From: Christian Koenig
The shader preemption on cayman doesn't work
correctly with multiple rings. So to be able to
still make use of the compute rings we use a
semaphore to make sure that only one IB can execute
at the same time.
This isn't as effective as shader
Try to remove or replace the cs_mutex with a
vm_mutex where it is still needed.
Signed-off-by: Christian K?nig
---
drivers/gpu/drm/radeon/radeon.h| 44 +---
drivers/gpu/drm/radeon/radeon_cs.c |7 ++---
drivers/gpu/drm/radeon/radeon_device.c |2
From: Christian Koenig
So we can skip the looking. Also renames sw_int to
ring_int, cause that better matches its purpose.
Signed-off-by: Christian Koenig
---
drivers/gpu/drm/radeon/evergreen.c | 32 -
drivers/gpu/drm/radeon/r100.c |
From: Christian Koenig
1. It is really dangerous to have more than one
spinlock protecting the same information.
2. radeon_irq_set sometimes wasn't called with lock
protection, so it can happen that more than one
CPU would tamper with the irq regs at the same
From: Christian Koenig
The spinlock was actually there to protect the
rptr, but rptr was read outside of the locked area.
v2: Keep the spinlock, looks like there is still something
in the ih that doesn't like irqs enabled.
Signed-off-by: Christian Koenig
---
From: Christian Koenig
Signed-off-by: Christian Koenig
---
drivers/gpu/drm/radeon/radeon.h |2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 3e83480..618df9a 100644
---
It is a rw_semaphore now and only write locked
while changing the clock. Also the lock is renamed
to better reflect what it is protecting.
v2: Keep the ttm_vm_ops on IGPs
Signed-off-by: Christian K?nig
---
drivers/gpu/drm/radeon/radeon.h|3 ++-
Move inter ring syncing with semaphores into the
existing ring allocations, with that we need to
lock the ring mutex only once.
Signed-off-by: Christian K?nig
---
drivers/gpu/drm/radeon/evergreen_blit_kms.c |3 +-
drivers/gpu/drm/radeon/r600.c |5 +-
Signed-off-by: Christian K?nig
---
drivers/gpu/drm/radeon/radeon.h | 23 ++-
drivers/gpu/drm/radeon/radeon_fence.c | 73 +
2 files changed, 85 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon.h
It is completely unnecessary to create fences
before they are emitted, so remove it and a bunch
of checks if fences are emitted or not.
Signed-off-by: Christian K?nig
---
drivers/gpu/drm/radeon/evergreen.c|2 +-
drivers/gpu/drm/radeon/ni.c |2 +-
On 31.05.2012 20:15, Alex Deucher wrote:
> On Thu, May 24, 2012 at 11:35 AM, Alex Deucher
> wrote:
>> On Thu, May 24, 2012 at 3:49 AM, Christian K?nig
>> wrote:
>>> From: Christian Koenig
>>>
>>> 1. It is really dangerous to have more than one
>>>spinlock protecting the same information.
On Thu, May 31, 2012 at 9:24 PM, Singh, Satyeshwar
wrote:
> Is there a reason for the 30 ms delay in the first place?
git blame says there is, it supposedely fixes a bug with tv detection.
-Daniel
--
Daniel Vetter
daniel.vetter at ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch
On Thu, May 31, 2012 at 11:34:37AM +0200, Lars-Peter Clausen wrote:
> >> + drm_helper_mode_fill_fb_struct(_cma->fb, mode_cmd);
> >> +
> >> + for (i = 0; i < num_planes; i++)
> >> + fb_cma->obj[i] = obj[i];
> >
> > Check for valid num_planes before this loop?
> >
>
> Hm, I think the
Does this by extension mean that stride changes should also not be allowed
while page flipping?
Thanks,
Satyeshwar
-Original Message-
From: dri-devel-bounces+satyeshwar.singh=intel.com at lists.freedesktop.org
[mailto:dri-devel-bounces+satyeshwar.singh=intel@lists.freedesktop.org]
Is there a reason for the 30 ms delay in the first place?
