aboutsummaryrefslogtreecommitdiffstats
path: root/fs/ocfs2
diff options
context:
space:
mode:
authorIdo Yariv <ido@wizery.com>2012-04-14 23:20:30 +0300
committerColin Cross <ccross@android.com>2012-04-30 13:14:58 -0700
commit6ae285d5333a9c5ebf2a448a73065730b24481f9 (patch)
treead2e51bfcd22c3cdcd635495979f009f94da174e /fs/ocfs2
parentfd8bc3774515bc6d2af3dfe567f902098cb0dea8 (diff)
downloadkernel_samsung_tuna-6ae285d5333a9c5ebf2a448a73065730b24481f9.zip
kernel_samsung_tuna-6ae285d5333a9c5ebf2a448a73065730b24481f9.tar.gz
kernel_samsung_tuna-6ae285d5333a9c5ebf2a448a73065730b24481f9.tar.bz2
arm: vfp: Fix memory corruption on PM suspend
Commit 36af2a47 ("ARM: vfp: Always save VFP state in vfp_pm_suspend") introduced a potential use-after-free bug. On SMP systems, vfp_current_hw_state might hold dangling pointers in case a task which used the VFP last migrates to another CPU and then exits. If vfp_pm_suspend is called while vfp_current_hw_state still holds a pointer to the freed thread_info, that memory location will be written, potentially overwriting a new object allocated there. The original problem is only relevant to UP systems in which the VFP state is stored lazily. Fix this by only storing the VFP state on UP systems, and avoid doing so on SMP ones. Change-Id: I8f7026eb735b340fcef4cf12fbd12b9a0ea08d3f Signed-off-by: Ido Yariv <ido@wizery.com> Signed-off-by: Eyal Shapira <eyal@wizery.com> Signed-off-by: Colin Cross <ccross@android.com>
Diffstat (limited to 'fs/ocfs2')
0 files changed, 0 insertions, 0 deletions