Revert "mm: madvise: skip unmapped vma holes passed to process_madvise"
authorCharan Teja Kalla <quic_charante@quicinc.com>
Fri, 1 Apr 2022 18:28:12 +0000 (11:28 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 1 Apr 2022 18:46:09 +0000 (11:46 -0700)
This reverts commit 08095d6310a7 ("mm: madvise: skip unmapped vma holes
passed to process_madvise") as process_madvise() fails to return the
exact processed bytes in other cases too.

As an example: if process_madvise() hits mlocked pages after processing
some initial bytes passed in [start, end), it just returns EINVAL
although some bytes are processed.  Thus making an exception only for
ENOMEM is partially fixing the problem of returning the proper advised
bytes.

Thus revert this patch and return proper bytes advised.

Link: https://lkml.kernel.org/r/e73da1304a88b6a8a11907045117cccf4c2b8374.1648046642.git.quic_charante@quicinc.com
Fixes: 08095d6310a7ce ("mm: madvise: skip unmapped vma holes passed to process_madvise")
Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/madvise.c

index b41858e..1873616 100644 (file)
@@ -1464,16 +1464,9 @@ SYSCALL_DEFINE5(process_madvise, int, pidfd, const struct iovec __user *, vec,
 
        while (iov_iter_count(&iter)) {
                iovec = iov_iter_iovec(&iter);
-               /*
-                * do_madvise returns ENOMEM if unmapped holes are present
-                * in the passed VMA. process_madvise() is expected to skip
-                * unmapped holes passed to it in the 'struct iovec' list
-                * and not fail because of them. Thus treat -ENOMEM return
-                * from do_madvise as valid and continue processing.
-                */
                ret = do_madvise(mm, (unsigned long)iovec.iov_base,
                                        iovec.iov_len, behavior);
-               if (ret < 0 && ret != -ENOMEM)
+               if (ret < 0)
                        break;
                iov_iter_advance(&iter, iovec.iov_len);
        }