linux-kernel - Re: [PATCH] riscv: fix multi-letter extension compare

Open Source and information security mailing list archives

Message-Id: <E75B9DE8-8943-4F40-8374-2E74C50D7A00@jrtc27.com>
Date:   Thu, 6 Oct 2022 17:10:37 +0100
From:   Jessica Clarke <jrtc27@...c27.com>
To:     Nikita Shubin <nikita.shubin@...uefel.me>
Cc:     Atish Patra <atishp@...osinc.com>,
        Albert Ou <aou@...s.berkeley.edu>,
        Yury Norov <yury.norov@...il.com>,
        Nikita Shubin <n.shubin@...ro.com>,
        Anup Patel <anup@...infault.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux <linux@...ro.com>,
        Tsukasa OI <research_trasio@....a4lg.com>,
        Sunil V L <sunilvl@...tanamicro.com>,
        Palmer Dabbelt <palmer@...belt.com>,
        Jisheng Zhang <jszhang@...nel.org>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        Guo Ren <guoren@...nel.org>, linux-riscv@...ts.infradead.org,
        Heiko Stuebner <heiko@...ech.de>
Subject: Re: [PATCH] riscv: fix multi-letter extension compare

On 6 Oct 2022, at 08:02, Nikita Shubin <nikita.shubin@...uefel.me> wrote:
> 
> From: Nikita Shubin <n.shubin@...ro.com>
> 
> Increment ext pointer to match letters after 'z' character, so it points
> to actual extension name.
> 
> Fixes: 4905ec2fb7e6 ("RISC-V: Add sscofpmf extension support")
> Signed-off-by: Nikita Shubin <n.shubin@...ro.com>
> ---
> Hello Atish,
> 
> Sorry to blame you, but i double checked this and i don't see a way 
> how extensions from device tree could be successfully matched with 
> encoded ones without incrementing the ext pointer or prepending 
> the encoded extensions with 'z' letter.
> 
> I have no idea how this could slip through review - am i missing something ?
> ---
> arch/riscv/kernel/cpufeature.c | 1 +
> 1 file changed, 1 insertion(+)
> 
> diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> index 3b5583db9d80..031546052dc0 100644
> --- a/arch/riscv/kernel/cpufeature.c
> +++ b/arch/riscv/kernel/cpufeature.c
> @@ -199,6 +199,7 @@ void __init riscv_fill_hwcap(void)
> 				this_hwcap |= isa2hwcap[(unsigned char)(*ext)];
> 				set_bit(*ext - 'a', this_isa);
> 			} else {
> +				ext++;
> 				SET_ISA_EXT_MAP("sscofpmf", RISCV_ISA_EXT_SSCOFPMF);
> 				SET_ISA_EXT_MAP("svpbmt", RISCV_ISA_EXT_SVPBMT);
> 				SET_ISA_EXT_MAP("zicbom", RISCV_ISA_EXT_ZICBOM);

This makes no sense. The single-letter prefix that’s part of the actual
extension name is already in the string there, whether s or z. Your
code means (a) you’d have to double up the prefix in the device tree
(which is obviously wrong) (b) you could use s or z interchangeably as
the first character (e.g. you’d make zzicbom and szicbom parsed as
zicbom).

Are you actually seeing a bug? Did you test this at all? This looks
like you misunderstood code and wrote an untested patch for a bug you
thought you saw.

Jess

> -- 
> 2.35.1
> 
> 
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@...ts.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv

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