Message-ID: <lcxrbfsd2b7prbxecdawlzht3m6knl3kopuomkqelbten5nkud@cyfqy26ndbzb>
Date: Mon, 18 Nov 2024 10:21:09 -0500
From: "Liam R. Howlett" <Liam.Howlett@...cle.com>
To: Aleksandr Nogikh <nogikh@...gle.com>
Cc: syzbot <syzbot+bc6bfc25a68b7a020ee1@...kaller.appspotmail.com>,
        akpm@...ux-foundation.org, jannh@...gle.com,
        linux-kernel@...r.kernel.org, linux-mm@...ck.org,
        lorenzo.stoakes@...cle.com, syzkaller-bugs@...glegroups.com,
        vbabka@...e.cz
Subject: Re: [syzbot] [mm?] general protection fault in mas_store_prealloc

* Aleksandr Nogikh <nogikh@...gle.com> [241118 08:13]:
> Hi Liam,
> 
> On Mon, Nov 18, 2024 at 3:49 AM 'Liam R. Howlett' via syzkaller-bugs
> <syzkaller-bugs@...glegroups.com> wrote:
> >
> > * syzbot <syzbot+bc6bfc25a68b7a020ee1@...kaller.appspotmail.com> [241117 05:42]:
> > > Hello,
> > >
> > > syzbot found the following issue on:
> > >
> > > HEAD commit:    8e9a54d7181b Merge remote-tracking branch 'iommu/arm/smmu'..
> > > git tree:       git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci
> > > console output: https://syzkaller.appspot.com/x/log.txt?x=17b0ace8580000
> > > kernel config:  https://syzkaller.appspot.com/x/.config?x=a1eb85a42cb8ccec
> > > dashboard link: https://syzkaller.appspot.com/bug?extid=bc6bfc25a68b7a020ee1
> > > compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> > > userspace arch: arm64
> > > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=147521a7980000
> > > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=102e14c0580000
> > >
> > > Downloadable assets:
> > > disk image: https://storage.googleapis.com/syzbot-assets/ad658fb4d0a2/disk-8e9a54d7.raw.xz
> > > vmlinux: https://storage.googleapis.com/syzbot-assets/1b7754fa8c67/vmlinux-8e9a54d7.xz
> > > kernel image: https://storage.googleapis.com/syzbot-assets/50315382fefb/Image-8e9a54d7.gz.xz
> > >
> > > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > > Reported-by: syzbot+bc6bfc25a68b7a020ee1@...kaller.appspotmail.com
> >
> > I was unable to get this reproducer to work on my own image, even using
> > the config and compiler specified in the report.  The injection was not
> > happening at the same location as the crash reports.
> >
> > After using the provided disk (which was tricky), I was able to get it
> > to work.
> 
> Just in case: did you follow the official syzbot instructions on
> reproducing the bugs from the attached assets? [1] If yes, what extra
> information could have made the process simpler for you?
> 
> [1] https://github.com/google/syzkaller/blob/master/docs/syzbot_assets.md

The instructions on getting the source we need (without adding yet
another remote and fetching the entire repo) would help, as well as
booting the kernel.

ie: --kernel <file> --append <root=...>

The missing part for me was that the commit listed was already gone, so
after Lorenzo responded and restested with code in the repo, I could
pull the kernel.  Maybe I could have just fetched the ref somehow?

So these instructions are good.  I didn't see them in the emails or on
the dashboard though?

Booting my own kernel ended up in panics about the rootfs missing, and
some fool around with that.

> 
> > Booting was painfully slow, and makes me wonder if there is a
> > better way for reproducing issues in the future.
> 
> I guess that unless you run it on an arm64 device with a nested
> virtualization support, this bug's reproduction will be slow anyway :(

The initial boot of arm64 seems to just sit around for a while thinking
about booting.  My installed vm was using uefi, which turned out to also
be a pain point.

> 
> If you recompile the kernel image without CONFIG_KASAN and
> CONFIG_KCOV, that should speed it up to some degree.

I think arm64 on amd64 is just slow, but thanks.  kasan seems necessary
a lot of the time.

Regards,
Liam

Powered by blists - more mailing lists