dri-devel August 2019

dri-devel@lists.freedesktop.org

318 participants
2338 discussions

[PATCH AUTOSEL 5.2 31/59] drm/amdgpu: fix error handling in amdgpu_cs_process_fence_dep
by Sasha Levin 06 Aug '19

06 Aug '19

From: Christian König <christian.koenig(a)amd.com> [ Upstream commit 67d0859e2758ef992fd32499747ce4b1038a63c0 ] We always need to drop the ctx reference and should check for errors first and then dereference the fence pointer. Signed-off-by: Christian König <christian.koenig(a)amd.com> Reviewed-by: Chunming Zhou <david1.zhou(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 26 ++++++++++++-------------- 1 file changed, 12 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 2f6239b6be6fe..fe028561dc0e6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -1093,29 +1093,27 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p, return r; } - fence = amdgpu_ctx_get_fence(ctx, entity, - deps[i].handle); + fence = amdgpu_ctx_get_fence(ctx, entity, deps[i].handle); + amdgpu_ctx_put(ctx); + + if (IS_ERR(fence)) + return PTR_ERR(fence); + else if (!fence) + continue; if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) { - struct drm_sched_fence *s_fence = to_drm_sched_fence(fence); + struct drm_sched_fence *s_fence; struct dma_fence *old = fence; + s_fence = to_drm_sched_fence(fence); fence = dma_fence_get(&s_fence->scheduled); dma_fence_put(old); } - if (IS_ERR(fence)) { - r = PTR_ERR(fence); - amdgpu_ctx_put(ctx); + r = amdgpu_sync_fence(p->adev, &p->job->sync, fence, true); + dma_fence_put(fence); + if (r) return r; - } else if (fence) { - r = amdgpu_sync_fence(p->adev, &p->job->sync, fence, - true); - dma_fence_put(fence); - amdgpu_ctx_put(ctx); - if (r) - return r; - } } return 0; } -- 2.20.1

1 0

[PATCH AUTOSEL 5.2 30/59] drm/amd/powerplay: fix null pointer dereference around dpm state relates
by Sasha Levin 06 Aug '19

06 Aug '19

From: Evan Quan <evan.quan(a)amd.com> [ Upstream commit 479156f2e5540077377a823eaf5a4263bd329063 ] DPM state relates are not supported on the new SW SMU ASICs. But still it's not OK to trigger null pointer dereference on accessing them. Signed-off-by: Evan Quan <evan.quan(a)amd.com> Reviewed-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 18 +++++++++++++----- drivers/gpu/drm/amd/powerplay/amdgpu_smu.c | 3 ++- 2 files changed, 15 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index abeaab4bf1bc2..d55519bc34e52 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -144,12 +144,16 @@ static ssize_t amdgpu_get_dpm_state(struct device *dev, struct amdgpu_device *adev = ddev->dev_private; enum amd_pm_state_type pm; - if (is_support_sw_smu(adev) && adev->smu.ppt_funcs->get_current_power_state) - pm = amdgpu_smu_get_current_power_state(adev); - else if (adev->powerplay.pp_funcs->get_current_power_state) + if (is_support_sw_smu(adev)) { + if (adev->smu.ppt_funcs->get_current_power_state) + pm = amdgpu_smu_get_current_power_state(adev); + else + pm = adev->pm.dpm.user_state; + } else if (adev->powerplay.pp_funcs->get_current_power_state) { pm = amdgpu_dpm_get_current_power_state(adev); - else + } else { pm = adev->pm.dpm.user_state; + } return snprintf(buf, PAGE_SIZE, "%s\n", (pm == POWER_STATE_TYPE_BATTERY) ? "battery" : @@ -176,7 +180,11 @@ static ssize_t amdgpu_set_dpm_state(struct device *dev, goto fail; } - if (adev->powerplay.pp_funcs->dispatch_tasks) { + if (is_support_sw_smu(adev)) { + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.user_state = state; + mutex_unlock(&adev->pm.mutex); + } else if (adev->powerplay.pp_funcs->dispatch_tasks) { amdgpu_dpm_dispatch_task(adev, AMD_PP_TASK_ENABLE_USER_STATE, &state); } else { mutex_lock(&adev->pm.mutex); diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c index eec329ab60370..61a6d183c153f 100644 --- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c @@ -63,7 +63,8 @@ int smu_get_power_num_states(struct smu_context *smu, /* not support power state */ memset(state_info, 0, sizeof(struct pp_states_info)); - state_info->nums = 0; + state_info->nums = 1; + state_info->states[0] = POWER_STATE_TYPE_DEFAULT; return 0; } -- 2.20.1

