]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
Fix mis-uses of 'cc-option' for warning disablement
authorLinus Torvalds <torvalds@linux-foundation.org>
Wed, 23 Apr 2025 17:08:29 +0000 (10:08 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 29 May 2025 09:14:09 +0000 (11:14 +0200)
commit45a1b9694529f100015fd8973ab607d00d0c64d6
tree11dc85d09b3d816bb243b406352c1f0125496769
parent9e01c6ff221f0d1df7dc94b71b9ae6cab48c72d5
Fix mis-uses of 'cc-option' for warning disablement

commit a79be02bba5c31f967885c7f3bf3a756d77d11d9 upstream.

This was triggered by one of my mis-uses causing odd build warnings on
sparc in linux-next, but while figuring out why the "obviously correct"
use of cc-option caused such odd breakage, I found eight other cases of
the same thing in the tree.

The root cause is that 'cc-option' doesn't work for checking negative
warning options (ie things like '-Wno-stringop-overflow') because gcc
will silently accept options it doesn't recognize, and so 'cc-option'
ends up thinking they are perfectly fine.

And it all works, until you have a situation where _another_ warning is
emitted.  At that point the compiler will go "Hmm, maybe the user
intended to disable this warning but used that wrong option that I
didn't recognize", and generate a warning for the unrecognized negative
option.

Which explains why we have several cases of this in the tree: the
'cc-option' test really doesn't work for this situation, but most of the
time it simply doesn't matter that ity doesn't work.

The reason my recently added case caused problems on sparc was pointed
out by Thomas Weißschuh: the sparc build had a previous explicit warning
that then triggered the new one.

I think the best fix for this would be to make 'cc-option' a bit smarter
about this sitation, possibly by adding an intentional warning to the
test case that then triggers the unrecognized option warning reliably.

But the short-term fix is to replace 'cc-option' with an existing helper
designed for this exact case: 'cc-disable-warning', which picks the
negative warning but uses the positive form for testing the compiler
support.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Link: https://lore.kernel.org/all/20250422204718.0b4e3f81@canb.auug.org.au/
Explained-by: Thomas Weißschuh <linux@weissschuh.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Makefile
arch/loongarch/kernel/Makefile
arch/loongarch/kvm/Makefile
arch/riscv/kernel/Makefile
scripts/Makefile.extrawarn