Re: Emboss not migrating due to autopkgtest error on s390x

Hi,

On 2024-04-14 16:42, Andreas Tille wrote:
> according to tracker[1] emboss is not migrating due to a test suite
> error on s390x[2].  IMHO the appropriate thing to do is to also remove
> s390x architecture - but we also need to care for all rdepends (again
> after removing these for 32bit).  Any volunteer to file the according
> bugs?

emboss autopkgtest on s390x fails with SIGSEGV. GDB for 'gdb seqret' says:

(gdb) run
Starting program: /usr/bin/seqret
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/s390x-linux-gnu/libthread_db.so.1".

Program received signal SIGSEGV, Segmentation fault.
cvt_s (code=<optimized out>, ap=0x3ffffff99c0, put=0x3fff77b79b8 
<fmtAppend>, cl=0x3ffffff9e70, flags=0x3ffffff99c8, width=-2147483648, 
precision=-2147483648) at ajfmt.c:352
352	    if(str)

GDB localises the issue in a string formatting function. I have a hunch 
this might be related to time_64 transition, but cannot say more, alas. 
Removal is the easiest solution for now.

> I guess adding architecture-is-little-endian to Build-Depends will
> to the trick to prevent building on s390x.

Limiting to architecture-is-little-endian is probably a good idea. We do 
not have any evidence seqret works on any big-endian architecture.

I volunteer to limit the architectures and file RM bugs.

> [1] https://tracker.debian.org/pkg/emboss
> [2] https://ci.debian.net/packages/e/emboss/testing/s390x/43624748/

Best,
Andrius