The 2.6.23 kernel built for mips with the attached .config works fine for me
under qemu (both big endian and little endian), but a 2.6.24 mips kernel
segfaults initializing the ne2k driver (again when run under qemu).
I've traced it to this commit:
http://kernel.org/hg/linux-2.6/rev/74258
Version 74257 works, 74258 does not.
The qemu invocation is:
qemu-system-mips -M mips -no-reboot -nographic -hda /dev/zero -kernel \
vmlinux -append "panic=1 console=ttyS0"
The panic is:
Linux version 2.6.24-rc2 (landley@driftwood) (libc/sysdeps/linux/mips/crt1.S:
(.text+0x1c): undefined reference to `main') #2 Mon Jan 28 02:08:00 CST 2008
console [early0] enabled
CPU revision is: 00019300 (MIPS 24K)
FPU revision is: 00739300
Determined physical RAM map:
memory: 08000000 @ 00000000 (usable)
Built 1 zonelists in Zone order, mobility grouping on. Total pages: 32512
Kernel command line: panic=1 console=ttyS0
Primary instruction cache 2kB, VIPT, 2-way, linesize 16 bytes.
Primary data cache 2kB, 2-way, VIPT, no aliases, linesize 16 bytes
Synthesized clear page handler (13 instructions).
Synthesized copy page handler (22 instructions).
Synthesized TLB refill handler (20 instructions).
Synthesized TLB load handler fastpath (32 instructions).
Synthesized TLB store handler fastpath (32 instructions).
Synthesized TLB modify handler fastpath (31 instructions).
Cache parity protection enabled
PID hash table entries: 512 (order: 9, 2048 bytes)
Using 100.000 MHz high precision timer.
Dentry cache hash table entries: 16384 (order: 4, 65536 bytes)
Inode-cache hash table entries: 8192 (order: 3, 32768 bytes)
CPU 0 Unable to handle kernel paging request at virtual address 00000004, epc
== 80053414, ra == 80054960
Oops[#1]:
Cpu 0
$ 0 : 00000000 10008400 ffff7fff 10000400
$ 4 : 00000017 00000000 801db96c 00000011
$ 8 : 10008400 1000001f 00000005 81100000
$12 : 801e0000 801e0000 3b9aca00 801b1868
$16 : 801b64d8 00000017 00000200 00000000
$20 : ffffffff 00007e20 0000001f 00015c00
$24 : 00091a6f 801b1a50
$28 : 801ae000 801afe08 00008000 80054960
Hi : 03b9aca0
Lo : 00000000
epc : 80053414 Not tainted
ra : 80054960 Status: 10000402 KERNEL EXL
Cause : 40008408
BadVA : 00000004
PrId : 00019300 (MIPS 24K)
Process swapper (pid: 0, threadinfo=801ae000, task=801b0160)
Stack : 801dc407 801dc7d4 801aff3c 801c0000 801b64d8 00000017 00000200
00000000
ffffffff 80054960 10000000 801b0000 000004b2 000004b2 00000000
0000001f
80010b6c 000005b8 80031348 8003129c 801aff2c 801d0000 00000000
800111e4
801dc414 8019f9c4 801dc413 801d0000 00000040 801b0000 00000000
10008400
00000000 fffffbff 00000001 81015e20 00000001 00000011 81015e40
81015c00
...
Call Trace:[<801c0000>][<80054960>][<80010b6c>][<80031348>][<8003129c>]
[<801d0000>][<800111e4>][<801d0000>][<8005d5b8>][<801ca414>][<801cb820>]
[<801cb744>][<801cb5a0>][<801d0000>][<801c3314>][<801c9b10>][<801cd79c>]
[<801d0000>][<801bfb30>][<801bf118>]
Code: afb3001c afb20018 afb10014 <8ca20004> 00a08021 30420020 14400009
0080a021 40016000
Kernel panic - not syncing: Fatal exception in interrupt
Rebooting in 1 seconds..
The final 2.6.24 release panics slightly differently, it waits until the ne2k
init and then loses it. I'm not quite sure this is the same problem (since
the error is now occuring in a different place), but I haven't found any
kernel _after_ that which won't panic during init. Only the placement of the
panic changes.
I fiddled around a bit in the debug menu, but didn't figure out how to get
actual function names of out a mips trace...
The way I reproduce this is with my http://landley.net/hg/firmware/ build
script, download the most recent version (links to tarballs at the top),
run "./build.sh mips" and let it grind a lot, then extract the
build/qemu-image-mips.tar.bz2 tarball, cd qemu-image-mips and
run "./run-mips.sh" in that directory. (I have qemu 0.9.1 installed.)
More details in my blog at http://landley.net/notes-2008.html#27-01-2008
Rob
--
"One of my most productive days was throwing away 1000 lines of code."
- Ken Thompson.
