Re: [PATCH 7/7] Revert "drm/i915: Enable semaphores on BDW"

[Daniel Vetter <daniel@ffwll.ch>]

On Mon, Aug 04, 2014 at 11:15:19AM -0700, Rodrigo Vivi wrote:
> This reverts commit 521e62e49a42661a4ee0102644517dbe2f100a23.
> 
> Although POST_SYNC brought a bit of stability to Semaphores on BDW
> it didn't solved all issues and some hungs can still occour when
> semaphores are enabled on BDW. Also some sloweness can be found on some
> igt tests, althoguth it apparently doesn't affect real workloads.
> 
> Besides that, no real performance gain was found on our tests with different
> and even multiple workloads.
> 
> Let's disable it again for now. At least until we are sure it is safe
> to re-enable it.
> 
> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
> ---
>  drivers/gpu/drm/i915/i915_drv.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> index 6c4b25c..ec96f9a 100644
> --- a/drivers/gpu/drm/i915/i915_drv.c
> +++ b/drivers/gpu/drm/i915/i915_drv.c
> @@ -481,6 +481,10 @@ bool i915_semaphore_is_enabled(struct drm_device *dev)
>  	if (i915.semaphores >= 0)
>  		return i915.semaphores;
>  
> +	/* Until we get further testing... */
> +	if (IS_GEN8(dev))
> +		return false;

Aside: With this we can't test the code any more at all. The usual
approach for adjusting defaults when it's not the same on all platforms is
to set the module option to -1 (per-platform defaults) and have a
sanitize_foo function call to set it to the correct default. Would be nice
on top of the revert.
-Daniel

> +
>  #ifdef CONFIG_INTEL_IOMMU
>  	/* Enable semaphores on SNB when IO remapping is off */
>  	if (INTEL_INFO(dev)->gen == 6 && intel_iommu_gfx_mapped)
> -- 
> 1.9.3
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch