* Andre Berger <[EMAIL PROTECTED]>, 2005-01-31 19:25 +0100: > * Paolo Lucente <[EMAIL PROTECTED]>, 2005-01-30 14:41 +0100: > > Hello Andre, > > the exception 'EXC_BAD_ACCESS' makes me think that the issue could be > > caused by an unaligned memory access. What follows is a fragment of the > > TODO document: > > > > + Alignment issues; configure now does the proper test for > > unaligned accesses; a set of macros to copy 8/16/32 bit > > chunks has also been created: it *should* be sufficient > > as framework. Alignment issues have to be incrementally > > resolved. > > > > I've verified such issues long time ago over MIPS hardware and IRIX. > > However 0.8.0 and 0.8.1 have worked just fine there. Now, what you can > > do: > > > > - tell me what is the response, at './configure' time, of the > > 'checking unaligned accesses ...' line. > > ./configure --enable-pgsql --enable-debug --prefix=/usr/local > > checking unaligned accesses... ok > > PLATFORM ..... : Power Macintosh > OS ........... : Darwin 7.7.0 (hvk.local) > COMPILER ..... : gcc > CFLAGS ....... : -O0 -g -W -Wall -I/usr/local/pgsql/include > LIBS ......... : -lpcap -L/usr/local/pgsql/lib -lpq -lm -lz > LDFLAGS ...... : > > > Hasn't crashed since. I'm suspecting a hardware defect, b/c mutt also > crashed on me which had never happened before. So I dis-/reassembled > the iBook, and all is stable again since.
"Oops, it did it again", and from hints in the syslog, I think I know why. In order to conserve memory, I had allowed for only a small number of simultaneous PostgreSQL connections. I increased that number, and everything is back to stable. -Andre
