dri-devel April 2020

dri-devel@lists.freedesktop.org

353 participants
944 discussions

Re: [PATCH 0/1] drm/i915: Fix a deadlock that only affects 5.4
by Greg KH 07 Apr '20

07 Apr '20

On Mon, Apr 06, 2020 at 11:26:21PM -0700, Sultan Alsawaf wrote: > From: Sultan Alsawaf <sultan(a)kerneltoast.com> > > Hi, > > There's a mutex lock deadlock in i915 that only affects 5.4, but was fixed in > 5.5. Normally, I would send a backport of the fix from 5.5, but the patch set > that fixes the deadlock involves massive changes that are neither feasible nor > desirable for backporting [1][2][3]. Therefore, I've made a small patch that > only addresses the deadlock specifically for 5.4. This paragraph needs to go into the patch itself, otherwise just looking at that doesn't make any sense. And you do not need a cover letter for a single patch. Please fix up and resend. thanks, greg k-h

1 0

[PATCH v2 1/2] drm/i915/dp_mst: Cast intel_connector->port as drm_dp_mst_port
by Lyude Paul 07 Apr '20

07 Apr '20

The only reason for having this cast as void * before was because we originally needed to use drm_dp_mst_get_port_validated() and friends in order to (attempt to) safely access MST ports. However, we've since improved how reference counting works with ports and mstbs such that we can now rely on drm_dp_mst_port structs remaining in memory for as long as the driver needs. This means we don't really need to cast this as void* anymore, and can just access the struct directly. We'll also need this for the next commit, so that we can remove drm_dp_mst_port_has_audio(). Signed-off-by: Lyude Paul <lyude(a)redhat.com> Reviewed-by: Sean Paul <sean(a)poorly.run> --- drivers/gpu/drm/i915/display/intel_display_types.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h index 5a0adf14ebef..0ddc98afe252 100644 --- a/drivers/gpu/drm/i915/display/intel_display_types.h +++ b/drivers/gpu/drm/i915/display/intel_display_types.h @@ -438,7 +438,7 @@ struct intel_connector { state of connector->polled in case hotplug storm detection changes it */ u8 polled; - void *port; /* store this opaque as its illegal to dereference it */ + struct drm_dp_mst_port *port; struct intel_dp *mst_port; -- 2.25.1

2 2

[PATCH v2 00/17] drm: subsytem-wide debugfs functions refactor
by Wambui Karuga 07 Apr '20