1 0

[PATCH AUTOSEL 5.2 29/59] drm/amdkfd: Fix byte align on VegaM
by Sasha Levin 06 Aug '19

06 Aug '19

From: Kent Russell <kent.russell(a)amd.com> [ Upstream commit d65848657c3da5c0d4b685f823d0230f151ab34e ] This was missed during the addition of VegaM support Reviewed-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Kent Russell <kent.russell(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 4b192e0ce92f4..ed7977d0dd018 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1148,7 +1148,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( adev->asic_type != CHIP_FIJI && adev->asic_type != CHIP_POLARIS10 && adev->asic_type != CHIP_POLARIS11 && - adev->asic_type != CHIP_POLARIS12) ? + adev->asic_type != CHIP_POLARIS12 && + adev->asic_type != CHIP_VEGAM) ? VI_BO_SIZE_ALIGN : 1; mapping_flags = AMDGPU_VM_PAGE_READABLE; -- 2.20.1

1 0

[PATCH AUTOSEL 5.2 23/59] drm/bridge: tc358764: Fix build error
by Sasha Levin 06 Aug '19

06 Aug '19

From: YueHaibing <yuehaibing(a)huawei.com> [ Upstream commit e1ae72a21e5f0d1846e26e3f5963930664702071 ] If CONFIG_DRM_TOSHIBA_TC358764=y but CONFIG_DRM_KMS_HELPER=m, building fails: drivers/gpu/drm/bridge/tc358764.o:(.rodata+0x228): undefined reference to `drm_atomic_helper_connector_reset' drivers/gpu/drm/bridge/tc358764.o:(.rodata+0x240): undefined reference to `drm_helper_probe_single_connector_modes' drivers/gpu/drm/bridge/tc358764.o:(.rodata+0x268): undefined reference to `drm_atomic_helper_connector_duplicate_state' drivers/gpu/drm/bridge/tc358764.o:(.rodata+0x270): undefined reference to `drm_atomic_helper_connector_destroy_state' Like TC358767, select DRM_KMS_HELPER to fix this, and change to select DRM_PANEL to avoid recursive dependency. Reported-by: Hulk Robot <hulkci(a)huawei.com> Fixes: f38b7cca6d0e ("drm/bridge: tc358764: Add DSI to LVDS bridge driver") Signed-off-by: YueHaibing <yuehaibing(a)huawei.com> Reviewed-by: Laurent Pinchart <laurent.pinchart(a)ideasonboard.com> Reviewed-by: Neil Armstrong <narmstrong(a)baylibre.com> Signed-off-by: Neil Armstrong <narmstrong(a)baylibre.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190729090520.25968-1-yuehai… Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/bridge/Kconfig | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/bridge/Kconfig b/drivers/gpu/drm/bridge/Kconfig index cc62603b87c59..e4e22bbae2a7c 100644 --- a/drivers/gpu/drm/bridge/Kconfig +++ b/drivers/gpu/drm/bridge/Kconfig @@ -117,9 +117,10 @@ config DRM_THINE_THC63LVD1024 config DRM_TOSHIBA_TC358764 tristate "TC358764 DSI/LVDS bridge" - depends on DRM && DRM_PANEL depends on OF select DRM_MIPI_DSI + select DRM_KMS_HELPER + select DRM_PANEL help Toshiba TC358764 DSI/LVDS bridge driver. -- 2.20.1

1 0

[PATCH AUTOSEL 5.2 22/59] drm/bridge: lvds-encoder: Fix build error while CONFIG_DRM_KMS_HELPER=m
by Sasha Levin 06 Aug '19

