Hi all,
After merging the mm tree, today's linux-next build (powerpc
ppc64_defconfig) failed like this:
ld: warning: discarding dynamic section .glink
ld: warning: discarding dynamic section .plt
ld: linkage table error against `hugetlb_vma_lock_release'
ld: stubs don't match calculated size
ld: can not build stubs: bad value
ld: fs/hugetlbfs/inode.o: in function `.remove_inode_hugepages':
inode.c:(.text+0x2538): undefined reference to `.hugetlb_vma_lock_release'
ld: inode.c:(.text+0x25f4): undefined reference to `.hugetlb_vma_lock_release'
Caused by commits
d56100904028 ("hugetlb: use new vma_lock for pmd sharing synchronization")
47bc61e7caf9 ("hugetlb: add vma based lock for pmd sharing")
The definition of hugetlb_vma_lock_release() is guarded by
CONFIG_ARCH_WANT_HUGE_PMD_SHARE, but the usage in fs/hugetlbfs/inode.c
is not.
I have reverted commit d56100904028 (and the following commit) for
today.
--
Cheers,
Stephen Rothwell
On 09/15/22 17:07, Stephen Rothwell wrote:
> Hi all,
>
> After merging the mm tree, today's linux-next build (powerpc
> ppc64_defconfig) failed like this:
>
> ld: warning: discarding dynamic section .glink
> ld: warning: discarding dynamic section .plt
> ld: linkage table error against `hugetlb_vma_lock_release'
> ld: stubs don't match calculated size
> ld: can not build stubs: bad value
> ld: fs/hugetlbfs/inode.o: in function `.remove_inode_hugepages':
> inode.c:(.text+0x2538): undefined reference to `.hugetlb_vma_lock_release'
> ld: inode.c:(.text+0x25f4): undefined reference to `.hugetlb_vma_lock_release'
>
> Caused by commits
>
> d56100904028 ("hugetlb: use new vma_lock for pmd sharing synchronization")
> 47bc61e7caf9 ("hugetlb: add vma based lock for pmd sharing")
>
> The definition of hugetlb_vma_lock_release() is guarded by
> CONFIG_ARCH_WANT_HUGE_PMD_SHARE, but the usage in fs/hugetlbfs/inode.c
> is not.
Sorry about that! When thinking about the need for a stub, I only
considered the !CONFIG_HUGETLB_PAGE case. :(
Andrew, the following change to 47bc61e7caf9 ("hugetlb: add vma based lock for
pmd sharing") will fix the issue. Would you like to just add the fix, or
would you like another version of 47bc61e7caf9, or perhaps another version of
the series? Happy to do whatever is easier for you.
From 8b3031350154e8e401ccfbc5e71cb95ef654d017 Mon Sep 17 00:00:00 2001
From: Mike Kravetz <[email protected]>
Date: Thu, 15 Sep 2022 09:33:44 -0700
Subject: [PATCH] hugetlb: fix build issue for missing hugetlb_vma_lock_release
Add a stub for hugetlb_vma_lock_release to build in the case
CONFIG_HUGETLB_PAGE && !CONFIG_ARCH_WANT_HUGE_PMD_SHARE.
Signed-off-by: Mike Kravetz <[email protected]>
---
mm/hugetlb.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 2207300791e5..cc7877da18d7 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -7008,6 +7008,10 @@ void hugetlb_vma_assert_locked(struct vm_area_struct *vma)
{
}
+void hugetlb_vma_lock_release(struct kref *kref)
+{
+}
+
static void hugetlb_vma_lock_free(struct vm_area_struct *vma)
{
}
--
2.37.2
On Thu, 15 Sep 2022 09:37:08 -0700 Mike Kravetz <[email protected]> wrote:
> Andrew, the following change to 47bc61e7caf9 ("hugetlb: add vma based lock for
> pmd sharing") will fix the issue. Would you like to just add the fix, or
> would you like another version of 47bc61e7caf9, or perhaps another version of
> the series? Happy to do whatever is easier for you.
This simple fix is easiest for me. More importantly, I think it's
easier for reviewers, who don't have to plough through a whole new
series unsure what might have changed.