On Thu, Aug 01, 2002 at 11:07:02AM -0400, Christopher C. Chimelis wrote: > I just posted a follow-up to that bug report, complete with some > experimentation results. To summarise, either double your shmmax or > recompile libmm11 with --with-shm=MMFILE. Looks like there's a bug > somewhere WRT IPCSHM (could be kernel, could be libmm...dunno yet..hell, > it could be PHP). I'm going to rebuild forcing shm to MMFILE anyway since it'll avoid problems with the shm method randomly changing depending on which buildd it gets built on. > I am going to try libmm12 to see if that helps (not looking forward to > recompiling php4, but I might as well). I wouldn't bother, it doesn't achieve much and since this bug keeps appearing and disappearing it needn't prove anything. I did try this already but since I can't reproduce on this PowerPC machine in the first place I'm not sure it's achieved anything. > Mark, anything else that you'd like me to try? Adding diagnostics to php4 so it'd actually say what went wrong (MM provides mm_error() giving a text-form error) might be helpful. -- "You grabbed my hand and we fell into it, like a daydream - or a fever."
Description: PGP signature