Re:Re: [PATCH] amdgpu: fixes memleak issue when init failed
发件人:"Christian König" 发送日期:2020-04-21 22:53:47 收件人:"赵军奎" 抄送人:Alex Deucher ,"David (ChunMing) Zhou" ,David Airlie ,Daniel Vetter ,Tom St Denis ,Ori Messinger ,Sam Ravnborg ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org,opensource.ker...@vivo.com 主题:Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am 21.04.20 um 15:39 schrieb 赵军奎: >> 发件人:"Christian König" >> 发送日期:2020-04-21 21:02:27 >> 收件人:"赵军奎" >> 抄送人:Alex Deucher ,"David (ChunMing) Zhou" >> ,David Airlie ,Daniel Vetter >> ,Tom St Denis ,Ori Messinger >> ,Sam Ravnborg >> ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org,opensource.ker...@vivo.com >> 主题:Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am 21.04.20 um >> 14:09 schrieb 赵军奎: From: "Christian König" Date: 2020-04-21 19:22:49 To: Bernard Zhao ,Alex Deucher ,"David (ChunMing) Zhou" ,David Airlie ,Daniel Vetter ,Tom St Denis ,Ori Messinger ,Sam Ravnborg ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org Cc: opensource.ker...@vivo.com Subject: Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am 21.04.20 um 13:17 schrieb Bernard Zhao: >> VRAM manager and DRM MM when init failed, there is no operaction >> to free kzalloc memory & remove device file. >> This will lead to memleak & cause stability issue. > NAK, failure to create sysfs nodes are not critical. > > Christian. > OK, get it. By the way, should i modify this patch to just handle in error branch, or that it is also unnecessary? >>> What you can do is to drop the "return ret" if anything with the sysfs >>> nodes goes wrong and instead print the error code. >> Emmm, for this part, i am not sure, my modify first print the error, secone >> release not free memory, >> and last return error, make everything clear to the system. >> I think it`s the same with what you mentioned, is there something that I >> misunderstood? > >Yes, maybe an example makes it more clear what to do here. Currently we >print and error and return when something with the sysfs files goes wrong: > >if (ret) { > DRM_ERROR("Failed to create device file mem_info_vram_total\n"); > return ret; >} > >But what we should do instead is just to print an error and continue and >in the end return success status: > >if (ret) > DRM_ERROR("Failed to create device file mem_info_vram_total >(%d)\n", r); > >... >return 0; > >Regards, >Christian. > Emmm, i am still confused about two points: 1 Does that mean there is no failed case in this function? 2 There is no need to free the kzmalloc space(no possibility of memory leak )? Regards, Bernard >> >>> It's really annoying that loading, unloading and loading the driver >>> again sometimes fails because we have a bug in the sysfs files cleanup. >>> >>> We certainly should fix those bugs as well, but they are just not >>> critical for correct driver functionality. >>> >>> Regards, >>> Christian. >> Regards, Bernard >> Signed-off-by: Bernard Zhao >> --- >> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 24 >> >> 1 file changed, 19 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> index 82a3299e53c0..4c5fb153e6b4 100644 >> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> @@ -175,30 +175,44 @@ static int amdgpu_vram_mgr_init(struct >> ttm_mem_type_manager *man, >> ret = device_create_file(adev->dev, >> &dev_attr_mem_info_vram_total); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vram_total\n"); >> -return ret; >> +goto VRAM_TOTAL_FAIL; >> } >> ret = device_create_file(adev->dev, >> &dev_attr_mem_info_vis_vram_total); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vis_vram_total\n"); >> -return ret; >> +goto VIS_VRAM_TOTA_FAIL; >> } >> ret = device_create_file(adev->dev, >> &dev_attr_mem_info_vram_used); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vram_used\n"); >> -return ret; >> +goto VRAM_USED_FAIL; >> } >> ret = device_create_file(adev->dev, >> &dev_attr_mem_info_vis_vram_used); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vis_vram_used\n"); >> -return ret; >> +goto VIS_VRAM_USED_FAIL; >>
Re:Re: [PATCH] amdgpu: fixes memleak issue when init failed
发件人:"Christian König" 发送日期:2020-04-21 21:02:27 收件人:"赵军奎" 抄送人:Alex Deucher ,"David (ChunMing) Zhou" ,David Airlie ,Daniel Vetter ,Tom St Denis ,Ori Messinger ,Sam Ravnborg ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org,opensource.ker...@vivo.com 主题:Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am 21.04.20 um 14:09 schrieb 赵军奎: >> From: "Christian König" >> Date: 2020-04-21 19:22:49 >> To: Bernard Zhao ,Alex Deucher >> ,"David (ChunMing) Zhou" >> ,David Airlie ,Daniel Vetter >> ,Tom St Denis ,Ori Messinger >> ,Sam Ravnborg >> ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org >> Cc: opensource.ker...