[PATCH] drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

[PATCH] drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

[Michel Dänzer]

From: Michel Dänzer <michel.daenzer@amd.com>

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84627
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
---
 drivers/gpu/drm/radeon/radeon_object.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c
index 33e6c7a..d4f472b 100644
--- a/drivers/gpu/drm/radeon/radeon_object.c
+++ b/drivers/gpu/drm/radeon/radeon_object.c
@@ -233,6 +233,13 @@ int radeon_bo_create(struct radeon_device *rdev,
 	if (!(rdev->flags & RADEON_IS_PCIE))
 		bo->flags &= ~(RADEON_GEM_GTT_WC | RADEON_GEM_GTT_UC);
 
+#ifdef CONFIG_X86_32
+	/* XXX: Write-combined CPU mappings of GTT seem broken on 32-bit
+	 * See https://bugs.freedesktop.org/show_bug.cgi?id=84627
+	 */
+	bo->flags &= ~RADEON_GEM_GTT_WC;
+#endif
+
 	radeon_ttm_placement_from_domain(bo, domain);
 	/* Kernel allocation are uninterruptible */
 	down_read(&rdev->pm.mclk_lock);
-- 
2.1.3

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH] drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

[Alex Deucher]

On Thu, Nov 27, 2014 at 4:00 AM, Michel Dänzer <michel@daenzer.net> wrote:
> From: Michel Dänzer <michel.daenzer@amd.com>
>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84627
> Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>

Applied to my -fixes tree.

Thanks,

Alex

> ---
>  drivers/gpu/drm/radeon/radeon_object.c | 7 +++++++
>  1 file changed, 7 insertions(+)
>
> diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c
> index 33e6c7a..d4f472b 100644
> --- a/drivers/gpu/drm/radeon/radeon_object.c
> +++ b/drivers/gpu/drm/radeon/radeon_object.c
> @@ -233,6 +233,13 @@ int radeon_bo_create(struct radeon_device *rdev,
>         if (!(rdev->flags & RADEON_IS_PCIE))
>                 bo->flags &= ~(RADEON_GEM_GTT_WC | RADEON_GEM_GTT_UC);
>
> +#ifdef CONFIG_X86_32
> +       /* XXX: Write-combined CPU mappings of GTT seem broken on 32-bit
> +        * See https://bugs.freedesktop.org/show_bug.cgi?id=84627
> +        */
> +       bo->flags &= ~RADEON_GEM_GTT_WC;
> +#endif
> +
>         radeon_ttm_placement_from_domain(bo, domain);
>         /* Kernel allocation are uninterruptible */
>         down_read(&rdev->pm.mclk_lock);
> --
> 2.1.3
>
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH] drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

[Dave Airlie]

On 1 December 2014 at 14:02, Alex Deucher <alexdeucher@gmail.com> wrote:
> On Thu, Nov 27, 2014 at 4:00 AM, Michel Dänzer <michel@daenzer.net> wrote:
>> From: Michel Dänzer <michel.daenzer@amd.com>
>>
>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84627
>> Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
>
> Applied to my -fixes tree.

Btw while I'm okay to merge this as a workaround, I'd really like to know
if someone cares enough to fix it.

Because there is no reason WC mappings should be broken on 32-bit,
and its quite likely they are broken on 64-bit too just more subtly.

Dave.
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH] drm/radeon: Ignore RADEON_GEM_GTT_WC on 32-bit x86

[Michel Dänzer]

On 03.12.2014 08:28, Dave Airlie wrote:
> On 1 December 2014 at 14:02, Alex Deucher <alexdeucher@gmail.com> wrote:
>> On Thu, Nov 27, 2014 at 4:00 AM, Michel Dänzer <michel@daenzer.net> wrote:
>>> From: Michel Dänzer <michel.daenzer@amd.com>
>>>
>>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84627
>>> Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
>>
>> Applied to my -fixes tree.
>
> Btw while I'm okay to merge this as a workaround, I'd really like to know
> if someone cares enough to fix it.
>
> Because there is no reason WC mappings should be broken on 32-bit,
> and its quite likely they are broken on 64-bit too just more subtly.

It would definitely be nice to get to the bottom of this, but I'm not 
sure where to dig next. Given that basically the same 64-bit kernel 
seems to fail when built on a 32-bit host but works fine when built on a 
64-bit host, I suspect it might be something from the host toolchain / 
build environment leaking into the resulting binaries somehow.


-- 
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel