24 May 2002


Rusty made a comment that sparc32 doesn’t even compile in 2.5… This worried me a little, as my main mail server is a sparc 10. I copied the code to wayne (the sparc), and did a make config; make dep… I didn’t et far:

[root@wayne linux-2.5.14]# make dep
gcc -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -o scripts/mkdep scripts/mkdep.c
make -C arch/sparc/kernel check_asm
make[1]: Entering directory `/usr/src/linux-2.5.14/arch/sparc/kernel'
gcc -E -D__KERNEL__ -I/usr/src/linux-2.5.14/include tmp.c -o tmp.i
In file included from /usr/src/linux-2.5.14/include/linux/spinlock.h:7,
                 from /usr/src/linux-2.5.14/include/linux/tqueue.h:16,
                 from /usr/src/linux-2.5.14/include/linux/sched.h:10,
                 from tmp.c:3:
/usr/src/linux-2.5.14/include/linux/thread_info.h:11: asm/thread_info.h: No such file or directory
In file included from tmp.c:3:
/usr/src/linux-2.5.14/include/linux/sched.h:353: warning: `get_task_struct' redefined
/usr/src/linux-2.5.14/include/asm/processor.h:199: warning: this is the location of the previous definition
make[1]: *** [check_asm] Error 1
make[1]: Leaving directory `/usr/src/linux-2.5.14/arch/sparc/kernel'
make: *** [check_asm] Error 2
[root@wayne linux-2.5.14]#

The first step seems to me to understand what to put in this file, so I copied the one from include/asm-sparc64 into include/asm-sparc, and tried to recompile…

It would seem that a whole bunch of definitions in processor.h which have moved into thread_info.h — started moving these across to reduce the compile errors.

Hmmmmmmmm, it would seem that spinlock_t isn’t defined whereever oplib.h needs it to be. Next step, find it!

CLUG meeting last night. Noone seems to have an answer about how to make docbook less ugly in it’s rendered form. Thye talk about the SQL database on top of tdb went well…


23 May 2002


Docbook questions for the day…

  • Where do I get less ugly docbook style sheets for jade?
  • How do I generate one big html page instead of the many small files?


22 May 2002


Hmmmm, no diary entry for a while. I’ve been doing heaps of marking of assignments, as well as doing some assignments myself. The really good thing for the week — Andrew slept through the night Sunday night…


15 May 2002


Still sleeping in shifts, so still quite tired. I have had a suggestion from someone on IRC… I am going to progressively installer increasingly older kernels until APM starts to work again. This will at least minimize the number of differences between the working kernel and the not working kernel. So I have downloaded 2.4.17, and am compiling it at the moment…


13 May 2002


Andrew, Catherine and I are all out of hospital again. We’re trialling a new scheme where I go to bed really early, and then it is my job to wake up to Andrew during the night and early in the morning. It is actually pretty good, because I get at least a few hours of sleep before I am first woken up. Catherine gets to stay up to do the late (midnight feed), and then gets to sleep for the rest of the night, so hopefully she wont get so tired either.

Still trying to debug this APM on 2.4.18 problem. My notes so far read something like:

I think my initial problem was some undefined symbols. I am not sure why
these weren't defined, I should try to find out why sometime...

[root@localhost kernel]# insmod apm
Using /lib/modules/2.4.18/kernel/arch/i386/kernel/apm.o
/lib/modules/2.4.18/kernel/arch/i386/kernel/apm.o: unresolved symbol machine_real_restart_Rsmp_3da1b07a
/lib/modules/2.4.18/kernel/arch/i386/kernel/apm.o: unresolved symbol default_idle_Rsmp_92897e3d

After commenting these function calls out, recompiling the module and
rebooting with an apm=debug, I get:

apm: BIOS version 152.172 Flags 0xd231 (Driver version 1.16)
apm: no 32 bit BIOS support

Which is interesting, because the 2.4.2 kernel says the following
(with apm-debug):

apm: BIOS version 1.2 Flags 0x03 (Driver version 1.14)
apm: entry f000:3f7f cseg16 f000 dseg 40 cseg len ffff, dseg len ffff cseg16 len ffff
Starting kswapd v1.8
apm: Connection version 1.2
apm: AC off line, battery status high, battery life 76%
apm: battery flag 0x01, battery life 81 minutes

The APM_BIOS_INFO #define in setup.c has not changed between the versions,
and seems to be copied straight to apm_info.bios.

I then printed out the value of the pointer to the apm_info.bios structure
for 2.4.18, and 2.4.2:

Mikal: apm.bios points to 0xc3a398ac   <---- 2.4.18

These values don't change across multiple insmods, but this isn't conclusive.

03 May 2002


Got very little sleep last night. Andrew was up all the time crying, for no descernable reason. I have to write a conference paper proposal for AUUG 2002 before Friday next week…