Including Suravee's patch for named default masks, and another patch to make all Intel default unit masks unique. -Andi ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev
I always get /home/ak/pmu/obj/pe_profiling/../../oprofile/pe_profiling/operf.cpp:1185: undefined reference to `op_netburst_get_perf_encoding' -- -- ak <at> linux.intel.com -- Speaking for myself only. ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev
From: Suravee Suthikulpanit <suravee.suthikulpanit <at> amd.com> This patch implements alternative solution for the patch: [PATCH 3/3] Don't do uniqueness checking for default masks http://marc.info/?l=oprofile-list&m=136279276004115&w=2 OVERVIEW ======== In case of Intel events with named-mask, the tool expect user to use named mask (instead of numerical mask) when configuring events. When no mask is specified, the default mask is used instead. The issue arises when the event has duplicated numerical mask. If user does not specify named mask, and fall back to using default mask which is duplicated, the logic will gives error. For example, for the following unit mask, ophelp gives error. name:uops_retired type:exclusive default:0x1 0x1 extra: all Actually retired uops. 0x2 extra: retire_slots Retirement slots used. 0x1 extra:pebs all_ps Actually retired uops. (Precise Event - PEBS) 0x2 extra:pebs retire_slots_ps Retirement slots used. (Precise Event - PEBS) 0x1 extra:cmask=1,inv stall_cycles Cycles without actually retired uops. 0x1 extra:cmask=10,inv total_cycles Cycles with less than 10 actually retired uops. 0x1 extra:cmask=1,inv,any core_stall_cycles Cycles without actually retired uops. # ophelp --check-events -c i386/haswell -e uops_retired:10000000 Unit mask (0x1) is non unique.(Continue reading)
Fix 'Invalid argument' error with 'opcontrol --start' in Timer mode When a processor type does not support oprofile event-based profiling, the oprofile kernel driver writes "timer" to the /dev/oprofile/cpu_type file. Usually (perhaps "always"), the architecture-specific oprofile kernel driver does not set up the standard oprofile function pointers that are defined in 'struct oprofile_operations oprofile_ops'. One of those function pointers is 'oprofile_ops.backtrace'. When opcontrol "--start" option is being executed, one of the actions taken is to write the CALLGRAPH value from the daemonrc file to /dev/oprofile/backtrace_depth. The kernel driver's filesystem function defined to respond to writes to this file is drivers/oprofile/oprofile_files.c:depth_write(). In that function, if 'oprofile_ops.backtrace' is not set, it returns -EINVAL, resulting in the following opcontrol error: opcontrol: line 1172: echo: write error: Invalid argument This patch detects when the system is in timer mode and handles this error appropriately -- if simply writing '0' to backtrace_depth, the error is ignored; otherwise, print a message that call graph is not supported on this system in TIMER mode. Signed-off-by: Maynard Johnson <maynardj <at> us.ibm.com> --- utils/opcontrol | 13 ++++++++++++- 1 files changed, 12 insertions(+), 1 deletions(-) diff --git a/utils/opcontrol b/utils/opcontrol index 038e0db..20c6c6a 100644 --- a/utils/opcontrol +++ b/utils/opcontrol(Continue reading)
On oprof_start startup "unit mask" sub-window sometimes overlaps "event list" one ( confusing appearance ). This problem occurs if oprof_start is built with Qt4 / Qt3Support. This patch fixes this. Signed-off-by: Gilles Allard <gilles.allard21 <at> orange.fr> --- gui/oprof_start.cpp | 2 -- 1 file changed, 2 deletions(-) diff --git a/gui/oprof_start.cpp b/gui/oprof_start.cpp index a4aaef5..78df87e 100644 --- a/gui/oprof_start.cpp +++ b/gui/oprof_start.cpp <at> <at> -834,8 +834,6 <at> <at> void oprof_start::setup_unit_masks(op_event_descr const & descr) check->show(); } - unit_mask_group->setMinimumSize(unit_mask_group->sizeHint()); - setup_config_tab->setMinimumSize(setup_config_tab->sizeHint()); } -- -- 1.8.1.5 ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows:(Continue reading)
The "kernel profile" & "user binaries profile" buttons of GUI are not correctly
set from daemonrc file. The 2 values are reversed & opcontrol handles a
wrong value of "--event" argument.
This patch fixes that bug.
Added: a few code optimization.
Signed-off-by: Gilles Allard <gilles.allard21 <at> orange.fr>
---
gui/oprof_start.cpp | 51 ++++++++++++++++++++++++++-------------------------
1 file changed, 26 insertions(+), 25 deletions(-)
diff --git a/gui/oprof_start.cpp b/gui/oprof_start.cpp
index 914e356..a4aaef5 100644
--- a/gui/oprof_start.cpp
+++ b/gui/oprof_start.cpp
<at> <at> -219,6 +219,7 <at> <at> void oprof_start::fill_events()
// from its name.
