[patch] gma500: restore the wb attribute for the scratch page

[patch] gma500: restore the wb attribute for the scratch page

[Suresh Siddha]

Alan, Found this while checking who is using set_pages_uc() in the
current mainline code. I am also looking to see if we should add bug
check in __free_page() to avoid this kind of attribute leak or better
yet, restore the wb attribute in __free_page() itself (perhaps using
something like arch_free_page).

---
From: Suresh Siddha <suresh.b.siddha@intel.com>
Subject: gma500: restore the wb attribute for the scratch page

psb_driver_load() is changing the scratch_page attribute to uncached.
Restore the write-back attribute before freeing the page in the unload
routine.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
---
 drivers/staging/gma500/psb_drv.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/staging/gma500/psb_drv.c b/drivers/staging/gma500/psb_drv.c
index b2cdce7..cb45ad7 100644
--- a/drivers/staging/gma500/psb_drv.c
+++ b/drivers/staging/gma500/psb_drv.c
@@ -285,6 +285,7 @@ static int psb_driver_unload(struct drm_device *dev)
 		}
 		psb_gtt_takedown(dev);
 		if (dev_priv->scratch_page) {
+			set_pages_wb(dev_priv->scratch_page, 1);
 			__free_page(dev_priv->scratch_page);
 			dev_priv->scratch_page = NULL;
 		}

Re: [patch] gma500: restore the wb attribute for the scratch page

[Alan Cox]

> From: Suresh Siddha <suresh.b.siddha@intel.com>
> Subject: gma500: restore the wb attribute for the scratch page
> 
> psb_driver_load() is changing the scratch_page attribute to uncached.
> Restore the write-back attribute before freeing the page in the unload
> routine.

Thanks will apply.