aboutsummaryrefslogtreecommitdiff
path: root/tools/perf/util
AgeCommit message (Collapse)AuthorFilesLines
2011-05-22perf tools: Propagate event parse error handlingFrederic Weisbecker2-8/+20
Better handle event parsing error by propagating the details in upper layers or by dumping some failure message. So that the user knows he has some crazy events in the batch. Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]>
2011-05-22perf tools: Robustify dynamic sample content fetchFrederic Weisbecker1-0/+26
Ensure the size of the dynamic fields such as callchains or raw events don't overlap the whole event boundaries. This prevents from dereferencing junk if the given size of the callchain goes too eager. Reported-by: Linus Torvalds <[email protected]> Reported-by: Ingo Molnar <[email protected]> Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]>
2011-05-22perf tools: Pre-check sample size before parsingFrederic Weisbecker6-4/+38
Check that the total size of the sample fields having a fixed size do not exceed the one of the whole event. This robustifies the sample parsing. Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]>
2011-05-22perf tools: Move evlist sample helpers to evlist areaFrederic Weisbecker4-33/+34
These APIs should belong to evlist.c as they may not be exclusively tied to the headers. Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]
2011-05-22perf tools: Remove junk code in mmap size handlingFrederic Weisbecker1-3/+0
size is overriden later and used only then. Those lines are only junk, probably a leftover. Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]>
2011-05-22perf tools: Check we are able to read the event size on mmapFrederic Weisbecker1-0/+7
Check we have enough mmaped space to read the current event size from its headers, otherwise we may dereference some hell there. Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Stephane Eranian <[email protected]>
2011-05-20sanitize <linux/prefetch.h> usageLinus Torvalds1-1/+1
Commit e66eed651fd1 ("list: remove prefetching from regular list iterators") removed the include of prefetch.h from list.h, which uncovered several cases that had apparently relied on that rather obscure header file dependency. So this fixes things up a bit, using grep -L linux/prefetch.h $(git grep -l '[^a-z_]prefetchw*(' -- '*.[ch]') grep -L 'prefetchw*(' $(git grep -l 'linux/prefetch.h' -- '*.[ch]') to guide us in finding files that either need <linux/prefetch.h> inclusion, or have it despite not needing it. There are more of them around (mostly network drivers), but this gets many core ones. Reported-by: Stephen Rothwell <[email protected]> Signed-off-by: Linus Torvalds <[email protected]>
2011-05-19Merge branch 'core-rcu-for-linus' of ↵Linus Torvalds1-1/+0
git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip * 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (78 commits) Revert "rcu: Decrease memory-barrier usage based on semi-formal proof" net,rcu: convert call_rcu(prl_entry_destroy_rcu) to kfree batman,rcu: convert call_rcu(softif_neigh_free_rcu) to kfree_rcu batman,rcu: convert call_rcu(neigh_node_free_rcu) to kfree() batman,rcu: convert call_rcu(gw_node_free_rcu) to kfree_rcu net,rcu: convert call_rcu(kfree_tid_tx) to kfree_rcu() net,rcu: convert call_rcu(xt_osf_finger_free_rcu) to kfree_rcu() net/mac80211,rcu: convert call_rcu(work_free_rcu) to kfree_rcu() net,rcu: convert call_rcu(wq_free_rcu) to kfree_rcu() net,rcu: convert call_rcu(phonet_device_rcu_free) to kfree_rcu() perf,rcu: convert call_rcu(swevent_hlist_release_rcu) to kfree_rcu() perf,rcu: convert call_rcu(free_ctx) to kfree_rcu() net,rcu: convert call_rcu(__nf_ct_ext_free_rcu) to kfree_rcu() net,rcu: convert call_rcu(net_generic_release) to kfree_rcu() net,rcu: convert call_rcu(netlbl_unlhsh_free_addr6) to kfree_rcu() net,rcu: convert call_rcu(netlbl_unlhsh_free_addr4) to kfree_rcu() security,rcu: convert call_rcu(sel_netif_free) to kfree_rcu() net,rcu: convert call_rcu(xps_dev_maps_release) to kfree_rcu() net,rcu: convert call_rcu(xps_map_release) to kfree_rcu() net,rcu: convert call_rcu(rps_map_release) to kfree_rcu() ...