-Satyeshwar
-Original Message-
From: dri-devel-bounces+satyeshwar.singh=intel.com at lists.freedesktop.org
[mailto:dri-devel-bounces+satyeshwar.singh=intel@lists.freedesktop.org] On
Behalf Of Daniel Vetter
Sent: Thursday, May
From: Alex Deucher
Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and
simplify render backend map. This now properly sets up the backend map
on r6xx-9xx which should improve 3D performance.
Signed-off-by: Alex Deucher
Signed-off-by: Jerome Glisse
From: Jerome Glisse
Bugs that should be fixed by this patch :
https://bugs.freedesktop.org/show_bug.cgi?id=49792
https://bugzilla.kernel.org/show_bug.cgi?id=43207
https://bugs.freedesktop.org/show_bug.cgi?id=39282
Signed-off-by: Jerome Glisse
---
From: Alex Deucher
Need to program an additional VM register. This doesn't not currently
cause any problems, but allows us to program the proper backend
map in a subsequent patch which should improve performance on these
asics.
Signed-off-by: Alex Deucher
Cc: stable
From: Alex Deucher
While there are cards with more than 8 mem banks, the max
number of banks from a tiling perspective is 8, so cap
the tiling config at 8 banks.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=43448
Signed-off-by: Alex Deucher
Cc: stable at
A page flip is not a mode set, changing the frame buffer pixel format
doesn't make sense and isn't handled by most drivers anyway. Disallow
it.
Signed-off-by: Laurent Pinchart
---
drivers/gpu/drm/drm_crtc.c |8
1 files changed, 8 insertions(+), 0 deletions(-)
diff --git
On Thu, May 31, 2012 at 4:15 PM, Christian K?nig
wrote:
> Signed-off-by: Christian K?nig
> ---
> ?drivers/gpu/drm/radeon/radeon.h ? ? ? | ? 23 ++-
> ?drivers/gpu/drm/radeon/radeon_fence.c | ? 73
> +
> ?2 files changed, 85 insertions(+), 11 deletions(-)
>
On Thu, May 31, 2012 at 4:16 PM, Christian K?nig
wrote:
> From: Christian Koenig
>
> 1. It is really dangerous to have more than one
> ? spinlock protecting the same information.
>
> 2. radeon_irq_set sometimes wasn't called with lock
> ? protection, so it can happen that more than one
> ? CPU
On Thu, May 31, 2012 at 3:44 PM, Singh, Satyeshwar
wrote:
> Does this by extension mean that stride changes should also not be allowed
> while page flipping?
Everything beyond a crtc base address change should require a full modeset.
Alex
> Thanks,
> Satyeshwar
>
> -Original Message-
On Thu, May 31, 2012 at 2:15 PM, Alex Deucher wrote:
> On Thu, May 24, 2012 at 11:35 AM, Alex Deucher
> wrote:
>> On Thu, May 24, 2012 at 3:49 AM, Christian K?nig
>> wrote:
>>> From: Christian Koenig
>>>
>>> 1. It is really dangerous to have more than one
>>> ? spinlock protecting the same
Hi Linus,
as mentioned previously these are the extra bits of drm that relied on the
dma-buf pull to work, the first three just stub out the mmap interface,
and the next set provide vmap export to i915/radeon/nouveau and vmap
import to udl.
Dave.
