Impact: remove spurious WARN on legacy SMP percpu allocator

Commit f2a8205c4ef1af917d175c36a4097ae5587791c8 incorrectly added too
tight WARN_ON_ONCE() on alignments for UP and legacy SMP percpu
allocator.  Commit e317603694bfd17b28a40de9d65e1a4ec12f816e fixed it
for UP but legacy SMP allocator was forgotten.  Fix it.

Signed-off-by: Tejun Heo <t...@kernel.org>
Reported-by: Sachin P. Sant <sach...@in.ibm.com>
---
(RESEND: cc'ing Ingo. :-)

Oops, that was a stupid omission.  This patch should fix it.  Ingo,
please pull from the following git vector to receive the first first
four patches from the use-dynamic-percpu-allocator-by-default patchset
(without the actual conversion which can disrupt archs) + this patch.
I moved the actual conversion patch into #tj-percpu-exp branch, so the
pull should be safe.

  git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc.git tj-percpu

Thanks.

 mm/allocpercpu.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/allocpercpu.c b/mm/allocpercpu.c
index 3653c57..1882923 100644
--- a/mm/allocpercpu.c
+++ b/mm/allocpercpu.c
@@ -120,7 +120,7 @@ void *__alloc_percpu(size_t size, size_t align)
         * on it.  Larger alignment should only be used for module
         * percpu sections on SMP for which this path isn't used.
         */
-       WARN_ON_ONCE(align > __alignof__(unsigned long long));
+       WARN_ON_ONCE(align > SMP_CACHE_BYTES);
 
        if (unlikely(!pdata))
                return NULL;
-- 
1.6.0.2

_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev

Reply via email to