Re: [PATCH v9 02/11] powerpc: prepare string/mem functions for KASAN

[Daniel Axtens]

Christophe Leroy <christophe.leroy@xxxxxx> writes:

> Le 04/03/2019 Ã 06:26, Daniel Axtens a ÃcritÂ:
>> Hi Christophe,
>>> diff --git a/arch/powerpc/include/asm/kasan.h b/arch/powerpc/include/asm/kasan.h
>>> new file mode 100644
>>> index 000000000000..c3161b8fc017
>>> --- /dev/null
>>> +++ b/arch/powerpc/include/asm/kasan.h
>>> @@ -0,0 +1,15 @@
>>> +/* SPDX-License-Identifier: GPL-2.0 */
>>> +#ifndef __ASM_KASAN_H
>>> +#define __ASM_KASAN_H
>>> +
>>> +#ifdef CONFIG_KASAN
>>> +#define _GLOBAL_KASAN(fn)	.weak fn ; _GLOBAL(__##fn) ; _GLOBAL(fn)
>>> +#define _GLOBAL_TOC_KASAN(fn)	.weak fn ; _GLOBAL_TOC(__##fn) ; _GLOBAL_TOC(fn)
>>> +#define EXPORT_SYMBOL_KASAN(fn)	EXPORT_SYMBOL(__##fn) ; EXPORT_SYMBOL(fn)
>> 
>> I'm having some trouble with this. I get warnings like this:
>
> I don't have such problem, neither with ppc32 nor with ppc64e_defconfig.
>

Fascinating, I'll dig into that more.

> What config are you using ?

Attached, it's based on the one provided with the SDK for the T4240RDB.

> Another (unrelated) question I have:
> On the initial arm64 implementation (39d114ddc682 arm64: add KASAN 
> support) they made KASAN implementation depend on SPARSEMEM_VMEMMAP 
> (allthough they later removed that dependency with commit e17d8025f07e 
> arm64/mm/kasan: don't use vmemmap_populate() to initialize shadow)
>
> So I'm wondering why on your side, KASAN depends on !SPARSEMEM_VMEMMAP

I use the vmemmap area as the shadow region for kasan, in a way that
takes absolutely no account of any other use. It's very possible that I
could instead do something similar to what arm64 used to do - I think
one of the previous ppc64 approaches did something similar too.

Regards,
Daniel

Attachment: .config
Description: Binary data