linux-kernel - Re: [PATCH v3 2/2] arm64: Support systems without FP/ASIMD

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]

Date:   Tue, 15 Nov 2016 10:42:41 +0000
From:   Suzuki K Poulose <suzuki.poulose@....com>
To:     Catalin Marinas <catalin.marinas@....com>
CC:     <linux-arm-kernel@...ts.infradead.org>, <mark.rutland@....com>,
        <ard.biesheuvel@...aro.org>, <marc.zyngier@....com>,
        <will.deacon@....com>, <linux-kernel@...r.kernel.org>,
        <kvmarm@...ts.cs.columbia.edu>,
        "Christoffer Dall" <christoffer.dall@...aro.org>, <nd@....com>
Subject: Re: [PATCH v3 2/2] arm64: Support systems without FP/ASIMD

On 14/11/16 11:48, Catalin Marinas wrote:
> Hi Suzuki,
>

>> +static inline bool system_supports_fpsimd(void)
>> +{
>> +	return !cpus_have_const_cap(ARM64_HAS_NO_FPSIMD);
>> +}
>
> Any particular reason why using negation instead of a ARM64_HAS_FPSIMD?
> A potential problem would be the default cpus_have_const_cap()
> implementation and the default static key having a slight performance
> impact.

The negation was chosen to avoid hotpatching in the most common case.
But as you said, it has an impact on the other side. I think doing
a one time hotpatching at boot time is more optimal than penalising
a bunch of other users throughout the execution. I will take a look
at changing it back to a ARM64_HAS_FPSIMD.

>>  	},
>> +	{
>> +		/* FP/SIMD is not implemented */
>> +		.capability = ARM64_HAS_NO_FPSIMD,
>> +		.def_scope = SCOPE_SYSTEM,
>> +		.min_field_value = 0,
>> +		.matches = has_no_fpsimd,
>> +	},
>
> If we go for negation, I don't think we need a min_field_value at all,
> the matching is done by the has_no_fpsimd() function.

You're right.

Suzuki