Subrahmanian, Raj
2006-Aug-21 05:38 UTC
[Xen-devel] NUMA patches and ES7000, partial success..
Ryan,
Applied the patch. Well, I was able to bring up and down machines. Our
tests ran ok. In that respect, it''s good news.
But when I tried to do xm top, there was segfault and a message "failed
to receive statistics from libxenstat.
I am sending a hg diff to you off-list so you can see if I patched it
right.
Raj
(XEN) *** Serial input -> Xen (type ''CTRL-a'' three times to
switch input
to DOM0).
(XEN) ''H'' pressed -> dumping heap info (now-0xBD:2C763C31)
(XEN) heap[0][0][2]-> 4 pages
(XEN) heap[0][0][3]-> 8 pages
(XEN) heap[0][0][4]-> 16 pages
(XEN) heap[0][0][6]-> 64 pages
(XEN) heap[0][0][7]-> 128 pages
(XEN) heap[0][0][8]-> 512 pages
(XEN) heap[0][0][9]-> 512 pages
(XEN) heap[0][0][10]-> 2048 pages
(XEN) heap[1][0][0]-> 10 pages
(XEN) heap[1][0][1]-> 14 pages
(XEN) heap[1][0][2]-> 20 pages
(XEN) heap[1][0][3]-> 32 pages
(XEN) heap[1][0][4]-> 64 pages
(XEN) heap[1][0][5]-> 128 pages
(XEN) heap[1][0][6]-> 128 pages
(XEN) heap[1][0][7]-> 384 pages
(XEN) heap[1][0][8]-> 256 pages
(XEN) heap[1][0][10]-> 2048 pages
(XEN) heap[1][0][11]-> 6144 pages
(XEN) heap[1][0][12]-> 4096 pages
(XEN) heap[1][0][13]-> 8192 pages
(XEN) heap[1][0][14]-> 16384 pages
(XEN) heap[1][0][16]-> 65536 pages
(XEN) heap[1][0][18]-> 262144 pages
(XEN) heap[1][0][20]-> 7340032 pages
(XEN) heap[2][0][1]-> 2 pages
(XEN) heap[2][0][2]-> 4 pages
(XEN) heap[2][0][3]-> 8 pages
(XEN) heap[2][0][4]-> 16 pages
(XEN) heap[2][0][5]-> 32 pages
(XEN) heap[2][0][6]-> 64 pages
(XEN) heap[2][0][7]-> 128 pages
(XEN) heap[2][0][10]-> 1024 pages
(XEN) heap[2][0][11]-> 2048 pages
(XEN) heap[2][0][12]-> 4096 pages
(XEN) heap[2][0][14]-> 16384 pages
(XEN) heap[2][0][17]-> 131072 pages
(XEN) heap[2][0][18]-> 262144 pages
(XEN) ''u'' pressed -> dumping numa info (now-0xBD:CE47F688)
(XEN) idx0 -> NODE0 start->0 size->8421376
(XEN) phys_to_nid(1000) -> 0 should be 0
(XEN) CPU0 -> NODE0
(XEN) CPU1 -> NODE0
(XEN) CPU2 -> NODE0
(XEN) CPU3 -> NODE0
[-- MARK -- Sun Aug 20 13:00:00 2006]
[-- MARK -- Sun Aug 20 14:00:00 2006]
(XEN) Scheduler: SMP Credit Scheduler (credit)
(XEN) info:
(XEN) ncpus = 4
(XEN) master = 0
(XEN) credit = 120
(XEN) credit balance = 0
(XEN) weight = 0
(XEN) runq_sort = 41875
(XEN) tick = 10ms
(XEN) tslice = 30ms
(XEN) accounting period = 30ms
(XEN) default-weight = 256
(XEN) idlers: 0xd
(XEN) stats:
(XEN) vcpu_init = 2305
(XEN) vcpu_sleep = 11059
(XEN) vcpu_wake_running = 921
(XEN) vcpu_wake_onrunq = 21
(XEN) vcpu_wake_runnable = 975730
(XEN) vcpu_wake_not_runnable = 0
(XEN) dom_destroy = 80
(XEN) schedule = 2389830
(XEN) tickle_local_idler = 782760
(XEN) tickle_local_over = 7500
(XEN) tickle_local_under = 0
(XEN) tickle_local_other = 0
(XEN) acct_run = 41875
(XEN) acct_no_work = 164917
(XEN) acct_balance = 1034
(XEN) acct_reorder = 136
(XEN) acct_min_credit = 235
(XEN) acct_vcpu_active = 14723
(XEN) acct_vcpu_idle = 14723
(XEN) acct_vcpu_credit_min = 0
(XEN) vcpu_migrate = 88489
(XEN) load_balance_idle = 1434334
(XEN) load_balance_over = 16672
(XEN) load_balance_other = 0
(XEN) steal_trylock_failed = 93607
(XEN) steal_peer_down = 0
(XEN) steal_peer_idle = 965974
(XEN) steal_peer_running = 1019
(XEN) steal_peer_pinned = 0
(XEN) tickle_idlers_none = 125230
(XEN) tickle_idlers_some = 61521
(XEN) active vcpus:
(XEN) NOW=0x00000656F0B89381
(XEN) CPU[00] tick=620378, sort=41875
(XEN) run: [32767.0] pri=-64 cpu=0
(XEN) CPU[01] tick=631734, sort=41871
(XEN) run: [0.1] pri=-1 cpu=1 credit=0 (50+30) {a=1489 i=1489 w=256}
(XEN) 1: [32767.1] pri=-64 cpu=1
(XEN) CPU[02] tick=630062, sort=41875
(XEN) run: [32767.2] pri=-64 cpu=2
(XEN) CPU[03] tick=633571, sort=41875
(XEN) run: [32767.3] pri=-64 cpu=3
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
Ryan Harper
2006-Aug-21 22:04 UTC
[Xen-devel] Re: NUMA patches and ES7000, partial success..
* Subrahmanian, Raj <raj.subrahmanian@unisys.com> [2006-08-21 00:39]:> Ryan, > Applied the patch. Well, I was able to bring up and down machines. Our > tests ran ok. In that respect, it''s good news. > But when I tried to do xm top, there was segfault and a message "failed > to receive statistics from libxenstat. > I am sending a hg diff to you off-list so you can see if I patched it > right.Glad to hear the tests went well. The segfault that you saw was what I saw on my machine as well prior to my updating the patch. The patch you applied (diff.gz) now works for me. My guess is that /usr/sbin/xentop is stale. I''d try manually copying xentop from the build dir to /usr/sbin/xentop. Also you can try executing xentop from the tree to make sure it works: tools/xenstat/xentop/xentop. -- Ryan Harper Software Engineer; Linux Technology Center IBM Corp., Austin, Tx (512) 838-9253 T/L: 678-9253 ryanh@us.ibm.com _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel