Re: strace build time tests fail on mips64el - missing syscall support?

James Cowgill <[email protected]> Tue, 9 Aug 2016 10:18:46 +0100

Newsgroups	gmane.linux.debian.ports.mips
Message-ID	<[email protected]>

On 09/08/16 00:25, Aurelien Jarno wrote:
> On 2016-08-08 14:39, James Cowgill wrote:
>> Hi Steve,
>>
>> On 08/08/16 14:35, Steve McIntyre wrote:
>>> On Wed, Jun 01, 2016 at 02:53:35PM +0100, Steve McIntyre wrote:
>>>> Hi folks,
>>>>
>>>> I'm seeing a fair number of test suite failures for strace on
>>>> mips64el, both in the most recent upload[1] and in current upstream
>>>> git HEAD[2]. It's looking like a lot of ENOSYS errors. Could I get
>>>> some porter help with this please? Otherwise I'm tempted to remove
>>>> mips64el from the arch list for the package.
>>>>
>>>> [1] https://buildd.debian.org/status/fetch.php?pkg=strace&arch=mips64el&ver=4.11-1&stamp=1452274105
>>>> [2] http://www.einval.com/debian/strace/build-logs/mips64el/
>>>
>>> Ping? Any help on this please?
>>
>> I'm having a look at this.
>>
>> In a preliminary check I found that about half the testsuite failures
>> were caused by a kernel bug in the ptrace handling of any syscalls
>> returning ENOSYS. The bug was fixed in 4.3 so testing strace on kernels
>> older than that probably isn't a good idea. There are still some other
>> failures though
> 
> Do you have a pointer to the fix? If it is small enough and easily
> backportable, we can try to get it in the jessie kernel.

This is the fix:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=d218af78492a36a4ae607c08fedfb59258440314

It seems like Markos actually did want the patch to go into the stable
tree, but it looks like that never happened:
https://www.linux-mips.org/archives/linux-mips/2015-09/msg00293.html

James

signature.asc (application/pgp-signature, 819 B)