[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#526961: confget_1.02-1(mips/unstable): FTBFS on mips. Regression test fails.



On Sat, May 09, 2009 at 05:34:41PM +0300, Peter Pentchev wrote:
> On Mon, May 04, 2009 at 08:55:07PM +0300, Peter De Schrijver wrote:
[short summary: my confget-1.02 package FTBFS on the mips buildd]

Hi,

A couple of days ago, I sent this message as an answer to an RC bug
report that Peter De Schrijver filed.  I Cc'd the debian-mips mailing
list with a request for help (see below); unfortunately, I've had
no response so far.

Today, the mipsel buildd seems to have got 'round to building confget,
and there's the same FTBFS from
https://buildd.debian.org/build.php?pkg=confget

As noted below, I'm starting to suspect a toolchain bug of some sort.
Would it be possible for you (the wb team) to take a look at this,
tell me what the differences are between the 98 and 99.999 versions of
sbuild that are installed on the various architectures, what
the differences are between the preinstalled packages on mips/mipsel
and all the other architectures, and possibly run a couple of test
builds as described at the end of my message below?

Sorry for bothering you with that, and thanks in advance for any help! :(
Unfortunately, I have no access to mips/mipsel hardware myself :(

G'luck,
Peter

> And from the actual buildd logs, all the subtests that ought to have
> fetched a value from the config file fail with one of these messages:
> 
> INTERNAL ERROR: backend 'ini' does not define a openfile routine
> INTERNAL ERROR: backend 'http_get' does not define a openfile routine
> 
> Now... that's strange.  Actually, that's passing strange - it's out
> there, right next to unbelievable :)  I'm the upstream author of
> confget, and I know what these messages mean; and for the two
> "standard" backends, these messages mean that a statically-defined
> structure with a member that is statically initialized with a pointer
> to a static function in the same object file comes up with a null
> pointer instead!  To be honest, I'm wondering about a bug in
> the toolchain, something to do with memory alignment or something.
> 
> Unfortunately, I don't have access to mips hardware, and according to
> https://buildd.debian.org/build.php?pkg=confget it builds just fine
> on all the other arches, so I'm a bit stumped.  Now, there's two questions
> (or maybe requests) that could help me along a bit:
> 
> First, from the logs of the mips buildd it seems that 1. it is a different
> version from the other sbuilds, and 2. it has more packages preinstalled.
> Could somebody give me a list of the packages installed on the mips
> buildds, just to see if something comes to mind?
> 
> Second, could somebody with mips hardware get the confget source from
> unstable and give it a plain vanilla "debuild -uc -us" or something?
> If that fails in the same way as it fails on the buildd, maybe try it
> again with DEB_BUILD_OPTIONS='noopt nostrip nohardening'?
> 
> Thanks in advance for any assistance; I'm really stumped here :)

-- 
Peter Pentchev	roam@ringlet.net    roam@space.bg    roam@FreeBSD.org
PGP key:	http://people.FreeBSD.org/~roam/roam.key.asc
Key fingerprint	FDBA FD79 C26F 3C51 C95E  DF9E ED18 B68D 1619 4553
The rest of this sentence is written in Thailand, on

Attachment: pgpqB9imtdq1h.pgp
Description: PGP signature


Reply to: