Re: [Qemu-devel] [PULL 00/25] pc, pci, virtio: features, fixes, cleanups

[Jason Wang <jasowang@redhat.com>]

On 2018年01月16日 14:28, Jason Wang wrote:
>
>
> On 2018年01月16日 10:48, Michael S. Tsirkin wrote:
>> On Tue, Jan 09, 2018 at 12:10:10PM +1100, David Gibson wrote:
>>> On Mon, Jan 08, 2018 at 08:10:23PM +0200, Michael S. Tsirkin wrote:
>>>> On Mon, Jan 08, 2018 at 09:14:41AM +0000, Peter Maydell wrote:
>>>>> On 21 December 2017 at 23:43, Michael S. Tsirkin <mst@redhat.com> 
>>>>> wrote:
>>>>>>> you also need to drop "hw/pci: remove obsolete PCIDevice->init()"
>>>>>>> (applied in your tree as 18951fce55d5aa58cd1629b4cb704ee51bab4420)
>>>>>>> else the not-yet-QOM'ified NVME won't work anymore.
>>>>>> Done that too, thanks a lot
>>>>> Hi. I'm afraid this pullreq (merge of 880b1ff) has new runtime
>>>>> errors from the clang sanitizer:
>>>>>
>>>>> /home/petmay01/linaro/qemu-for-merges/hw/net/net_tx_pkt.c:201:27:
>>>>> runtime error: member access within misaligned address 0x55b4ffee8246
>>>>> for type 'struct ip_header', which requires 4 byte alignment
>>>>> 0x55b4ffee8246: note: pointer points here
>>>>>   01 00 00 00 45 00  01 a9 01 00 00 00 40 11  78 45 00 00 00 00 ff ff
>>>>> ff ff 00 00 00 00 00 00  00 00
>>>>>               ^
>>>>> /home/petmay01/linaro/qemu-for-merges/hw/net/net_tx_pkt.c:201:27:
>>>>> runtime error: load of misaligned address 0x55b4ffee8246 for type
>>>>> 'uint8_t' (aka 'unsigned char'), which requires 4 byte alignment
>>>>> 0x55b4ffee8246: note: pointer points here
>>>>>   01 00 00 00 45 00  01 a9 01 00 00 00 40 11  78 45 00 00 00 00 ff ff
>>>>> ff ff 00 00 00 00 00 00  00 00
>>>>>               ^
>>>>> /home/petmay01/linaro/qemu-for-merges/hw/net/net_tx_pkt.c:208:65:
>>>>> runtime error: member access within misaligned address 0x55b4ffee8246
>>>>> for type 'struct ip_header', which requires 4 byte alignment
>>>>> 0x55b4ffee8246: note: pointer points here
>>>>>   01 00 00 00 45 00  01 a9 01 00 00 00 40 11  78 45 00 00 00 00 ff ff
>>>>> ff ff 00 00 00 00 00 00  00 00
>>>>>               ^
>>>>> /home/petmay01/linaro/qemu-for-merges/hw/net/net_tx_pkt.c:210:13:
>>>>> runtime error: member access within misaligned address 0x55b4ffee8246
>>>>> for type 'struct ip_header', which requires 4 byte alignment
>>>>> 0x55b4ffee8246: note: pointer points here
>>>>>   01 00 00 00 45 00  01 a9 01 00 00 00 40 11  78 45 00 00 00 00 ff ff
>>>>> ff ff 00 00 00 00 00 00  00 00
>>>>>               ^
>>>>> /home/petmay01/linaro/qemu-for-merges/hw/net/net_tx_pkt.c:210:13:
>>>>> runtime error: load of misaligned address 0x55b4ffee8246 for type
>>>>> 'uint8_t' (aka 'unsigned char'), which requires 4 byte alignment
>>>>> 0x55b4ffee8246: note: pointer points here
>>>>>   01 00 00 00 45 00  01 a9 01 00 00 00 40 11  78 45 00 00 00 00 ff ff
>>>>> ff ff 00 00 00 00 00 00  00 00
>>>>>               ^
>>>>>
>>>>> Sorry for the late report, I didn't quite have time
>>>>> to process the pull before Christmas holidays.
>>>>>
>>>>> thanks
>>>>> -- PMM
>>>> So I bisected and the reason is this commit:
>>>>
>>>> commit 18b20bb43a2f37f0c8ae23a3e9b3d9a4a05b6bd4
>>>> Author: David Gibson <david@gibson.dropbear.id.au>
>>>> Date:   Tue Dec 19 15:45:22 2017 +1100
>>>>
>>>>      tests/pxe-test: Add some extra tests
>>>>           Previously virtio-net was only tested for ppc64 in "slow" 
>>>> mode.  That
>>>>      doesn't make much sense since virtio-net is used much more 
>>>> often in
>>>>      practice than the spapr-vlan device which was tested always.  
>>>> So, move
>>>>      virtio-net to always be tested on ppc64.
>>>>           We had no tests at all for the q35 machine, which doesn't 
>>>> seem wise
>>>>      given its increasing prominence.  Add a couple of tests for it,
>>>>      including testing the newer e1000e adapter.
>>>>           Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
>>>>      Reviewed-by: Thomas Huth <thuth@redhat.com>
>>>>      Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
>>>>      Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
>>>>
>>>> :040000 040000 5a982bfea24b9ac3c651b84425a39b3c85f4871e 
>>>> 771af3fdfb2778c6d6ed6b1098d1e79c181d6fb0 M      tests
>>>>
>>>> Pls either fix or drop e1000e test
>>> Feel free to drop this patch for now.  I'll debug and repost it when I
>>> have a chance.
>>>
>>> -- 
>>> David Gibson            | I'll have my music baroque, and my code
>>> david AT gibson.dropbear.id.au    | minimalist, thank you. NOT _the_ 
>>> _other_
>>>                 | _way_ _around_!
>>> http://www.ozlabs.org/~dgibson
>> BTw, Jason, you might want to take a look.
>> I suspect it's a bug somewhere in the e1000e emulation.
>>
>
> Will have a look.
>
> cc Dmitry for more thoughts.
>
> Thanks
>

Cc again with the correct address.