re: netbsd-5 branch now considered mostly stable for SMP

[matthew green <mrg%eterna.com.au@localhost>]

failure is expected :-)  thanks for testing.

> netbsd:esp_dma_setup+0x1c esp_dma_setup(0xf3d82000, 0xf3d82148, 0xf3d8214c, 
> 0x0, 0xf03898c4, 0xf03dcfc8) at netbsd:ncr53c9x_intr+0x12b8
> ncr53c9x_intr(0x0, 0xf00cc184, 0x700, 0x408000e6, 0xf03efcb4, 0x2) at 
> netbsd:sparc_interrupt44c+0x160

OK, i'm convinced i need to put a disk into my ss20.  :-)  i usually
run everything diskless..

> == 2nd panic ==
> sparc20c# panic: tp->t_lastm == NULL
> Begin traceback...
> 0x0(0xc, 0x3c, 0x48, 0x4000, 0x4000, 0x8000) at netbsd:tcp_usrreq+0x404
> tcp_usrreq(0xf0afaab0, 0x9, 0xf0b9ad00, 0x0, 0xf474d5a0, 0x0) at 
> netbsd:tcp_usrreq_wrapper+0x4a39c
> tcp_usrreq_wrapper(0xf0bc3b80, 0x9, > 0xf0b9ad00, 0x0, 0x0, 0xf474d5a0) at 
> netbsd:sosend+0x49c
> sosend(0x0, > 0xc8, 0xf48ade20, 0x0, 0x0, 0x0) at netbsd:soo_write+0x2c
[ ... ]

this one i don't know about at all, and have never seen it.

i'm curious if you have or can test -current kernels?  thanks.


in other news, the rmind-uvmplock branch is appearing to be very stable
for SMP sparc machines, more so than -current.  we can't back port that
branch to netbsd-5, but maybe we can use it to figure out the missing
parts.


.mrg.