[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZpHCSKuKDjzuSmXx@yury-ThinkPad>
Date: Fri, 12 Jul 2024 16:54:48 -0700
From: Yury Norov <yury.norov@...il.com>
To: Kees Cook <kees@...nel.org>
Cc: Brian Norris <briannorris@...omium.org>,
Nathan Chancellor <nathan@...nel.org>, llvm@...ts.linux.dev,
linux-kernel@...r.kernel.org, Bill Wendling <morbo@...gle.com>,
Justin Stitt <justinstitt@...gle.com>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Rasmus Villemoes <linux@...musvillemoes.dk>
Subject: Re: [PATCH v2] bitmap: Switch from inline to __always_inline
On Fri, Jul 12, 2024 at 11:01:13AM -0700, Kees Cook wrote:
> On Fri, Jul 12, 2024 at 10:02:55AM -0700, Yury Norov wrote:
> > Hi Brian,
> >
> > Thanks for taking over this!
> >
> > + Kees Cook for GCC
> > [...]
> > But I'm not sure about that and don't know how to check what happens
> > under the compilers' hood. Can compiler gurus please clarify?
>
> I don't know much about GCC internals. I just ask GCC devs nicely to
> help us where they can. :)
My concern here is that this __always_inline may hide a compiler's
inability to inline things like that properly. In that case, we
shouldn't convert bitmaps, and should file a bug to compilers.
>From your and LLVM people comments, it seems it's OK to convert
kernel code. I just want to make it explicit before moving forward.
> > > Subject: [PATCH 1/3] bitmap: switch from inline to __always_inline
>
> We always expected them to be inline, and if we need to hit the
> compilers harder with __always_inline, that seems sensible to me.
>
> Reviewed-by: Kees Cook <kees@...nel.org>
Is that for bitmaps only, or for all files in the patch?
Powered by blists - more mailing lists