No problem.
Always related to using pfmon on QS22. What can be that prevents pfmon from
sampling?
To be precise, when I try sampling, like:
> pfmon --with-header --long-smpl-periods=10000 -e MFC_TLB_MISS -- ./noploop 10
I get
noploop for 10 seconds
#
# date: Thu Aug 26 15:20:30 2010
#
# hostname: c1b1
#
# kernel version: Linux 2.6.25.14-108.20090129bsc.ppc64 #1 SMP Thu Jan 29
12:38:56 CET 2009
#
# pfmon version: 3.9
# kernel perfmon version: 2.8
#
#
# page size: 65536 bytes
#
# CLK_TCK: 1000000000 ticks/second
#
# CPU configured: 4
# CPU online : 4
#
# physical memory : 16858808320 bytes (16077.8 MB)
# physical memory available: 20643840 bytes (19.7 MB)
#
# CPU name: Cell Broadband Engine.
#
#
# information for set0:
# PMD0: MFC_TLB_MISS = user
#
#
# pfmon command line: ../bin/pfmon --with-header --long-smpl-periods=10000 -e
MFC_TLB_MISS -- ./noploop 10
#
# monitoring mode : per-thread
# command line : /export/home/danieles/tmp/pfmon-3.9/tests/noploop
#
# process id(tgid) : 24956
# thread id(pid) : 24956
# parent process id: 24951
# exec count : 1
#
#
#
# sampling information for set0:
# recorded PMDs when MFC_TLB_MISS overflows:
#
# short sampling rates (base/mask/seed):
# MFC_TLB_MISS 10000
#
# long sampling rates (base/mask/seed):
# MFC_TLB_MISS 10000
#
#
# description of columns:
# column 0: number of samples for event 0 (sorting key)
# column 1: relative percentage of samples for event 0
# column 2: cumulative percentage for event 0
# other columns are self-explanatory
#
#
# results for [24956:24956<-[24951]]
(/export/home/danieles/tmp/pfmon-3.9/tests/noploop)
# total samples : 0
# total buffer overflows : 0
#
# event00
# counts %self %cum code addr
same if I run changing event:
> pfmon --with-header --long-smpl-periods=10000 -- ./noploop 10
I get
noploop for 10 seconds
#
# date: Thu Aug 26 15:43:19 2010
#
# hostname: c1b1
#
# kernel version: Linux 2.6.25.14-108.20090129bsc.ppc64 #1 SMP Thu Jan 29
12:38:56 CET 2009
#
# pfmon version: 3.9
# kernel perfmon version: 2.8
#
#
# page size: 65536 bytes
#
# CLK_TCK: 1000000000 ticks/second
#
# CPU configured: 4
# CPU online : 4
#
# physical memory : 16858808320 bytes (16077.8 MB)
# physical memory available: 21692416 bytes (20.7 MB)
#
# CPU name: Cell Broadband Engine.
#
#
# information for set0:
# PMD0: CYCLES = user
#
#
# pfmon command line: ../bin/pfmon --with-header --long-smpl-periods=10000 --
./noploop 10
#
# monitoring mode : per-thread
# command line : /export/home/danieles/tmp/pfmon-3.9/tests/noploop
#
# process id(tgid) : 24963
# thread id(pid) : 24963
# parent process id: 24958
# exec count : 1
#
#
#
# sampling information for set0:
# recorded PMDs when CYCLES overflows:
#
# short sampling rates (base/mask/seed):
# CYCLES 10000
#
# long sampling rates (base/mask/seed):
# CYCLES 10000
#
#
# description of columns:
# column 0: number of samples for event 0 (sorting key)
# column 1: relative percentage of samples for event 0
# column 2: cumulative percentage for event 0
# other columns are self-explanatory
#
#
# results for [24963:24963<-[24958]]
(/export/home/danieles/tmp/pfmon-3.9/tests/noploop)
# total samples : 0
# total buffer overflows : 0
#
# event00
# counts %self %cum code addr
And the same empty table I get trying with many other combinations of:
privilege levels, programs to monitor, sampling periods, events, etc..
Any help would be really appreciated.
Best regards,
Daniele
On Aug 26, 2010, at 3:33 PM, stephane eranian wrote:
> Hi,
>
> Thanks for reporting the problem and giving the solution.
> I have updated the pfmon code in CVS now.
>
>
> On Thu, Aug 26, 2010 at 2:01 PM, Daniele G. Spampinato
> <daniele.spampin...@inf.ethz.ch> wrote:
>> Hi,
>> I wanted to try pfmon 3.9 on a IBM QS22 blade.
>> In order to compile it properly I had to add/modify the following + marked
>> lines in
>> pfmon/pfmon_utils_cell.c: (function pfmon_get_dlopen_hook)
>> add: unsigned int version;
>> change line 205:
>> from vbprintf("[%d] dlopen hook on %s\n", sdesc->pid,
>> sdesc->new_cmdline);
>> to vbprintf("[%d] dlopen hook on %s\n", sdesc->pid, sdesc->cmd);
>> change line 210:
>> from ret = find_sym_addr("_r_debug", options.primary_syms,
>> PFMON_DATA_SYMBOL, &start, NULL);
>> to ret = find_sym_addr("_r_debug", version, sdesc->syms, &start,
>> NULL);
>>
>>
>> Best,
>> Daniele Spampinato
>> ------------------------------------------------------------------------------
>> Sell apps to millions through the Intel(R) Atom(Tm) Developer Program
>> Be part of this innovative community and reach millions of netbook users
>> worldwide. Take advantage of special opportunities to increase revenue and
>> speed time-to-market. Join now, and jumpstart your future.
>> http://p.sf.net/sfu/intel-atom-d2d
>> _______________________________________________
>> perfmon2-devel mailing list
>> perfmon2-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
>>
>>
------------------------------------------------------------------------------
Sell apps to millions through the Intel(R) Atom(Tm) Developer Program
Be part of this innovative community and reach millions of netbook users
worldwide. Take advantage of special opportunities to increase revenue and
speed time-to-market. Join now, and jumpstart your future.
http://p.sf.net/sfu/intel-atom-d2d
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
