kmemleak report: 5.15.0-rc3: nouveau_fence_new

* kmemleak report: 5.15.0-rc3: nouveau_fence_new
@ 2021-09-29 19:36 ` John Ogness
  0 siblings, 0 replies; 2+ messages in thread
From: John Ogness @ 2021-09-29 19:36 UTC (permalink / raw)
  To: David Airlie, Daniel Vetter; +Cc: dri-devel, nouveau

Hello,

With 5.15.0-rc3 on my ppc64 (PowerMac G5) I am seeing kmemleak
reports. They are always 96 bytes and with the same stacktrace.

unreferenced object 0xc000000011d2a7e0 (size 96):
  comm "X", pid 1743, jiffies 4295010075 (age 5457.040s)
  hex dump (first 32 bytes):
    c0 00 00 00 0b 9f f0 00 c0 00 3d 00 00 b0 85 90  ..........=.....
    00 00 00 a9 77 41 30 23 c0 00 00 00 08 db b7 c8  ....wA0#........
  backtrace:
    [<000000006f102108>] .nouveau_fence_new+0x4c/0x120 [nouveau]
    [<00000000395e0a83>] .nouveau_bo_move+0x4f0/0x870 [nouveau]
    [<00000000f17bc6da>] .ttm_bo_handle_move_mem+0xb4/0x1e0 [ttm]
    [<00000000fb36762f>] .ttm_bo_validate+0x144/0x230 [ttm]
    [<00000000a84dc7b3>] .nouveau_bo_validate+0x70/0xc0 [nouveau]
    [<00000000b4e870a2>] .nouveau_gem_ioctl_pushbuf+0x6e0/0x1a90 [nouveau]
    [<000000007b7c5c38>] .drm_ioctl_kernel+0x104/0x180 [drm]
    [<000000000af76e30>] .drm_ioctl+0x244/0x490 [drm]
    [<00000000ebb759e8>] .nouveau_drm_ioctl+0x78/0x140 [nouveau]
    [<00000000263274a7>] .__se_sys_ioctl+0xfc/0x160
    [<0000000088c39f3d>] .system_call_exception+0x178/0x2a0
    [<000000000cfdf34f>] system_call_common+0xec/0x250

If I decode this stacktrace using decode_stacktrace.sh so that the line
numbers can be seen, I get the following:

.nouveau_fence_new+0x4c/0x120 [nouveau]
linux-5.15-rc3/include/linux/slab.h:591
linux-5.15-rc3/include/linux/slab.h:721
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_fence.c:424

.nouveau_bo_move+0x4f0/0x870 [nouveau]
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_bo.c:821
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_bo.c:1032

.ttm_bo_handle_move_mem+0xb4/0x1e0 [ttm]
linux-5.15-rc3/drivers/gpu/drm/ttm/ttm_bo.c:197

.ttm_bo_validate+0x144/0x230 [ttm]
linux-5.15-rc3/drivers/gpu/drm/ttm/ttm_bo.c:904
linux-5.15-rc3/drivers/gpu/drm/ttm/ttm_bo.c:981

.nouveau_bo_validate+0x70/0xc0 [nouveau]
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_bo.c:647

.nouveau_gem_ioctl_pushbuf+0x6e0/0x1a90 [nouveau]
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_gem.c:548
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_gem.c:605
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_gem.c:799

.drm_ioctl_kernel+0x104/0x180 [drm]
linux-5.15-rc3/drivers/gpu/drm/drm_ioctl.c:795

.drm_ioctl+0x244/0x490 [drm]
linux-5.15-rc3/include/linux/thread_info.h:185
linux-5.15-rc3/include/linux/thread_info.h:218
linux-5.15-rc3/include/linux/uaccess.h:199
linux-5.15-rc3/drivers/gpu/drm/drm_ioctl.c:899

.nouveau_drm_ioctl+0x78/0x140 [nouveau]
linux-5.15-rc3/drivers/gpu/drm/nouveau/nouveau_drm.c:1163

.__se_sys_ioctl+0xfc/0x160
linux-5.15-rc3/fs/ioctl.c:51
linux-5.15-rc3/fs/ioctl.c:874
linux-5.15-rc3/fs/ioctl.c:860

.system_call_exception+0x178/0x2a0
.system_call_exception
linux-5.15-rc3/arch/powerpc/kernel/interrupt.c:233

system_call_common+0xec/0x250
linux-5.15-rc3/arch/powerpc/kernel/interrupt_64.S:314

Here are all enabled DRM and NOUVEAU configs in my kernel:

CONFIG_DRM=m
CONFIG_DRM_KMS_HELPER=m
CONFIG_DRM_FBDEV_EMULATION=y
CONFIG_DRM_FBDEV_OVERALLOC=100
CONFIG_DRM_TTM=m
CONFIG_DRM_TTM_HELPER=m
CONFIG_DRM_NOUVEAU=m
CONFIG_NOUVEAU_DEBUG=5
CONFIG_NOUVEAU_DEBUG_DEFAULT=3
CONFIG_DRM_NOUVEAU_BACKLIGHT=y
CONFIG_DRM_PANEL=y
CONFIG_DRM_BRIDGE=y
CONFIG_DRM_PANEL_BRIDGE=y
CONFIG_DRM_PANEL_ORIENTATION_QUIRKS=m

And lspci output:

0000:f0:10.0 VGA compatible controller: NVIDIA Corporation NV34 [GeForce FX 5200 Ultra] (rev a1)

I have been running 5.12 on my machine without these reports. So it
might be something that showed up in 5.13 or 5.14 as well.

I do not know if this is a good channel for reporting this, so please
let me know if I should report it somewhere else. Also let me know if
you need any additional information from me.

John Ogness

^ permalink raw reply	[flat|nested] 2+ messages in thread