Re: [GIT PULL] asm-generic changes for 6.15

From: Nathan Chancellor
Date: Wed Mar 26 2025 - 18:19:54 EST


On Wed, Mar 26, 2025 at 10:32:39PM +0100, Arnd Bergmann wrote:
> On Wed, Mar 26, 2025, at 22:08, Jann Horn wrote:
> > On Wed, Mar 26, 2025 at 8:43 PM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> >> The following changes since commit 2014c95afecee3e76ca4a56956a936e23283f05b:
> >>
> >> Linux 6.14-rc1 (2025-02-02 15:39:26 -0800)
> >>
> >> are available in the Git repository at:
> >>
> >> https://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic.git tags/asm-generic-6.15
> >>
> >> for you to fetch changes up to ece69af2ede103e190ffdfccd9f9ec850606ab5e:
> >>
> >> rwonce: handle KCSAN like KASAN in read_word_at_a_time() (2025-03-25 17:50:38 +0100)
> > [...]
> >> Jann Horn (1):
> >> rwonce: handle KCSAN like KASAN in read_word_at_a_time()
> >
> > Uh, sorry about this...
> >
> > Nathan Chancellor just pointed out that my commit "rwonce: handle
> > KCSAN like KASAN in read_word_at_a_time()" breaks the arm64 build when
> > LTO is enabled (<https://lore.kernel.org/all/20250326203926.GA10484@ax162/>).
> > I just posted a patch that undoes the buggy part of my patch at
> > <https://lore.kernel.org/r/20250326-rwaat-fix-v1-1-600f411eaf23@xxxxxxxxxx>.
> >
> > @Linus: Sorry for throwing a spanner in the works here... maybe you
> > should only pull up to the commit before mine (luckily mine's the last
> > in the stack, and it's not important), or wait for Arnd to give his
> > opinion.
>
> I've already tagged a tags/asm-generic-6.15-2 with your fix included
> and the same tag description.
>
> I don't think it's worth doing a partial pull here.
>
> Linus, if you have already pulled the tags/asm-generic-6.15 tag,
> I suggest you just apply the fix directly yourself, otherwise
> you can use the tags/asm-generic-6.15-2 tag instead, or hold off
> for today and wait for me to send a new pull request after
> I get an ok from Nathan.

Jann's patch resolves the issue for me. My two arm64 boxes happily
boot with it applied to next-20250326 and the virtual test case passes
as well.

Cheers,
Nathan