linux-kernel - [PATCH] drm: i915: ensure objects are allocated below 4GB on PAE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20090526162717.GC14808@bombadil.infradead.org>
Date:	Tue, 26 May 2009 12:27:17 -0400
From:	Kyle McMartin <kyle@...artin.ca>
To:	airlied@...hat.com
Cc:	dri-devel@...ts.sf.net, linux-kernel@...r.kernel.org,
	jbarnes@...tuousgeek.org, eric@...olt.net, stable@...nel.org,
	hugh.dickins@...cali.co.uk, linux-mm@...ck.org,
	shaohua.li@...el.com
Subject: [PATCH] drm: i915: ensure objects are allocated below 4GB on PAE

From: Kyle McMartin <kyle@...hat.com>

Ensure we allocate GEM objects below 4GB on PAE machines, otherwise
misery ensues. This patch is based on a patch found on dri-devel by
Shaohua Li, but Keith P. expressed reticence that the changes unfairly
penalized other hardware.

(The mm/shmem.c hunk is necessary to ensure the DMA32 flag isn't used
 by the slab allocator via radix_tree_preload, which will hit a
 WARN_ON.)

Signed-off-by: Kyle McMartin <kyle@...hat.com>
---

We're shipping a variant of this in Fedora 11 to fix a myriad of bugs on
PAE hardware.

cheers, Kyle

---
diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
index 4984aa8..ae52edc 100644
--- a/drivers/gpu/drm/drm_gem.c
+++ b/drivers/gpu/drm/drm_gem.c
@@ -142,6 +142,9 @@ drm_gem_object_alloc(struct drm_device *dev, size_t size)
 		return NULL;
 	}
 
+	if (dev->gem_flags)
+		mapping_set_gfp_mask(obj->filp->f_mapping, dev->gem_flags);
+
 	kref_init(&obj->refcount);
 	kref_init(&obj->handlecount);
 	obj->size = size;
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 53d5445..c89ae3d 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1153,12 +1153,12 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 	}
 
 #ifdef CONFIG_HIGHMEM64G
-	/* don't enable GEM on PAE - needs agp + set_memory_* interface fixes */
-	dev_priv->has_gem = 0;
-#else
+	/* avoid allocating buffers above 4GB on PAE */
+	dev->gem_flags = GFP_USER | GFP_DMA32;
+#endif
+
 	/* enable GEM by default */
 	dev_priv->has_gem = 1;
-#endif
 
 	dev->driver->get_vblank_counter = i915_get_vblank_counter;
 	if (IS_GM45(dev))
diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index c8c4221..3744c1f 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1019,6 +1019,7 @@ struct drm_device {
 	uint32_t gtt_total;
 	uint32_t invalidate_domains;    /* domains pending invalidation */
 	uint32_t flush_domains;         /* domains pending flush */
+	gfp_t gem_flags;		/* object allocation flags */
 	/*@} */
 
 };
diff --git a/mm/shmem.c b/mm/shmem.c
index b25f95c..e615887 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1241,7 +1241,7 @@ repeat:
 		 * Try to preload while we can wait, to not make a habit of
 		 * draining atomic reserves; but don't latch on to this cpu.
 		 */
-		error = radix_tree_preload(gfp & ~__GFP_HIGHMEM);
+		error = radix_tree_preload(gfp & ~(__GFP_HIGHMEM|__GFP_DMA32));
 		if (error)
 			goto failed;
 		radix_tree_preload_end();
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/