Dear comunity, I use ar2317 mips wisoc board and i got 2 problems: netbsd-5 - ar2317 wlan doesn't work netbsd-current - ABI is broken(I suppose you know) and wifi is not working of course. http://dpaste.com/181471 -- -- Sergeyli
On Fri, Apr 09, 2010 at 10:42:42PM +0400, Jon Shadow wrote: > Dear comunity, > > I use ar2317 mips wisoc board and i got 2 problems: > > netbsd-5 - ar2317 wlan doesn't work IIRC, the MIPS kernel ABI changed in -5 with the SMP improvements, however, -5 still uses the binary HAL, so there is an ABI mismatch. > netbsd-current - ABI is broken(I suppose you know) and wifi is not > working of course. The ABI is broken? Will you elaborate on that? In -current, the HAL is built from sources. Dave -- -- David Young OJC Technologies dyoung <at> ojctech.com Urbana, IL * (217) 278-3933
When I boot with a new kernel, I get ABI is set to O32 or something like that and It crashes into ddb console. -- -- Sergeyli
On Sat, Apr 10, 2010 at 08:40:19AM +0400, Jon Shadow wrote: > When I boot with a new kernel, I get ABI is set to O32 or something > like that and It crashes into ddb console. The ABI is probably not the issue there as O32 is what's been in place all along. What's the stack trace from the crash? -- -- David A. Holland dholland <at> netbsd.org
That's with wifi on the current :
db> bt
PC 0x0: not in kernel space
0+0 (a87f0000,b0000000,0,1000004) ra 0 sz 0
User-level: pid 0.1
pid 1(init): ABI set to O32 (e_flags=0x1007)
panic: init died (signal 0, exit 1)
Stopped in pid 1.1 (init) at netbsd:cpu_Debugger+0x4: jr ra
bdslot: nop
db> bt
cpu_Debugger+4 (80fff000,b1100000,0,c2389e1c) ra 8016e300 sz 0
panic+260 (80fff000,0,1,7dfac494) ra 8010a6e4 sz 48
exit1+58 (80fff000,0,1,7dfac494) ra 8010addc sz 120
sys_exit+60 (80fff000,0,1,7dfac494) ra 8017dcd4 sz 32
syscall+1c4 (80f3b5e0,0,1,7dfac494) ra 801311d0 sz 104
mips32_SystemCall+dc (80f3b5e0,0,1,7dfac494) ra 7dfac494 sz 0
PC 0x7dfac494: not in kernel space
0+7dfac494 (80f3b5e0,0,1,7dfac494) ra 0 sz 0
User-level: pid 1.1
db> ps
PID LID S CPU FLAGS STRUCT LWP * NAME WAIT
1 > 1 7 0 0 80f3b5e0 init
0pid 1(init): trap: address error (store) in kernel mode
status=0x2, cause=0x50009014, epc=0x8009f6fc, vaddr=0xc2389bb1
tf=0xc2389af0 ksp=0xc2389b50 ra=0x80c
Stopped in pid 1.1 (init) at netbsd:db_read_bytes+0x2c: sw v
0,0(s1)
db>
ABI stuff:
(Continue reading)
Okay, wifi fails in function ar5312MacReset in ~/netbsd/attempt/src/sys/external/isc/atheros_hal/dist/ar5312 -- -- Sergeyli
file name art5312_reset.c 2010/4/10 Jon Shadow <shadowpilot34 <at> gmail.com>: > Okay, wifi fails in function ar5312MacReset in > ~/netbsd/attempt/src/sys/external/isc/atheros_hal/dist/ar5312 > > -- > Sergeyli > -- -- Sergeyli
On Sat, Apr 10, 2010 at 02:15:32PM +0400, Jon Shadow wrote: > pid 1(init): ABI set to O32 (e_flags=0x1007) > panic: init died (signal 0, exit 1) That doesn't make much sense - AFAICT the only way to get init to exit(1) once it starts is to send it SIGHUP, and that is not very plausible here. Perhaps something bogus is going on with the dynamic linker? -- -- David A. Holland dholland <at> netbsd.org
Okay, that was with new kernel and netbsd5 userspace. With all current I only get pid 1(init): ABI set to O32 (e_flags=0x1007) and that's it. 2010/4/11 David Holland <dholland-mips <at> netbsd.org>: > On Sat, Apr 10, 2010 at 02:15:32PM +0400, Jon Shadow wrote: > > pid 1(init): ABI set to O32 (e_flags=0x1007) > > panic: init died (signal 0, exit 1) > > That doesn't make much sense - AFAICT the only way to get init to > exit(1) once it starts is to send it SIGHUP, and that is not very > plausible here. > > Perhaps something bogus is going on with the dynamic linker? > > -- > David A. Holland > dholland <at> netbsd.org > -- -- Sergeyli
On Sun, Apr 11, 2010 at 05:14:15PM +0400, Jon Shadow wrote: > Okay, that was with new kernel and netbsd5 userspace. With all current > I only get pid 1(init): ABI set to O32 (e_flags=0x1007) and that's it. It just sits there? Marvelous. I guess this is the point where I'd start loading in debug printfs. :-/ ddb can probably tell you something about what's happening to init, but probably not enough to be useful. Make sure /dev/console exists and is the correct device number, too. If it's completely not there init will try to mount and populate an mfs, but that can easily go wrong, and if it's there but the wrong device all manner of creative things can happen. Compiling init statically might also make it easier to diagnose what's up, or if the problem involves the dynamic linker might make the system come up far enough to make the rest tractable. -- -- David A. Holland dholland <at> netbsd.org
RSS Feed