linux-kernel - Re: [PATCH v2 2/2] arm64: Use static keys for CPU features

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] [day] [month] [year] [list]

Date:   Thu, 8 Sep 2016 14:40:25 +0100
From:   Catalin Marinas <catalin.marinas@....com>
To:     Jason Baron <jbaron@...mai.com>
Cc:     Will Deacon <will.deacon@....com>, peterz@...radead.org,
        Suzuki.Poulose@....com, linux-kernel@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, corbet@....net
Subject: Re: [PATCH v2 2/2] arm64: Use static keys for CPU features

On Wed, Sep 07, 2016 at 12:59:52PM -0400, Jason Baron wrote:
> On 09/05/2016 01:25 PM, Catalin Marinas wrote:
> > This patch adds static keys transparently for all the cpu_hwcaps
> > features by implementing an array of default-false static keys and
> > enabling them when detected. The cpus_have_cap() check uses the static
> > keys if the feature being checked is a constant, otherwise the compiler
> > generates the bitmap test.
> > 
> > Because of the early call to static_branch_enable() via
> > check_local_cpu_errata() -> update_cpu_capabilities(), the jump labels
> > are initialised in cpuinfo_store_boot_cpu().
> 
> Was there a reason the jump_label_init() couldn't be moved
> earlier in the common code?

No particular reason, only that I wasn't sure what the arch requirements
to be able to initialise the jump labels early are (for example,
jump_label_init() calls arch_jump_label_transform_static(); there don't
seem to be any issues at a first look but I don't have the hardware to
test and confirm). Therefore I followed the powerpc idea of calling
jump_label_init() directly earlier.

We also don't know how early it needs to be to benefit other
architectures (powerpc seems to call it on a very early path via
early_setup()).

-- 
Catalin