Re: UVM Fault

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Wed, Jun 11, 2008 at 07:59:28AM -0600, John R. Shannon wrote:
> Manuel Bouyer wrote:
> >On Tue, Jun 10, 2008 at 04:30:05AM -0600, John R. Shannon wrote:
> >>proxy# uvm_fault(0xcd370af4, 0, 1) -> 0xe
> >>fatal page fault in supervisor mode
> >>trap type 6 code 0 eip c046eb0e cs 9 eflags 10282 cr2 0 ilevel 0
> >>kernel: supervisor trap page fault, code=0
> >>Stopped in pid 334.1 (pinger) at        netbsd:rb_tree_insert_node+0xc: 
> >>movl    0
> >>(%eax),%eax
> >>db> bt
> >>rb_tree_insert_node(c05099f8,c075b18c,cd44a9dc,c0342e6b,c05099e0,c075b18c,cd44a9
> >>cc,c03071f9,c075b18c,0) at netbsd:rb_tree_insert_node+0xc
> >>uvm_pageinsert_after(c075b18c,0,cd44aa2c,c0307186,c075b18c,0,1,0,cd44aa0c,c07594
> >>1c) at netbsd:uvm_pageinsert_after+0x24
> >>uvm_pageinsert(c075b18c,0,1,0,cd44aa0c,c075941c,2,cd44aa0c,0,0) at 
> >>netbsd:uvm_pa
> >>geinsert+0x19
> >>uvm_pagealloc_strat(c05099e0,0,0,0,2,0,0,c0436096,0,1) at 
> >>netbsd:uvm_pagealloc_s
> >>trat+0x283
> >>uvm_loanzero(cd44aad0,cd44aaec,2,cd63e8b0,cd370af4,804d000,cd44aae4,804e000,2000
> >>,cd3c9310) at netbsd:uvm_loanzero+0x76
> >>uvm_loanentry(cd44aad0,cd44aaec,2,0,0,cd6c4000,cd44aafc,cd370af4,804d000,3000)
> >> 
> >>a
> >>t netbsd:uvm_loanentry+0x10d
> >>uvm_loan(cd370af4,804d000,3000,c0b41658,2,c0b41600,cd44ab4c,c043613c,0,c0b41600)
> >> at netbsd:uvm_loan+0x61
> >>sosend_loan(c0c0d610,cd44ac3c,c0b41600,1f44,20f302f0,0,cd3d8a60,724b,1,0) 
> >>at net
> >>bsd:sosend_loan+0xe5
> >>sosend(c0c0d610,0,cd44ac3c,0,0,0,cd3d8a60,c0437136,0,c04d5e40) at 
> >>netbsd:sosend+
> >>0x360
> >>do_sys_sendmsg(cd3d8a60,1,cd44acb8,0,cd44ad24,bfbfee64,8,0,cd3d8a60,0) 
> >>at netbsd
> >>:do_sys_sendmsg+0x2fd
> >>sys_sendto(cd3d8a60,cd44acfc,cd44ad24,cd44ad00,c0356c85,cd3d8a60,1,804d620,1f44,
> >>0) at netbsd:sys_sendto+0x71
> >>syscall(cd44ad48,1f,1f,1f,1f,50,14,bfbfeba8,bbbaff38,56) at 
> >>netbsd:syscall+0x134
> >
> >
> >I suspect it's an MI problem. Did you try a newer kernel ? I think I seen
> >some code commited and then backed out, which may be related to this.
> >
> 
> From source pulled 2 hours ago:
> 
> uvm_fault(0xcd36daf4, 0, 1) -> 0xe
> fatal page fault in supervisor mode
> trap type 6 code 0 eip c046e8de cs 9 eflags 10282 cr2 0 ilevel 0
> kernel: supervisor trap page fault, code=0
> Stopped in pid 334.1 (pinger) at        netbsd:rb_tree_insert_node+0xc: 
> movl    0
> (%eax),%eax
> db> bt
> rb_tree_insert_node(c0509838,c074c07c,cd4329dc,c0342e2f,c0509820,c074c07c,cd4329
> cc,c0307295,c074c07c,0) at netbsd:rb_tree_insert_node+0xc
> uvm_pageinsert_after(c074c07c,0,cd432a2c,c0307222,c074c07c,0,1,0,cd432a0c,c07431
> 58) at netbsd:uvm_pageinsert_after+0x24
> uvm_pageinsert(c074c07c,0,1,0,cd432a0c,c0743158,2,cd432a0c,0,0) at 
> netbsd:uvm_pa
> geinsert+0x19
> uvm_pagealloc_strat(c0509820,0,0,0,2,0,0,c0435e56,0,1) at 
> netbsd:uvm_pagealloc_s
> trat+0x283
> uvm_loanzero(cd432ad0,cd432aec,2,cd5ae390,cd36daf4,804d000,cd432ae4,804e000,2000
> ,cd3b4e6c) at netbsd:uvm_loanzero+0x76
> uvm_loanentry(cd432ad0,cd432aec,2,0,0,cd683000,cd432afc,cd36daf4,804d000,3000)
>  
> a
> t netbsd:uvm_loanentry+0x10d
> uvm_loan(cd36daf4,804d000,3000,c0b41758,2,c0b41700,cd432b4c,c0435efc,0,c0b41700)
>  at netbsd:uvm_loan+0x61
> sosend_loan(c0c0d610,cd432c3c,c0b41700,1f3d,3c00e6b4,0,cd3b5a60,6a,1,0) 
> at netbs
> d:sosend_loan+0xe5
> sosend(c0c0d610,0,cd432c3c,0,0,0,cd3b5a60,c0436ef6,0,c04d5c80) at 
> netbsd:sosend+
> 0x360
> do_sys_sendmsg(cd3b5a60,1,cd432cb8,0,cd432d24,bfbfee64,8,0,cd3b5a60,0) 
> at netbsd
> :do_sys_sendmsg+0x2fd
> sys_sendto(cd3b5a60,cd432cfc,cd432d24,cd432d00,c0356a69,cd3b5a60,1,804d620,1f3d,
> 0) at netbsd:sys_sendto+0x71
> syscall(cd432d48,1f,1f,1f,1f,49,14,bfbfde28,bbbaff38,4f) at 
> netbsd:syscall+0x134

Maybe send this to current-users, or send a PR. It doesn't look Xen-specific.

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--