aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/Documentation/perf-list.txt
diff options
context:
space:
mode:
authorDavid S. Miller <davem@davemloft.net>2010-05-19 02:01:55 -0400
committerDavid S. Miller <davem@davemloft.net>2010-05-19 02:01:55 -0400
commit2ec8c6bb5d8f3a62a79f463525054bae1e3d4487 (patch)
treefa7f8400ac685fb52e96f64997c7c682fc2aa021 /tools/perf/Documentation/perf-list.txt
parent7b39f90fabcf9e2af0cd79d0a60440d821e22b56 (diff)
parent537b60d17894b7c19a6060feae40299d7109d6e7 (diff)
Merge branch 'master' of /home/davem/src/GIT/linux-2.6/
Conflicts: include/linux/mod_devicetable.h scripts/mod/file2alias.c
Diffstat (limited to 'tools/perf/Documentation/perf-list.txt')
-rw-r--r--tools/perf/Documentation/perf-list.txt33
1 files changed, 32 insertions, 1 deletions
diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt
index 8290b9422668..43e3dd284b90 100644
--- a/tools/perf/Documentation/perf-list.txt
+++ b/tools/perf/Documentation/perf-list.txt
@@ -15,6 +15,35 @@ DESCRIPTION
15This command displays the symbolic event types which can be selected in the 15This command displays the symbolic event types which can be selected in the
16various perf commands with the -e option. 16various perf commands with the -e option.
17 17
18RAW HARDWARE EVENT DESCRIPTOR
19-----------------------------
20Even when an event is not available in a symbolic form within perf right now,
21it can be encoded in a per processor specific way.
22
23For instance For x86 CPUs NNN represents the raw register encoding with the
24layout of IA32_PERFEVTSELx MSRs (see [Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide] Figure 30-1 Layout
25of IA32_PERFEVTSELx MSRs) or AMD's PerfEvtSeln (see [AMD64 Architecture Programmer’s Manual Volume 2: System Programming], Page 344,
26Figure 13-7 Performance Event-Select Register (PerfEvtSeln)).
27
28Example:
29
30If the Intel docs for a QM720 Core i7 describe an event as:
31
32 Event Umask Event Mask
33 Num. Value Mnemonic Description Comment
34
35 A8H 01H LSD.UOPS Counts the number of micro-ops Use cmask=1 and
36 delivered by loop stream detector invert to count
37 cycles
38
39raw encoding of 0x1A8 can be used:
40
41 perf stat -e r1a8 -a sleep 1
42 perf record -e r1a8 ...
43
44You should refer to the processor specific documentation for getting these
45details. Some of them are referenced in the SEE ALSO section below.
46
18OPTIONS 47OPTIONS
19------- 48-------
20None 49None
@@ -22,4 +51,6 @@ None
22SEE ALSO 51SEE ALSO
23-------- 52--------
24linkperf:perf-stat[1], linkperf:perf-top[1], 53linkperf:perf-stat[1], linkperf:perf-top[1],
25linkperf:perf-record[1] 54linkperf:perf-record[1],
55http://www.intel.com/Assets/PDF/manual/253669.pdf[Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide],
56http://support.amd.com/us/Processor_TechDocs/24593.pdf[AMD64 Architecture Programmer’s Manual Volume 2: System Programming]