linux-kernel - Re: RESEND: GCC asm goto outputs workaround (Was: "Re: [PATCH 1/3] arm64: start using 'asm goto' for get

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAHk-=wjrj-qLVxvtxf+uOBPNp8dJv1R-64E0s7k7uTpKk-1B9w@mail.gmail.com>
Date: Wed, 17 Jul 2024 11:29:18 -0700
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Mark Rutland <mark.rutland@....com>
Cc: Linux ARM <linux-arm-kernel@...ts.infradead.org>, linux-kernel@...r.kernel.org, 
	Will Deacon <will@...nel.org>, Catalin Marinas <catalin.marinas@....com>, 
	Jakub Jelinek <jakub@....gnu.org>, Peter Zijlstra <peterz@...radead.org>, 
	Sean Christopherson <seanjc@...gle.com>, Alex Coplan <alex.coplan@....com>, 
	Szabolcs Nagy <szabolcs.nagy@....com>
Subject: Re: RESEND: GCC asm goto outputs workaround (Was: "Re: [PATCH 1/3]
 arm64: start using 'asm goto' for get_user() when") available

On Wed, 17 Jul 2024 at 09:28, Mark Rutland <mark.rutland@....com> wrote:
>
> [resending as I messed up the LKML address; sorry!]

Bah. And my reply was to your original.

I'll just leave the lore link to the reply (that did go to the arm
list successfully) here for anybody that only gets lkml..

    https://lore.kernel.org/all/CAHk-=wghzGt7J9XaQgcmLniYrQMtoXGcv+FvdGcyspkb+FxUsw@mail.gmail.com/

but the gist of it probably boils down to the final few sentences:

 "Maybe we just need to give up on the workaround and say that "asm goto
  with outputs only works when ASM_GOTO_OUTPUT_WORKAROUND is not set".

  The set of broken gcc versions will get progressively smaller as time goes on"

which is sad, but not the end of the world.

Unless the gcc people can maybe come up with some variation of our
current workaround that actually works (because the "workaround at
target label" model looks very inconvenient).

           Linus