[U-Boot] Travis test/py sandbox_spl test fail

[Stephen Warren]

On 8/18/19 7:01 PM, Kever Yang wrote:
> Hi Simon, Stephen,
> 
>      Could you help to comment on my other mail, which patch cause this 
> failure.

If you run "git bisect", you should be able to track down which patch 
introduced the problem.

> 
> 
> Thanks,
> 
> - Kever
> 
> On 2019/8/14 下午11:49, Stephen Warren wrote:
>> On 8/13/19 8:06 PM, Kever Yang wrote:
>>> Hi Stephen,
>>>
>>> On 2019/8/14 上午4:54, Stephen Warren wrote:
>>>> On 8/13/19 3:39 AM, Simon Glass wrote:
>>>>> +Stephen
>>>>>
>>>>> Hi Kever,
>>>>>
>>>>> On Tue, 13 Aug 2019 at 03:35, Kever Yang 
>>>>> <kever.yang at rock-chips.com> wrote:
>>>>>>
>>>>>> Hi Simon,
>>>>>>
>>>>>>      I got fail in test/py sandbox_spl, and the log says:
>>>>>>
>>>>>> E OSError: [Errno 5] Input/output error
>>>>>>
>>>>>> I have no idea about what's wrong in source code, could you help
>>>>>>
>>>>>> to take a look?
>>>>>>
>>>>>> Travis job:
>>>>>>
>>>>>> https://travis-ci.org/keveryang/u-boot/jobs/571125119
>>>>>
>>>>> When I've seen this (ugly) error it is normally because U-Boot
>>>>> crashed, e.g. with a segfault.
>>>>
>>>> Yes, that's the typical reason. If you run test.py locally you'll be 
>>>> able to access the log file (which Travis doesn't save), which will 
>>>> likely give you more details about the crash, and/or you could 
>>>> attach gdb to the sandbox process to trap the problem too.
>>>
>>> I got:
>>>
>>> $ ./u-boot
>>> bloblist_init() Existing bloblist not found: creating new bloblist
>>> [1]    958 segmentation fault (core dumped)  ./u-boot
>>>
>>> And no more logs, is there any other log file I can check?
>>
>> There probably isn't any more log information beyond that. I think the 
>> next step would be to run U-Boot sandbox under gdb, reproduce the 
>> problem, and then debug it.
> 
>