The following changes since commit
? ?rdev->irq.pflip[crtc] = true;
>> ? ? ? ? ? ? ? ?radeon_irq_set(rdev);
>> ? ? ? ?}
>> - ? ? ? spin_unlock_irqrestore(>irq.pflip_lock[crtc], irqflags);
>> + ? ? ? spin_unlock_irqrestore(>irq.lock, irqflags);
>> ?}
>>
>> ?void radeon_irq_kms_pflip_irq_put(struct radeon_device *rdev, int crtc)
>> @@ -260,12 +268,52 @@ void radeon_irq_kms_pflip_irq_put(struct radeon_device
>> *rdev, int crtc)
>> ? ? ? ?if (crtc < 0 || crtc >= rdev->num_crtc)
>> ? ? ? ? ? ? ? ?return;
>>
>> - ? ? ? spin_lock_irqsave(>irq.pflip_lock[crtc], irqflags);
>> + ? ? ? spin_lock_irqsave(>irq.lock, irqflags);
>> ? ? ? ?BUG_ON(rdev->ddev->irq_enabled && rdev->irq.pflip_refcount[crtc] <=
>> 0);
>> ? ? ? ?if (rdev->ddev->irq_enabled && (--rdev->irq.pflip_refcount[crtc] ==
>> 0)) {
>> ? ? ? ? ? ? ? ?rdev->irq.pflip[crtc] = false;
>> ? ? ? ? ? ? ? ?radeon_irq_set(rdev);
>> ? ? ? ?}
>> - ? ? ? spin_unlock_irqrestore(>irq.pflip_lock[crtc], irqflags);
>> + ? ? ? spin_unlock_irqrestore(>irq.lock, irqflags);
>> +}
>> +
>> +void radeon_irq_kms_enable_afmt(struct radeon_device *rdev, int block)
>> +{
>> + ? ? ? unsigned long irqflags;
>> +
>> + ? ? ? spin_lock_irqsave(>irq.lock, irqflags);
>> + ? ? ? rdev->irq.afmt[block] = true;
>> + ? ? ? radeon_irq_set(rdev);
>> + ? ? ? spin_unlock_irqrestore(>irq.lock, irqflags);
>> +
>> +}
>> +
>> +void radeon_irq_kms_disable_afmt(struct radeon_device *rdev, int block)
>> +{
>> + ? ? ? unsigned long irqflags;
>> +
>> + ? ? ? spin_lock_irqsave(>irq.lock, irqflags);
>> + ? ? ? rdev->irq.afmt[block] = false;
>> + ? ? ? radeon_irq_set(rdev);
>> + ? ? ? spin_unlock_irqrestore(>irq.lock, irqflags);
>> ?}
>>
>
> Should probably also add radeon_irq_kms_[en|dis]able_hpd() function
> and call replaced the calls to *_irq_set() in the *_hpd_init() and
> *_hpd_fini() callbacks for display hotplug.
See attached follow on patch.
Alex
-- next part --
A non-text attachment was scrubbed...
Name: 0001-drm-radeon-apply-Murphy-s-law-to-the-kms-hpd-irq-cod.patch
Type: text/x-patch
Size: 13299 bytes
Desc: not available
URL:
<http://lists.freedesktop.org/archives/dri-devel/attachments/20120531/2386658a/attachment-0001.bin>
2012/5/30 Dave Jones :
> On Wed, May 30, 2012 at 05:58:48PM -0400, Dave Jones wrote:
> ?> On Wed, May 30, 2012 at 11:51:54PM +0200, Daniel Vetter wrote:
> ?> ?> On Wed, May 30, 2012 at 11:31 PM, Dave Jones
> wrote:
> ?> ?> > On this hardware:
> ?> ?> >
> ?> ?> > 00:02.0 VGA compatible
Hi Sascha,
On Wednesday 30 May 2012 18:40:56 Sascha Hauer wrote:
> On Wed, May 30, 2012 at 02:32:59PM +0200, Laurent Pinchart wrote:
> > The SH Mobile LCD controller (LCDC) DRM driver supports the main
> > graphics plane in RGB and YUV formats, as well as the overlay planes (in
> > alpha-blending
Hi Sascha,
On Wednesday 30 May 2012 18:28:12 Sascha Hauer wrote:
> On Wed, May 30, 2012 at 05:40:13PM +0200, Laurent Pinchart wrote:
> > Hi Sascha,
> >
> > Thank you for the patch. I've successfully tested the helper with the new
> > SH Mobile DRM driver. Just a couple of comments below in
Hi Sascha,
Thanks for the patch. Here's a bit more extensive review.
On Thursday 31 May 2012 10:08:54 Sascha Hauer wrote:
> Many embedded drm devices do not have a IOMMU and no dedicated
> memory for graphics. These devices use cma (Contiguous Memory
> Allocator) backed graphics memory. This
On 05/31/2012 10:13 AM, Sascha Hauer wrote:
> On Wed, May 30, 2012 at 05:30:15PM +0200, Lars-Peter Clausen wrote:
>> This patch introduces a set of helper function for implementing the KMS
>> framebuffer layer for drivers which use the drm gem CMA helper function.
