On Tue, Feb 28, 2017, at 10:07, Jan Engelhardt wrote:
On Tuesday 2017-02-28 15:47, Greg Ward wrote:
The bus errors are also a strong hint: it's been many years and many OS/architecture combinations since I saw a bus error, but as I recall it was caused by alignment problems.
SIGBUS, or more specifically CPU exceptions, are generated on e.g. SPARC and (I might have heard) x86's SSE, but not the "normal" x86 or ppc instructions. It just handles it behind the scenes [likely at a slower pace - but still].
I can confirm from experience that SIGBUS happens with alignment errors on MIPS chips from the 1990s. Doubt anyone cares about that now, though!
The lack of _FILE_OFFSET_BITS=64 in 32-bit compilations would also normally manifest in (seemingly-spurious) error returns (EOVERFLOW/"Value too large"), not so much asynchronous signals.
Indeed, I saw EOVERFLOW from getdents() system calls. So the _FILE_OFFSET_BITS hypothesis looks pretty strong. Anyways, I'm not going to dig any deeper: I'm just going to conclude "32-bit SL 6 in a chroot under 64-bit CentOS 7 does not work" and move on. I'm certainly not going to recompile an entire OS with different preprocessor flags. Thanks for your help! Greg -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org