Attachment 114795 Details for Bug 89857 – patch to revert

[patch] patch to revert

0001-Revert-drm-i915-Performed-deferred-clflush-inside-se.patch (text/plain), 5.17 KB, created by sudip on 2015-04-01 03:57:39 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: sudip

Created: 2015-04-01 03:57:39 UTC

Size: 5.17 KB

patch

obsolete

>From bf95bea0267cd14116a62691a78d4687ed308b6e Mon Sep 17 00:00:00 2001
>From: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
>To: Daniel Vetter <daniel.vetter@intel.com>
>To: Jani Nikula <jani.nikula@linux.intel.com>
>To: David Airlie <airlied@linux.ie>
>To: Chris Wilson <chris@chris-wilson.co.uk>
>To: Ville SyrjÃ¤lÃ¤ <ville.syrjala@linux.intel.com>
>To: huax.lu@intel.com
>To: Daniel Vetter <daniel.vetter@ffwll.ch>
>Cc: intel-gfx@lists.freedesktop.org
>Cc: dri-devel@lists.freedesktop.org
>Cc: linux-kernel@vger.kernel.org
>Date: Tue, 31 Mar 2015 20:01:51 +0530
>Subject: [PATCH] Revert "drm/i915: Performed deferred clflush inside set-cache-level"
>
>This reverts commit <0f71979ab7fbd0c71c41c2798de3d33937915434>.
>
>my display was getting garbled for a moment very frequently. it looked
>like when the screen was getting refreshed then something was going
>wrong.
>git bisect gave this as the first bad commit, and after reverting it
>now display is not having that problem.
>
>Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
>---
>
>my system is x86_64.
>lspci -k gives:
>"VGA compatible controller: Intel Corporation Xeon E3-1200 v2/3rd Gen Core processor Graphics Controller (rev 09)
>Subsystem: Foxconn International, Inc. Device 0d74
>Kernel driver in use: i915"
>
>This is my bisect log:
># bad: [e42391cd048809d903291d07f86ed3934ce138e9] Linux 4.0-rc6
># good: [b7392d2247cfe6771f95d256374f1a8e6a6f48d6] Linux 3.19-rc2
>git bisect start 'v4.0-rc6' 'v3.19-rc2' '--' 'drivers/gpu/drm/i915/'
># good: [d2182a660808d9053a605e3ebc8c46a323ec6e5d] drm/i915: Don't register HDMI connectors for eDP ports on VLV/CHV
>git bisect good d2182a660808d9053a605e3ebc8c46a323ec6e5d
># bad: [36d21f4c557a2b18ed7c9d254060d4ca07a6c5c7] drm/i915/dsi: remove unnecessary dsi device callbacks
>git bisect bad 36d21f4c557a2b18ed7c9d254060d4ca07a6c5c7
># good: [72f95afa5faaf899f7344879b6ccd5f0cb271b28] drm/i915: Removed duplicate members from submit_request
>git bisect good 72f95afa5faaf899f7344879b6ccd5f0cb271b28
># good: [2844a9214759901f382086644842e39ad6f7d894] drm/i915: Use pipe_name() in the get_plane_config() functions
>git bisect good 2844a9214759901f382086644842e39ad6f7d894
># bad: [1b842c89bd8eb0e9619e1aba071c9a5529b7a179] drm/i915: Fix kzalloc() smatch warnings in get_initial_plane_config()
>git bisect bad 1b842c89bd8eb0e9619e1aba071c9a5529b7a179
># good: [8d360dffd6d8634868e433128d5178bea14cc42c] drm/i915: Specify bsd rings through exec flag
>git bisect good 8d360dffd6d8634868e433128d5178bea14cc42c
># good: [1197b4f230fb7c8fe3a9b549596fe130b09a0db2] drm/i915: Balance context pinning on reset cleanup
>git bisect good 1197b4f230fb7c8fe3a9b549596fe130b09a0db2
># bad: [0f71979ab7fbd0c71c41c2798de3d33937915434] drm/i915: Performed deferred clflush inside set-cache-level
>git bisect bad 0f71979ab7fbd0c71c41c2798de3d33937915434
># good: [a7cbedec8317a5cacecb567674fdbc1c3fb22de8] drm/i915: Rename unpin_count to pin_count
>git bisect good a7cbedec8317a5cacecb567674fdbc1c3fb22de8
>
> drivers/gpu/drm/i915/i915_drv.h |  1 -
> drivers/gpu/drm/i915/i915_gem.c | 31 ++++++++++++++++++++++---------
> 2 files changed, 22 insertions(+), 10 deletions(-)
>
>diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
>index 8727086..abe6c16 100644
>--- a/drivers/gpu/drm/i915/i915_drv.h
>+++ b/drivers/gpu/drm/i915/i915_drv.h
>@@ -2054,7 +2054,6 @@ struct drm_i915_gem_object {
> 	 */
> 	unsigned long gt_ro:1;
> 	unsigned int cache_level:3;
>-	unsigned int cache_dirty:1;
> 
> 	unsigned int has_dma_mapping:1;
> 
>diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
>index 27ea6bd..e9341e9 100644
>--- a/drivers/gpu/drm/i915/i915_gem.c
>+++ b/drivers/gpu/drm/i915/i915_gem.c
>@@ -3647,14 +3647,11 @@ i915_gem_clflush_object(struct drm_i915_gem_object *obj,
> 	 * snooping behaviour occurs naturally as the result of our domain
> 	 * tracking.
> 	 */
>-	if (!force && cpu_cache_is_coherent(obj->base.dev, obj->cache_level)) {
>-		obj->cache_dirty = true;
>+	if (!force && cpu_cache_is_coherent(obj->base.dev, obj->cache_level))
> 		return false;
>-	}
> 
> 	trace_i915_gem_object_clflush(obj);
> 	drm_clflush_sg(obj->pages);
>-	obj->cache_dirty = false;
> 
> 	return true;
> }
>@@ -3836,11 +3833,27 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
> 		vma->node.color = cache_level;
> 	obj->cache_level = cache_level;
> 
>-	if (obj->cache_dirty &&
>-	    obj->base.write_domain != I915_GEM_DOMAIN_CPU &&
>-	    cpu_write_needs_clflush(obj)) {
>-		if (i915_gem_clflush_object(obj, true))
>-			i915_gem_chipset_flush(obj->base.dev);
>+	if (cpu_write_needs_clflush(obj)) {
>+		u32 old_read_domains, old_write_domain;
>+
>+		/* If we're coming from LLC cached, then we haven't
>+		 * actually been tracking whether the data is in the
>+		 * CPU cache or not, since we only allow one bit set
>+		 * in obj->write_domain and have been skipping the clflushes.
>+		 * Just set it to the CPU cache for now.
>+		 */
>+		i915_gem_object_retire(obj);
>+		WARN_ON(obj->base.write_domain & ~I915_GEM_DOMAIN_CPU);
>+
>+		old_read_domains = obj->base.read_domains;
>+		old_write_domain = obj->base.write_domain;
>+
>+		obj->base.read_domains = I915_GEM_DOMAIN_CPU;
>+		obj->base.write_domain = I915_GEM_DOMAIN_CPU;
>+
>+		trace_i915_gem_object_change_domain(obj,
>+						    old_read_domains,
>+						    old_write_domain);
> 	}
> 
> 	return 0;
>-- 
>1.8.1.2
>

Actions: View | Splinter Review

Attachments on bug 89857: 114795 | 114801 | 114802 | 116649

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.