Re: SMP support for sparc64 is now in -current

[Chris Ross <cross%distal.com@localhost>]

On Mar 14, 2008, at 10:56 PM, Chris Ross wrote:
>  Okay.  I built current as of this afternoon, and tried bringing it  
> up on a dual-200Mhz UltraSPARC Ultra2 (yes, that's with UltraSPARC-1  
> processors).  Network boot, NFS root and swap.

  Okay.  I've gotten back to this again.  A current updated as of  
yesterday afternoon (27-March), running GENERIC.MP on a dual- 
UltraSPARC, network boot, NFS root and swap.

  Most times the machine doesn't get all the way to getty.  It always  
fails with the following data fault:

cpu0: data fault: pc=100bdc8 addr=0

  Sometimes this is on cpu0, sometimes on cpu1.  The pc is always  
0x100bdc8, at least for this build of the kernel.  When it's on cpu0  
it just hangs hard at this point.  When on cpu1, I get to ddb, and  
backtrace showed (the twice I tried it) the following

  Anything I can do to help at this point?  Is this a known problem?

                               - Chris


cpu1: data fault: pc=100bdc8 addr=0
kernel trap 30: data access exception
1 tt=60 tstate=4448000602 tpc=0x143cadc tnpc=0x143cae0
Stopped in pid 103.1 (syslogd) at       netbsd:sparc64_ipi_save_fpstate 
+0x10:   ldx              [%g3 + 0x40], %g3
db{1}> bt
sleepq_remove(bee8000, c85d1c0, 142, 166edb8, 2, 1c14000) at  
netbsd:sleepq_remove+0x248
sleepq_wake(c85d1c0, c7a1968, 1, ffffffffffffffff, 0, f4240) at  
netbsd:sleepq_wake+0x8c
cv_broadcast(c7a1968, 0, ffffffffffffffff, 18cb960, 0, c9b1500) at  
netbsd:cv_broadcast+0x64
nfs_rcvunlock(c7a1960, c9b1500, c9b1500, 70, 3, c7a1960) at  
netbsd:nfs_rcvunlock+0x20
nfs_reply(0, c9b1500, 24c2800, c7fe3d0, c9b1500, 15030b0) at  
netbsd:nfs_reply+0x464
nfs_request(0, 24c3400, 18c5000, c9b1500, b807ec0, 6) at  
netbsd:nfs_request+0x2c8
nfs_readrpc(0, c8ef498, 18c5000, c8ef398, 0, b82ecc0) at  
netbsd:nfs_readrpc+0x394
nfs_doio(b832ea0, c9b1500, 0, 1667b30, 2, 1c14000) at netbsd:nfs_doio 
+0x27c
VOP_STRATEGY(b82ecc0, 24edc00, c8ef7e8, c8ef7d0, c8ef7f8, 0) at  
netbsd:VOP_STRATEGY+0x40
genfs_getpages(c8ef998, c8ef748, 0, 1, 0, c8ef748) at  
netbsd:genfs_getpages+0x8
nfs_getpages(c8ef998, c8ef021, 343, c8ef8c0, 1, b832ea0) at  
netbsd:nfs_getpages+0x74
VOP_GETPAGES(b82ecc0, dc000, c8efce0, c8efcfc, 0, 4) at  
netbsd:VOP_GETPAGES+0x60
uvn_get(0, dc000, c8efce0, c8efcfc, 0, 4) at netbsd:uvn_get+0xd4
uvm_fault_internal(409e4000, 40, 4, 5, c8efc28, 1) at  
netbsd:uvm_fault_internal+0x134c
text_access_fault(c8efed0, 64, 409dcb00, 0, 0, 5e6b490) at  
netbsd:text_access_fault+0x68
?(2001, 107800, 14, ffffffffbf3d494f, ffffffffffffffff, ff00) at  
0x1008a3c



cpu1: data fault: pc=100bdc8 addr=0
kernel trap 30: data access exception
1 tt=60 tstate=4448000601 tpc=0x143cadc tnpc=0x143cae0
Stopped in pid 102.1 (sh) at    netbsd:sparc64_ipi_save_fpstate 
+0x10:   ldx         [%g3 + 0x40], %g3
db{1}> bt
sleepq_remove(bee8000, b8303c0, 142, 166edb8, 0, 0) at  
netbsd:sleepq_remove+0x248
sleepq_wake(b8303c0, b821a78, 1, 166fb98, 27a2, 1) at  
netbsd:sleepq_wake+0x8c
cv_broadcast(b821a78, c861c40, ff0a0000000001, 18cb9a0, 0, 0) at  
netbsd:cv_broadcast+0x64
exit1(c861c40, b821a40, 0, ffffffffffffa8f8, 18cac00, 0) at  
netbsd:exit1+0x570
sys_exit(c861c40, c8b7dc0, c8b7e00, 1, 222f80, 0) at netbsd:sys_exit 
+0x4c
syscall_plain(c8b7ed0, 1, 4093c228, c8b7dc0, 1, 4093c228) at  
netbsd:syscall_plain+0x124
?(0, 0, 223800, ffffffffffffff88, 11e000, 223400) at 0x1008da8
db{1}> reboot