Thread (2 messages) 2 messages, 2 authors, 2015-09-21

Re: [PATCH] fs: fix data race on mnt.mnt_flags

From: Kirill A. Shutemov <hidden>
Date: 2015-09-21 14:06:19
Also in: linux-fsdevel, lkml

On Mon, Sep 21, 2015 at 02:16:47PM +0200, Dmitry Vyukov wrote:
do_remount() does:

mnt_flags |= mnt->mnt.mnt_flags & ~MNT_USER_SETTABLE_MASK;
mnt->mnt.mnt_flags = mnt_flags;

This can easily be compiled as:

mnt->mnt.mnt_flags &= ~MNT_USER_SETTABLE_MASK;
mnt->mnt.mnt_flags |= mnt_flags;

(also 2 memory accesses, less register pressure)
The flags are being concurrently read by e.g. do_mmap_pgoff()
which does:

if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)

As the result we can allow to mmap a MNT_NOEXEC mount
as VM_EXEC.

Use WRITE_ONCE() to set new flags.

The data race was found with KernelThreadSanitizer (KTSAN).

Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Acked-by: Kirill A. Shutemov <redacted>

-- 
 Kirill A. Shutemov

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help