>>
>> Signed-off-by: Lars-Peter
On Thu, May 24, 2012 at 09:26:49PM +0200, Daniel Vetter wrote:
> A 30 ms delay is simply way too big to waste cpu cycles on.
>
> Signed-off-by: Daniel Vetter
I've queued this patch here for -next with Chris' irc ack added.
-Daniel
--
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365
2012/5/31 Chris Wilson :
> Before that commit we had no idea that we had run out of property slots.
> I think the WARN is genuine, but maybe we should just bump the count set
> it to WARN_ONCE and hope the conversion to lists arrives sooner rather
> than latter.
> -Chris
>
Chris is right: this is
On Wed, May 30, 2012 at 05:30:15PM +0200, Lars-Peter Clausen wrote:
> This patch introduces a set of helper function for implementing the KMS
> framebuffer layer for drivers which use the drm gem CMA helper function.
>
> Signed-off-by: Lars-Peter Clausen
>
> ---
> Changes since v1:
> *
Many embedded drm devices do not have a IOMMU and no dedicated
memory for graphics. These devices use cma (Contiguous Memory
Allocator) backed graphics memory. This patch provides helper
functions to be able to share the code.
Signed-off-by: Sascha Hauer
---
changes since v1:
- map whole
Hello all,
here are some warning I got after resuming from suspend.
It is not really clear to me how I could debug further. Please tell me if I can
do something else.
I also attached the whole dmesg.
Best Regards,
[170934.181984] [ cut here ]
[170934.181991] WARNING:
On Thu, May 31, 2012 at 3:09 AM, Daniel Vetter wrote:
> On Wed, May 30, 2012 at 11:21:32AM -0400, Sean Paul wrote:
>> On Wed, May 30, 2012 at 10:16 AM, Adam Jackson wrote:
>> > On 5/30/12 8:05 AM, Sean Paul wrote:
>> >
>> >> Yes, definitely. The reason I can't set it via xrandr (easily) is
>> >>
On Wed, May 30, 2012 at 11:21:32AM -0400, Sean Paul wrote:
> On Wed, May 30, 2012 at 10:16 AM, Adam Jackson wrote:
> > On 5/30/12 8:05 AM, Sean Paul wrote:
> >
> >> Yes, definitely. The reason I can't set it via xrandr (easily) is
> >> because we look for lvds downclock modes (in i915) on the
On Thu, May 31, 2012 at 12:35 AM, Eric Anholt wrote:
>
> Did you want pointer for timeout in the userspace api? ?I don't feel
> strongly about it, I just didn't see a use. ?The equivalent API I could
> think of was select(), where apparently linux returns time unwaited,
> while "everyone else"
On Wed, 30 May 2012 18:26:51 -0400, Dave Jones wrote:
> On Wed, May 30, 2012 at 05:58:48PM -0400, Dave Jones wrote:
> > On Wed, May 30, 2012 at 11:51:54PM +0200, Daniel Vetter wrote:
> > > On Wed, May 30, 2012 at 11:31 PM, Dave Jones
> wrote:
> > > > On this hardware:
> > > >
> > > >
https://bugzilla.kernel.org/show_bug.cgi?id=43295
--- Comment #4 from Thanigaivelan 2012-05-31
02:51:44 ---
My System hangs when panic occurred switching back to text console. I have to
restart it again. it happen many time. Please tell me what is the problem in
kernel file.
I am using
https://bugs.freedesktop.org/show_bug.cgi?id=49981
--- Comment #2 from Brad Campbell 2012-05-30 19:15:29
PDT ---
I can replicate this on all radeon enabled kernels including current.
It appears than when more than one head is enabled, a separate set of power
profiles is used which attempts to
https://bugs.freedesktop.org/show_bug.cgi?id=49981
--- Comment #1 from Brad Campbell 2012-05-30 19:13:12
PDT ---
Created attachment 62305
--> https://bugs.freedesktop.org/attachment.cgi?id=62305
The hack I use to enable 3 heads on my iMac without torching the Radeon
--
Configure bugmail:
On Wed, May 30, 2012 at 11:31 PM, Dave Jones wrote:
> On this hardware:
>
> 00:02.0 VGA compatible controller: Intel Corporation 82945G/GZ Integrated
> Graphics Controller (rev 02)
>
> I get this every boot with Linus current tree (up to
> af56e0aa35f3ae2a4c1a6d1000702df1dd78cb76)
Just a quick
On Tue, May 29, 2012 at 6:25 AM, Laurent Pinchart
wrote:
> Hi Tomasz,
Hi Tomasz, Laurent, Mauro,
>
> On Wednesday 23 May 2012 14:10:14 Tomasz Stanislawski wrote:
>> Hello everyone,
>> This patchset adds support for DMABUF [2] importing to V4L2 stack.