2011-05-19Merge branch 'perf-core-for-linus' of ↵Linus Torvalds9-397/+534
git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip * 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (107 commits) perf stat: Add more cache-miss percentage printouts perf stat: Add -d -d and -d -d -d options to show more CPU events ftrace/kbuild: Add recordmcount files to force full build ftrace: Add self-tests for multiple function trace users ftrace: Modify ftrace_set_filter/notrace to take ops ftrace: Allow dynamically allocated function tracers ftrace: Implement separate user function filtering ftrace: Free hash with call_rcu_sched() ftrace: Have global_ops store the functions that are to be traced ftrace: Add ops parameter to ftrace_startup/shutdown functions ftrace: Add enabled_functions file ftrace: Use counters to enable functions to trace ftrace: Separate hash allocation and assignment ftrace: Create a global_ops to hold the filter and notrace hashes ftrace: Use hash instead for FTRACE_FL_FILTER ftrace: Replace FTRACE_FL_NOTRACE flag with a hash of ignored functions perf bench, x86: Add alternatives-asm.h wrapper x86, 64-bit: Fix copy_[to/from]_user() checks for the userspace address limit x86, mem: memset_64.S: Optimize memset by enhanced REP MOVSB/STOSB x86, mem: memmove_64.S: Optimize memmove by enhanced REP MOVSB/STOSB ...
2011-05-18perf bench, x86: Add alternatives-asm.h wrapperIngo Molnar1-0/+8
perf bench needs this to build the kernel's memcpy routine: In file included from bench/mem-memcpy-x86-64-asm.S:2:0: bench/../../../arch/x86/lib/memcpy_64.S:7:33: fatal error: asm/alternative-asm.h: No such file or directory Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Steven Rostedt <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-05-17perf: Fix multi-event parsing bugStephane Eranian1-0/+3
This patch fixes an issue with event parsing. The following commit appears to have broken the ability to specify a comma separated list of events: commit ceb53fbf6dbb1df26d38379a262c6981fe73dd36 Author: Ingo Molnar <[email protected]> Date: Wed Apr 27 04:06:33 2011 +0200 perf stat: Fail more clearly when an invalid modifier is specified This patch fixes this while preserving the desired effect: $ perf stat -e instructions:u,instructions:k ls /dev/null /dev/null Performance counter stats for 'ls /dev/null': 365956 instructions:u # 0.00 insns per cycle 731806 instructions:k # 0.00 insns per cycle 0.001108862 seconds time elapsed $ perf stat -e task-clock-msecs true invalid event modifier: '-msecs' Run 'perf list' for a list of valid events and modifiers Signed-off-by: Stephane Eranian <[email protected]> Cc: [email protected] Cc: [email protected] Cc: [email protected] Link: http://lkml.kernel.org/r/20110517133619.GA6999@quad Signed-off-by: Ingo Molnar <[email protected]>
2011-05-15perf evlist: Fix per thread mmap setupArnaldo Carvalho de Melo3-47/+109
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using --pid when monitoring multithreaded apps, as we can only share a ring buffer for events on the same thread if not doing per cpu. Fix it by using per thread ring buffers. Tested with: [root@felicio ~]# tuna -t 26131 -CP | nl 1 thread ctxt_switches 2 pid SCHED_ rtpri affinity voluntary nonvoluntary cmd 3 26131 OTHER 0 0,1 10814276 2397830 chromium-browse 4 642 OTHER 0 0,1 14688 0 chromium-browse 5 26148 OTHER 0 0,1 713602 115479 chromium-browse 6 26149 OTHER 0 0,1 801958 2262 chromium-browse 7 26150 OTHER 0 0,1 1271128 248 chromium-browse 8 26151 OTHER 0 0,1 3 0 chromium-browse 9 27049 OTHER 0 0,1 36796 9 chromium-browse 10 618 OTHER 0 0,1 14711 0 chromium-browse 11 661 OTHER 0 0,1 14593 0 chromium-browse 12 29048 OTHER 0 0,1 28125 0 chromium-browse 13 26143 OTHER 0 0,1 2202789 781 chromium-browse [root@felicio ~]# So 11 threads under pid 26131, then: [root@felicio ~]# perf record -F 50000 --pid 26131 [root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl 1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] [root@felicio ~]# 11 mmaps, one per thread since we didn't specify any CPU list, so we need one mmap per thread and: [root@felicio ~]# perf record -F 50000 --pid 26131 ^M ^C[ perf record: Woken up 79 times to write data ] [ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ] [root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl 1 371310 26131 2 96516 26148 3 95694 26149 4 95203 26150 5 7291 26143 6 87 27049 7 76 661 8 60 29048 9 47 618 10 43 642 [root@felicio ~]# Ok, one of the threads, 26151 was quiescent, so no samples there, but all the others are there. Then, if I specify one CPU: [root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1 ^C[ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ] [root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl 1 8444 26131 2 2584 26149 3 2518 26148 4 2324 26150 5 123 26143 6 9 661 7 9 29048 [root@felicio ~]# This machine has two cores, so fewer threads appeared on the radar, and: [root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] [root@felicio ~]# Just one mmap, as now we can use just one per-cpu buffer instead of the per-thread needed in the previous case. For global profiling: [root@felicio ~]# perf record -F 50000 -a ^C[ perf record: Woken up 26 times to write data ] [ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ] [root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl 1 7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] 2 7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] [root@felicio ~]# It uses per-cpu buffers. For just one thread: [root@felicio ~]# perf record -F 50000 --tid 26148 ^C[ perf record: Woken up 2 times to write data ] [ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ] [root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl 1 9969 26148 [root@felicio ~]# [root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl 1 7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event] [root@felicio ~]# Tested-by: David Ahern <[email protected]> Tested-by: Lin Ming <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-05-15perf tools: Honour the cpu list parameter when also monitoring a thread listArnaldo Carvalho de Melo1-1/+1
The perf_evlist__create_maps was discarding the --cpu parameter when a --pid or --tid was specified, fix that. Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-05-10perf probe: Fix the missed parameter initializationLin Ming1-0/+1
pubname_callback_param::found should be initialized to 0 in fastpath lookup, the structure is on the stack and uninitialized otherwise. Signed-off-by: Lin Ming <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-05-10perf: Only include annotate.h once in tools/perf/util/ui/browsers/annotate.cJesper Juhl1-1/+0
Including "../../annotate.h" once in tools/perf/util/ui/browsers/annotate.c is enough. No need to do it twice. Signed-off-by: Jesper Juhl <[email protected]> Signed-off-by: Jiri Kosina <[email protected]>
2011-05-05rcu: move TREE_RCU from softirq to kthreadPaul E. McKenney1-1/+0
If RCU priority boosting is to be meaningful, callback invocation must be boosted in addition to preempted RCU readers. Otherwise, in presence of CPU real-time threads, the grace period ends, but the callbacks don't get invoked. If the callbacks don't get invoked, the associated memory doesn't get freed, so the system is still subject to OOM. But it is not reasonable to priority-boost RCU_SOFTIRQ, so this commit moves the callback invocations to a kthread, which can be boosted easily. Also add comments and properly synchronized all accesses to rcu_cpu_kthread_task, as suggested by Lai Jiangshan. Signed-off-by: Paul E. McKenney <[email protected]> Signed-off-by: Paul E. McKenney <[email protected]> Reviewed-by: Josh Triplett <[email protected]>
2011-04-29perf list: Fix max event string sizeIngo Molnar1-12/+13
Recent stalled-cycles event names were larger than the 40 chars printout used by perf list. Extend that, make it robust for future extensions and also adjust alignments in face of wider event names. Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-29perf stat: Analyze front-end and back-end stall countsIngo Molnar1-3/+4
Sample output: Performance counter stats for './loop_1b': 873.691065 task-clock # 1.000 CPUs utilized 1 context-switches # 0.000 M/sec 1 CPU-migrations # 0.000 M/sec 96 page-faults # 0.000 M/sec 2,012,637,222 cycles # 2.304 GHz (66.58%) 1,001,397,911 stalled-cycles-frontend # 49.76% frontend cycles idle (66.58%) 7,523,398 stalled-cycles-backend # 0.37% backend cycles idle (66.76%) 2,004,551,046 instructions # 1.00 insns per cycle # 0.50 stalled cycles per insn (66.80%) 1,001,304,992 branches # 1146.063 M/sec (66.76%) 39,453 branch-misses # 0.00% of all branches (66.64%) 0.874046121 seconds time elapsed Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-29perf tools: Add front-end and back-end stalled cycles supportIngo Molnar2-19/+22
Update perf tooling to deal with front-end and back-end stalled cycles events. Add both the default 'perf stat' output. Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-26perf stat: Add stalled cycles to the default outputIngo Molnar1-5/+6
