linux-kernel - Re: [PATCH] x86: fix early boot crash on gcc-10

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:   Mon, 16 Mar 2020 22:12:51 +0000
From:   Sergei Trofimovich <slyfox@...too.org>
To:     Jakub Jelinek <jakub@...hat.com>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        linux-kernel@...r.kernel.org, Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        "H. Peter Anvin" <hpa@...or.com>,
        Andy Lutomirski <luto@...nel.org>, x86@...nel.org
Subject: Re: [PATCH] x86: fix early boot crash on gcc-10

On Mon, 16 Mar 2020 14:26:48 +0100
Jakub Jelinek <jakub@...hat.com> wrote:

> > > +# smpboot's init_secondary initializes stack canary.
> > > +# Make sure we don't emit stack checks before it's
> > > +# initialized.
> > > +nostackp := $(call cc-option, -fno-stack-protector)
> > > +CFLAGS_smpboot.o := $(nostackp)  
> > 
> > What makes GCC10 insert this while GCC9 does not. Also, I would much  
> 
> My bet is different inlining decisions.
> If somebody hands me over the preprocessed source + gcc command line, I can
> have a look in detail (which exact change and why).

In case you are still interested in preprocessed files and results I've collected
all the bits in a single tarball:
    https://dev.gentoo.org/~slyfox/bugs/linux-gcc-10-boot-2020-03-14.tar.gz
Same available in separate files in:
    https://dev.gentoo.org/~slyfox/bugs/linux-gcc-10-boot-2020-03-14/

Specifically:
- gcc-v.gcc-{9,10}: gcc-v output of both compilers. Note --enable-default-pie --enable-default-ssp.
- config.gcc-{9,10}: note, they are not identical as Kbuild does not recognize gcc-10's
  plugin support. I don't use it though.
- boot-crash-gcc-10.jpg: picture of a full boot crash
- command.gcc-{9,10} called to generate .s files (it's almost the same when building .o files)
- arch-x86-kernel-smpboot.s-gcc-{9,10}: asm files, gennerated with 'make arch/x86/kernel/smpboot.s V=1'
- arch-x86-kernel-smpboot.c.c-gcc-{9,10}: preprocessed files, generated from command by changing -S to -E.

Another observation: kernel built by gcc-10 boots as-is in qemu without patches.
I wonder if the following boot line right before the crash has something to do wit it:
    "random: get_random_bgtes called from start_secondary+0x105/0x1a0 with crng_init=0"
I hope it's not a race of async canary initialization and canary use.
Only one CPU is booted at that time, yes?

-- 

  Sergei