s, comments below. Excellent to
> have
> pictures included. I wrote this reply over several days, sorry if
> it's
> not quite coherent.
>
>
>
>
> > +
> > +
> > +Overview and background
> > +===
> > +
> > +I highly recommend y
mically (v2)")
Suggested-by: Felix Kuehling
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/amd/amdkfd/kfd_crat.c | 11 +++
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
index 8cac497c2c45..a56
On Mon, Jan 11, 2021 at 09:21:48PM +0100, Christian König wrote:
> Am 11.01.21 um 17:40 schrieb Jeremy Cline:
> > check_unmap() is producing a warning about a missing map error check.
> > The return value from dma_map_page() should be checked for an error, not
> > the cal
check_unmap() is producing a warning about a missing map error check.
The return value from dma_map_page() should be checked for an error, not
the caller-provided dma_addr.
Fixes: d099fc8f540a ("drm/ttm: new TT backend allocation pool v3")
Signed-off-by: Jeremy Cline
---
drivers/g
Hi,
On Mon, Jan 11, 2021 at 09:53:56AM +0100, Christian König wrote:
> Am 08.01.21 um 22:58 schrieb Jeremy Cline:
> > dcn20_resource_construct() includes a number of kzalloc(GFP_KERNEL)
> > calls which can sleep, but kernel_fpu_begin() disables preemption and
> > sleepi
On Fri, Jan 08, 2021 at 06:46:17PM -0500, Felix Kuehling wrote:
> Am 2021-01-08 um 11:31 a.m. schrieb Jeremy Cline:
> > KASAN reported a slab-out-of-bounds read of size 1 in
> > kdf_create_vcrat_image_cpu().
> >
> > This occurs when, for example, when on an x86
,max}_fill_clk_mhz. Narrow the scope to just these two parts to
avoid sleeping while using the FPU.
Fixes: 7a8a3430be15 ("amdgpu: Wrap FPU dependent functions in dc20")
Cc: Timothy Pearson
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c | 8 --
ad and hopefully makes the
pointer arithmetic clearer. It should have no functional change beyond
removing the out-of-bounds read.
Fixes: b7b6c38529c9 ("drm/amdkfd: Calculate CPU VCRAT size dynamically (v2)")
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/amd/amdkfd
On Thu, Dec 10, 2020 at 08:56:08AM +0100, Thomas Zimmermann wrote:
> Hi
>
> Am 09.12.20 um 19:04 schrieb Jeremy Cline:
> > Hi,
> >
> > On Tue, Dec 01, 2020 at 11:35:35AM +0100, Thomas Zimmermann wrote:
> > > Using struct drm_device.pdev is d
Hi,
On Tue, Dec 01, 2020 at 11:35:35AM +0100, Thomas Zimmermann wrote:
> Using struct drm_device.pdev is deprecated. Convert nouveau to struct
> drm_device.dev. No functional changes.
>
> Signed-off-by: Thomas Zimmermann
> Cc: Ben Skeggs
> ---
> drivers/gpu/drm/nouveau/dispnv04/arb.c | 12
memory freed by nouveau_bo_del_ttm().
Fixes: 019cbd4a4feb ("drm/nouveau: Initialize GEM object before TTM object")
Cc: Thierry Reding
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_gem.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/gpu/d
Thus, I've broken this out so the fix can be
easily backported.
Cc: sta...@vger.kernel.org
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 30 +++
1 file changed, 30 insertions(+)
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c
b/drivers
e
list and potentially lock their mutex, which is the same class as the
lock protecting the entire list.
Cc: sta...@vger.kernel.org
Signed-off-by: Jeremy Cline
---
Changes since v1:
- Add a mutex_destroy() call when destroying the device struct
drivers/gpu/drm/nouveau/nouveau_drm.c | 10 ++
d
truly
support hot-unplugging.
Cc: sta...@vger.kernel.org
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c
b/drivers/gpu/drm/nouveau/nouveau_drm.c
index d141a5f004af..4fe4d66
postclose() handler. While this does not enable
hot-unplug support for nouveau, it's enough to avoid crashing the kernel
and leads to all the core_hotunplug tests to pass.
This series reroll addresses a missing mutex_destroy() call and a typo
in a commit message.
