[PATCH] generic_file_buffered_write(): deadlock on vectored write
authorVladimir V. Saveliev <vs@namesys.com>
Tue, 27 Jun 2006 09:53:57 +0000 (02:53 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Wed, 28 Jun 2006 00:32:39 +0000 (17:32 -0700)
generic_file_buffered_write() prefaults in user pages in order to avoid
deadlock on copying from the same page as write goes to.

However, it looks like there is a problem when write is vectored:
fault_in_pages_readable brings in current segment or its part (maxlen).
OTOH, filemap_copy_from_user_iovec is called to copy number of bytes
(bytes) which may exceed current segment, so filemap_copy_from_user_iovec
switches to the next segment which is not brought in yet.  Pagefault is
generated.  That causes the deadlock if pagefault is for the same page
write goes to: page being written is locked and not uptodate, pagefault
will deadlock trying to lock locked page.

[akpm@osdl.org: somewhat rewritten]
Cc: Neil Brown <neilb@suse.de>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
mm/filemap.c

index 9c7334bafda8da5150f4a9f0a7556bda113b8a19..d504d6e98886f2d0d2f5307c7bbe962b8af41e72 100644 (file)
@@ -2095,14 +2095,21 @@ generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov,
        do {
                unsigned long index;
                unsigned long offset;
-               unsigned long maxlen;
                size_t copied;
 
                offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
                index = pos >> PAGE_CACHE_SHIFT;
                bytes = PAGE_CACHE_SIZE - offset;
-               if (bytes > count)
-                       bytes = count;
+
+               /* Limit the size of the copy to the caller's write size */
+               bytes = min(bytes, count);
+
+               /*
+                * Limit the size of the copy to that of the current segment,
+                * because fault_in_pages_readable() doesn't know how to walk
+                * segments.
+                */
+               bytes = min(bytes, cur_iov->iov_len - iov_base);
 
                /*
                 * Bring in the user page that we will copy from _first_.
@@ -2110,10 +2117,7 @@ generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov,
                 * same page as we're writing to, without it being marked
                 * up-to-date.
                 */
-               maxlen = cur_iov->iov_len - iov_base;
-               if (maxlen > bytes)
-                       maxlen = bytes;
-               fault_in_pages_readable(buf, maxlen);
+               fault_in_pages_readable(buf, bytes);
 
                page = __grab_cache_page(mapping,index,&cached_page,&lru_pvec);
                if (!page) {