[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHk-=wgvG1X60i5DQdExmS9DB3-0v3ByDyzLZdewvy6ptf1-+w@mail.gmail.com>
Date: Fri, 13 Oct 2023 13:07:51 -0700
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Uros Bizjak <ubizjak@...il.com>
Cc: Sean Christopherson <seanjc@...gle.com>, x86@...nel.org,
linux-kernel@...r.kernel.org, Nadav Amit <namit@...are.com>,
Ingo Molnar <mingo@...nel.org>,
Andy Lutomirski <luto@...nel.org>,
Brian Gerst <brgerst@...il.com>,
Denys Vlasenko <dvlasenk@...hat.com>,
"H . Peter Anvin" <hpa@...or.com>,
Peter Zijlstra <peterz@...radead.org>,
Thomas Gleixner <tglx@...utronix.de>,
Josh Poimboeuf <jpoimboe@...hat.com>
Subject: Re: [PATCH tip] x86/percpu: Rewrite arch_raw_cpu_ptr()
On Fri, 13 Oct 2023 at 12:30, Uros Bizjak <ubizjak@...il.com> wrote:
>
> There are some other issues when memory access to the percpu area is
> implemented with an asm. An ongoing analysis shows that compilers
> can't CSE asm over basic-block boundaries, the CSE of asm is a very
> simple pattern matching through the BB.
Ahh. That explains the odd "partial CSE".
I was assuming that CSE was done on the SSA level with some dominance
analysis. Which presumably all the load simplification ends up doing.
Linus
Powered by blists - more mailing lists