linux-kernel - Re: [PATCH] ARM: pxa: fix building with clang

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: <9a4d34d4-f78b-49f0-a57a-b80432ad652b@app.fastmail.com>
Date:   Sun, 18 Dec 2022 12:00:00 +0100
From:   "Arnd Bergmann" <arnd@...db.de>
To:     "Robert Jarzmik" <robert.jarzmik@...e.fr>,
        "Arnd Bergmann" <arnd@...nel.org>
Cc:     "Daniel Mack" <daniel@...que.org>,
        "Haojian Zhuang" <haojian.zhuang@...il.com>, soc@...nel.org,
        "Russell King" <linux@...linux.org.uk>,
        "Nathan Chancellor" <nathan@...nel.org>,
        "Nick Desaulniers" <ndesaulniers@...gle.com>,
        "Tom Rix" <trix@...hat.com>, linux-arm-kernel@...ts.infradead.org,
        linux-kernel@...r.kernel.org, llvm@...ts.linux.dev
Subject: Re: [PATCH] ARM: pxa: fix building with clang

On Sat, Dec 17, 2022, at 21:45, Robert Jarzmik wrote:
> Arnd Bergmann <arnd@...nel.org> writes:
>
>> From: Arnd Bergmann <arnd@...db.de>
>>
>> The integrated assembler in clang does not understand the xscale
>> specific mra/mar instructions:
>>
>> arch/arm/mach-pxa/pxa27x.c:136:15: error: unsupported 
>> architectural extension: xscale
>>         asm volatile(".arch_extension xscale\n\t"
>> arch/arm/mach-pxa/pxa27x.c:136:40: error: invalid instruction, 
>> did you mean: mcr, mla, mrc, mrs, msr?
>>         mra r2, r3, acc0
>>
>> Since these are coprocessor features, the same can be expressed 
>> using
>> mrrc/mcrr, so use that for builds with IAS.
> Ok Arnd, but "mrrc" is an ARMv6 instruction if I'm not mistaken.
> Yet PXA27X is an ARMv5 XScale instruction set IP.
>
> Is that patch correct therefore, or is it just to make clang happy 
> even if it's not correct ?

According to

https://developer.arm.com/documentation/dui0231/b/arm-instruction-reference/coprocessor-instructions/mrrc-and-mrrc2?lang=en

"MRRC is available in ARMv6 and above, and E variants of ARMv5
excluding xP variants". I'm not entire sure what "xP variants"
means, but we do build for ARMv5E, so I think this is actually
correct.

    Arnd