[PATCH] drm/i915: fixup hangman rebase goof-up

[PATCH] drm/i915: fixup hangman rebase goof-up

[Daniel Vetter]

I've added a bit of logic such that running the hangman test on chips
without any hw reset support at all doesn't wedge the gpu because the
reset failed. This relied on checking for non-null stop_rings.
Unfortunately I've botched a rebase somewhere and stop_rings is still
cleared at the old place before the reset code.

Fix this up so that running the i-g-t tests on gen2/3 doesn't result
in a wedged gpu.

Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.c |    2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index e41aade..f0bd30a 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -860,6 +860,8 @@ int i915_reset(struct drm_device *dev)
 	if (!mutex_trylock(&dev->struct_mutex))
 		return -EBUSY;
 
+	dev_priv->stop_rings = 0;
+
 	i915_gem_reset(dev);
 
 	ret = -ENODEV;
-- 
1.7.10

Re: [PATCH] drm/i915: fixup hangman rebase goof-up

[Chris Wilson]

On Sat, 16 Jun 2012 16:42:05 +0200, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> I've added a bit of logic such that running the hangman test on chips
> without any hw reset support at all doesn't wedge the gpu because the
> reset failed. This relied on checking for non-null stop_rings.
> Unfortunately I've botched a rebase somewhere and stop_rings is still
> cleared at the old place before the reset code.
> 
> Fix this up so that running the i-g-t tests on gen2/3 doesn't result
> in a wedged gpu.

dev_priv->stop_rings = 0; is there on both dif and dinq. An unpushed
mistake perhaps?
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: [PATCH] drm/i915: fixup hangman rebase goof-up

[Daniel Vetter]

On Sat, Jun 16, 2012 at 4:57 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> On Sat, 16 Jun 2012 16:42:05 +0200, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
>> I've added a bit of logic such that running the hangman test on chips
>> without any hw reset support at all doesn't wedge the gpu because the
>> reset failed. This relied on checking for non-null stop_rings.
>> Unfortunately I've botched a rebase somewhere and stop_rings is still
>> cleared at the old place before the reset code.
>>
>> Fix this up so that running the i-g-t tests on gen2/3 doesn't result
>> in a wedged gpu.
>
> dev_priv->stop_rings = 0; is there on both dif and dinq. An unpushed
> mistake perhaps?

Well, the hangman got merged for 3.5 and has been broken since then.
Yeah, shame on me for not noticing earlier :(

This came about that stop_rings = 0 as removed by this patch was the
earlier place if reset it. But to not wedge the gpu I need to still
know whether this is a simulated gpu hang after the reset code ran and
failed with -ENODEV (indicating the missing reset code). The right
code is in intel_gpu_reset. I've tested before submitting the patches,
but somehow managed to slip in the old hunk somehow in a last-minute
rebase.

This patch just kills this spurious hunk.
-Daniel
-- 
Daniel Vetter
daniel.vetter@ffwll.ch - +41 (0) 79 365 57 48 - http://blog.ffwll.ch

Re: [PATCH] drm/i915: fixup hangman rebase goof-up

[Chris Wilson]

On Sat, 16 Jun 2012 17:04:27 +0200, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> On Sat, Jun 16, 2012 at 4:57 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> > On Sat, 16 Jun 2012 16:42:05 +0200, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> >> I've added a bit of logic such that running the hangman test on chips
> >> without any hw reset support at all doesn't wedge the gpu because the
> >> reset failed. This relied on checking for non-null stop_rings.
> >> Unfortunately I've botched a rebase somewhere and stop_rings is still
> >> cleared at the old place before the reset code.
> >>
> >> Fix this up so that running the i-g-t tests on gen2/3 doesn't result
> >> in a wedged gpu.
> >
> > dev_priv->stop_rings = 0; is there on both dif and dinq. An unpushed
> > mistake perhaps?
> 
> Well, the hangman got merged for 3.5 and has been broken since then.
> Yeah, shame on me for not noticing earlier :(
> 
> This came about that stop_rings = 0 as removed by this patch was the
> earlier place if reset it. But to not wedge the gpu I need to still
> know whether this is a simulated gpu hang after the reset code ran and
> failed with -ENODEV (indicating the missing reset code). The right
> code is in intel_gpu_reset. I've tested before submitting the patches,
> but somehow managed to slip in the old hunk somehow in a last-minute
> rebase.
> 
> This patch just kills this spurious hunk.

...but this patches adds the existing line...

* confused.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

[PATCH] drm/i915: fixup hangman rebase goof-up

[Daniel Vetter]

I've added a bit of logic such that running the hangman test on chips
without any hw reset support at all doesn't wedge the gpu because the
reset failed. This relied on checking for non-null stop_rings.
Unfortunately I've botched a rebase somewhere and stop_rings is still
cleared at the old place before the reset code.

Fix this up so that running the i-g-t tests on gen2/3 doesn't result
in a wedged gpu.

v2: Actually remove the lines instead of adding them twice ... my git
license should be revoked immediately.

Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.c |    2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index f0bd30a..e41aade 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -860,8 +860,6 @@ int i915_reset(struct drm_device *dev)
 	if (!mutex_trylock(&dev->struct_mutex))
 		return -EBUSY;
 
-	dev_priv->stop_rings = 0;
-
 	i915_gem_reset(dev);
 
 	ret = -ENODEV;
-- 
1.7.10

Re: [PATCH] drm/i915: fixup hangman rebase goof-up

[Daniel Vetter]

On Sat, Jun 16, 2012 at 06:29:44PM +0200, Daniel Vetter wrote:
> I've added a bit of logic such that running the hangman test on chips
> without any hw reset support at all doesn't wedge the gpu because the
> reset failed. This relied on checking for non-null stop_rings.
> Unfortunately I've botched a rebase somewhere and stop_rings is still
> cleared at the old place before the reset code.
> 
> Fix this up so that running the i-g-t tests on gen2/3 doesn't result
> in a wedged gpu.
> 
> v2: Actually remove the lines instead of adding them twice ... my git
> license should be revoked immediately.
> 
> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I've queued this one for -next.
-Daniel
-- 
Daniel Vetter
Mail: daniel@ffwll.ch
Mobile: +41 (0)79 365 57 48