Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <CABXGCsM8gjvh8x9QyzbrbnSb-NmWpmmEUJnFiFv5MaPkcG=Ubg@mail.gmail.com>
 <CABXGCsMcS+H9jxEotmM3u8F28zhPobvT3+Hd_iWstFQN744v6Q@mail.gmail.com> <alpine.DEB.2.21.1904141948200.4917@nanos.tec.linutronix.de>
In-Reply-To: <alpine.DEB.2.21.1904141948200.4917@nanos.tec.linutronix.de>
From:   Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Date:   Mon, 15 Apr 2019 01:07:28 +0500
Message-ID: <CABXGCsNaw=mb0YF4Bkd5_CTwMfU=dNZ=RTV96dwO=UQy=n_gOQ@mail.gmail.com>
Subject: Re: [5.0.0 rc3 BUG] possible irq lock inversion dependency detected
To:     Thomas Gleixner <tglx@linutronix.de>
Cc:     Linux List Kernel Mailing <linux-kernel@vger.kernel.org>,
        thomas.lendacky@amd.com
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, 14 Apr 2019 at 22:51, Thomas Gleixner <tglx@linutronix.de> wrote:
> Because mails fall through the cracks occasionally.
>
> Does the patch below cure your problem?
>
> Thanks,
>
>         tglx
>
> 8<------------------
> --- a/arch/x86/kernel/process.c
> +++ b/arch/x86/kernel/process.c
> @@ -426,6 +426,8 @@ static __always_inline void __speculatio
>         u64 msr = x86_spec_ctrl_base;
>         bool updmsr = false;
>
> +       lockdep_assert_irqs_disabled();
> +
>         /*
>          * If TIF_SSBD is different, select the proper mitigation
>          * method. Note that if SSBD mitigation is disabled or permanentely
> @@ -477,10 +479,12 @@ static unsigned long speculation_ctrl_up
>
>  void speculation_ctrl_update(unsigned long tif)
>  {
> +       unsigned long flags;
> +
>         /* Forced update. Make sure all relevant TIF flags are different */
> -       preempt_disable();
> +       local_irq_save(flags);
>         __speculation_ctrl_update(~tif, tif);
> -       preempt_enable();
> +       local_irq_restore(flags);
>  }
>
>  /* Called from seccomp/prctl update */

Thanks, with this patch problem was gone.
We have time land it in 5.1?

--
Best Regards,
Mike Gavrilov.