Re: Linux-1.3.9

Temptation (
Wed, 12 Jul 1995 11:16:31 -0400 (EDT)

Hmmm. I have the SMC Power PCI combo card, DEC chipset, it would crash
only when FTPing to another one machine running NE2000 and any kernel.
but to another machine using the same type of card Ne2000 running
1.2.10, no problems, the 1.2.10 is in the middle, and connects fine, I
can connect the second machine from the SMC, and run programs, etc, I
can ping -f and no errors, just ftping files large then 1 meg, freeze
the machine.

Now with 1.3.9, seemed to solve this problem, I was ftping an 80meg
tar.gz file, got up to 70megs, while that was going, I opened another
console, and was doing ls -al , on the file, making sure the system was
crashing. Then I got

Process update (pid: 7, process nr: 6, stackpage=01f1f000)
Stack: .............

Call Trace: ..........
Code: ......
ls -algeneral protection: 00000
more stuff

Process Swapper (pid: 0, process nr: 0, stackpage=001c8e14)
Stack: ......
more stuff again

Aiee, killing interrupt handler
kfree of non-kmalloced memory: 001cae0c, next= 000000000, order=0
task[0] (swapper) killed: unable to recover
kernel panic: Trying to free up swapper memory space
In swapper task - not syncing

I never get this message with any other kernel, nor with ftping to the
other 1.2.10 ne2000 machine, and I ftp about 5 gigs of files from it.
I am using the base slackware-ELF-beta, which I updated with all of HJ
gcc.2.7.0 binutils etc... with the newest ones in /gcc/private
I also add support for soudblaster, which it has. it's an ASUS MB, with
IntelDx66, onboard NCR scsi, and either 1.gig seagate or/and 9gig seagate,
and of course the SMC pci card.

On another note, 1.3.6-1.3.9 (different machine then above ones)
1.3.6 works fine, but 1.3.7-1.3.9, run VERY SLOW.
it's also running all the newest ELF stuff, but this one I did most of
all the compiling to ELF(with some help here and there) everything is ELF
but AccelX and Netscape, but doesn't matter. I could load the ELF Xfree,
and have the same problem.

11:08am up 8:07, 13 users, load average: 0.14, 0.06, 0.07
67 processes: 66 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 6.3% user, 33.4% system, 1.0% nice, 60.4% idle
Mem: 31156K av, 30140K used, 1016K free, 34272K shrd, 6512K buff
Swap: 58364K av, 0K used, 58364K free

this is what top says

total used free shared buffers
Mem: 31156 30260 896 34472 6512
-/+ buffers: 23748 7408
Swap: 58364 0 58364

and free. both about the same. Machine is P120, and with 1.3.7-1.3.9, it
takes for ever to run things, even doing "w" takes much longer then if I
ran and kernel before .7-.9
The shared seems very high to me. the Strange thing is it sorta always
high with other kernels, but after a day or so up time it would drop down
to 20-25megs, even tho the same programs where running, and nothing was
changed by me. I was going to report this long time ago, but I talked to
a few other people who said the didn't have this problem, but now it's
getting worse, and I'm getting the same thing on more then one brand and
time of machine. Also the more memory I add to the machine, the more
Linux uses. I could see a slow down if it used the swap, but it never
seems to use it, unless I run more then my standard config.


On Tue, 11 Jul 1995, Linus Torvalds wrote:

> I also hope lance ethernet chips will work again now..
> Linus