07 Apr '20

This series includes work on various debugfs functions both in drm/core and across various drivers in the subsystem. Since commit 987d65d01356 (drm: debugfs: make drm_debugfs_create_files() never fail), drm_debugfs_create_files() does not fail and only returns zero. This series therefore removes the left over error handling and checks for its return value across drm drivers. As a result of these changes, most drm_debugfs functions are converted to return void in this series. This also enables the drm_driver, debugfs_init() hook to be changed to return void. v2: individual driver patches have been converted to have debugfs functions return 0 instead of void to prevent breaking individual driver builds. The last patch then converts the .debugfs_hook() and its users across all drivers to return void. Wambui Karuga (17): drm/tegra: remove checks for debugfs functions return value drm/tilcdc: remove check for return value of debugfs functions. drm/v3d: make v3d_debugfs_init() return 0 drm/vc4: remove check of return value of drm_debugfs functions drm/arc: make arcgpu_debugfs_init() return 0. drm/arm: make hdlcd_debugfs_init() return 0 drm/etnaviv: remove check for return value of drm_debugfs_create_files() drm/msm: remove checks for return value of drm_debugfs_create_files() drm/sti: remove use of drm_debugfs functions as return values drm/vram-helper: make drm_vram_mm_debugfs_init() return 0 drm/nouveau: make nouveau_drm_debugfs_init() return 0 drm/pl111: make pl111_debugfs_init return 0 drm/omap: remove checks for return value of drm_debugfs functions drm/i915: have *_debugfs_init() functions return void. drm: make various debugfs_init() functions return 0 drm/debugfs: remove checks for return value of drm_debugfs functions. drm: convert .debugfs_init() hook to return void. drivers/gpu/drm/arc/arcpgu_drv.c | 7 +-- drivers/gpu/drm/arm/hdlcd_drv.c | 7 +-- drivers/gpu/drm/arm/malidp_drv.c | 3 +- drivers/gpu/drm/drm_atomic.c | 8 ++-- drivers/gpu/drm/drm_client.c | 8 ++-- drivers/gpu/drm/drm_crtc_internal.h | 2 +- drivers/gpu/drm/drm_debugfs.c | 45 +++++-------------- drivers/gpu/drm/drm_framebuffer.c | 8 ++-- drivers/gpu/drm/drm_gem_vram_helper.c | 14 ++---- drivers/gpu/drm/drm_internal.h | 2 +- drivers/gpu/drm/drm_mipi_dbi.c | 6 +-- drivers/gpu/drm/etnaviv/etnaviv_drv.c | 18 ++------ .../drm/i915/display/intel_display_debugfs.c | 8 ++-- .../drm/i915/display/intel_display_debugfs.h | 4 +- drivers/gpu/drm/i915/i915_debugfs.c | 8 ++-- drivers/gpu/drm/i915/i915_debugfs.h | 4 +- drivers/gpu/drm/msm/adreno/a5xx_debugfs.c | 18 +++----- drivers/gpu/drm/msm/adreno/a5xx_gpu.h | 2 +- drivers/gpu/drm/msm/disp/mdp5/mdp5_kms.c | 14 ++---- drivers/gpu/drm/msm/msm_debugfs.c | 23 +++------- drivers/gpu/drm/msm/msm_debugfs.h | 2 +- drivers/gpu/drm/msm/msm_gpu.h | 2 +- drivers/gpu/drm/nouveau/nouveau_debugfs.c | 26 +++++------ drivers/gpu/drm/nouveau/nouveau_debugfs.h | 8 ++-- drivers/gpu/drm/omapdrm/omap_debugfs.c | 29 +++--------- drivers/gpu/drm/omapdrm/omap_drv.h | 2 +- drivers/gpu/drm/pl111/pl111_debugfs.c | 8 ++-- drivers/gpu/drm/pl111/pl111_drm.h | 2 +- drivers/gpu/drm/qxl/qxl_debugfs.c | 21 +++------ drivers/gpu/drm/qxl/qxl_drv.h | 13 +++--- drivers/gpu/drm/qxl/qxl_ttm.c | 6 +-- drivers/gpu/drm/sti/sti_compositor.c | 6 +-- drivers/gpu/drm/sti/sti_compositor.h | 4 +- drivers/gpu/drm/sti/sti_crtc.c | 2 +- drivers/gpu/drm/sti/sti_cursor.c | 14 +++--- drivers/gpu/drm/sti/sti_drv.c | 16 ++----- drivers/gpu/drm/sti/sti_dvo.c | 13 +++--- drivers/gpu/drm/sti/sti_gdp.c | 7 +-- drivers/gpu/drm/sti/sti_hda.c | 13 +++--- drivers/gpu/drm/sti/sti_hdmi.c | 13 +++--- drivers/gpu/drm/sti/sti_hqvdp.c | 12 ++--- drivers/gpu/drm/sti/sti_mixer.c | 10 ++--- drivers/gpu/drm/sti/sti_mixer.h | 2 +- drivers/gpu/drm/sti/sti_tvout.c | 13 +++--- drivers/gpu/drm/sti/sti_vid.c | 8 ++-- drivers/gpu/drm/sti/sti_vid.h | 2 +- drivers/gpu/drm/tegra/dc.c | 11 +---- drivers/gpu/drm/tegra/drm.c | 8 ++-- drivers/gpu/drm/tegra/dsi.c | 11 +---- drivers/gpu/drm/tegra/hdmi.c | 11 +---- drivers/gpu/drm/tegra/sor.c | 11 +---- drivers/gpu/drm/tilcdc/tilcdc_drv.c | 17 ++----- drivers/gpu/drm/v3d/v3d_debugfs.c | 8 ++-- drivers/gpu/drm/v3d/v3d_drv.h | 2 +- drivers/gpu/drm/vc4/vc4_debugfs.c | 11 ++--- drivers/gpu/drm/vc4/vc4_drv.h | 2 +- drivers/gpu/drm/virtio/virtgpu_debugfs.c | 3 +- drivers/gpu/drm/virtio/virtgpu_drv.h | 2 +- include/drm/drm_client.h | 2 +- include/drm/drm_debugfs.h | 16 +++---- include/drm/drm_drv.h | 2 +- include/drm/drm_gem_vram_helper.h | 2 +- include/drm/drm_mipi_dbi.h | 2 +- 63 files changed, 204 insertions(+), 380 deletions(-) -- 2.25.1

