Re: pae MP on cherry-xenmp

["Michael L. Hitch" <mhitch%lightning.msu.montana.edu@localhost>]

On Tue, 30 Aug 2011, Jeff Rizzo wrote:

> I did fairly quickly have the same panic Michael did (note that this is in a 
> kernel including the per-cpu MMU queue commit you made about 45m ago).:

  With my last version of my patch, I was able to run with 4 vcpus for a 
while.  I was having some things going on - most noticable was a number of
my cacti graphs weren't being drawn correctly, if at all.  I then updated 
with Cherry's last commit and started with 4 vcpus again.  My cacti graphs 
looked much better at that point, but I got this same panic fairly 
quickly.

> panic: kernel diagnostic assertion "gnt_entries[last_gnt_entry] == 
> XENGNT_NO_ENTRY" failed: file 
> "/Users/riz/Documents/code/netbsd/xenmpsrc/sys/arch/xen/xen/xengnt.c", line 
> 208

  I just took a quick look at this code, and it looks like it's a problem 
of MP-unsafe code.  It's using splvm() to protect sections of code, which 
isn't going to work for MP.  I'm going to see if I can add a mutex and see 
if that works better.

> fatal breakpoint trap in supervisor mode
> trap type 1 code 0 eip c0136934 cs 9 eflags 246 cr2 caff6000 ilevel 6
> Stopped in pid 75.1 (ssh) at    netbsd:breakpoint+0x4:  popl    %ebp
> breakpoint(c047ee33,cb87fb88,ca705440,c0420927,1,cb41a2c0,cb87fb7c,c021fd42,0,4)
> at netbsd:breakpoint+0x4
> panic(c04bd818,c047807d,c04ba60c,c04ba63c,d0,cb41a2c0,cb87fbbc,c0425e98,c047807d
> ,c04ba63c) at netbsd:panic+0x20c
> kern_assert(c047807d,c04ba63c,d0,c04ba60c,974e9000,1,cb87fbdc,c0425f24,0,cb41a2c
> 0) at netbsd:kern_assert+0x39
> xengnt_get_entry(0,cb41a2c0,0,1e6f323e,4,30e,cb87fc5c,c01c0934,0,974e9000) at 
> ne
> tbsd:xengnt_get_entry+0x168
> xengnt_grant_access(0,974e9000,1,4,cb0a325c,0,1e5f3,0,0,0) at 
> netbsd:xengnt_gran
> t_access+0x24
> xennet_softstart(cb0a2004,0,4,5,cb87fd00,0,cb87fc98,ca73a074,ca73a154,ca705440) 
> a
> t netbsd:xennet_softstart+0x514
> softint_overlay(5,cb095040,bb230000,30,cb095040,1,cb87fd28,cb41a2c0,cb41b87c,0) 
> a
> t netbsd:softint_overlay+0x483
> lwp_userret(cb41a2c0,cb87fd00,cb87fd28,bb25f000,bb25f000,cb41b87c,4,5,bb230000,3
> 0) at netbsd:lwp_userret+0x215
> syscall(cb87fd48,bb4000b3,1f,bf7f001f,bb60001f,bb230000,30,bf7fbf68,bb63b398,0) 
> a
> t netbsd:syscall+0x15f
> ds          cb0a0011
> es          c04b0011    copyright+0x383d1
> fs          31
> gs          cb870011
> edi         cb0a325c
> esi         c04bd818    copyright+0x45bd8
> ebp         cb87fb3c
> ebx         104
> edx         0
> ecx         c04bd818    copyright+0x45bd8
> eax         1
> eip         c0136934    breakpoint+0x4
> cs          9
> eflags      246
> esp         cb87fb3c
> ss          11
> netbsd:breakpoint+0x4:  popl    %ebp
> db{1}>


Mike

--
Michael L. Hitch                        mhitch%montana.edu@localhost
Computer Consultant
Information Technology Center
Montana State University        Bozeman, MT     USA