xenoprof: oprofiled segfaulting upon --dump

[Jan B. <ja...@br...>]

hello everybody,

i got stuck here trying to do some research using xenoprof in an active 
domain profiling scenario with 1 active domain:

dom0 (xen 3.4.2, kernel 2.6.26-1-xen-amd64) on 8 core intel 64 : 
Intel(R) Xeon(R) CPU           E5450  @ 3.00GHz

opcontrol --start-daemon --separate=library --active-domains=1 
--xen=/boot/xen-syms-3.4.2 --event=LLC_MISSES:100000:0x41:0:1 
--event=CPU_CLK_UNHALTED:100000:0:0:1

domU (same kernel):

opcontrol --start-daemon --separate=library 
--event=LLC_MISSES:100000:0x41:0:1 --event=CPU_CLK_UNHALTED:100000:0:0:1

i frequently run opcontrol --dump on the domU during the profiling 
session. for example, i tried the "watch" example from the FAQ.

after a few (seems pretty random) calls to opcontrol --dump, i get a 
segfault in oprofiled

is there any way I can ship around this problem (by using --save instead 
of --dump ?) or otherwise avoid this? anything i can provide to help 
rectify this problem?

the output is:

dump fail: daemon died during last run ?
Unable to complete dump of oprofile data: is the oprofile daemon running?
dump fail: daemon died during last run ?
Signalling daemon... /usr/local/bin/opcontrol: line 1755: kill: (3173) - 
No such process
done

domU dmesg says:

[  374.418932] xenoprofile_init: ret 0, events 2, xenoprof_is_primary 0
[  417.718795] oprofiled[3173]: segfault at fd7015f0 ip 40c171 sp 
7fffb3380170 error 4 in oprofiled[400000+17000]
[  417.748351] ------------[ cut here ]------------
[  417.748360] WARNING: at 
arch/x86/oprofile/../../../drivers/xen/xenoprof/xenoprofile.c:349 
xenoprof_shutdown+0x2f/0x86 [oprofile]()
[  417.748363] Modules linked in: oprofile evdev nfs nfs_acl lockd 
sunrpc ipv6 sata_nv ata_piix libata scsi_mod dock ide_disk ide_generic 
ide_core
[  417.748406] Pid: 4999, comm: opjitconv Not tainted 2.6.26-1-xen-amd64 #1
[  417.748413]
[  417.748413] Call Trace:
[  417.748429]  [<ffffffff8022cba2>] warn_on_slowpath+0x51/0x78
[  417.748439]  [<ffffffff8023c57a>] flush_cpu_workqueue+0x85/0xa3
[  417.748446]  [<ffffffff8026874c>] __free_pages_ok+0x194/0x311
[  417.748455]  [<ffffffffa016740e>] :oprofile:xenoprof_shutdown+0x2f/0x86
[  417.748464]  [<ffffffffa016510e>] :oprofile:oprofile_shutdown+0x35/0x57
[  417.748473]  [<ffffffffa0165f76>] :oprofile:event_buffer_release+0xe/0x3c
[  417.748478]  [<ffffffff8028b1aa>] __fput+0xa1/0x16b
[  417.748483]  [<ffffffff802888c3>] filp_close+0x5d/0x65
[  417.748487]  [<ffffffff8022eaeb>] put_files_struct+0x66/0xc4
[  417.748491]  [<ffffffff8022ff02>] do_exit+0x274/0x6ce
[  417.748496]  [<ffffffff80230402>] do_group_exit+0xa6/0xdc
[  417.748503]  [<ffffffff8023044a>] sys_exit_group+0x12/0x17
[  417.748506]  [<ffffffff8020b528>] system_call+0x68/0x6d
[  417.748512]  [<ffffffff8020b4c0>] system_call+0x0/0x6d
[  417.748519]
[  417.748525] ---[ end trace 20bf62ebc741e4ca ]---


on the dom0 xm dmesg says:

xenoprof: operation 11 failed for dom 1 (status : -1)


thank you very much

yours,

jan