2019-07-12 09:06:00

by Arnd Bergmann

[permalink] [raw]
Subject: [PATCH] slab: work around clang bug #42570

Clang gets rather confused about two variables in the same special
section when one of them is not initialized, leading to an assembler
warning later:

/tmp/slab_common-18f869.s: Assembler messages:
/tmp/slab_common-18f869.s:7526: Warning: ignoring changed section attributes for .data..ro_after_init

Adding an initialization to kmalloc_caches is rather silly here
but does avoid the issue.

Link: https://bugs.llvm.org/show_bug.cgi?id=42570
Signed-off-by: Arnd Bergmann <[email protected]>
---
We might decide to wait until this is fixed in clang, but
so far all versions targetting x86 seem to be affected.
---
mm/slab_common.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 6c49dbb3769e..807490fe217a 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1028,7 +1028,8 @@ struct kmem_cache *__init create_kmalloc_cache(const char *name,
}

struct kmem_cache *
-kmalloc_caches[NR_KMALLOC_TYPES][KMALLOC_SHIFT_HIGH + 1] __ro_after_init;
+kmalloc_caches[NR_KMALLOC_TYPES][KMALLOC_SHIFT_HIGH + 1] __ro_after_init =
+{ /* initialization for https://bugs.llvm.org/show_bug.cgi?id=42570 */ };
EXPORT_SYMBOL(kmalloc_caches);

/*
--
2.20.0


2019-07-12 17:31:28

by David Rientjes

[permalink] [raw]
Subject: Re: [PATCH] slab: work around clang bug #42570

On Fri, 12 Jul 2019, Arnd Bergmann wrote:

> Clang gets rather confused about two variables in the same special
> section when one of them is not initialized, leading to an assembler
> warning later:
>
> /tmp/slab_common-18f869.s: Assembler messages:
> /tmp/slab_common-18f869.s:7526: Warning: ignoring changed section attributes for .data..ro_after_init
>
> Adding an initialization to kmalloc_caches is rather silly here
> but does avoid the issue.
>
> Link: https://bugs.llvm.org/show_bug.cgi?id=42570
> Signed-off-by: Arnd Bergmann <[email protected]>

Acked-by: David Rientjes <[email protected]>

Let me followup on the clang bug as well.