struct list_head * pos;
struct list_head * events = op_events(cpu_type);
+ event_setting *cfg;
list_for_each(pos, events) {
struct op_event * event = list_entry(pos, struct op_event,
event_next);
<at> <at> -255,12 +256,13 <at> <at> void oprof_start::fill_events()
count = descr.min_count * 100;
}
(Continue reading)
From: Suravee Suthikulpanit <suravee.suthikulpanit <at> amd.com> This patch implements alternative solution for the patch: [PATCH 3/3] Don't do uniqueness checking for default masks http://marc.info/?l=oprofile-list&m=136279276004115&w=2 OVERVIEW ======== In case of Intel events with named-mask, the tool expect user to use named mask (instead of numerical mask) when configuring events. When no mask is specified, the default mask is used instead. The issue arises when the event has duplicated numerical mask. If user does not specify named mask, and fall back to using default mask which is duplicated, the logic will gives error. For example, for the following unit mask, ophelp gives error. name:uops_retired type:exclusive default:0x1 0x1 extra: all Actually retired uops. 0x2 extra: retire_slots Retirement slots used. 0x1 extra:pebs all_ps Actually retired uops. (Precise Event - PEBS) 0x2 extra:pebs retire_slots_ps Retirement slots used. (Precise Event - PEBS) 0x1 extra:cmask=1,inv stall_cycles Cycles without actually retired uops. 0x1 extra:cmask=10,inv total_cycles Cycles with less than 10 actually retired uops. 0x1 extra:cmask=1,inv,any core_stall_cycles Cycles without actually retired uops. # ophelp --check-events -c i386/haswell -e uops_retired:10000000 Unit mask (0x1) is non unique.(Continue reading)
Will Deacon has generously volunteered his ARM expertise to serve officially in the role of ARM architecture maintainer for oprofile. The "Current maintainers" list (see http://oprofile.sourceforge.net/credits/) has been updated to reflect this. Thanks, Will! -Maynard ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev
*** To all oprofile users *** SourceForge has done some project reorganizations/upgrades, resulting in a change to our Git repo URL. I have updated oprofile's "Download and Git" page (http://oprofile.sourceforge.net/download/) to reflect the new URL (used for read-only checkouts). Please see https://sourceforge.net/p/forge/community-docs/Repository%20Upgrade%20FAQ/ for information about migrating your current local repo to point to the proper url; otherwise, just do a fresh checkout using the new URL. Thank you. -Maynard ------------------------------------------------------------------------------ How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j
From: Suravee Suthikulpanit <suravee.suthikulpanit <at> amd.com> This patch implements alternative solution for the patch: [PATCH 3/3] Don't do uniqueness checking for default masks http://marc.info/?l=oprofile-list&m=136279276004115&w=2 OVERVIEW ======== In case of Intel events with named-mask, the tool expect user to use named mask (instead of numerical mask) when configuring events. When no mask is specified, the default mask is used instead. The issue arises when the event has duplicated numerical mask. If user does not specify named mask, and fall back to using default mask which is duplicated, the logic will gives error. For example, for the following unit mask, ophelp gives error. name:uops_retired type:exclusive default:0x1 0x1 extra: all Actually retired uops. 0x2 extra: retire_slots Retirement slots used. 0x1 extra:pebs all_ps Actually retired uops. (Precise Event - PEBS) 0x2 extra:pebs retire_slots_ps Retirement slots used. (Precise Event - PEBS) 0x1 extra:cmask=1,inv stall_cycles Cycles without actually retired uops. 0x1 extra:cmask=10,inv total_cycles Cycles with less than 10 actually retired uops. 0x1 extra:cmask=1,inv,any core_stall_cycles Cycles without actually retired uops. # ophelp --check-events -c i386/haswell -e uops_retired:10000000 Unit mask (0x1) is non unique.(Continue reading)
The GUI component of oprofile needs a lot of tender loving care if it is to survive. The oprofile project used to have a sub-maintainer (Richard Purdie) assigned to handling GUI-related patches and, generally, ensuring the GUI worked properly. Richard is apparently no longer involved with the project -- recent messages sent to his email address have bounced. I have not taken much personal interest in this part of oprofile myself, but some recent patches sent by Gilles (on cc) have led me to test drive it. I find the GUI to be antiquated, hard to use, and buggy (I won't even try to list all the problems here). The GUI is supposed to support qt3 and qt4 (via the qt3Support in the qt4 package). Perhaps this attempt at supporting multiple qt versions is one of the reasons for its problems, but it's certainly not the only issue. I am asking for someone in the community who has an interest in the oprofile GUI to step forward and volunteer to be its maintainer. If no one volunteers, we may have no choice but to deprecate the GUI component. Thanks for your consideration. -Maynard ------------------------------------------------------------------------------ How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j
RSS Feed51 | |
|---|---|
75 | |
128 | |
65 | |
63 | |
94 | |
89 | |
42 | |
45 | |
58 | |
72 | |
151 | |
89 | |
103 | |
36 | |
39 | |
62 | |
105 | |
95 | |
88 | |
115 | |
58 | |
54 | |
55 | |
82 | |
143 | |
121 | |
121 | |
54 | |
63 | |
57 | |
50 | |
84 | |
83 | |
69 | |
35 | |
65 | |
85 | |
50 | |
88 | |
59 | |
71 | |
79 | |
88 | |
147 | |
80 | |
75 | |
155 | |
153 | |
203 | |
197 | |
157 | |
85 | |
131 | |
113 | |
114 | |
162 | |
88 | |
89 | |
206 | |
99 | |
114 | |
125 | |
48 | |
67 | |
104 | |
52 | |
111 | |
102 | |
40 | |
54 | |
184 | |
105 | |
145 | |
124 | |
43 | |
119 | |
63 | |
105 | |
101 | |
87 | |
99 | |
44 | |
36 | |
60 | |
66 | |
95 | |
65 | |
117 | |
102 | |
60 | |
77 | |
55 | |
16 | |
87 | |
35 | |
35 | |
69 | |
151 | |
120 | |
42 | |
25 | |
33 | |
114 | |
71 | |
87 | |
89 | |
32 | |
34 | |
92 | |
72 | |
58 | |
93 | |
99 | |
49 | |
99 | |
102 | |
41 | |
95 | |
86 | |
58 | |
83 | |
103 | |
187 | |
131 |