>> The support for DMABUF exporting was moved
On Thu, May 31, 2012 at 12:35 AM, Eric Anholt e...@anholt.net wrote:
Did you want pointer for timeout in the userspace api? I don't feel
strongly about it, I just didn't see a use. The equivalent API I could
think of was select(), where apparently linux returns time unwaited,
while everyone
On Wed, May 30, 2012 at 11:21:32AM -0400, Sean Paul wrote:
On Wed, May 30, 2012 at 10:16 AM, Adam Jackson a...@redhat.com wrote:
On 5/30/12 8:05 AM, Sean Paul wrote:
Yes, definitely. The reason I can't set it via xrandr (easily) is
because we look for lvds downclock modes (in i915) on the
On Wed, 30 May 2012 18:26:51 -0400, Dave Jones da...@redhat.com wrote:
On Wed, May 30, 2012 at 05:58:48PM -0400, Dave Jones wrote:
On Wed, May 30, 2012 at 11:51:54PM +0200, Daniel Vetter wrote:
On Wed, May 30, 2012 at 11:31 PM, Dave Jones da...@redhat.com wrote:
On this hardware:
Many embedded drm devices do not have a IOMMU and no dedicated
memory for graphics. These devices use cma (Contiguous Memory
Allocator) backed graphics memory. This patch provides helper
functions to be able to share the code.
Signed-off-by: Sascha Hauer s.ha...@pengutronix.de
---
changes since
On Wed, May 30, 2012 at 05:30:15PM +0200, Lars-Peter Clausen wrote:
This patch introduces a set of helper function for implementing the KMS
framebuffer layer for drivers which use the drm gem CMA helper function.
Signed-off-by: Lars-Peter Clausen l...@metafoo.de
---
Changes since v1:
On Thu, May 24, 2012 at 09:26:49PM +0200, Daniel Vetter wrote:
A 30 ms delay is simply way too big to waste cpu cycles on.
Signed-off-by: Daniel Vetter daniel.vet...@ffwll.ch
I've queued this patch here for -next with Chris' irc ack added.
-Daniel
--
Daniel Vetter
Mail: dan...@ffwll.ch
Hi Sascha,
Thanks for the patch. Here's a bit more extensive review.
On Thursday 31 May 2012 10:08:54 Sascha Hauer wrote:
Many embedded drm devices do not have a IOMMU and no dedicated
memory for graphics. These devices use cma (Contiguous Memory
Allocator) backed graphics memory. This patch
On 05/31/2012 10:13 AM, Sascha Hauer wrote:
On Wed, May 30, 2012 at 05:30:15PM +0200, Lars-Peter Clausen wrote:
This patch introduces a set of helper function for implementing the KMS
framebuffer layer for drivers which use the drm gem CMA helper function.
Signed-off-by: Lars-Peter Clausen
Hi Sascha,
On Wednesday 30 May 2012 18:28:12 Sascha Hauer wrote:
On Wed, May 30, 2012 at 05:40:13PM +0200, Laurent Pinchart wrote:
Hi Sascha,
Thank you for the patch. I've successfully tested the helper with the new
SH Mobile DRM driver. Just a couple of comments below in addition to
Hi Sascha,
On Wednesday 30 May 2012 18:40:56 Sascha Hauer wrote:
On Wed, May 30, 2012 at 02:32:59PM +0200, Laurent Pinchart wrote:
The SH Mobile LCD controller (LCDC) DRM driver supports the main
graphics plane in RGB and YUV formats, as well as the overlay planes (in
alpha-blending mode
2012/5/31 Chris Wilson ch...@chris-wilson.co.uk:
Before that commit we had no idea that we had run out of property slots.