The new default output looks like this: Performance counter stats for './loop_1b_instructions': 236.010686 task-clock # 0.996 CPUs utilized 0 context-switches # 0.000 M/sec 0 CPU-migrations # 0.000 M/sec 99 page-faults # 0.000 M/sec 756,487,646 cycles # 3.205 GHz 354,938,996 stalled-cycles # 46.92% of all cycles are idle 1,001,403,797 instructions # 1.32 insns per cycle # 0.35 stalled cycles per insn 100,279,773 branches # 424.895 M/sec 12,646 branch-misses # 0.013 % of all branches 0.236902540 seconds time elapsed We dropped cache-refs and cache-misses and added stalled-cycles - this is a more generic "how well utilized is the CPU" metric. If the stalled-cycles ratio is too high then more specific measurements can be taken to figure out the source of the inefficiency. Acked-by: Peter Zijlstra <[email protected]> Acked-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-26perf stat: Make all displayed event names parseable as wellIngo Molnar1-2/+2
Right now we display this by default: 0.202204 task-clock-msecs # 0.282 CPUs 0 context-switches # 0.000 M/sec 0 CPU-migrations # 0.000 M/sec 85 page-faults # 0.420 M/sec The task-clock-msecs event cannot actually be passed back as an event name, the event name we recognize is 'task-clock'. So change the output of the cpu-clock and task-clock events to be idempotent. ( Units should be printed out in the right-side column, if needed. ) Acked-by: Peter Zijlstra <[email protected]> Acked-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-26perf stat: Fail more clearly when an invalid modifier is specifiedIngo Molnar1-11/+22
Currently we fail without printing any error message on "perf stat -e task-clock-msecs". The reason is that the task-clock event is matched and the "-msecs" postfix is assumed to be an event modifier - but is not recognized. This patch changes the code to be more informative: $ perf stat -e task-clock-msecs true invalid event modifier: '-msecs' Run 'perf list' for a list of valid events and modifiers And restructures the return value of parse_event_modifier() to allow the printing of all variants of invalid event modifiers. Acked-by: Peter Zijlstra <[email protected]> Acked-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-26perf tools: Accept case-insensitive symbolic event variantsIngo Molnar1-3/+5
We currently fail on something like '-e CPU-migrations', with: invalid or unsupported event: 'CPU-migrations' While 'CPU-migrations' is how we actually print out the event in the default perf stat output: Performance counter stats for 'true': 0.202204 task-clock-msecs # 0.282 CPUs 0 context-switches # 0.000 M/sec 0 CPU-migrations # 0.000 M/sec So change the matching to be case-insensitive. Acked-by: Peter Zijlstra <[email protected]> Acked-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-26perf events: Add stalled cycles generic event - PERF_COUNT_HW_STALLED_CYCLESIngo Molnar2-0/+2
The new PERF_COUNT_HW_STALLED_CYCLES event tries to approximate cycles the CPU does nothing useful, because it is stalled on a cache-miss or some other condition. Acked-by: Peter Zijlstra <[email protected]> Acked-by: Arnaldo Carvalho de Melo <[email protected]> Cc: Frederic Weisbecker <[email protected]> Link: http://lkml.kernel.org/n/[email protected] Signed-off-by: Ingo Molnar <[email protected]>
2011-04-20perf script: improve validation of sample attributes for output fieldsDavid Ahern2-0/+15
Check for required sample attributes using evsel rather than sample_type in the session header. If the attribute for a default field is not present for the event type (e.g., new command operating on file from older kernel) the field is removed from the output list. Expected event types must exist. For example, if a user specifies -f trace:time,trace -f sw:time,cpu,sym the perf.data file must contain both tracepoints and software events (ie., it is an error if either does not exist in the file). Attribute checking is done once at the beginning of perf-script rather than for each sample. v1 -> v2: - addressed comments from acme Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: David Ahern <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-19perf symbols: Give more useful names to 'self' parametersArnaldo Carvalho de Melo2-346/+361
One more installment on an area that is mostly dormant. Suggested-by: Thomas Gleixner <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: Tom Zanussi <[email protected]> LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-19Merge branch 'perf/urgent' into perf/coreIngo Molnar12-102/+153
Merge reason: we'll be queueing up dependent changes. Signed-off-by: Ingo Molnar <[email protected]>
2011-04-15perf evsel: Fix use of inheritArnaldo Carvalho de Melo4-31/+25
perf stat doesn't mmap and its perfectly fine for it to use task-bound counters with inheritance. So set the attr.inherit on the caller and leave the syscall itself to validate it. When the mmap fails perf_evlist__mmap will just emit a warning if this is the failure reason. Reported-by: Peter Zijlstra <[email protected]> Acked-by: Peter Zijlstra <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-15perf hists browser: Fix seg fault when annotate null symbolLin Ming2-3/+5
In hists browser, press hotkey 'a' to annotate current symbol. Now it causes segment fault if 'a' is pressed on a null symbol. Here are 2 small bugs: - In perf_evsel__hists_browse, the condition check after 'a' is pressed is not correct, we should check ->sym instead of ->map. - In symbol__tui_annotate we must check whether sym is NULL or not before getting annotation structure. This patch fixes above 2 small bugs. Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Lin Ming <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-08perf: Fix a build error with some GCC versionsEric Dumazet1-1/+1
Fix this: util/cgroup.c: In function ‘open_cgroup’: util/cgroup.c:16:16: error: ‘saved_ptr’ may be used uninitialized in this function util/cgroup.c:16:16: note: ‘saved_ptr’ was declared here Apparently newer GCC (4.6) can figure out that this variable is properly initialized - but some versions of GCC (such as 4.5.2) need help. Signed-off-by: Eric Dumazet <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <[email protected]>
2011-04-07Merge branches 'x86-fixes-for-linus', 'sched-fixes-for-linus', ↵Linus Torvalds2-65/+113
'timers-fixes-for-linus', 'irq-fixes-for-linus' and 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip * 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: x86-32, fpu: Fix FPU exception handling on non-SSE systems x86, hibernate: Initialize mmu_cr4_features during boot x86-32, NUMA: Fix ACPI NUMA init broken by recent x86-64 change x86: visws: Fixup irq overhaul fallout * 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: sched: Clean up rebalance_domains() load-balance interval calculation * 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: x86/mrst/vrtc: Fix boot crash in mrst_rtc_init() rtc, x86/mrst/vrtc: Fix boot crash in rtc_read_alarm() * 'irq-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: genirq: Fix cpumask leak in __setup_irq() * 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: perf probe: Fix listing incorrect line number with inline function perf probe: Fix to find recursively inlined function perf probe: Fix multiple --vars options behavior perf probe: Fix to remove redundant close perf probe: Fix to ensure function declared file
2011-04-07Merge branch 'for-linus2' of git://git.profusion.mobi/users/lucas/linux-2.6Linus Torvalds1-1/+1
* 'for-linus2' of git://git.profusion.mobi/users/lucas/linux-2.6: Fix common misspellings
2011-04-05perf probe: Fix listing incorrect line number with inline functionMasami Hiramatsu1-53/+81
Fix a bug showing incorrect line number when a probe is put on the head of an inline function. This patch updates find_perf_probe_point() and introduces new rules to get correct line number. - If debuginfo doesn't have a correct file name, we shouldn't return line number too, because, without file name, line number is meaningless. - If the address is in a function, it stores the function name and the offset from the function entry. - If the address is on a line, it tries to get the relative line number from the function entry line, except for the address is same as the entry address of the function (in this case, the relative line number should be 0). - If the address is in an inline function entry (call-site), it uses the inline function call line number as the line on which the address is. - If the address is in an inline function body, it stores the inline function name and offset from the inline function call site instead of the (non-inlined) function. Cc: [email protected] Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Lin Ming <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-05perf probe: Fix to find recursively inlined functionMasami Hiramatsu1-1/+14
Fix die_find_inlinefunc() to return correct innermost inlined function at given address. Without this fix, it returns the outermost inlined function. Cc: [email protected] Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Lin Ming <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-05perf probe: Fix multiple --vars options behaviorMasami Hiramatsu1-8/+8
Fix a bug that perf-probe fails to initialize libdwfl and shows incorrect error when user gives multiple --vars options. Cc: [email protected] Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Lin Ming <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-05perf probe: Fix to remove redundant closeMasami Hiramatsu2-3/+2
Since dwfl_end() closes given fd with dwfl, caller doesn't need to close its fd when finishing process. Cc: [email protected] Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Lin Ming <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-04-05perf probe: Fix to ensure function declared fileMasami Hiramatsu1-0/+8
