(&bdev->fence_lock); - ret = driver->sync_obj_wait(sync_obj, + ret = driver->sync_obj_wait(bo->sync_obj, lazy, interruptible); if (unlikely(ret != 0)) { - driver->sync_obj_unref(&sync_obj); - spin_lock(&bdev->fence_lock); return ret; } - spin_lock(&bdev->fence_lock); - if (likely(bo->sync_obj == sync_obj)) { - void *tmp_obj = bo->sync_obj; - bo->sync_obj = NULL; - clear_bit(TTM_BO_PRIV_FLAG_MOVING, - &bo->priv_flags); - spin_unlock(&bdev->fence_lock); - driver->sync_obj_unref(&sync_obj); - driver->sync_obj_unref(&tmp_obj); - spin_lock(&bdev->fence_lock); - } else { - spin_unlock(&bdev->fence_lock); - driver->sync_obj_unref(&sync_obj); - spin_lock(&bdev->fence_lock); - } + driver->sync_obj_unref(&bo->sync_obj); + clear_bit(TTM_BO_PRIV_FLAG_MOVING, + &bo->priv_flags); } return 0; } @@ -1799,9 +1771,7 @@ static int ttm_bo_swapout(struct ttm_mem_shrink *shrink) * Wait for GPU, then move to system cached. */ - spin_lock(&bo->bdev->fence_lock); ret = ttm_bo_wait(bo, false, false, false); - spin_unlock(&bo->bdev->fence_lock); if (unlikely(ret != 0)) goto out; diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index d80d1e8..84d6e01 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -622,7 +622,6 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, struct ttm_buffer_object *ghost_obj; void *tmp_obj = NULL; - spin_lock(&bdev->fence_lock); if (bo->sync_obj) { tmp_obj = bo->sync_obj; bo->sync_obj = NULL; @@ -630,7 +629,6 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, bo->sync_obj = driver->sync_obj_ref(sync_obj); if (evict) { ret = ttm_bo_wait(bo, false, false, false); - spin_unlock(&bdev->fence_lock); if (tmp_obj) driver->sync_obj_unref(&tmp_obj); if (ret) @@ -653,7 +651,6 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, */ set_bit(TTM_BO_PRIV_FLAG_MOVING, &bo->priv_flags); - spin_unlock(&bdev->fence_lock); if (tmp_obj) driver->sync_obj_unref(&tmp_obj); diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c index a877813..55718c2 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_vm.c +++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c @@ -122,17 +122,14 @@ static int ttm_bo_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf) * move. */ - spin_lock(&bdev->fence_lock); if (test_bit(TTM_BO_PRIV_FLAG_MOVING, &bo->priv_flags)) { ret = ttm_bo_wait(bo, false, true, false); - spin_unlock(&bdev->fence_lock); if (unlikely(ret != 0)) { retval = (ret != -ERESTARTSYS) ? VM_FAULT_SIGBUS : VM_FAULT_NOPAGE; goto out_unlock; } - } else - spin_unlock(&bdev->fence_lock); + } ret = ttm_mem_io_lock(man, true); if (unlikely(ret != 0)) { diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c index 721886e..51b5e97 100644 --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c @@ -207,7 +207,6 @@ void ttm_eu_fence_buffer_objects(struct list_head *list, void *sync_obj) driver = bdev->driver; glob = bo->glob; - spin_lock(&bdev->fence_lock); spin_lock(&glob->lru_lock); list_for_each_entry(entry, list, head) { @@ -218,7 +217,6 @@ void ttm_eu_fence_buffer_objects(struct list_head *list, void *sync_obj) entry->reserved = false; } spin_unlock(&glob->lru_lock); - spin_unlock(&bdev->fence_lock); list_for_each_entry(entry, list, head) { if (entry->old_sync_obj) diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index ed3c1e7..e038c9a 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -251,13 +251,10 @@ static void vmw_dummy_query_bo_prepare(struct vmw_private *dev_priv) volatile SVGA3dQueryResult *result; bool dummy; int ret; - struct ttm_bo_device *bdev = &dev_priv->bdev; struct ttm_buffer_object *bo = dev_priv->dummy_query_bo; ttm_bo_reserve(bo, false, false, false, 0); - spin_lock(&bdev->fence_lock); ret = ttm_bo_wait(bo, false, false, false); - spin_unlock(&bdev->fence_lock); if (unlikely(ret != 0)) (void) vmw_fallback_wait(dev_priv, false, true, 0, false, 10*HZ); diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 816d9b1..6c69224 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -222,6 +222,8 @@ struct ttm_buffer_object { struct file *persistent_swap_storage; struct ttm_tt *ttm; bool evicted; + void *sync_obj; + unsigned long priv_flags; /** * Members protected by the bdev::lru_lock. @@ -242,16 +244,6 @@ struct ttm_buffer_object { atomic_t reserved; /** - * Members protected by struct buffer_object_device::fence_lock - * In addition, setting sync_obj to anything else - * than NULL requires bo::reserved to be held. This allows for - * checking NULL while reserved but not holding the mentioned lock. - */ - - void *sync_obj; - unsigned long priv_flags; - - /** * Members protected by the bdev::vm_lock */ diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index 0c8c3b5..aac101b 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -515,8 +515,6 @@ struct ttm_bo_global { * * @driver: Pointer to a struct ttm_bo_driver struct setup by the driver. * @man: An array of mem_type_managers. - * @fence_lock: Protects the synchronizing members on *all* bos belonging - * to this device. * @addr_space_mm: Range manager for the device address space. * lru_lock: Spinlock that protects the buffer+device lru lists and * ddestroy lists. @@ -539,7 +537,6 @@ struct ttm_bo_device { struct ttm_bo_driver *driver; rwlock_t vm_lock; struct ttm_mem_type_manager man[TTM_NUM_MEM_TYPES]; - spinlock_t fence_lock; /* * Protected by the vm lock. */

