linux-kernel - Re: [PATCH] arm64: Unconditionally support ARCH_HAVE_NMI_SAFE

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:   Fri, 15 Sep 2017 12:15:38 -0700
From:   Stephen Boyd <sboyd@...eaurora.org>
To:     James Morse <james.morse@....com>
Cc:     Will Deacon <will.deacon@....com>,
        Catalin Marinas <catalin.marinas@....com>,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        Tyler Baicar <tbaicar@...eaurora.org>,
        Jonathan Zhang <zjzhang@...eaurora.org>,
        Tom Zanussi <tom.zanussi@...ux.intel.com>,
        Steven Rostedt <rostedt@...dmis.org>
Subject: Re: [PATCH] arm64: Unconditionally support ARCH_HAVE_NMI_SAFE_CMPXCHG

On 09/15, James Morse wrote:
> On 15/09/17 02:19, Stephen Boyd wrote:
> > From what I can see there isn't anything about ACPI_APEI_SEA that
> > means the arm64 architecture can or cannot support NMI safe
> > cmpxchg, so the if condition here is not important. 
> 
> Yup, it was to match 'HAVE_NMI', which was new with ACPI_APEI_SEA and pulls in
> some printk() stuff.
> 
> ... how come you don't need to change HAVE_NMI too?

I didn't change HAVE_NMI because it wasn't needed to get hist
triggers working. Sorry, I didn't even notice that HAVE_NMI
was also there, probably both of these configs should drop the
condition. I'll update the patch to do so.

> 
> 
> > Let's remove
> > it. Doing that allows us to support ftrace histograms via
> > CONFIG_HIST_TRIGGERS that depends on the arch having this config
> > selected.
> 
> What does CONFIG_HIST_TRIGGERS need this for? I can't see any
> cmpxchg use in kernel/trace/trace_events_hist.c

True. I think this isn't necessary for CONFIG_HIST_TRIGGERS, but
more for CONFIG_TRACING_MAP that is selected and used by
CONFIG_HIST_TRIGGERS. We put the dependency in both places so we
don't get the trigger config selecting a config that has unmet
dependencies. Note there's a cmpxchg in the tracing map code too.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project