8 32

[PULL] nouveau-fixes 5.7
by Ben Skeggs 07 Apr '20

07 Apr '20

Hey Dave, A couple of misc fixes/workarounds for some issues that are causing a lot of pain for people. Of most interest are the PCI power management and GR init WARs, which effect a rather significant number of laptop systems that are in use today. Ben. The following changes since commit 0e7e6198af28c1573267aba1be33dd0b7fb35691: Merge branch 'ttm-transhuge' of git://people.freedesktop.org/~thomash/linux into drm-next (2020-04-03 09:07:49 +1000) are available in the Git repository at: git://github.com/skeggsb/linux linux-5.7 for you to fetch changes up to 374b55802cd567e9f340b7f27d2c5e20b069ac3c: drm/nouveau/kms/nv50-: wait for FIFO space on PIO channels (2020-04-07 14:37:50 +1000) ---------------------------------------------------------------- Ben Skeggs (5): drm/nouveau/ttm: evict other IO mappings when running out of BAR1 space drm/nouveau/gr/gp107,gp108: implement workaround for HW hanging during init drm/nouveau/nvif: access PTIMER through usermode class, if available drm/nouveau/nvif: protect waits against GPU falling off the bus drm/nouveau/kms/nv50-: wait for FIFO space on PIO channels Karol Herbst (1): drm/nouveau: workaround runpm fail by disabling PCI power management on certain intel bridges Ralph Campbell (3): drm/nouveau/svm: fix vma range check for migration drm/nouveau/svm: check for SVM initialized before migrating drm/nouveau/svm: remove useless SVM range check Wambui Karuga (1): drm/nouveau: remove checks for return value of debugfs functions drivers/gpu/drm/nouveau/dispnv04/dac.c | 3 ++- drivers/gpu/drm/nouveau/dispnv04/hw.c | 1 + drivers/gpu/drm/nouveau/dispnv50/base507c.c | 1 + drivers/gpu/drm/nouveau/dispnv50/core507d.c | 1 + drivers/gpu/drm/nouveau/dispnv50/corec37d.c | 2 ++ drivers/gpu/drm/nouveau/dispnv50/curs507a.c | 21 ++++++++++++++++++--- drivers/gpu/drm/nouveau/dispnv50/cursc37a.c | 9 ++++++--- drivers/gpu/drm/nouveau/dispnv50/disp.c | 1 + drivers/gpu/drm/nouveau/dispnv50/ovly827e.c | 2 ++ drivers/gpu/drm/nouveau/dispnv50/wndw.h | 1 + drivers/gpu/drm/nouveau/include/nvif/device.h | 21 --------------------- drivers/gpu/drm/nouveau/include/nvif/timer.h | 35 +++++++++++++++++++++++++++++++++++ drivers/gpu/drm/nouveau/include/nvif/user.h | 1 + drivers/gpu/drm/nouveau/nouveau_bo.c | 9 +++++++-- drivers/gpu/drm/nouveau/nouveau_debugfs.c | 20 ++++++++------------ drivers/gpu/drm/nouveau/nouveau_drm.c | 63 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/nouveau/nouveau_drv.h | 2 ++ drivers/gpu/drm/nouveau/nouveau_svm.c | 9 ++++++--- drivers/gpu/drm/nouveau/nvif/Kbuild | 1 + drivers/gpu/drm/nouveau/nvif/device.c | 14 +++++++++----- drivers/gpu/drm/nouveau/nvif/timer.c | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ drivers/gpu/drm/nouveau/nvif/userc361.c | 14 ++++++++++++++ drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c | 26 ++++++++++++++++++++++++++ 23 files changed, 263 insertions(+), 50 deletions(-) create mode 100644 drivers/gpu/drm/nouveau/include/nvif/timer.h create mode 100644 drivers/gpu/drm/nouveau/nvif/timer.c

1 0

[PATCH v8 05/19] virtgpu: pull in uaccess.h
by Michael S. Tsirkin 07 Apr '20

07 Apr '20

In preparation to virtio header changes, include uaccess.h directly as this file is using copy to/from user. Signed-off-by: Michael S. Tsirkin <mst(a)redhat.com> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 205ec4abae2b..289dabbce477 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -27,6 +27,7 @@ #include <linux/file.h> #include <linux/sync_file.h> +#include <linux/uaccess.h> #include <drm/drm_file.h> #include <drm/virtgpu_drm.h> -- MST

