Re: 32 bit qemu regression from v6.5 tip pull [6c480f222128 x86/alternative: Rewrite optimize_nops() some]

From: Thomas Gleixner <tglx@linutronix.de>
To: Paul Gortmaker <paul.gortmaker@windriver.com>,
	Peter Zijlstra <peterz@infradead.org>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>,
	Borislav Petkov <bp@alien8.de>,
	x86@kernel.org, linux-kernel@vger.kernel.org
Subject: Re: 32 bit qemu regression from v6.5 tip pull [6c480f222128 x86/alternative: Rewrite optimize_nops() some]
Date: Mon, 30 Oct 2023 19:24:27 +0100	[thread overview]
Message-ID: <877cn4ynms.ffs@tglx> (raw)
In-Reply-To: <ZT/Lmj3xAdwvLE7R@windriver.com>

On Mon, Oct 30 2023 at 11:28, Paul Gortmaker wrote:
> [Re: 32 bit qemu regression from v6.5 tip pull [6c480f222128 x86/alternative: Rewrite optimize_nops() some]] On 30/10/2023 (Mon 12:44) Peter Zijlstra wrote:
>
>> Thomas was looking at this and wondered if something like the below
>> would help?
>
> I tested this on a vanilla v6.5.7 baseline, for lack of a better choice
> and got six failures in 136 boots - everything else unchanged - even the
> shell instance that builds the kernel.

While the sync_core() invocation is definitely at the wrong place, I did
not really expect that this cures it.

Can you add "debug-alternative" to the kernel command line and log both
a working and the non-working kernel output. It's noisy :)

Also do you have a .config and the qemu command line handy?

Thanks,

        tglx