06 Aug '19

From: YueHaibing <yuehaibing(a)huawei.com> [ Upstream commit f4cc743a98136df3c3763050a0e8223b52d9a960 ] If DRM_LVDS_ENCODER=y but CONFIG_DRM_KMS_HELPER=m, build fails: drivers/gpu/drm/bridge/lvds-encoder.o: In function `lvds_encoder_probe': lvds-encoder.c:(.text+0x155): undefined reference to `devm_drm_panel_bridge_add' Reported-by: Hulk Robot <hulkci(a)huawei.com> Fixes: dbb58bfd9ae6 ("drm/bridge: Fix lvds-encoder since the panel_bridge rework.") Signed-off-by: YueHaibing <yuehaibing(a)huawei.com> Reviewed-by: Neil Armstrong <narmstrong(a)baylibre.com> Signed-off-by: Neil Armstrong <narmstrong(a)baylibre.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190729071216.27488-1-yuehai… Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/bridge/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/bridge/Kconfig b/drivers/gpu/drm/bridge/Kconfig index ee777469293a4..cc62603b87c59 100644 --- a/drivers/gpu/drm/bridge/Kconfig +++ b/drivers/gpu/drm/bridge/Kconfig @@ -48,6 +48,7 @@ config DRM_DUMB_VGA_DAC config DRM_LVDS_ENCODER tristate "Transparent parallel to LVDS encoder support" depends on OF + select DRM_KMS_HELPER select DRM_PANEL_BRIDGE help Support for transparent parallel to LVDS encoders that don't require -- 2.20.1

1 0

[PATCH v6 0/3] mm/gup: add make_dirty arg to put_user_pages_dirty_lock()
by john.hubbard＠gmail.com 06 Aug '19

06 Aug '19

From: John Hubbard <jhubbard(a)nvidia.com> Changes since v5: * Patch #1: Fixed a bug that I introduced in v4: drivers/infiniband/sw/siw/siw_mem.c needs to refer to umem->page_chunk[i].plist, rather than umem->page_chunk[i]. Changes since v4: * Christophe Hellwig's review applied: deleted siw_free_plist() and __qib_release_user_pages(), now that put_user_pages_dirty_lock() does what those routines were doing. * Applied Bjorn's ACK for net/xdp, and Christophe's Reviewed-by for patch #1. Changes since v3: * Fixed an unused variable warning in siw_mem.c Changes since v2: * Critical bug fix: remove a stray "break;" from the new routine. Changes since v1: * Instead of providing __put_user_pages(), add an argument to put_user_pages_dirty_lock(), and delete put_user_pages_dirty(). This is based on the following points: 1. Lots of call sites become simpler if a bool is passed into put_user_page*(), instead of making the call site choose which put_user_page*() variant to call. 2. Christoph Hellwig's observation that set_page_dirty_lock() is usually correct, and set_page_dirty() is usually a bug, or at least questionable, within a put_user_page*() calling chain. * Added the Infiniband driver back to the patch series, because it is a caller of put_user_pages_dirty_lock(). Unchanged parts from the v1 cover letter (except for the diffstat): Notes about the remaining patches to come: There are about 50+ patches in my tree [2], and I'll be sending out the remaining ones in a few more groups: * The block/bio related changes (Jerome mostly wrote those, but I've had to move stuff around extensively, and add a little code) * mm/ changes * other subsystem patches * an RFC that shows the current state of the tracking patch set. That can only be applied after all call sites are converted, but it's good to get an early look at it. This is part a tree-wide conversion, as described in commit fc1d8e7cca2d ("mm: introduce put_user_page*(), placeholder versions"). John Hubbard (3): mm/gup: add make_dirty arg to put_user_pages_dirty_lock() drivers/gpu/drm/via: convert put_page() to put_user_page*() net/xdp: convert put_page() to put_user_page*() drivers/gpu/drm/via/via_dmablit.c | 10 +- drivers/infiniband/core/umem.c | 5 +- drivers/infiniband/hw/hfi1/user_pages.c | 5 +- drivers/infiniband/hw/qib/qib_user_pages.c | 13 +-- drivers/infiniband/hw/usnic/usnic_uiom.c | 5 +- drivers/infiniband/sw/siw/siw_mem.c | 19 +--- include/linux/mm.h | 5 +- mm/gup.c | 115 +++++++++------------ net/xdp/xdp_umem.c | 9 +- 9 files changed, 64 insertions(+), 122 deletions(-) -- 2.22.0

