linux-kernel - Re: [PATCH] x86/idt: load idt early in start

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: <CALCETrVRWSXjSxV98rdFu1XVwS5kTfbjRRiz8vr7Cmfz0A7n-Q@mail.gmail.com>
Date:   Mon, 27 Nov 2017 19:41:20 -0800
From:   Andy Lutomirski <luto@...nel.org>
To:     Chunyu Hu <chuhu@...hat.com>
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...nel.org>,
        "H. Peter Anvin" <hpa@...or.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Andrew Lutomirski <luto@...nel.org>,
        Borislav Petkov <bp@...en8.de>,
        Steven Rostedt <rostedt@...dmis.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] x86/idt: load idt early in start_secondary

On Mon, Nov 27, 2017 at 6:21 AM, Chunyu Hu <chuhu@...hat.com> wrote:
> For ap, idt is first loaded in cpu_init() with load_current_idt(),
> that is to say, no exception can be handled before there. And then
> the idt_table has been completed by the bp.
>
> While there are some WARNs which needs the UD exception handling
> in the early boot code might be triggered when something uexpected
> happens during boot. In that case, cpu would fail to boot as the
> exception can't be handled. A WARNing during boot is not usually
> meaning the system could not boot.
>
> One use case is when ftrace=function is setup in kernel cmdline, the
> ftrace callback function will be called for every traced function.
> And in my case, the first traced function is load_ucode_ap. And there
> are WARN()s in function trace callback handling, it failed to reboot
> as one of the WARN()s is triggered before load_current_idt() executed.
>
> To make WARN()s can work earlier to ap, we load the idt_table early
> in start_secondary, and keep the second time idt load in cpu_init,
> as there is a load_ucode_ap() there.

Reviewed-by: Andy Lutomirski <luto@...nel.org>

Except that this is presumably missing a Cc: stable.