On Thu, May 13, 2004 at 05:28:01PM +0100, William Bailey wrote:
> 	Sorry about the delay have had strane few weeks :) Anyway i have the
> same problem on all of my FreeBSD boxes but i did try it on a debian box
> without any issues and the index built fine.
> 
> 	Here is the debug stuff from the different machines (sorry about the
> formatting). Can you think of any reason why FreeBSD is haveing these
> issues?

No idea.  I run swish-e 4.2.4 on FreeBSD 4.9-STABLE.  I assume I just
have different data to index instead of it being a problem with BSD
version.

Can you make this happen by indexing a small number of documents?

Jose, any ideas from the gdb output?  I doubt that we really need to
malloc 31MB.


> Program received signal SIGBUS, Bus error.
> 0x2828722d in compress3 (num=2139062143,
> ~    buffer=0x2536ce64
> (gdb) bt
> #0  progerr (
> ~    msgfmt=0x2829ae80 "Ran out of memory (could not allocate %lu more bytes)!")
> ~    at error.c:59
> #1  0x2828acb1 in emalloc (size=31370220) at mem.c:81
> #2  0x282879ed in compress_worddata (
> ~    wdata=0x24955000
> compress.c:654
> #3  0x0805eeba in write_worddata (sw=0x80c1000, ep=0x2829ae80, indexf=0x0)
> ~    at db_write.c:325
> #4  0x08056a98 in write_index (sw=0x80c1000, indexf=0x8158000) at
> index.c:2052
> #5  0x0804d32c in write_index_file (sw=0x80c1000, process_stopwords=1,
> ~    elapsedStart=1084443949.848747, cpuStart=0, merge=0) at swish.c:1629
> #6  0x0804cdae in cmd_index (sw=0x80c1000, params=0x814d040) at swish.c:1364
> #7  0x0804b98f in main (argc=8, argv=0xbfbff734) at swish.c:209
> #8  0x0804b855 in _start ()
> (gdb)

-- 
Bill Moseley
moseley@hank.org