linux-kernel - Re: [PATCH v6 06/18] asm-generic/tlb: Conditionally provide tlb_migrate

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <20190219124738.GD8501@fuggles.cambridge.arm.com>
Date:   Tue, 19 Feb 2019 12:47:38 +0000
From:   Will Deacon <will.deacon@....com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     aneesh.kumar@...ux.vnet.ibm.com, akpm@...ux-foundation.org,
        npiggin@...il.com, linux-arch@...r.kernel.org, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, linux@...linux.org.uk,
        heiko.carstens@...ibm.com, riel@...riel.com
Subject: Re: [PATCH v6 06/18] asm-generic/tlb: Conditionally provide
 tlb_migrate_finish()

On Tue, Feb 19, 2019 at 11:31:54AM +0100, Peter Zijlstra wrote:
> Needed for ia64 -- alternatively we drop the entire hook.
> 
> Cc: Will Deacon <will.deacon@....com>
> Cc: "Aneesh Kumar K.V" <aneesh.kumar@...ux.vnet.ibm.com>
> Cc: Andrew Morton <akpm@...ux-foundation.org>
> Cc: Nick Piggin <npiggin@...il.com>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
> ---
>  include/asm-generic/tlb.h |    2 ++
>  1 file changed, 2 insertions(+)
> 
> --- a/include/asm-generic/tlb.h
> +++ b/include/asm-generic/tlb.h
> @@ -539,6 +539,8 @@ static inline void tlb_end_vma(struct mm
>  
>  #endif /* CONFIG_MMU */
>  
> +#ifndef tlb_migrate_finish
>  #define tlb_migrate_finish(mm) do {} while (0)
> +#endif

Fine for now, but I agree that we should drop the hook altogether. AFAICT,
this only exists to help an ia64 optimisation which looks suspicious to
me since it uses:

    mm == current->active_mm && atomic_read(&mm->mm_users) == 1

to identify a "single-threaded fork()" and therefore perform only local TLB
invalidation. Even if this was the right thing to do, it's not clear to me
that tlb_migrate_finish() is called on the right CPU anyway.

So I'd be keen to remove this hook before it spreads, but in the meantime:

Acked-by: Will Deacon <will.deacon@....com>

Will