I think the WARN is genuine, but maybe we should just bump the count set
it to WARN_ONCE and hope the conversion to lists arrives sooner rather
than latter.
-Chris
Chris
On Thu, May 31, 2012 at 3:09 AM, Daniel Vetter dan...@ffwll.ch wrote:
On Wed, May 30, 2012 at 11:21:32AM -0400, Sean Paul wrote:
On Wed, May 30, 2012 at 10:16 AM, Adam Jackson a...@redhat.com wrote:
On 5/30/12 8:05 AM, Sean Paul wrote:
Yes, definitely. The reason I can't set it via xrandr
2012/5/30 Dave Jones da...@redhat.com:
On Wed, May 30, 2012 at 05:58:48PM -0400, Dave Jones wrote:
On Wed, May 30, 2012 at 11:51:54PM +0200, Daniel Vetter wrote:
On Wed, May 30, 2012 at 11:31 PM, Dave Jones da...@redhat.com wrote:
On this hardware:
00:02.0 VGA compatible
Hi Sascha,
it's good try to avoid demand paging way by page fault handler.
actually, we don't need demand paging way on non-iommu system. I
looked into your previous patch and I realized that patch was based on
old exynos driver. now patch looks almostly good to me and below is my
comments
Hi everybody,
Here are two small fixes that disallow format changes in page flip operations,
and perform a full mode set instead of a mode set base in the CRTC helper set
config handler if the pixel format changed.
Laurent Pinchart (2):
drm: Don't allow page flip to change pixel format
drm:
A page flip is not a mode set, changing the frame buffer pixel format
doesn't make sense and isn't handled by most drivers anyway. Disallow
it.
Signed-off-by: Laurent Pinchart laurent.pinch...@ideasonboard.com
---
drivers/gpu/drm/drm_crtc.c |8
1 files changed, 8 insertions(+), 0
Test whether the pixel format changes in the mode set handler, and
perform a full mode set instead of a mode set base if it does.
Signed-off-by: Laurent Pinchart laurent.pinch...@ideasonboard.com
---
drivers/gpu/drm/drm_crtc_helper.c |3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
On Thu, May 31, 2012 at 11:34:37AM +0200, Lars-Peter Clausen wrote:
+ drm_helper_mode_fill_fb_struct(fb_cma-fb, mode_cmd);
+
+ for (i = 0; i num_planes; i++)
+ fb_cma-obj[i] = obj[i];
Check for valid num_planes before this loop?
Hm, I think the callers already take
On Thu, May 24, 2012 at 11:35 AM, Alex Deucher alexdeuc...@gmail.com wrote:
On Thu, May 24, 2012 at 3:49 AM, Christian König
deathsim...@vodafone.de wrote:
From: Christian Koenig christian.koe...@amd.com
1. It is really dangerous to have more than one
spinlock protecting the same
On Thu, May 31, 2012 at 2:15 PM, Alex Deucher alexdeuc...@gmail.com wrote:
On Thu, May 24, 2012 at 11:35 AM, Alex Deucher alexdeuc...@gmail.com wrote:
On Thu, May 24, 2012 at 3:49 AM, Christian König
deathsim...@vodafone.de wrote:
From: Christian Koenig christian.koe...@amd.com
1. It is
On Thu, May 31, 2012 at 9:24 PM, Singh, Satyeshwar
satyeshwar.si...@intel.com wrote:
Is there a reason for the 30 ms delay in the first place?
git blame says there is, it supposedely fixes a bug with tv detection.
-Daniel
--
Daniel Vetter
daniel.vet...@ffwll.ch - +41 (0) 79 364 57 48 -
Does this by extension mean that stride changes should also not be allowed
while page flipping?
Thanks,
Satyeshwar
-Original Message-
From: dri-devel-bounces+satyeshwar.singh=intel@lists.freedesktop.org
[mailto:dri-devel-bounces+satyeshwar.singh=intel@lists.freedesktop.org] On
On Thu, May 31, 2012 at 3:44 PM, Singh, Satyeshwar
satyeshwar.si...@intel.com wrote:
Does this by extension mean that stride changes should also not be allowed
while page flipping?
Everything beyond a crtc base address change should require a full modeset.
