linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] fs: generic_copy_file_checks: Do not adjust count based on file size
@ 2021-01-26  5:50 Nicolas Boichat
  2021-01-26 23:38 ` Dave Chinner
  2021-01-26 23:50 ` Al Viro
  0 siblings, 2 replies; 6+ messages in thread
From: Nicolas Boichat @ 2021-01-26  5:50 UTC (permalink / raw)
  To: Darrick J . Wong
  Cc: Dave Chinner, Luis Lozano, Ian Lance Taylor, Nicolas Boichat,
	Alexander Viro, Amir Goldstein, Darrick J. Wong, Dave Chinner,
	linux-fsdevel, linux-kernel

copy_file_range (which calls generic_copy_file_checks) uses the
inode file size to adjust the copy count parameter. This breaks
with special filesystems like procfs/sysfs, where the file size
appears to be zero, but content is actually returned when a read
operation is performed.

This commit ignores the source file size, and makes copy_file_range
match the end of file behaviour documented in POSIX's "read",
where 0 is returned to mark EOF. This would allow "cp" and other
standard tools to make use of copy_file_range with the exact same
behaviour as they had in the past.

Fixes: 96e6e8f4a68d ("vfs: add missing checks to copy_file_range")
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
---
This can be reproduced with this simple test case:
 #define _GNU_SOURCE
 #include <fcntl.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <sys/stat.h>
 #include <unistd.h>

 int
 main(int argc, char **argv)
 {
   int fd_in, fd_out;
   loff_t ret;

   fd_in = open("/proc/version", O_RDONLY);
   fd_out = open("version", O_CREAT | O_WRONLY | O_TRUNC, 0644);

   do {
     ret = copy_file_range(fd_in, NULL, fd_out, NULL, 1024, 0);
     printf("%d bytes copied\n", (int)ret);
   } while (ret > 0);

   return 0;
 }

Without this patch, `version` output file is empty, and no bytes
are copied:
0 bytes copied

With this patch, the loop runs twice and the content of the file
is copied:
315 bytes copied
0 bytes copied

We hit this issue when upgrading Go compiler from 1.13 to 1.15 [1],
as we use Go's `io.Copy` to copy the content of
`/sys/kernel/debug/tracing/trace` to a temporary file.

Under the hood, Go 1.15 uses `copy_file_range` syscall to optimize the
copy operation. However, that fails to copy any content when the input
file is from sysfs/tracefs, with an apparent size of 0 (but there is
still content when you `cat` it, of course).

[1] http://issuetracker.google.com/issues/178332739

 fs/read_write.c | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/fs/read_write.c b/fs/read_write.c
index 75f764b43418..7236146f6ad7 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -1424,7 +1424,6 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
 	struct inode *inode_in = file_inode(file_in);
 	struct inode *inode_out = file_inode(file_out);
 	uint64_t count = *req_count;
-	loff_t size_in;
 	int ret;
 
 	ret = generic_file_rw_checks(file_in, file_out);
@@ -1442,13 +1441,6 @@ static int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
 	if (pos_in + count < pos_in || pos_out + count < pos_out)
 		return -EOVERFLOW;
 
-	/* Shorten the copy to EOF */
-	size_in = i_size_read(inode_in);
-	if (pos_in >= size_in)
-		count = 0;
-	else
-		count = min(count, size_in - (uint64_t)pos_in);
-
 	ret = generic_write_check_limits(file_out, pos_out, &count);
 	if (ret)
 		return ret;
-- 
2.30.0.280.ga3ce27912f-goog


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2021-02-12  4:49 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-26  5:50 [PATCH] fs: generic_copy_file_checks: Do not adjust count based on file size Nicolas Boichat
2021-01-26 23:38 ` Dave Chinner
2021-01-28  0:46   ` Nicolas Boichat
2021-01-28  5:57     ` Darrick J. Wong
2021-02-12  4:48       ` Nicolas Boichat
2021-01-26 23:50 ` Al Viro

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).