On 2013-02-22 16:49, Lev Serebryakov wrote:
You wrote 22 февраля 2013 г., 18:21:50:
DA> The default for FreeBSD on 32-bit x86 is i486, so maybe the problems are
DA> caused by the -march=geode setting. If you disable that, do the
DA> problems disappear?
Problem is, that code compiled with "-march=geode" works. Code
built without any "-march" at all (without CPUTYPE in configs) doesn't.
It looks like clang or use "build system" CPU as default "-march" or
issue some >= i686 commands without it. Or both :)
Clang defaults to i486 (that is, on i386-unknown-freebsdXX arch), unless
you specify -march= or -mcpu= on the command line.
Maybe samba, or any of its dependencies, attempts to be "smart", and
enables some custom CPU optimizations?
DA> In any case, can you attempt to figure out which exact instructions it
DA> dies on? If gdb does not work, like you said above, maybe you can use
DA> objdump to disassemble the executable in question, and find the address
DA> of the failing instruction.
I'm trying to do this with very last sources both as build system
and target sources.
As Joerg Sonnenberger mentioned to me, the address 0x10351d6 you show in
the gdb session seems to be quite high, possibly pointing to some shared
library. Maybe you can try to figure out which library it is?
-Dimitry
_______________________________________________
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"