3 5

Re: drm pull for v5.3-rc1
by Linus Torvalds 06 Aug '19

06 Aug '19

On Tue, Aug 6, 2019 at 12:38 AM Christoph Hellwig <hch(a)infradead.org> wrote: > > Seems like no one took this up. Below is a version which I think is > slightly better by also moving the mm_walk structure initialization > into the helpers, with an outcome of just a handful of added lines. Ack. Agreed, I think that's a nicer interface. In fact, I do note that a lot of the users don't actually use the "void *private" argument at all - they just want the walker - and just pass in a NULL private pointer. So we have things like this: > + if (walk_page_range(&init_mm, va, va + size, &set_nocache_walk_ops, > + NULL)) { and in a perfect world we'd have arguments with default values so that we could skip those entirely for when people just don't need it. I'm not a huge fan of C++ because of a lot of the complexity (and some really bad decisions), but many of the _syntactic_ things in C++ would be nice to use. This one doesn't seem to be one that the gcc people have picked up as an extension ;( Yes, yes, we could do it with a macro, I guess. #define walk_page_range(mm, start,end, ops, ...) \ __walk_page_range(mm, start, end, (NULL , ## __VA_ARGS__)) but I'm not sure it's worthwhile. Linus

2 1

[PATCH] gpu: drm: amd: display: dc: dcn20: Remove redudant condition
by Hariprasad Kelam 06 Aug '19

06 Aug '19

Remove redudant codition "dsc_cfg->dc_dsc_cfg.num_slices_v". fixes coverity defect 1451853 Signed-off-by: Hariprasad Kelam <hariprasad.kelam(a)gmail.com> --- drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c index e870caa..42133bd 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dsc.c @@ -302,7 +302,7 @@ static bool dsc_prepare_config(struct display_stream_compressor *dsc, const stru dsc_cfg->dc_dsc_cfg.linebuf_depth == 0))); ASSERT(96 <= dsc_cfg->dc_dsc_cfg.bits_per_pixel && dsc_cfg->dc_dsc_cfg.bits_per_pixel <= 0x3ff); // 6.0 <= bits_per_pixel <= 63.9375 - if (!dsc_cfg->dc_dsc_cfg.num_slices_v || !dsc_cfg->dc_dsc_cfg.num_slices_v || + if (!dsc_cfg->dc_dsc_cfg.num_slices_v || !(dsc_cfg->dc_dsc_cfg.version_minor == 1 || dsc_cfg->dc_dsc_cfg.version_minor == 2) || !dsc_cfg->pic_width || !dsc_cfg->pic_height || !((dsc_cfg->dc_dsc_cfg.version_minor == 1 && // v1.1 line buffer depth range: -- 2.7.4

3 3

[PATCH] fence: checking for NULL pointer after dereferencing it.
by Xinpeng Liu 06 Aug '19

06 Aug '19

drivers/dma-buf/dma-fence.c:131 in function dma_fence_signal_locked, if (WARN_ON(!fence)) should be before lockdep_assert_held(fence->lock); otherwise there is not any meaning. Signed-off-by: Xinpeng Liu <danielliu861(a)gmail.com> --- drivers/dma-buf/dma-fence.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/dma-buf/dma-fence.c b/drivers/dma-buf/dma-fence.c index 59ac96e..0b11063 100644 --- a/drivers/dma-buf/dma-fence.c +++ b/drivers/dma-buf/dma-fence.c @@ -131,10 +131,9 @@ int dma_fence_signal_locked(struct dma_fence *fence) struct dma_fence_cb *cur, *tmp; int ret = 0; - lockdep_assert_held(fence->lock); - if (WARN_ON(!fence)) return -EINVAL; + lockdep_assert_held(fence->lock); if (test_and_set_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags)) { ret = -EINVAL; -- 1.8.3.1

1 0

Re: [PATCH 03/15] nouveau: pass struct nouveau_svmm to nouveau_range_fault
by Jason Gunthorpe 06 Aug '19

06 Aug '19

On Tue, Aug 06, 2019 at 07:05:41PM +0300, Christoph Hellwig wrote: > We'll need the nouveau_svmm structure to improve the function soon. > For now this allows using the svmm->mm reference to unlock the > mmap_sem, and thus the same dereference chain that the caller uses > to lock and unlock it. > > Signed-off-by: Christoph Hellwig <hch(a)lst.de> > --- > drivers/gpu/drm/nouveau/nouveau_svm.c | 12 ++++++------ > 1 file changed, 6 insertions(+), 6 deletions(-) Reviewed-by: Jason Gunthorpe <jgg(a)mellanox.com> Jason

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel August 2019