malloc woes (was Re: Info needed: UNIX for 286/386....)

[Michael J. Young]

In article <717 at actnyc.UUCP> jhs at actnyc.UUCP (John Spicer) writes:
>When I have tried the -lmalloc version on Microport SYSV/AT 286 it dumps core.
>I tried using the standard version and it is SLOW.  I finally ended rewriting
>malloc so I could get a version that works, and works quickly.
>
>I hope Microport provides a decent version of malloc some day.

Dwight Leu had some interesting things to say about malloc(3x) on Bix a few
months ago.  This is taken without permission, but it was posted in a
public forum, so I'm assuming there is no problem.  If there is, I apologize.

- ==========
- microport/bugs #23, from dleu, 1042 chars, Sat Dec 12 17:04:46 1987
- Comment to 15. Comment(s). More refs to 15.
- ----------
- ...
- 
- someone mentioned that malloc(3X) wasn't working, so i took a
- look into this myself. malloc(2) is the traditional malloc(),
- and malloc(3X) is the faster version that was added later. because
- it's much faster, i was curious why it wasn't working.
- 
- lo and behold, the software is straight from vaxland! believe it
- or not, the lowest bit of the address pointers is used as a flag!
- grrrr. no wonder it doesn't work. so i fixed it, and noticed
- an improvement in speed up to a certain limit.
- 
- but after this limit, one still runs into the wall. so i started
- looking into the sbreak() kernel software, and that's where things
- lie. i wish i had more time for this! by the way, if you want to
- play around with the source code yourself, let me know via mail.
- we encourage our customers who want to fix such things. but there's
- some paperwork in order to keep ATT satisfied.
- 

The last time I logged onto the Microport BBS, I didn't notice any
new malloc(3x) code, but I may have missed it.  I don't log in very
frequently, because it's a long distance call.  An 800 number would
be nice!
-- 
Mike Young - Software Development Technologies, Inc., Sudbury MA 01776
UUCP     : {decvax,harvard,linus,mit-eddie}!necntc!necis!mrst!sdti!mjy
Internet : mjy%sdti.uucp at harvard.harvard.edu      Tel: +1 617 443 5779