1 0

[PATCH v7 05/19] virtgpu: pull in uaccess.h
by Michael S. Tsirkin 07 Apr '20

07 Apr '20

In preparation to virtio header changes, include uaccess.h directly as this file is using copy to/from user. Signed-off-by: Michael S. Tsirkin <mst(a)redhat.com> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 205ec4abae2b..2342a8baa5f8 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -27,6 +27,7 @@ #include <linux/file.h> #include <linux/sync_file.h> +#include <linux/uacess.h> #include <drm/drm_file.h> #include <drm/virtgpu_drm.h> -- MST

1 0

[PATCH AUTOSEL 5.4 28/32] drm/scheduler: fix rare NULL ptr race
by Sasha Levin 07 Apr '20

07 Apr '20

From: Yintian Tao <yttao(a)amd.com> [ Upstream commit 77bb2f204f1f0a53a602a8fd15816d6826212077 ] There is one one corner case at dma_fence_signal_locked which will raise the NULL pointer problem just like below. ->dma_fence_signal ->dma_fence_signal_locked ->test_and_set_bit here trigger dma_fence_release happen due to the zero of fence refcount. ->dma_fence_put ->dma_fence_release ->drm_sched_fence_release_scheduled ->call_rcu here make the union fled “cb_list” at finished fence to NULL because struct rcu_head contains two pointer which is same as struct list_head cb_list Therefore, to hold the reference of finished fence at drm_sched_process_job to prevent the null pointer during finished fence dma_fence_signal [ 732.912867] BUG: kernel NULL pointer dereference, address: 0000000000000008 [ 732.914815] #PF: supervisor write access in kernel mode [ 732.915731] #PF: error_code(0x0002) - not-present page [ 732.916621] PGD 0 P4D 0 [ 732.917072] Oops: 0002 [#1] SMP PTI [ 732.917682] CPU: 7 PID: 0 Comm: swapper/7 Tainted: G OE 5.4.0-rc7 #1 [ 732.918980] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014 [ 732.920906] RIP: 0010:dma_fence_signal_locked+0x3e/0x100 [ 732.938569] Call Trace: [ 732.939003] <IRQ> [ 732.939364] dma_fence_signal+0x29/0x50 [ 732.940036] drm_sched_fence_finished+0x12/0x20 [gpu_sched] [ 732.940996] drm_sched_process_job+0x34/0xa0 [gpu_sched] [ 732.941910] dma_fence_signal_locked+0x85/0x100 [ 732.942692] dma_fence_signal+0x29/0x50 [ 732.943457] amdgpu_fence_process+0x99/0x120 [amdgpu] [ 732.944393] sdma_v4_0_process_trap_irq+0x81/0xa0 [amdgpu] v2: hold the finished fence at drm_sched_process_job instead of amdgpu_fence_process v3: resume the blank line Signed-off-by: Yintian Tao <yttao(a)amd.com> Reviewed-by: Christian König <christian.koenig(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/scheduler/sched_main.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 2af64459b3d77..dfb29e6eeff1e 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -627,7 +627,9 @@ static void drm_sched_process_job(struct dma_fence *f, struct dma_fence_cb *cb) trace_drm_sched_process_job(s_fence); + dma_fence_get(&s_fence->finished); drm_sched_fence_finished(s_fence); + dma_fence_put(&s_fence->finished); wake_up_interruptible(&sched->wake_up_worker); } -- 2.20.1

1 0

[PATCH AUTOSEL 5.5 31/35] drm/scheduler: fix rare NULL ptr race
by Sasha Levin 07 Apr '20

