On 2/27/25 13:49, Pavel Begunkov wrote:
On 2/27/25 13:20, Arnd Bergmann wrote:
From: Arnd Bergmann <arnd@xxxxxxxx>
A code rework resulted in an uninitialized return code when COMPAT
mode is disabled:
As mentioned in the lkp report, it should be a false positive.
io_uring/net.c:722:6: error: variable 'ret' is used uninitialized whenever 'if' condition is true [-Werror,-Wsometimes-uninitialized]
722 | if (io_is_compat(req->ctx)) {
| ^~~~~~~~~~~~~~~~~~~~~~
io_uring/net.c:736:15: note: uninitialized use occurs here
736 | if (unlikely(ret))
| ^~~
Since io_is_compat() turns into a compile-time 'false', the #ifdef
here is completely unnecessary, and removing it avoids the warning.
I don't think __get_compat_msghdr() and other helpers are
compiled for !COMPAT. I'd just silence it like:
I guess we're relying on dead code elimination to prevent
linking against them, if that's a normal practise and we
do mandate compilers to do that, then it looks fine to me
if (io_is_compat(req->ctx)) {
ret = -EFAULT;
#ifdef CONFIG_COMPAT
...
#endif CONFIG_COMPAT
}
Let's see if Jens wants to fix it up in the tree.
--
Pavel Begunkov
