linux-kernel - RE: [PATCH V2 1/2] x86/msr: add msr_set/clear_bit_on

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:   Mon, 27 Mar 2017 17:12:48 +0000
From:   "Liang, Kan" <kan.liang@...el.com>
To:     Borislav Petkov <bp@...en8.de>
CC:     "peterz@...radead.org" <peterz@...radead.org>,
        "tglx@...utronix.de" <tglx@...utronix.de>,
        "mingo@...hat.com" <mingo@...hat.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "acme@...nel.org" <acme@...nel.org>,
        "eranian@...gle.com" <eranian@...gle.com>,
        "jolsa@...nel.org" <jolsa@...nel.org>,
        "ak@...ux.intel.com" <ak@...ux.intel.com>
Subject: RE: [PATCH V2 1/2] x86/msr: add msr_set/clear_bit_on_cpu/cpus
 access functions


> 
> On Mon, Mar 27, 2017 at 08:47:37AM -0700, kan.liang@...el.com wrote:
> > From: Kan Liang <Kan.liang@...el.com>
> >
> > Having msr_set/clear_bit on many cpus or given CPU can avoid extra
> > unnecessory IPIs
> 
> How does that happen?
>

My previous patch did a read-modify-write operation. Compared with the
single operation set/clear, it will has extra IPIs.
Sorry for the confusing wording.
I will change the description. 

> You have smp_call_function_many() sending IPIs to each CPU in the mask.
> Doesn't look like avoiding anything to me.
> 
> Now if you want to have interfaces set/clear_bit_on_cpu(s), that's a
> different story.
> 
> And those actually double the amount of IPIs the moment you do a read-
> modify-write operation on the MSR, i.e., you want to read *and* write
> afterwards.
> 
> If you only want to do a single operation - set or clear - like you're doing in
> your other patch, then I guess that's fine as it wraps the
> smp_call_function* boilerplate code.
> 
> > and simplify MSR content manipulation, when it only needs to flip a
> > bit.
> > There is already msr_set/clear_bit, but missing the _on_cpu and
> > _on_cpus version.
> >
> > Signed-off-by: Kan Liang <Kan.liang@...el.com>
> > ---
> >  arch/x86/include/asm/msr.h | 29 ++++++++++++++++++
> >  arch/x86/lib/msr-smp.c     | 76
> ++++++++++++++++++++++++++++++++++++++++++++++
> >  2 files changed, 105 insertions(+)
> >
> > diff --git a/arch/x86/include/asm/msr.h b/arch/x86/include/asm/msr.h
> > index 898dba2..9bc999b 100644
> > --- a/arch/x86/include/asm/msr.h
> > +++ b/arch/x86/include/asm/msr.h
> > @@ -20,6 +20,11 @@ struct msr {
> >  	};
> >  };
> >
> > +struct msr_bit_info {
> > +	u32 msr_no;
> > +	u8 bit;
> > +};
> 
> No, not *another* struct msr*info. Please reuse msr_info.
> 

OK.

Thanks,
Kan