Re: [PATCH v2] x86/pm: fix false positive kmemleak report in msr_build_context()

[Borislav Petkov <bp@alien8.de>]

On Tue, Apr 26, 2022 at 06:24:04PM +0200, Rafael J. Wysocki wrote:
> So can the comment be dropped entirely?

Looks like it to me. All the accesses in wakeup_64.S are done through
those offsets which are computed at build-time so they should always be
valid.

OTOH, I wouldn't mind having there some text making any future person
touching this, aware of where to look when making changes.

Some changes like removing a struct member are nicely caught, ofc,
see below. But for something else which is a lot more subtle having a
comment say "hey, have a look at where this is used in wakeup_64.S and
make sure everything is still kosher" is better than having no comment
at all. IMHO.

Thx.

In file included from arch/x86/kernel/asm-offsets.c:14:
arch/x86/kernel/asm-offsets_64.c: In function ‘main’:
./include/linux/stddef.h:16:33: error: ‘struct saved_context’ has no member named ‘gdt_desc’
   16 | #define offsetof(TYPE, MEMBER)  __builtin_offsetof(TYPE, MEMBER)
      |                                 ^~~~~~~~~~~~~~~~~~
./include/linux/kbuild.h:6:69: note: in definition of macro ‘DEFINE’
    6 |         asm volatile("\n.ascii \"->" #sym " %0 " #val "\"" : : "i" (val))
      |                                                                     ^~~
./include/linux/kbuild.h:11:21: note: in expansion of macro ‘offsetof’
   11 |         DEFINE(sym, offsetof(struct str, mem))
      |

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette