Bug#458118: ICE on arm while building nemiver
Steve Langasek a écrit :
> On Mon, Dec 31, 2007 at 02:44:00PM +0100, Martin Michlmayr wrote:
>> * Steve Langasek <vorlon@debian.org> [2007-12-29 11:08]:
>>>>> nmv-sess-mgr.cc: In member function 'virtual void nemiver::SessMgr::store_session(nemiver::ISessMgr::Session&, nemiver::common::Transaction&)':
>>>>> nmv-sess-mgr.cc:343: internal compiler error: Segmentation fault
>
>>>> Works fine here. I think the package should be retried on ARM.
>
>>> Yes, reasonable since the failing buildd was elara; but then, it would still
>>> be good to know why gcc-4.2 is segfaulting on netwinder since we still have
>>> several of those in use.
>
>> I just tried on a Netwinder and I can successfully compile that source
>> file there too. Do those segfaults happen on all Netwinder buildds
>> or only on elara?
>
> elara and europa have both, IMO, shown themselves to be significantly
> unreliable over time. I don't know whether these are kernel issues or
> hardware peculiarities, but last I knew, they were still not running the
> same kernel as the other buildds for whatever reason.
Their chroots is broken more often than other build daemons, however, I
don't really know why. This mainly concerns C++ stuff.
> The "netwinder" buildd seems to also have its share of problems, though not
> necessarily the same; and the biggest problem is still that they're all
> substantially slower than the cats buildds...
>
The Thecus ones are even faster.
--
.''`. Aurelien Jarno | GPG: 1024D/F1BCDB73
: :' : Debian developer | Electrical Engineer
`. `' aurel32@debian.org | aurelien@aurel32.net
`- people.debian.org/~aurel32 | www.aurel32.net
Reply to: