Thanks,
Kuang
>
>
> linux-kernel-digest Wednesday, 25 June 1997 Volume 01 : Number 9=
> 65
>
> In this issue:
>
> =09A strange slowing down problem
> =09Re: Interrupt distribution on 2.1.42
> =09Re: menuconf
> =09Re: LINUX && MEMORY >>>
> =092.1.43 has killed me
> =09more on linux kernel 2.1.43 problems
> =09Re: System is too big.!!!=20
> =09me again (2.1.43 probs) i said the wrong card for some reason
> =09new make modules on 2.1.43 (FREEZE UP)
> =092.1.43 problems almost solved
> =09ifconfig with 2.1.43
> =09CLOSE_WAIT
> =09Re: How fast can linux do serial I/O to a Z8530
> =09Re: menuconf
> =09Re: Keeping File system in Sync on two servers
> =09Re: Keeping File system in Sync on two servers
> =09The stalls are hardware or low-level driver, most likely...
> =09Last try on buffer code
> =09Support for 3c905 network card.
>
> See the end of the digest for information on subscribing to the linux-kerne=
> l
> or linux-kernel-digest mailing lists.
>
> ----------------------------------------------------------------------
>
> From: Dong Chen <chen@ctpa04.mit.edu>
> Date: Tue, 24 Jun 1997 17:52:06 -0400 (EDT)
> Subject: A strange slowing down problem
>
> Hi,
>
> There is a strange problem when I run kernel 2.0.30 on my AST J50 P133.
> In short, when APM is enabled in the kernel and PCMCIA module
> i82365.o is loaded, any process running longer than 1 second are slowed dow=
> n.
> (This happens on other AST J series notebooks as well, not sure about
> other manufacturer's notebooks.)=20
>
> If there are keyboard/mouse activities at a frequency > 1Hz, the slowing
> down does not occur. If the kernel APM is disabled, or i82365.o
> is unloaded, no slowing down either.
>
> A summary table: (PCMCIA modules are compiled without APM support.)
>
> =09kernel APM=09i82365.o loaded=09=09slowing down
> =09no=09=09no=09=09=09no
> =09no=09=09yes=09=09=09no
> =09yes=09=09no=09=09=09no
> =09yes=09=09yes=09=09=09yes
> =09yes=09=09yes=09=09=09no if lots of keyboard typing
>
> The i82365.o loads an interrupt handler at IRQ 11. I am currently looking
> at arch/i386/kernel/process.c and drivers/char/apm_bios.c, but haven't
> seen anything suspicious.
>
> Can somebody shed some light on what might be the problem? Do other
> people have similar problems? Thanks a lot,
>
> Dong Chen
> chen@ctp.mit.edu
>
>
> - -----------------------------------------------------------------------
> PS. Sample code and timing
> - -----------------------------------------------------------------------
>
> /* loop_test.c
> *=09To compile:=09cc -o loop_test loop_test.c -lm
> *
> *=09Timing on AST J50, Pentium 133 MHz
> *
> *=09=09nloops=09=09seconds (user)
> *=09=0910=09=09 0.110
> *=09=0920=09=09 0.230
> *=09=0940=09=09 0.430
> *=09=0980=09=09 0.820
> *=09 100=09=09 1.070
> *=09 200=09=09 3.030=09=09(*)
> *=09 400=09=09 13.710=09=09(*)
> */
>
> #include <math.h>
>
> static volatile double x;
>
> main(int argc, char *argv[])
> {
> int i,j,k,n;
>
> if (argc < 2) {
> printf("Usage: %s nloops\n", argv[0]);=20
> =09exit(0);
> }
>
> sscanf(argv[1],"%d",&n);
>
> for (i=3D0; i<5; i++) {
> =09printf("%d\n", i+1);
> for (j=3D0; j<n; j++) {
> =09 for (k=3D0; k<1000; k++) {
> =09=09x =3D sin(1.0);
> =09=09x =3D cos(1.0);
> =09 }
> =09}
> }
> printf("done.\n");
> }
>
> ------------------------------
>
> From: Linus Torvalds <torvalds@transmeta.com>
> Date: Tue, 24 Jun 1997 14:53:39 -0700 (PDT)
> Subject: Re: Interrupt distribution on 2.1.42
>
> On Tue, 24 Jun 1997, Victor Yodaiken wrote:
> >
> > On my Dual Ppro Vectrex Motherboard,
> > only processor 1 gets I/O interrupts. Is this Linux or
> > is it the motherboard? The following patch prints the number of
> > i/o irqs handled in proc/cpuinfo.
>
> It's the motherboard (or rather, just the PC SMP standard layout). You
> _can_ do interrupt routing on the x86, but it has various nasty details
> that make it harder than it should be (somebody correct me if I'm wrong,
> but I think you generally can't route ISA interrupts except by bouncing
> them around between CPU's).
>
> The kernel can handle interrupts on different CPU's - it works that way on
> the sparcs, and in fact even on x86 we do do separate CPU's for the timer
> interrupts.
>
> =09=09Linus
>
>
> ------------------------------
>
> From: "William E. Roadcap" <roadcapw@titus.org>
> Date: Tue, 24 Jun 1997 19:34:49 -0400 (EDT)
> Subject: Re: menuconf
>
> On Tue, 24 Jun 1997, A.E. Brouwer wrote:
>
> >=20
> > o The console/window is "/usr/bin/reset" upon exit.
> >=20
> > A bad idea. Why would you want to do that?
> > This messes up my interrupt character etc.
>
> A bad I idea? Maybe.. It all depends on how you keep your interrupt set. =
>
> For some people, including me, it will make no difference. However, your
> point is well taken, and I should probably modify that change. Perhaps a
> simple 'tput rs2' or something similar?=20
>
> > Note that there are lots of `reset' programs or scripts around,
> > all intending to get a tty into more or less reasonable shape
> > after an unintended crash. Your program is buggy if it needs
> > reset.
>
> Ouch! That's kind of a harsh stance. Perhaps using /usr/bin/reset is not
> the best idea. However, choosing to reset the terminal doesn't
> automatically make a program "buggy". After all, Menuconfig can only be
> as stable and bug free as the ncurses library and the config.in script in
> the kernel tree. We all know that ncurses's stability varies from version
> to version, and the config.in scripts have been know to have errors during
> developent. I was simply trying to cover my ass in case ncurses, lxdialog
> or whatever left the terminal in an odd state. That's not bugginess, it's
> cleanliness.
>
>
> Thanks for your advice. I'll make the adjustment.
>
> __
> William E. Roadcap mailto:roadcapw@titus.org
> TITUS Software ftp://ftp.titus.org/pub
> Waynesboro, Va (USA) http://www.cfw.com/~roadcap=
> w
>
>
> ------------------------------
>
> From: robert@rocknet.net.au
> Date: Wed, 25 Jun 1997 12:49:06 +1000 (EST)
> Subject: Re: LINUX && MEMORY >>>
>
> >=20
> > robert@rocknet.net.au wrote:
> > >=20
> > > Hi,
> > >=20
> > > make zlilo ....
> > >=20
> > > Setup is 4336 bytes.
> > > System is 406 kB
> > > sync
> > > if [ -f /vmlinuz ]; then mv /vmlinuz /vmlinuz.old; fi
> > > if [ -f /System.map ]; then mv /System.map /System.old; fi
> > > cat zImage > /vmlinuz
> > > cp /usr/src/linux/System.map /
> > > if [ -x /sbin/lilo ]; then /sbin/lilo; else /etc/lilo/install; fi
> > > Added linux *
> > > kfree of non-kmalloced memory: 01a15618, next=3D 01e7f000, order=3D2
> > > kfree of non-kmalloced memory: 01a15498, next=3D 0110b000, order=3D2
> > > kfree of non-kmalloced memory: 01a27698, next=3D 00549000, order=3D2
> > > kfree of non-kmalloced memory: 01a27618, next=3D 00549000, order=3D2
> > > Added dos
> > > make[1]: Leaving directory `/usr/src/linux/arch/i386/boot'
> > >=20
> > > What are these "kfree of non-kmalloced memory ... " ...
> > > does it mean I have bad memory?
> >=20
> > Either that, or one of the drivers you have is behaving badly. As
> > far as I can tell, those numbers look just fine to me.=20
> >=20
> > I haven't had a look at the code just now, but I don't understand why
> > it would think that this memory was not-kmalloced.=20
> >=20
> > =09=09=09=09Roger.
> >=20
> ps: if replying please CC: robert@rocknet.net.au
>
> Ok, I got some more of those "kfree of ..." error messages just pop
> at random times, ... like as soon as I login in a virtual terminal,
> or run shell commands, ... I checked the 'numbers' I was getting
> this time: "1885e1g" to be memory in the third > 16MB simm bank,
> so I changed these edo ram chips ( 3rd and 4th bank ) for another two
> 8 MB chips ... and run a slightly modified version of the script
> on Roger's page
>
> #!/bin/bash
> log_dir=3D/root/adm
> cd /usr/src/linux
> echo -n "Running initial 'make zImage' ... "
> make zImage > $log_dir/zImage.init.log 2>&1
> echo "Done."
>
> for i in 0 1 2 3 4 5 6 7 8 9 ; do
> for j in 0 1 2 3 4 5 6 7 8 9 ; do
> echo -n "Running test: $i$j ... "
> make clean > $log_dir/clean.log.$i$j 2>&1
> make zImage > $log_dir/zImage.log.$i$j 2>&1
> echo "Done."
> done
> done
> # eof
>
> every zImage.log.XX file had the same size, so I "hope" it was
> _just_ the ram :) ....
>
> Anyway, thanks to all of you that have replied and given
> me sound advice.
>
> =09Rob -
>
>
> ------------------------------
>
> From: zeus <z6d41@TTACS.TTU.EDU>
> Date: Tue, 24 Jun 1997 22:26:19 -0500
> Subject: 2.1.43 has killed me
>
> This is a multi-part message in MIME format.
>
> - --------------7D185DD241998CE1B065C10
> Content-Type: text/plain; charset=3Dus-ascii
> Content-Transfer-Encoding: 7bit
>
> hello. I finally had the time to get back into the development kernels
> this past weekend. I downloaded all the new stuff needed for the 2.1.x
> kernels:
>
>
> - - Kernel modules modutils-2.1.34=20
>
> - - Binutils 2.8.1.0.1=20
> - - Linux C Library 5.4.23=20
> - - Dynamic Linker (ld.so) 1.8.5=20
> - - Linux C++ Library 2.7.2.1=20
> - - Procps 1.01=20
> - - Procinfo 0.11=20
> - - Mount 2.6g=20
> - - Net-tools 1.41=20
> - - Sh-utils 1.16=20
> - - Autofs 0.3.3=20
> =20
>
> I have a dual pentium (tyan tomcat) motherboard with 1 100mhz cpu. I
> have 32megs of ram, a 1.9gig scsi drive, ncr 8000 card, 1gig ide, 4x4
> nec cdrom. I am using (had already) Gnu C 2.7.2. I am running redhat
> linux 4.0. anyhow i installed all of this stuff and went to compile
> kernel 2.1.43. everything went find (make config; make dep; make
> clean... etc) until make zImage. then there is an error about SMP so I
> go into the toplevel Makefile and set SMP =3D 0 rather than SMP =3D 1.=20
> Everything compiles till i get to devices/char/rtc.c there is one to
> many ')' on line 185 ( i believe that is right ); I get rid of that
> brace and every thing compiles fine ( a few warnings here and there but
> hey i always see a few warnings ). i do a make zdisk it goes through
> fine does a dd and all seems well. I then do a make modules and a make
> modules_install. Actually it hung up the first time i did make modules.
> but i do not remeber where. then i cat arch/i386/zImage >
> /boot/zImage-2.1.43, and add the image to lilo.conf. i try running lilo
> ( version 19 ) but after it prints the first image name /boot/zImage
> (default) it hangs on the new image. I go ahead and reboot to try the
> new floppy but I get the following and it freezes:
>
>
> Lots of stuff...
> .
>