Fix to ensure function declared file matches given file name. This fixes a potential bug. As I've commented on Lin Ming's fastpath enhancement, decl_file should be checked on each probe point if user gives a probe point as func@file. Cc: [email protected] Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Lin Ming <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Masami Hiramatsu <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-31Fix common misspellingsLucas De Marchi1-1/+1
Fixes generated by 'codespell' and manually reviewed. Signed-off-by: Lucas De Marchi <[email protected]>
2011-03-30Merge commits 'ca6a42586fae' and 'c286c419c784' into perf/urgentIngo Molnar2-0/+11
Pick up these two commits from Arnaldo's perf/core tree: ca6a42586fae: perf tools: Emit clearer message for sys_perf_event_open ENOENT return c286c419c784: perf tools: Fixup exit path when not able to open events As they are really fixes we want to have sooner than laer. Signed-off-by: Ingo Molnar <[email protected]>
2011-03-29perf tools: Fix NO_NEWT=1 python build errorRobert Richter1-1/+6
Fix the following build error: GEN python/perf.so In file included from util/evsel.h:10, from util/python.c:6: util/hist.h:106:18: error: newt.h: No such file or directory error: command 'x86_64-pc-linux-gnu-gcc' failed with exit status 1 make: *** [python/perf.so] Error 1 by passing BASIC_CFLAGS to setup.py. BASIC_CFLAGS variable contains the -DNO_NEWT_SUPPORT switch which prevents building python c extension with newt. Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Robert Richter <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-29perf symbols: Properly align symbol_conf.priv_sizeDavid S. Miller1-0/+2
If symbol_conf.priv_size is not a multiple of "sizeof(u64)" we'll bus error on sparc64 in symbol__new because the "struct symbol *" pointer is computed by adding symbol_conf.priv_size to the memory allocated. We cannot isolate the fix to symbol__new and symbol__delete since the private area is computed by subtracting the priv_size value from a "struct symbol" pointer, so then the private area can still be potentially unaligned. So, simply align the symbol_conf.priv_size value in symbol__init() Cc: Ingo Molnar <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: David S. Miller <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-29perf probe: Add fastpath to do lookup by function nameLin Ming2-0/+74
v3 -> v2: - Make pubname_search_cb more generic - Add fastpath to find_probes also v2 -> v1: - Don't compare file names with cu_find_realpath(...), instead, compare them with the name returned by dwarf_decl_file(sp_die) The vmlinux file may have thousands of CUs. We can lookup function name from .debug_pubnames section to avoid the slow loop on CUs. 1. Improvement data for find_line_range ./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \ -s /home/mlin/linux-2.6 \ --line csum_partial_copy_to_user > tmp.log before patch applied ===================== 847,988,276 cycles 0.355075856 seconds time elapsed after patch applied ===================== 206,102,622 cycles 0.086883555 seconds time elapsed 2. Improvement data for find_probes ./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \ -s /home/mlin/linux-2.6 \ --vars csum_partial_copy_to_user > tmp.log before patch applied ===================== 848,490,844 cycles 0.355307901 seconds time elapsed after patch applied ===================== 205,684,469 cycles 0.086694010 seconds time elapsed Acked-by: Masami Hiramatsu <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Masami Hiramatsu <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: linux-kernel <[email protected]> LKML-Reference: <[email protected]> Signed-off-by: Lin Ming <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-29perf tools: Fixup exit path when not able to open eventsArnaldo Carvalho de Melo2-0/+11
We have to deal with the TUI mode in perf top, so that we don't end up with a garbled screen when, say, a non root user on a machine with a paranoid setting (the default) tries to use 'perf top'. Introduce a ui__warning_paranoid() routine shared by top and record that tells the user the valid values for /proc/sys/kernel/perf_event_paranoid. Cc: David Ahern <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-28perf symbols: Fix vsyscall symbol lookupAndrew Lutomirski2-1/+4
