[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: srange crash in virtio-gpu
From: |
Christian Borntraeger |
Subject: |
Re: srange crash in virtio-gpu |
Date: |
Thu, 24 Sep 2020 13:39:02 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.12.0 |
On 24.09.20 12:48, Daniel P. Berrangé wrote:
> On Thu, Sep 24, 2020 at 12:35:06PM +0200, Christian Borntraeger wrote:
>> Gerd,
>>
>> with current master build via a slightly fixed up fedora spec file I do get
>> a crash in virtio-gpu
>> when libvirt queries the qemu.
>> I can trigger that also via command line
>>
>> $ /usr/bin/qemu-system-s390x -device virtio-gpu-pci,help
>> qemu-system-s390x: -device virtio-gpu-pci,help: missing object type
>> 'virtio-gpu-device'
>> Aborted (core dumped)
>> $ /usr/bin/qemu-system-s390x -device virtio-gpu-ccw,help
>> qemu-system-s390x: -device virtio-gpu-ccw,help: missing object type
>> 'virtio-gpu-device'
>> Aborted (core dumped)
>>
>> Since I cannot reproduce the same when I build it without the spec file this
>> must
>> be something configuration specific.
>
> The Fedora spec builds using --enable-modules, and the error message
> shows that you don't have the .so for virtio-gpu on disk.
Yes, I needed to include hw-display-virtio-gpu.so as well. Thanks.
>
>> The call trace is:
>>
>> (gdb) bt
>> #0 0x000003ff9e349026 in raise () at /lib64/libc.so.6
>> #1 0x000003ff9e32b162 in abort () at /lib64/libc.so.6
>> #2 0x000002aa274abfca in object_initialize (data=<optimized out>,
>> size=<optimized out>, typename=0x2aa2754904e "virtio-gpu-device") at
>> ../qom/object.c:531
>
> This abort is evil though
>
> We need this to get into the "Device '<blah>' not found" code path
> to report a nice error.
Yes, a less catastrophic way would be good.