On Tue, May 10, 2016 at 4:41 PM, Samuel Holland <sam...@sholland.org> wrote:
> On 05/10/2016 05:15 PM, Alex Williamson wrote:
>
>> On Mon, May 9, 2016 at 4:46 PM, Damon Namod wrote:
>>
>>> Hi all,
>>>
>>> I just tried Linux 4.6.0-rc7 with qemu 2.6.0-rc4. Same error and
>>> behaviour. What the heck is this?
>>>
>>
>> Could you post 'sudo lspci -vvvs 01:00.0' from the host system? I
>> found possibly something more similar to your card and it has rather
>> poor alignment of the MSI-X structures, I wonder if your issue is
>> something similar. Unfortunately this HBA hosts the root disk on
>> the system I found, so I can't assign it unless I can figure out
>> something else to run the host from (and I don't have physical
>> access to run from USB).
>>
>
> I have a similar piece of hardware (it's the LSI branded card instead of
> the IBM, but the chip is the same). Attached is the lspci output from
> it, for comparison if nothing else. Unfortunately, this is also hosting
> my root filesystem, so I haven't had a chance to use it with a VM yet,
> so I can't say if it would work.
>
Interesting, yours has:
Capabilities: [c0] MSI-X: Enable+ Count=15 Masked-
Vector table: BAR=1 offset=00002000
PBA: BAR=1 offset=00003800
Mine had:
Capabilities: [c0] MSI-X: Enable+ Count=15 Masked-
Vector table: BAR=1 offset=0000e000
PBA: BAR=1 offset=0000f800
I'm guessing Damon's will be more similar to yours and the 0x2000 parameter
to the failing vfio_dma_map() is trying to map the region of BAR1 before
the MSI-X vector table, matching the 00002000 vector table offset on your
card. But that's not really anything peculiar. The PBA offset is poorly
aligned, but that's not unusual either and not the mapping we're failing.
Hmmm. Thanks for the data point.
_______________________________________________
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users
