[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240711090329.GI4587@noisy.programming.kicks-ass.net>
Date: Thu, 11 Jul 2024 11:03:29 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Pawan Gupta <pawan.kumar.gupta@...ux.intel.com>
Cc: Borislav Petkov <bp@...en8.de>,
Dave Hansen <dave.hansen@...ux.intel.com>,
linux-kernel@...r.kernel.org, x86@...nel.org,
Robert Gill <rtgill82@...il.com>,
Jari Ruusu <jariruusu@...tonmail.com>,
Brian Gerst <brgerst@...il.com>,
"Linux regression tracking (Thorsten Leemhuis)" <regressions@...mhuis.info>,
antonio.gomez.iglesias@...ux.intel.com,
daniel.sneddon@...ux.intel.com, stable@...r.kernel.org
Subject: Re: [PATCH v4] x86/entry_32: Use stack segment selector for VERW
operand
On Wed, Jul 10, 2024 at 12:06:47PM -0700, Pawan Gupta wrote:
> +/*
> + * Safer version of CLEAR_CPU_BUFFERS that uses %ss to reference VERW operand
> + * mds_verw_sel. This ensures VERW will not #GP for an arbitrary user %ds.
> + */
> +.macro CLEAR_CPU_BUFFERS_SAFE
> + ALTERNATIVE "jmp .Lskip_verw\@", "", X86_FEATURE_CLEAR_CPU_BUF
> + verw %ss:_ASM_RIP(mds_verw_sel)
> +.Lskip_verw\@:
> +.endm
I know this is somewhat of a common pattern, but I think it is silly in
this case. Since we already have the ALTERNATIVE() why not NOP the one
VERW instruction instead?
That is,
ALTERNATIVE("", "verw %ss:_ASM_RIP(mds_verw_sel)", X86_FEATURE_CLEAR_CPU_BUF)
and call it a day?
Powered by blists - more mailing lists