4 14

[Bug 44301] New: [i915g] regression 37b62cc35435a704d33c1814fbdbd0e807699e21 results in solid color fills in 3Dmark01 and flatout in wine
by bugzilla-daemon＠freedesktop.org 11 Mar '14

11 Mar '14

https://bugs.freedesktop.org/show_bug.cgi?id=44301 Bug #: 44301 Summary: [i915g] regression 37b62cc35435a704d33c1814fbdbd0e807699e21 results in solid color fills in 3Dmark01 and flatout in wine Classification: Unclassified Product: Mesa Version: git Platform: x86 (IA32) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/i915g AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: cme3000(a)gmail.com Commit 37b62cc35435a704d33c1814fbdbd0e807699e21 breaks 3DMark01 and flatout in wine - I915_LIE=1 (default) results in a black rectangle when 3dmark shows the logo while starting, it's rendered correctly with I915_LIE=0 - I915_LIE=1 results in a solid color fill when using 32bpp and a 32bit zbuffer in game test 1 (i haven't tried the other tests), choosing 16bpp or a 16 or 24bit zbuffer works fine. no rendering errors with I915_LIE=0 - flatout results in a solid color fill, no matter if with I915_LIE or without all those solid color fills look like if it picks the rgb values of one pixel at some corner. mesa ded02bd54b5925ee222d7f426850312236b4f59d wine-1.3.34-202-gb0f704d 00:02.0 VGA compatible controller: Intel Corporation Mobile 945GME Express Integrated Graphics Controller (rev 03) -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 1

[Bug 44499] New: r280 and xbmc - choppy menu and video playback
by bugzilla-daemon＠freedesktop.org 01 Mar '14

01 Mar '14

https://bugs.freedesktop.org/show_bug.cgi?id=44499 Bug #: 44499 Summary: r280 and xbmc - choppy menu and video playback Classification: Unclassified Product: Mesa Version: git Platform: x86 (IA32) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/DRI/r200 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: smoki00790(a)gmail.com After mesa commit b2596c36c8f73e8bb7a0b1679b491662aeb2f9d9 i can experianced choppy/micro/freezes in video playback and also going thorugh menu in xbmc and other video players using GL based output. This can be also seen in many GL games menus like in ppracer/etracer, pingus (from git it now have GL option), etc. Have 9d21b5dd265959b28ab1e4103fb381074ec51a08 currently and same problem. DDX from current git, xorg 1.10.4, kernel 3.1.6. card:1002:5960:17af:2020 kms. Debian. Color tilling enabled and 75 fps at 1024x768 in UT4.1 with correct rendering 50 in OpenArena. Drawing tests in gtkperf are first slow later faster. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 18

← Newer
1
...
33
34
35
36
37
38
39
...
71
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel May 2013