linux-kernel - Re: [PATCH v11 6/6] x86/tsc: use tsc early

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:   Thu, 21 Jun 2018 13:51:44 +0800
From:   Feng Tang <feng.tang@...el.com>
To:     Pavel Tatashin <pasha.tatashin@...cle.com>
Cc:     steven.sistare@...cle.com, daniel.m.jordan@...cle.com,
        linux@...linux.org.uk, schwidefsky@...ibm.com,
        heiko.carstens@...ibm.com, john.stultz@...aro.org,
        sboyd@...eaurora.org, x86@...nel.org, linux-kernel@...r.kernel.org,
        mingo@...hat.com, tglx@...utronix.de, hpa@...or.com,
        douly.fnst@...fujitsu.com, peterz@...radead.org, prarit@...hat.com,
        pmladek@...e.com, gnomes@...rguk.ukuu.org.uk
Subject: Re: [PATCH v11 6/6] x86/tsc: use tsc early

Hi Pavel,

On Wed, Jun 20, 2018 at 05:27:00PM -0400, Pavel Tatashin wrote:
> We want to get timestamps and high resultion clock available to us as early
> as possible in boot. But, native_sched_clock() outputs time based either on
> tsc after tsc_init() is called later in boot, or using jiffies when clock
> interrupts are enabled, which is also happens later in boot.
> 
> On the other hand, we know tsc frequency from as early as when
> tsc_early_delay_calibrate() is called. So, we use the early tsc calibration
> to output timestamps early. Later in boot when tsc_init() is called we
> calibrate tsc again using more precise methods, and start using that.
> 
> Since sched_clock() is in a hot path, we want to make sure that no
> regressions are introduced to this function after machine is booted, this
> is why we are using static branch that is enabled by default, but is
> disabled once we have initialized a permanent clock source.
> 
> Signed-off-by: Pavel Tatashin <pasha.tatashin@...cle.com>
> ---
>  arch/x86/kernel/tsc.c | 64 ++++++++++++++++++++++++++++++-------------
>  1 file changed, 45 insertions(+), 19 deletions(-)
> 
> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> index 654a01cc0358..1dd69612c69c 100644
> --- a/arch/x86/kernel/tsc.c
> +++ b/arch/x86/kernel/tsc.c
> @@ -39,6 +39,9 @@ EXPORT_SYMBOL(tsc_khz);
>  static int __read_mostly tsc_unstable;
>  
>  static DEFINE_STATIC_KEY_FALSE(__use_tsc);
> +static DEFINE_STATIC_KEY_TRUE(tsc_early_enabled);

Do we still need add a static_key? after Peter worked out the patch
to enable ealy jump_label_init?

Thanks,
Feng