koji rawhide ppc64 build failing due to strange error?

[Panu Matilainen]

On Sat, 12 Jul 2008, David Woodhouse wrote:

> On Sat, 2008-07-12 at 00:53 +0300, Panu Matilainen wrote:
>> On Fri, 11 Jul 2008, Jesse Keating wrote:
>>
>>> On Fri, 2008-07-11 at 15:27 -0400, Bill Nottingham wrote:
>>>> Mamoru Tasaka (mtasaka at ioa.s.u-tokyo.ac.jp) said:
>>>>> After koji began to accept jobs again, all dist-f10 ppc64 builds seem
>>>>> to be failing due to some strange error?
>>>>>
>>>>> http://koji.fedoraproject.org/koji/taskinfo?taskID=710447
>>>>> http://koji.fedoraproject.org/koji/taskinfo?taskID=710455
>>>>> http://koji.fedoraproject.org/koji/taskinfo?taskID=710471
>>>>>
>>>>> None of them have build.log and root.log say:
>>>>>
>>>>> DEBUG util.py:272:  Executing command: ['rpm', '-Uvh', '--nodeps', '/builddir/build/originals/xscreensaver-5.05.90.3-3.fc10.src.rpm']
>>>>> DEBUG util.py:250:  memory alloc (41 bytes) returned NULL.
>>>>>
>>>>> Would someone investigate what is happening here?
>>>>
>>>> Almost certianly the new rpm broke.
>>>>
>>>> Bill
>>>
>>> Yeah, that's coming from rpm in the chroot.  I can easily reproduce.
>>> I'm going to untag the new rpm until we sort this out.  It'll be a bit
>>> before builds start working again.
>>
>> Ugh :-/ Can you reproduce that outside of koji, and if so, how? Is this a
>> ppc64 specific thing or a more generic one?
>
> Let me know if you need an account on a suitable machine so you can try
> this in mock.

Thanks for the offer, Jarod Wilson already kindly provided me access to 
ppc64 host where I can mess around :)

The good news is that I can reproduce it, and it is indeed a ppc64-only 
problem. Here's what happens:

stat("/var/lib/rpm/Name", {st_mode=S_IFREG|0644, st_size=12288, ...}) = 0
open("/var/lib/rpm/Name", O_RDONLY)     = 6
fcntl(6, F_SETFD, FD_CLOEXEC)           = 0
read(6, 
"\0\0\0\0\0\0\0\1\0\0\0\0\0\6\25a\0\0\0\10\0\0\20\0\0\10\0\0\0\0\0\0"..., 
512) = 512
close(6)                                = 0
open("/var/lib/rpm/Name", O_RDONLY)     = 6
fcntl(6, F_SETFD, FD_CLOEXEC)           = 0
fstat(6, {st_mode=S_IFREG|0644, st_size=12288, ...}) = 0
rt_sigprocmask(SIG_BLOCK, ~[RTMIN RT_1], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
mmap(NULL, 171798695936, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, 
-1, 0) = -1 ENOMEM (Cannot allocate memory)

No wonder it fails... Now just to figure out where that comes from... but 
that'll have to wait till later today/tomorrow, family is demanding 
attention :)

 	- Panu -