replace_fd() returns the number of the new file descriptor through the return value of do_dup2(). However its callers never care about the specific number. In fact the caller in receive_fd_replace() treats any non-zero return value as an error and therefore never calls __receive_sock() for most file descriptors, which is a bug.
To fix the bug in receive_fd_replace() and to avoid the same issue happening in future callers, signal success through a plain zero.
Suggested-by: Al Viro viro@zeniv.linux.org.uk Link: https://lore.kernel.org/lkml/20250801220215.GS222315@ZenIV/ Fixes: 173817151b15 ("fs: Expand __receive_fd() to accept existing fd") Fixes: 42eb0d54c08a ("fs: split receive_fd_replace from __receive_fd") Cc: stable@vger.kernel.org Signed-off-by: Thomas Weißschuh thomas.weissschuh@linutronix.de --- Changes in v2: - Move the fix to replace_fd() (Al) - Link to v1: https://lore.kernel.org/r/20250801-fix-receive_fd_replace-v1-1-d46d600c74d6@... --- Untested, it stuck out while reading the code. --- fs/file.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/fs/file.c b/fs/file.c index 6d2275c3be9c6967d16c75d1b6521f9b58980926..f8a271265913951d755a5db559938d589219c4f2 100644 --- a/fs/file.c +++ b/fs/file.c @@ -1330,7 +1330,10 @@ int replace_fd(unsigned fd, struct file *file, unsigned flags) err = expand_files(files, fd); if (unlikely(err < 0)) goto out_unlock; - return do_dup2(files, file, fd, flags); + err = do_dup2(files, file, fd, flags); + if (err < 0) + goto out_unlock; + err = 0;
out_unlock: spin_unlock(&files->file_lock);
--- base-commit: d2eedaa3909be9102d648a4a0a50ccf64f96c54f change-id: 20250801-fix-receive_fd_replace-7fdd5ce6532d
Best regards,