[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <YsLvDHYlLgmfzP4n@hirez.programming.kicks-ass.net>
Date: Mon, 4 Jul 2022 15:45:48 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Guo Ren <guoren@...nel.org>
Cc: Palmer Dabbelt <palmer@...osinc.com>,
Arnd Bergmann <arnd@...db.de>, Ingo Molnar <mingo@...hat.com>,
Will Deacon <will@...nel.org>,
Waiman Long <longman@...hat.com>,
Boqun Feng <boqun.feng@...il.com>,
linux-riscv <linux-riscv@...ts.infradead.org>,
linux-arch <linux-arch@...r.kernel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Guo Ren <guoren@...ux.alibaba.com>
Subject: Re: [PATCH V7 4/5] asm-generic: spinlock: Add combo spinlock (ticket
& queued)
On Mon, Jul 04, 2022 at 09:13:40PM +0800, Guo Ren wrote:
> > Urggghhhh....
> >
> > I really don't think you want this in generic code. Also, I'm thinking
> > any arch that does this wants to make sure it doesn't inline any of this
> Your advice is the same with Arnd, I would move static_branch out of generic.
>
> > stuff. That is, said arch must not have ARCH_INLINE_SPIN_*
> What do you mean? I've tested with ARCH_INLINE_SPIN_* and it's okay
> with EXPORT_SYMBOL(use_qspinlock_key).
Well, with the static_branch and the two paths I just don't see the code
being sane/small enough to inline. I mean, sure, you can force it to
inline the thing, but I'm not sure that's wise.
Powered by blists - more mailing lists