[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <8bcf5f5a-a4e6-41cb-84a8-e2919b0785cb@lucifer.local>
Date: Thu, 29 May 2025 06:39:09 +0100
From: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: Lorenzo Stoakes via B4 Relay <devnull+lorenzo.stoakes.oracle.com@...nel.org>,
"Liam R. Howlett" <Liam.Howlett@...cle.com>,
Vlastimil Babka <vbabka@...e.cz>, Jann Horn <jannh@...gle.com>,
Pedro Falcato <pfalcato@...e.de>, linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] tools/testing/vma: add missing function stub
On Wed, May 28, 2025 at 03:52:18PM -0700, Andrew Morton wrote:
> On Wed, 28 May 2025 15:15:39 +0100 Lorenzo Stoakes via B4 Relay <devnull+lorenzo.stoakes.oracle.com@...nel.org> wrote:
>
> > From: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
> >
> > The hugetlb fix introduced in commit ee40c9920ac2 ("mm: fix copy_vma()
> > error handling for hugetlb mappings") mistakenly did not provide a stub for
> > the VMA userland testing, which results in a compile error when trying to
> > build this.
>
> Thanks, I'll add the Fixes: and the cc:stable (because ee40c9920ac2 had
> cc:stable).
Yeah I intentionally excluded those, as it'll lead to some backport pain
for something that isn't shipped.
I'm not sure if we generally backport test fixes as a rule? Though I
suppose it might be useful if somebody is investigating an issue in a
stable kernel.
>
> > Provide this stub to resolve the issue.
> >
> > ---
> > Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
>
> SOB goes above the "^---$", please ;)
b4 decided to do this :) first experiment with it. Will look at options, I also
pinged Konstantin.
>
> > --- a/tools/testing/vma/vma_internal.h
> > +++ b/tools/testing/vma/vma_internal.h
> > @@ -1461,4 +1461,9 @@ static inline int __call_mmap_prepare(struct file *file,
> > return file->f_op->mmap_prepare(desc);
> > }
> >
> > +static inline void fixup_hugetlb_reservations(struct vm_area_struct *vma)
> > +{
> > + (void)vma;
>
> <Curiousity> Was this statement found to be needed? Normally we just
> use an empty function body in such cases.
I _could have sworn_ I was getting warnings in the gcc version I was using
the past by not doing this.
At any rate, it's consistent-ish with other functions here.
Since this is a churn-friendly file, I"ll probably do a pass to make all
the signatures consistent (by just removing variable names probably) at
some point.
>
> > +}
> > +
> > #endif /* __MM_VMA_INTERNAL_H */
>
Powered by blists - more mailing lists