07 Apr '20

From: Yintian Tao <yttao(a)amd.com> [ Upstream commit 3c0fdf3302cb4f186c871684eac5c407a107e480 ] There is one one corner case at dma_fence_signal_locked which will raise the NULL pointer problem just like below. ->dma_fence_signal ->dma_fence_signal_locked ->test_and_set_bit here trigger dma_fence_release happen due to the zero of fence refcount. ->dma_fence_put ->dma_fence_release ->drm_sched_fence_release_scheduled ->call_rcu here make the union fled “cb_list” at finished fence to NULL because struct rcu_head contains two pointer which is same as struct list_head cb_list Therefore, to hold the reference of finished fence at drm_sched_process_job to prevent the null pointer during finished fence dma_fence_signal [ 732.912867] BUG: kernel NULL pointer dereference, address: 0000000000000008 [ 732.914815] #PF: supervisor write access in kernel mode [ 732.915731] #PF: error_code(0x0002) - not-present page [ 732.916621] PGD 0 P4D 0 [ 732.917072] Oops: 0002 [#1] SMP PTI [ 732.917682] CPU: 7 PID: 0 Comm: swapper/7 Tainted: G OE 5.4.0-rc7 #1 [ 732.918980] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014 [ 732.920906] RIP: 0010:dma_fence_signal_locked+0x3e/0x100 [ 732.938569] Call Trace: [ 732.939003] <IRQ> [ 732.939364] dma_fence_signal+0x29/0x50 [ 732.940036] drm_sched_fence_finished+0x12/0x20 [gpu_sched] [ 732.940996] drm_sched_process_job+0x34/0xa0 [gpu_sched] [ 732.941910] dma_fence_signal_locked+0x85/0x100 [ 732.942692] dma_fence_signal+0x29/0x50 [ 732.943457] amdgpu_fence_process+0x99/0x120 [amdgpu] [ 732.944393] sdma_v4_0_process_trap_irq+0x81/0xa0 [amdgpu] v2: hold the finished fence at drm_sched_process_job instead of amdgpu_fence_process v3: resume the blank line Signed-off-by: Yintian Tao <yttao(a)amd.com> Reviewed-by: Christian König <christian.koenig(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/scheduler/sched_main.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 3c57e84222ca9..5bb9feddbfd6b 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -632,7 +632,9 @@ static void drm_sched_process_job(struct dma_fence *f, struct dma_fence_cb *cb) trace_drm_sched_process_job(s_fence); + dma_fence_get(&s_fence->finished); drm_sched_fence_finished(s_fence); + dma_fence_put(&s_fence->finished); wake_up_interruptible(&sched->wake_up_worker); } -- 2.20.1

1 0

[PATCH 0/4] drm/dp_mst: drm_dp_check_act_status() fixes
by Lyude Paul 06 Apr '20

06 Apr '20

Noticed this while fixing some unrelated issues with NAKs being dropped - we don't wait nearly long enough to receive ACTs from MST hubs in some situations. Also, we take the time to refactor this function a bit. This fixes some ACT timeouts I observed on an EVGA MST hub with i915. Lyude Paul (4): drm/dp_mst: Improve kdocs for drm_dp_check_act_status() drm/dp_mst: Reformat drm_dp_check_act_status() a bit drm/dp_mst: Increase ACT retry timeout to 3s drm/dp_mst: Print errors on ACT timeouts drivers/gpu/drm/drm_dp_mst_topology.c | 50 ++++++++++++++++++--------- 1 file changed, 34 insertions(+), 16 deletions(-) -- 2.25.1

2 12

[PATCH v7 0/4] drm: Add support for integrated privacy screen
by Rajat Jain 06 Apr '20

06 Apr '20

This patchset adds: - (Generic) integrated privacy screen support in drm - Ability in i915 to attach to ACPI nodes - an implmentation for ACPI privacy screen in i915 Rajat Jain (4): intel_acpi: Rename drm_dev local variable to dev drm/connector: Add support for privacy-screen property drm/i915: Lookup and attach ACPI device node for connectors drm/i915: Add support for integrated privacy screen drivers/gpu/drm/drm_atomic_uapi.c | 4 + drivers/gpu/drm/drm_connector.c | 56 ++++++ drivers/gpu/drm/i915/Makefile | 3 +- drivers/gpu/drm/i915/display/intel_acpi.c | 28 ++- drivers/gpu/drm/i915/display/intel_atomic.c | 1 + .../drm/i915/display/intel_display_types.h | 5 + drivers/gpu/drm/i915/display/intel_dp.c | 43 ++++- .../drm/i915/display/intel_privacy_screen.c | 175 ++++++++++++++++++ .../drm/i915/display/intel_privacy_screen.h | 27 +++ drivers/gpu/drm/i915/i915_drv.h | 2 + include/drm/drm_connector.h | 25 +++ 11 files changed, 365 insertions(+), 4 deletions(-) create mode 100644 drivers/gpu/drm/i915/display/intel_privacy_screen.c create mode 100644 drivers/gpu/drm/i915/display/intel_privacy_screen.h -- 2.25.1.481.gfbce0eb801-goog

2 8

← Newer
1
...
85
86
87
88
89
90
91
...
95
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel April 2020