Jeremy Cline (3):
d
On Wed, Nov 25, 2020 at 01:37:06PM -0500, Lyude Paul wrote:
> On Tue, 2020-11-03 at 14:49 -0500, Jeremy Cline wrote:
> > Rather than protecting the nouveau_drm clients list with the lock within
> > the "client" nouveau_cli, add a dedicated lock to serialize access to
&
On Fri, Nov 06, 2020 at 02:31:44PM +0100, Karol Herbst wrote:
> On Fri, Nov 6, 2020 at 3:17 AM Jeremy Cline wrote:
> >
> > Make use of the devm_drm_dev_alloc() API to bind the lifetime of
> > nouveau_drm structure to the drm_device. This is important because a
> >
On Fri, Nov 06, 2020 at 02:31:44PM +0100, Karol Herbst wrote:
> On Fri, Nov 6, 2020 at 3:17 AM Jeremy Cline wrote:
> >
> > Make use of the devm_drm_dev_alloc() API to bind the lifetime of
> > nouveau_drm structure to the drm_device. This is important because a
> >
currently occurs.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 44 ---
drivers/gpu/drm/nouveau/nouveau_drv.h | 10 --
2 files changed, 26 insertions(+), 28 deletions(-)
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c
b/drivers/gpu/drm
Start on documentation for the Nouveau device structure and the NVIF
client structure it uses. This documentation is not complete as the
structures are non-trivial and I am not familiar with large portions of
them.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drv.h | 67
tions before the drm_device is
destroyed, so I suspect much of the current cleanup code in Nouveau
would benefit from some refactoring to use this.
Finally, although not *strictly* necessary for this series, I included
some documentation for structures I investigated for this work.
Jeremy Cline (3):
patch should introduce no functional changes.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/dispnv04/crtc.c | 10 +++---
drivers/gpu/drm/nouveau/dispnv50/base.c | 2 +-
drivers/gpu/drm/nouveau/dispnv50/base507c.c | 7 ++--
drivers/gpu/drm/nouveau/dispnv50/core.c | 2
e
list and potentially lock their mutex, which is the same class as the
lock protecting the entire list.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 9 +
drivers/gpu/drm/nouveau/nouveau_drv.h | 5 +
2 files changed, 10 insertions(+), 4 deletions(-)
diff --git a/d
postclose() handler. While this does not enable
hot-unplug support for nouveau, it's enough to avoid crashing the kernel
and leads to all the core_hotunplug tests to pass.
Jeremy Cline (3):
drm/nouveau: use drm_dev_unplug() during device removal
drm/nouveau: Add a dedicated mutex for the cli
truly
support hot-unplugging.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c
b/drivers/gpu/drm/nouveau/nouveau_drm.c
index d141a5f004af..4fe4d664c5f2 100644
--- a/drivers/gp
hange. Thus, I've broken this out so the fix can be
easily backported.
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/nouveau/nouveau_drm.c | 30 +++
1 file changed, 30 insertions(+)
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c
b/drivers/gpu/drm/nouveau/nou
On Wed, Sep 16, 2020 at 10:03:22PM +0200, Karol Herbst wrote:
> On Wed, Sep 16, 2020 at 10:01 PM Karol Herbst wrote:
> >
> > On Wed, Sep 16, 2020 at 9:47 PM Jeremy Cline wrote:
> > >
> > > The temp_get() function currently returns negative error numbers or
clear what the
unit is.
Signed-off-by: Jeremy Cline
---
.../nouveau/include/nvkm/subdev/bios/therm.h | 13 ++
.../drm/nouveau/include/nvkm/subdev/therm.h | 26 +++
drivers/gpu/drm/nouveau/nouveau_hwmon.c | 2 +-
.../gpu/drm/nouveau/nvkm/subdev/therm/ba
some, like gp100_temp_get(), do not.
Rather than relying on implementations remembering to clamp values,
dedicate the function return value to error codes and accept a pointer
to an integer where the temperature reading should be stored.
Signed-off-by: Jeremy Cline
---
.../drm/nouveau/include/nvkm
r than adding a
separate helper, but was torn about whether this is clearer than a
separate helper function.
* I added a WARN_ON in places that previously called
nvkm_therm_temp_get() and didn't check the return value for an error.
This may not be a reasonable error handling method.
Jer
On Wed, Sep 09, 2020 at 10:22:14AM +0200, Karol Herbst wrote:
> On Wed, Sep 9, 2020 at 6:06 AM Ben Skeggs wrote:
> >
> > On Thu, 13 Aug 2020 at 06:50, Jeremy Cline wrote:
> > >
> > > Commit d32656373857 ("drm/nouveau/therm/gp100: initial implementation
p_get() to report finer-grain
temperatures, which would be inconvenient for other users of the
function, a second interface has been added to line up with hwmon's
native unit of temperature.
Signed-off-by: Jeremy Cline
---
.../drm/nouveau/include/nvkm/subdev/therm.h | 18 +
Hi folks,
A Fedora user, Mikhail (CCed), reported[0] a regression with their Vega
56 AMD GPU introduced by bfdec2340478 ("drm/amd/display: Implement
dm_pp_get_clock_levels_by_type_with_latency"). The symptom is that
nothing is displayed and looking at the logs (attached to the Bugzilla),
I see gno
c90b6 ("drm/qxl: reapply cursor after resetting primary")
Cc: sta...@vger.kernel.org
Signed-off-by: Jeremy Cline
---
drivers/gpu/drm/qxl/qxl_display.c | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/qxl/qxl_display.c
b/drivers/gpu/drm/qxl/qxl_
34 matches
Mail list logo