The patch titled
mm: pagecache write deadlocks EFAULT fix
has been added to the -mm tree. Its filename is
mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix.patch
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: mm: pagecache write deadlocks EFAULT fix
From: Nick Piggin <npiggin@xxxxxxx>
Fix buffered writes to handle -EFAULT properly.
Also, pass 'seglen' rather than 'bytes' to fault_in_pages_readable in the
DEBUG_VM case, which matches !DEBUG_VM and is the correct thing to do.
Signed-off-by: Nick Piggin <npiggin@xxxxxxx>
Acked-by: Badari Pulavarty <pbadari@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
---
include/linux/pagemap.h | 5 +++--
mm/filemap.c | 12 ++++++++++--
2 files changed, 13 insertions(+), 4 deletions(-)
diff -puN include/linux/pagemap.h~mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix include/linux/pagemap.h
--- a/include/linux/pagemap.h~mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix
+++ a/include/linux/pagemap.h
@@ -213,7 +213,7 @@ static inline int fault_in_pages_writeab
return ret;
}
-static inline void fault_in_pages_readable(const char __user *uaddr, int size)
+static inline int fault_in_pages_readable(const char __user *uaddr, int size)
{
volatile char c;
int ret;
@@ -224,8 +224,9 @@ static inline void fault_in_pages_readab
if (((unsigned long)uaddr & PAGE_MASK) !=
((unsigned long)end & PAGE_MASK))
- __get_user(c, end);
+ ret = __get_user(c, end);
}
+ return ret;
}
#endif /* _LINUX_PAGEMAP_H */
diff -puN mm/filemap.c~mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix mm/filemap.c
--- a/mm/filemap.c~mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix
+++ a/mm/filemap.c
@@ -2108,8 +2108,14 @@ retry_noprogress:
* Bring in the user page that we will copy from _first_, this
* minimises the chance we have to break into the slowpath
* below.
+ *
+ * Must check for -EFAULT here, because an unhandled pagefault
+ * from the atomic copy below may not actually be due to an
+ * unmapped user region.
*/
- fault_in_pages_readable(buf, seglen);
+ status = fault_in_pages_readable(buf, seglen);
+ if (unlikely(status))
+ break;
#endif
page = __grab_cache_page(mapping,index,&cached_page,&lru_pvec);
@@ -2200,7 +2206,9 @@ retry_noprogress:
}
} else {
#ifdef CONFIG_DEBUG_VM
- fault_in_pages_readable(buf, bytes);
+ status = fault_in_pages_readable(buf, seglen);
+ if (unlikely(status))
+ break;
#endif
/*
* OK, we took a fault without making progress. Fall
_
Patches currently in -mm which might be from npiggin@xxxxxxx are
mm-comment-mmap_sem--lock_page-lockorder.patch
mm-only-mm-debug-write-deadlocks.patch
mm-fix-pagecache-write-deadlocks.patch
mm-fix-pagecache-write-deadlocks-comment.patch
mm-fix-pagecache-write-deadlocks-mm-pagecache-write-deadlocks-efault-fix.patch
fs-prepare_write-fixes.patch
fs-prepare_write-fixes-fuse-fix.patch
oom-dont-kill-unkillable-children-or-siblings.patch
oom-cleanup-messages.patch
oom-less-memdie.patch
mm-incorrect-vm_fault_oom-returns-from-drivers.patch
mm-add-arch_alloc_page.patch
radix-tree-rcu-lockless-readside.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
