Re: [Qemu-devel] [PULL 09/23] gdbstub: Fix vCont behaviour

[Jan Kiszka]

On 2018-02-19 19:15, Claudio Imbrenda wrote:
> On Sat, 17 Feb 2018 10:07:38 +0100
> Jan Kiszka <address@hidden> wrote:
> 
> [...]
> 
>>> Seems like no one is doing guest debugging with kvm on x86 except
>>> me, and I'm only doing it too infrequently now: This one broke that
>>> use case for SMP guests long ago. How was it tested?
>>>
>>> To reproduce the bug: set up an x86-64 guest kernel with > 1 core,
>>> break on some prominent syscall entry (e.g. sys_execve), continue
>>> the guest on hit and it will quickly lock up, even after disabling
>>> the breakpoint again. Kernel version doesn't matter (was my first
>>> guess), gdb is 7.7.50.20140604-cvs (OpenSUSE) here.
> 
> [...]
> 
>> Sending packet: $Hg1#e0...Ack
>> Packet received: OK
>> Sending packet: $mffffc90000c0bf30,8#83...Ack
>> Packet received: ae180081ffffffff
>> Sending packet: $mffffc90000c0bf30,8#83...Ack
>> Packet received: ae180081ffffffff
>> Breakpoint 1, SyS_execve (filename=7029648, argv=7583376,
>> envp=7598304) at ../fs/exec.c:1923 (gdb) c
>> Continuing.
>> Sending packet: $vCont;s:1#23...Ack
>>
>> ...and now the guest is dead. I can still interrupt it, but it's
>> otherwise not working properly.
> 
> I tried, but I could not reproduce this bug neither on s390x nor on
> amd64. in both cases I used recent enough versions of qemu so that
> they have the patch you mentioned, and qemu was started with several
> cpus (I tried 64 on s390x, and 4 on amd64).
> 
> in particular on amd64:
> host: 4.4.0-112-generic (ubuntu 16.04)
> QEMU version 2.10.0  (vanilla from git)
> gdb (Ubuntu 7.11.1-0ubuntu1~16.5) 7.11.1
> 
> I also used the gdb script you provided but everything worked both with
> -smp 1 and with -smp 4
> 
> the only issue I had was that I had to disable kaslr in the guest to be
> able to do anything, but that does not seem to be the problem you have.
> 
> my primary suspicion at this point would be an issue in KVM, and not
> in qemu. have you tried running it without KVM? what is the version of
> qemu and kernel in the host?

Yes, debugging works without KVM. But that is a completely different
setup for various reasons (no parallelism, no guest-visible
modifications by soft-breakpoints and continuations). But I would not
exclude that the issue is caused by KVM, and your patch just surfaced it.

Jan

signature.asc
Description: OpenPGP digital signature