Perf can't currently trace into the vsyscall page. It looks like it was meant to work. Tested on 2.6.38 and today's -git. The bug is easy to reproduce. Compile this: int main() { int i; struct timespec t; for(i = 0; i < 10000000; i++) clock_gettime(CLOCK_MONOTONIC, &t); return 0; } and run it through perf record; perf report. The top entry shows "[unknown]" and you can't zoom in. It looks like there are two issues. The first is a that a test for user mode executing in kernel space is backwards. (That's the first hunk below). The second (I think) is that something's wrong with the code that generates lots of little struct dso objects for different sections -- when it runs on vmlinux it results in bogus long_name values which cause objdump to fail. Cc: Ingo Molnar <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> LPU-Reference: <[email protected]> Signed-off-by: Andy Lutomirski <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-23perf symbols: Look at .dynsym again if .symtab not foundArnaldo Carvalho de Melo1-12/+13
The original intent of the code was to repeat the search with want_symtab = 0. But as the code stands now, we never hit the "default" case of the switch statement. Which means we never repeat the search. Tested-by: Srikar Dronamraju <[email protected]> Reported-by: Arun Sharma <[email protected]> Reported-by: Srikar Dronamraju <[email protected]> Cc: Dave Martin <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-23perf build-id: Add quirk to deal with perf.data file format breakageArnaldo Carvalho de Melo1-1/+56
The a1645ce1 changeset: "perf: 'perf kvm' tool for monitoring guest performance from host" Added a field to struct build_id_event that broke the file format. Since the kernel build-id is the first entry, process the table using the old format if the well known '[kernel.kallsyms]' string for the kernel build-id has the first 4 characters chopped off (where the pid_t sits). Reported-by: Han Pingtian <[email protected]> Cc: Avi Kivity <[email protected]> Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> Cc: Zhang Yanmin <[email protected]> Cc: [email protected] LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-23perf session: Pass evsel in event_ops->sample()Arnaldo Carvalho de Melo8-4/+34
Resolving the sample->id to an evsel since the most advanced tools, report and annotate, and the others will too when they evolve to properly support multi-event perf.data files. Good also because it does an extra validation, checking that the ID is valid when present. When that is not the case, the overhead is just a branch + function call (perf_evlist__id2evsel). Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Mike Galbraith <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Stephane Eranian <[email protected]> Cc: Tom Zanussi <[email protected]> LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-16perf tools: Version incorrect with some versions of grepJosh Hunt1-4/+4
Some versions of grep don't treat '\s' properly. When building perf on such systems and using a kernel tarball the perf version is unable to be determined from the main kernel Makefile and the user is left with a version of '..'. Replacing the use of '\s' with '[[:space:]]', which should work in all grep versions, gives a usable version number. Reported-by: Tapan Dhimant <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Tapan Dhimant <[email protected]> Cc: [email protected] Cc: [email protected] LKML-Reference: <[email protected]> Signed-off-by: Josh Hunt <[email protected]> Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
2011-03-16Merge branch 'perf/core' of ↵Ingo Molnar14-167/+179
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 into perf/urgent
2011-03-15Merge branch 'perf-core-for-linus' of ↵Linus Torvalds57-1821/+5072
git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip * 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (184 commits) perf probe: Clean up probe_point_lazy_walker() return value tracing: Fix irqoff selftest expanding max buffer tracing: Align 4 byte ints together in struct tracer tracing: Export trace_set_clr_event() tracing: Explain about unstable clock on resume with ring buffer warning ftrace/graph: Trace function entry before updating index ftrace: Add .ref.text as one of the safe areas to trace tracing: Adjust conditional expression latency formatting. tracing: Fix event alignment: skb:kfree_skb tracing: Fix event alignment: mce:mce_record tracing: Fix event alignment: kvm:kvm_hv_hypercall tracing: Fix event alignment: module:module_request tracing: Fix event alignment: ftrace:context_switch and ftrace:wakeup tracing: Remove lock_depth from event entry perf header: Stop using 'self' perf session: Use evlist/evsel for managing perf.data attributes perf top: Don't let events to eat up whole header line perf top: Fix events overflow in top command ring-buffer: Remove unused #include <linux/trace_irq.h> tracing: Add an 'overwrite' trace_option. ...
Powered by cgit, Gruvboxed by Blaster4385.