@vivo.com >> Subject: Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am >> 21.04.20 um 13:17 schrieb Bernard Zhao: VRAM manager and DRM MM when init failed, there is no operaction to free kzalloc memory & remove device file. This will lead to memleak & cause stability issue. >>> NAK, failure to create sysfs nodes are not critical. >>> >>> Christian. >>> >> OK, get it. >> By the way, should i modify this patch to just handle in error >> branch, or that it is also unnecessary? > >What you can do is to drop the "return ret" if anything with the sysfs >nodes goes wrong and instead print the error code. Emmm, for this part, i am not sure, my modify first print the error, secone release not free memory, and last return error, make everything clear to the system. I think it`s the same with what you mentioned, is there something that I misunderstood? > >It's really annoying that loading, unloading and loading the driver >again sometimes fails because we have a bug in the sysfs files cleanup. > >We certainly should fix those bugs as well, but they are just not >critical for correct driver functionality. > >Regards, >Christian. >> >> Regards, >> Bernard >> Signed-off-by: Bernard Zhao --- drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 24 1 file changed, 19 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 82a3299e53c0..4c5fb153e6b4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -175,30 +175,44 @@ static int amdgpu_vram_mgr_init(struct ttm_mem_type_manager *man, ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_total); if (ret) { DRM_ERROR("Failed to create device file mem_info_vram_total\n"); - return ret; + goto VRAM_TOTAL_FAIL; } ret = device_create_file(adev->dev, &dev_attr_mem_info_vis_vram_total); if (ret) { DRM_ERROR("Failed to create device file mem_info_vis_vram_total\n"); - return ret; + goto VIS_VRAM_TOTA_FAIL; } ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_used); if (ret) { DRM_ERROR("Failed to create device file mem_info_vram_used\n"); - return ret; + goto VRAM_USED_FAIL; } ret = device_create_file(adev->dev, &dev_attr_mem_info_vis_vram_used); if (ret) { DRM_ERROR("Failed to create device file mem_info_vis_vram_used\n"); - return ret; + goto VIS_VRAM_USED_FAIL; } ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_vendor); if (ret) { DRM_ERROR("Failed to create device file mem_info_vram_vendor\n"); - return ret; + goto VRAM_VERDOR_FAIL; } return 0; + +VRAM_VERDOR_FAIL: + device_remove_file(adev->dev, &dev_attr_mem_info_vis_vram_used); +VIS_VRAM_USED_FAIL: + device_remove_file(adev->dev, &dev_attr_mem_info_vram_used); +RVAM_USED_FAIL: + device_remove_file(adev->dev, &dev_attr_mem_info_vis_vram_total); +VIS_VRAM_TOTA_FAIL: + device_remove_file(adev->dev, &dev_attr_mem_info_vram_total); +VRAM_TOTAL_FAIL: + kfree(mgr); + man->priv = NULL; + + return ret; } /** >> > ___ amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Re:Re: [PATCH] amdgpu: fixes memleak issue when init failed
From: "Christian König" Date: 2020-04-21 19:22:49 To: Bernard Zhao ,Alex Deucher ,"David (ChunMing) Zhou" ,David Airlie ,Daniel Vetter ,Tom St Denis ,Ori Messinger ,Sam Ravnborg ,amd-gfx@lists.freedesktop.org,dri-de...@lists.freedesktop.org,linux-ker...@vger.kernel.org Cc: opensource.ker...@vivo.com Subject: Re: [PATCH] amdgpu: fixes memleak issue when init failed>Am 21.04.20 um 13:17 schrieb Bernard Zhao: >> VRAM manager and DRM MM when init failed, there is no operaction >> to free kzalloc memory & remove device file. >> This will lead to memleak & cause stability issue. > >NAK, failure to create sysfs nodes are not critical. > >Christian. > OK, get it. By the way, should i modify this patch to just handle in error branch, or that it is also unnecessary? Regards, Bernard >> >> Signed-off-by: Bernard Zhao >> --- >> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 24 >> 1 file changed, 19 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> index 82a3299e53c0..4c5fb153e6b4 100644 >> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >> @@ -175,30 +175,44 @@ static int amdgpu_vram_mgr_init(struct >> ttm_mem_type_manager *man, >> ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_total); >> if (ret) { >> DRM_ERROR("Failed to create device file mem_info_vram_total\n"); >> -return ret; >> +goto VRAM_TOTAL_FAIL; >> } >> ret = device_create_file(adev->dev, &dev_attr_mem_info_vis_vram_total); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vis_vram_total\n"); >> -return ret; >> +goto VIS_VRAM_TOTA_FAIL; >> } >> ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_used); >> if (ret) { >> DRM_ERROR("Failed to create device file mem_info_vram_used\n"); >> -return ret; >> +goto VRAM_USED_FAIL; >> } >> ret = device_create_file(adev->dev, &dev_attr_mem_info_vis_vram_used); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vis_vram_used\n"); >> -return ret; >> +goto VIS_VRAM_USED_FAIL; >> } >> ret = device_create_file(adev->dev, &dev_attr_mem_info_vram_vendor); >> if (ret) { >> DRM_ERROR("Failed to create device file >> mem_info_vram_vendor\n"); >> -return ret; >> +goto VRAM_VERDOR_FAIL; >> } >> >> return 0; >> + >> +VRAM_VERDOR_FAIL: >> +device_remove_file(adev->dev, &dev_attr_mem_info_vis_vram_used); >> +VIS_VRAM_USED_FAIL: >> +device_remove_file(adev->dev, &dev_attr_mem_info_vram_used); >> +RVAM_USED_FAIL: >> +device_remove_file(adev->dev, &dev_attr_mem_info_vis_vram_total); >> +VIS_VRAM_TOTA_FAIL: >> +device_remove_file(adev->dev, &dev_attr_mem_info_vram_total); >> +VRAM_TOTAL_FAIL: >> +kfree(mgr); >> +man->priv = NULL; >> + >> +return ret; >> } >> >> /** > ___ amd-gfx mailing list amd-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/amd-gfx