Re: [linux-linus test] 179607: regressions - trouble: fail/pass/starved

[Jan Beulich <jbeulich@xxxxxxxx>]

On 23.03.2023 16:55, Juergen Gross wrote:
> On 14.03.23 14:27, Juergen Gross wrote:
>> On 14.03.23 13:52, osstest service owner wrote:
>>> flight 179607 linux-linus real [real]
>>> http://logs.test-lab.xenproject.org/osstest/logs/179607/
>>>
>>> Regressions :-(
>>>
>>> Tests which did not succeed and are blocking,
>>> including tests which could not be run:
>>>   test-amd64-amd64-freebsd12-amd64 13 guest-start          fail REGR. vs. 
>>> 178042
>>>   test-amd64-amd64-xl-credit1  19 guest-saverestore.2      fail REGR. vs. 
>>> 178042
>>>   test-amd64-amd64-xl-shadow   14 guest-start              fail REGR. vs. 
>>> 178042
>>
>> ...
>>
>> In the logs [1] I'm seeing errors like:
>>
>> Mar 13 23:51:26.275421 [  271.713740] xenbr0: port 2(vif1.0) entered 
>> forwarding 
>> state
>> Mar 13 23:51:26.287346 (XEN) common/grant_table.c:2982:d0v3 copy beyond page 
>> area
>> Mar 13 23:51:48.115383 (XEN) common/grant_table.c:2982:d0v3 copy beyond page 
>> area
>> Mar 13 23:51:49.123347 (XEN) common/grant_table.c:2982:d0v3 copy beyond page 
>> area
>> Mar 13 23:51:49.459367 (XEN) common/grant_table.c:2982:d0v3 copy beyond page 
>> area
>>
>> Given the vif related message directly before those errors the chance is high
>> this problem is related to netback.
>>
>> Ross, your patch "xen/netback: Ensure protocol headers don't fall in the
>> non-linear area" (upstream commit ad7f402ae4f4666) did the most recent 
>> changes
>> in netback affecting GNTTABOP_copy operations. There are probably page 
>> boundary
>> checks (probably on netback side) missing. Could you please have a look?
> 
> Okay, I've had a try (see attached patch).

Just one remark: Don't you need to use XEN_PAGE_SIZE in place of PAGE_SIZE?

Jan