1.3.94 problems seemed genuine....

Chris Evans (chris@ferret.lmh.ox.ac.uk)
Wed, 24 Apr 1996 23:03:12 +0100 (BST)


On Tue, 23 Apr 1996, Alan Cox wrote:

> > Other 1.3.94 problems I've found include a large number of unresolved
> > symbols (every module I compiled in fact), and the failure of affs to
> > compile (a module option for this would be nice).
>
> AFFS doesnt compile. I've had no other apparent module trouble with this
> release.

I probably failed to do a make mrproper and start again.... but I've
dropped to 1.2.13 for now (which was hassle since I need 3c59x and
aic7xxx drivers!!)

> > Running sdoom on the console still reliably freezes the dumb terminal
> > attatched to ttyS2 (2nd serial card), sometime causing an oops in doom
> > when you hammer the keyboard on this terminal in its freezed state. Other
> > console games (abuse for example) cause no such problems.
>
> Sounds like you have an XGA compatible video card or an XGA probe going on.
> Some (many) ISA cards decode only the lower 9bits and the ISA stuff for
> XGA confuses it. Have you changed your SVGAlib at all ?

Card is DS64 2Mb DRAM, ark2000pv chipset. Changed my svgalib because of
this a few days ago... 1.2.10. (which recognises chipset, but seems a
little unstable and offers only 640x480x256 as yet). 1.2.13 dumb
terminal hangs while doom runs too, (but can't be persuaded to crash by
over-enthusiastic hammering on dumb terminal keyboard).

> [Real looking OOPS omitted]

All the oopses were probably NOT faulty hardware I can reveal, 1.2.13
hasn't barfed once yet, and I've compiled a few kernels. 1.3.94 tended to
throw out ~3 oopses per kernel compile attempt... statistically this
would suggest 1.3.9x problems.

Incidentally, the reproducible badblocks oops has disappeared in 1.2.13.
I can now run badblocks /dev/sda1 72261 (swap partition, deactivated of
course) with no oops.

Device Boot Begin Start End Blocks Id System
/dev/sda1 1 1 9 72261 82 Linux swap

In 1.3.9[1,3,4], the badblocks command would invariably oops, (timewise)
near the end of its check. The oops was posted here before.

Since my problems don't seem to be hardware related, I would treat you to
my last few oopses before I finally got shot of 1.3.94, but my old
System.map got hosed. Basically, I was seeing a lot of "free_one_pmd: bad
directory entry 0000000" variety of message. I also saw the occasional
swap free: weirdness and even a "swap-space map bad". Note that swap is
rarely touched.

Cheers,
-- Chris