B132L+ crash with mpg123
Hi fellows !
I've been experiencing several crashes on B132L+ while playing mp3s with mpg123, the box was then running 2.4.18-pa21.
I've tried recently with 2.4.19-pa18, and it crashed again. The crash happens at random time, on random files.
What's worth mentionning is that files are played from a remote NFS filesystem. (BTW, the console shows "svc: unknown version (0)" when mounting the NFS system).
So I took my little fingers and I reproduced here what the console showed me. I hope there is now mistake in this "hand dump", though I read it twice ;)
HTH,
Thibaut VARENE
PA/Linux ESIEE Team
http://pateam.esiee.fr/
WARNING! Stack pointer and cr30 do not correspond! Dumping virtual address stack instead
Kernel Fault: Code=15 regs=11ddc640 (Addr=00277e00)
YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000011000000000000001111 Not tainted
r00-03 00000000 00000023 101feac4 00000200 r04-07 1035c1fc fffff000 1035c010 00000000 r08-11 1035c1fc 00001200 1035c010 1035c010 r12-15 00029f00 1035c25c faf01590 00017000 r16-19 00028678 00028678 00002000 00000e00 r20-23 00001000 00000000 00000001 00001200 r24-27 000001ff 00029f01 00277e00 10290010 r28-31 00000000 ffffffff 11ddc640 101fea8c sr0-3 00000000 00000246 00000000 00000246 sr4-7 00000000 00000000 00000000 00000000
IASQ: 00000000 0000000 IAOQ: 10263c78 10263c70
IIR: 0f411222 ISR: 00000000 IOR: 00277e00
CPU: 0 CR30: 11ddc000 CR31: 10338000
ORIG_R28: 11ddc7b8
Quoted 2 lines · Thibaut VARENE
I've been experiencing several crashes on B132L+ while playing mp3s with mpg123, the box was then running 2.4.18-pa21.
Harmony driver definitely has some bugs.
My B180 is stable until I listen to MP3s as well.
System crashes after ~1 hour of play.
Quoted 3 lines
What's worth mentionning is that files are played from a remote NFS filesystem. (BTW, the console shows "svc: unknown version (0)" when mounting the NFS system).
I'm doing everything local.
Quoted 5 lines
Kernel Fault: Code=15 regs=11ddc640 (Addr=00277e00)
YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000011000000000000001111 Not tainted
r00-03 00000000 00000023101feac400000200
...
Quoted 1 line
IASQ: 00000000 0000000 IAOQ:
10263c7810263c70
...
what symbols do 101feac4 and 10263c78 refer to?
grant
Quoted 9 lines · Grant Grundler
Quoted 5 lines
Kernel Fault: Code=15 regs=11ddc640 (Addr=00277e00)
YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000011000000000000001111 Not tainted
r00-03 00000000 00000023101feac400000200...
Quoted 1 line
IASQ: 00000000 0000000 IAOQ:
10263c7810263c70...
what symbols do101feac4and10263c78refer to?
10263c78 is in copy_from_user, 101feac4 is in harmony_audio_write(). the only point at which we call copy_from_user is:
/* Copy the page to an aligned buffer */ copy_from_user(played_buf.addr + (HARMONY_BUF_SIZE*buf_to_fill)
+ harmony.play_offset,
buffer+count, frame_size);
Signature · 1 line
Revolutions do not require corporate support.
Quoted 10 lines
Quoted 9 lines
Quoted 5 lines
Kernel Fault: Code=15 regs=11ddc640 (Addr=00277e00)
YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000011000000000000001111 Not tainted
r00-03 00000000 00000023101feac400000200...
Quoted 1 line
IASQ: 00000000 0000000 IAOQ:
10263c7810263c70...
what symbols do101feac4and10263c78refer to?
Quoted 8 lines
10263c78is incopy_from_user,101feac4is inharmony_audio_write(). the only point at which we callcopy_from_useris:/* Copy the page to an aligned buffer */ copy_from_user(played_buf.addr + (HARMONY_BUF_SIZE*buf_to_fill)
+ harmony.play_offset,
buffer+count,frame_size);
Data TLB miss fault / Data page fault.
Looks like it walked a little too far into some address that wasn't mapped?
c.
Le mercredi, 18 sep 2002, =E0 03:02 Europe/Paris, Carlos O'Donell a = =E9crit=20
:
Quoted 20 lines
Quoted 10 lines
Quoted 9 lines
Quoted 5 lines
Kernel Fault: Code=3D15 regs=3D11ddc640 (Addr=3D00277e00)
YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000011000000000000001111 Not tainted
r00-03 00000000 00000023101feac400000200...
Quoted 1 line
IASQ: 00000000 0000000 IAOQ:
10263c7810263c70...
what symbols do101feac4and10263c78refer to?Quoted 9 lines
10263c78is incopy_from_user,101feac4is inharmony_audio_write(). the only point at which we callcopy_from_useris:/* Copy the page to an aligned buffer */ copy_from_user(played_buf.addr +=20 (HARMONY_BUF_SIZE*buf_to_fill) + harmony.play_offset, buffer+count, frame_size);
Quoted 4 lines
Data TLB miss fault / Data page fault.
Looks like it walked a little too far into some address that wasn't mapped?
Yeah,
I caught Matthieu Delahaye lately, he told me that it was probably a failing "modulo" on the calculation of the remaining size of buffer to be sent to the audio card, or something like that...
Quoted 2 lines
c.
Thibaut VARENE
PA/Linux ESIEE Team
http://pateam.esiee.fr/=