uprobes: Fix mmap_region()'s mm->mm_rb corruption if uprobe_mmap() fails
authorOleg Nesterov <oleg@redhat.com>
Sun, 19 Aug 2012 17:10:42 +0000 (19:10 +0200)
committerIngo Molnar <mingo@kernel.org>
Tue, 21 Aug 2012 09:48:12 +0000 (11:48 +0200)
commitc7a3a88c938fbe3d70c2278e082b80eb830d1c58
treee25070dc5141c2614fbcfaae543de4237418a623
parent9160338de92c0305329be5163a76f849806e83de
uprobes: Fix mmap_region()'s mm->mm_rb corruption if uprobe_mmap() fails

This patch fixes:

  https://bugzilla.redhat.com/show_bug.cgi?id=843640

If mmap_region()->uprobe_mmap() fails, unmap_and_free_vma path
does unmap_region() but does not remove the soon-to-be-freed vma
from rb tree. Actually there are more problems but this is how
William noticed this bug.

Perhaps we could do do_munmap() + return in this case, but in
fact it is simply wrong to abort if uprobe_mmap() fails. Until
at least we move the !UPROBE_COPY_INSN code from
install_breakpoint() to uprobe_register().

For example, uprobe_mmap()->install_breakpoint() can fail if the
probed insn is not supported (remember, uprobe_register()
succeeds if nobody mmaps inode/offset), mmap() should not fail
in this case.

dup_mmap()->uprobe_mmap() is wrong too by the same reason,
fork() can race with uprobe_register() and fail for no reason if
it wins the race and does install_breakpoint() first.

And, if nothing else, both mmap_region() and dup_mmap() return
success if uprobe_mmap() fails. Change them to ignore the error
code from uprobe_mmap().

Reported-and-tested-by: William Cohen <wcohen@redhat.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: <stable@vger.kernel.org> # v3.5
Cc: Anton Arapov <anton@redhat.com>
Cc: William Cohen <wcohen@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20120819171042.GB26957@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/fork.c
mm/mmap.c