Alex
Thanks,
Satyeshwar
On 31.05.2012 20:15, Alex Deucher wrote:
On Thu, May 24, 2012 at 11:35 AM, Alex Deucheralexdeuc...@gmail.com wrote:
On Thu, May 24, 2012 at 3:49 AM, Christian König
deathsim...@vodafone.de wrote:
From: Christian Koenigchristian.koe...@amd.com
1. It is really dangerous to have more than one
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/radeon.h | 23 ++-
drivers/gpu/drm/radeon/radeon_fence.c | 73 +
2 files changed, 85 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon.h
It is completely unnecessary to create fences
before they are emitted, so remove it and a bunch
of checks if fences are emitted or not.
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/evergreen.c|2 +-
drivers/gpu/drm/radeon/ni.c |2
From: Christian Koenig christian.koe...@amd.com
Signed-off-by: Christian Koenig christian.koe...@amd.com
---
drivers/gpu/drm/radeon/radeon.h |2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 3e83480..618df9a 100644
---
Move inter ring syncing with semaphores into the
existing ring allocations, with that we need to
lock the ring mutex only once.
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/evergreen_blit_kms.c |3 +-
drivers/gpu/drm/radeon/r600.c |5 +-
From: Christian Koenig christian.koe...@amd.com
The spinlock was actually there to protect the
rptr, but rptr was read outside of the locked area.
v2: Keep the spinlock, looks like there is still something
in the ih that doesn't like irqs enabled.
Signed-off-by: Christian Koenig
From: Christian Koenig christian.koe...@amd.com
1. It is really dangerous to have more than one
spinlock protecting the same information.
2. radeon_irq_set sometimes wasn't called with lock
protection, so it can happen that more than one
CPU would tamper with the irq regs at the same
From: Christian Koenig christian.koe...@amd.com
So we can skip the looking. Also renames sw_int to
ring_int, cause that better matches its purpose.
Signed-off-by: Christian Koenig christian.koe...@amd.com
---
drivers/gpu/drm/radeon/evergreen.c | 32 -
Try to remove or replace the cs_mutex with a
vm_mutex where it is still needed.
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/radeon.h| 44 +---
drivers/gpu/drm/radeon/radeon_cs.c |7 ++---
From: Christian Koenig christian.koe...@amd.com
The shader preemption on cayman doesn't work
correctly with multiple rings. So to be able to
still make use of the compute rings we use a
semaphore to make sure that only one IB can execute
at the same time.
This isn't as effective as shader
It is a rw_semaphore now and only write locked
while changing the clock. Also the lock is renamed
to better reflect what it is protecting.
v2: Keep the ttm_vm_ops on IGPs
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/radeon.h|3 ++-
On Thu, May 31, 2012 at 4:15 PM, Christian König
deathsim...@vodafone.de wrote:
Signed-off-by: Christian König deathsim...@vodafone.de
---
drivers/gpu/drm/radeon/radeon.h | 23 ++-
drivers/gpu/drm/radeon/radeon_fence.c | 73
+
2 files
From: Alex Deucher alexander.deuc...@amd.com
While there are cards with more than 8 mem banks, the max
number of banks from a tiling perspective is 8, so cap
the tiling config at 8 banks.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=43448
Signed-off-by: Alex Deucher
From: Alex Deucher alexander.deuc...@amd.com
Need to program an additional VM register. This doesn't not currently
cause any problems, but allows us to program the proper backend
map in a subsequent patch which should improve performance on these
asics.
Signed-off-by: Alex Deucher
From: Jerome Glisse jgli...@redhat.com
Bugs that should be fixed by this patch :
https://bugs.freedesktop.org/show_bug.cgi?id=49792
https://bugzilla.kernel.org/show_bug.cgi?id=43207
https://bugs.freedesktop.org/show_bug.cgi?id=39282
Signed-off-by: Jerome Glisse jgli...@redhat.com
---
From: Alex Deucher alexander.deuc...@amd.com
Tiling group size is always 256bits on r6xx/r7xx/r8xx/9xx. Also fix and
simplify render backend map. This now properly sets up the backend map
on r6xx-9xx which should improve 3D performance.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
84 matches
Mail list logo