linux-kernel - Re: [PATCH 8/9] perf/breakpoint: Split breakpoint "check" and "commit"

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]

Message-ID: <20180515135259.GB24389@lerouge>
Date:   Tue, 15 May 2018 15:53:00 +0200
From:   Frederic Weisbecker <frederic@...nel.org>
To:     Joel Fernandes <joel.opensrc@...il.com>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        jolsa@...hat.com, namhyung@...nel.org,
        Peter Zijlstra <peterz@...radead.org>,
        torvalds@...ux-foundation.org, ysato@...rs.sourceforge.jp,
        benh@...nel.crashing.org,
        Catalin Marinas <catalin.marinas@....com>, chris@...kel.net,
        paulus@...ba.org, Thomas Gleixner <tglx@...utronix.de>,
        Will Deacon <will.deacon@....com>, mpe@...erman.id.au,
        dalias@...c.org, Ingo Molnar <mingo@...nel.org>,
        mark.rutland@....com, alexander.shishkin@...ux.intel.com,
        luto@...nel.org, acme@...nel.org, jcmvbkbc@...il.com
Subject: Re: [PATCH 8/9] perf/breakpoint: Split breakpoint "check" and
 "commit"

On Mon, May 07, 2018 at 12:46:06AM +0000, Joel Fernandes wrote:
> On Sun, May 6, 2018 at 12:22 PM Frederic Weisbecker <frederic@...nel.org>
> wrote:
> 
> > arch_validate_hwbkpt_settings() mixes up attribute check and commit into
> > a single code entity. Therefore the validation may return an error due to
> > incorrect atributes while still leaving halfway modified architecture
> > breakpoint struct.
> 
> > Now that we have split its logic on all archs, we can remove this
> > misdesigned function and call directly the arch check and commit
> > functions instead. This allows us to later avoid commiting
> > a breakpoint to architecture when its slot couldn't be allocated.
> 
> [...]
> > diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c
> > index 6e28d28..6896ceeb 100644
> > --- a/kernel/events/hw_breakpoint.c
> > +++ b/kernel/events/hw_breakpoint.c
> > @@ -402,11 +402,12 @@ int dbg_release_bp_slot(struct perf_event *bp)
> 
> >   static int validate_hw_breakpoint(struct perf_event *bp)
> >   {
> > -       int ret;
> > +       int err;
> 
> > -       ret = arch_validate_hwbkpt_settings(bp);
> > -       if (ret)
> > -               return ret;
> > +       err = hw_breakpoint_arch_check(bp, &bp->attr);
> > +       if (err)
> > +               return err;
> > +       hw_breakpoint_arch_commit(bp);
> 
> minor nit:
> To preserve bisectability, shouldn't this be the following in this and
> earlier patches?
> 
>         err = hw_breakpoint_arch_check(bp, &bp->attr);
>         hw_breakpoint_arch_commit(bp);
>         if (err)
>                 return err;
> 
> And then in patch 9/9 you can fix it the right way?

I don't see how it was breaking bisectability.
Anyway I'm rewriting it entirely to use:

    struct arch_hw_breakpoint hw;
    int err;

    err = hw_breakpoint_arch_parse(bp, attr, &hw);
    if (err)
        return err;

    .....

    bp->hw.info = hw;


Thanks.