From d2d4edbebe07ddb77980656abe7b9bc7a9e0cdf7 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 11 Jan 2017 14:59:38 +0900 Subject: [PATCH 1/4] perf probe: Fix to show correct locations for events on modules Fix to show correct locations for events on modules by relocating given address instead of retrying after failure. This happens when the module text size is big enough, bigger than sh_addr, because the original code retries with given address + sh_addr if it failed to find CU DIE at the given address. Any address smaller than sh_addr always fails and it retries with the correct address, but addresses bigger than sh_addr will get a CU DIE which is on the given address (not adjusted by sh_addr). In my environment(x86-64), the sh_addr of ".text" section is 0x10030. Since i915 is a huge kernel module, we can see this issue as below. $ grep "[Tt] .*\[i915\]" /proc/kallsyms | sort | head -n1 ffffffffc0270000 t i915_switcheroo_can_switch [i915] ffffffffc0270000 + 0x10030 = ffffffffc0280030, so we'll check symbols cross this boundary. $ grep "[Tt] .*\[i915\]" /proc/kallsyms | grep -B1 ^ffffffffc028\ | head -n 2 ffffffffc027ff80 t haswell_init_clock_gating [i915] ffffffffc0280110 t valleyview_init_clock_gating [i915] So setup probes on both function and see what happen. $ sudo ./perf probe -m i915 -a haswell_init_clock_gating \ -a valleyview_init_clock_gating Added new events: probe:haswell_init_clock_gating (on haswell_init_clock_gating in i915) probe:valleyview_init_clock_gating (on valleyview_init_clock_gating in i915) You can now use it in all perf tools, such as: perf record -e probe:valleyview_init_clock_gating -aR sleep 1 $ sudo ./perf probe -l probe:haswell_init_clock_gating (on haswell_init_clock_gating@gpu/drm/i915/intel_pm.c in i915) probe:valleyview_init_clock_gating (on i915_vga_set_decode:4@gpu/drm/i915/i915_drv.c in i915) As you can see, haswell_init_clock_gating is correctly shown, but valleyview_init_clock_gating is not. With this patch, both events are shown correctly. $ sudo ./perf probe -l probe:haswell_init_clock_gating (on haswell_init_clock_gating@gpu/drm/i915/intel_pm.c in i915) probe:valleyview_init_clock_gating (on valleyview_init_clock_gating@gpu/drm/i915/intel_pm.c in i915) Committer notes: In my case: # perf probe -m i915 -a haswell_init_clock_gating -a valleyview_init_clock_gating Added new events: probe:haswell_init_clock_gating (on haswell_init_clock_gating in i915) probe:valleyview_init_clock_gating (on valleyview_init_clock_gating in i915) You can now use it in all perf tools, such as: perf record -e probe:valleyview_init_clock_gating -aR sleep 1 # perf probe -l probe:haswell_init_clock_gating (on i915_getparam+432@gpu/drm/i915/i915_drv.c in i915) probe:valleyview_init_clock_gating (on __i915_printk+240@gpu/drm/i915/i915_drv.c in i915) # # readelf -SW /lib/modules/4.9.0+/build/vmlinux | egrep -w '.text|Name' [Nr] Name Type Address Off Size ES Flg Lk Inf Al [ 1] .text PROGBITS ffffffff81000000 200000 822fd3 00 AX 0 0 4096 # So both are b0rked, now with the fix: # perf probe -m i915 -a haswell_init_clock_gating -a valleyview_init_clock_gating Added new events: probe:haswell_init_clock_gating (on haswell_init_clock_gating in i915) probe:valleyview_init_clock_gating (on valleyview_init_clock_gating in i915) You can now use it in all perf tools, such as: perf record -e probe:valleyview_init_clock_gating -aR sleep 1 # perf probe -l probe:haswell_init_clock_gating (on haswell_init_clock_gating@gpu/drm/i915/intel_pm.c in i915) probe:valleyview_init_clock_gating (on valleyview_init_clock_gating@gpu/drm/i915/intel_pm.c in i915) # Both looks correct. Signed-off-by: Masami Hiramatsu Tested-by: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/148411436777.9978.1440275861947194930.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-finder.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-) diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index df4debe564da..0278fe1a4cc6 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1543,16 +1543,12 @@ int debuginfo__find_probe_point(struct debuginfo *dbg, unsigned long addr, Dwarf_Addr _addr = 0, baseaddr = 0; const char *fname = NULL, *func = NULL, *basefunc = NULL, *tmp; int baseline = 0, lineno = 0, ret = 0; - bool reloc = false; -retry: + /* We always need to relocate the address for aranges */ + if (debuginfo__get_text_offset(dbg, &baseaddr) == 0) + addr += baseaddr; /* Find cu die */ if (!dwarf_addrdie(dbg->dbg, (Dwarf_Addr)addr, &cudie)) { - if (!reloc && debuginfo__get_text_offset(dbg, &baseaddr) == 0) { - addr += baseaddr; - reloc = true; - goto retry; - } pr_warning("Failed to find debug information for address %lx\n", addr); ret = -EINVAL; From 3e96dac7c956089d3f23aca98c4dfca57b6aaf8a Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 11 Jan 2017 15:00:47 +0900 Subject: [PATCH 2/4] perf probe: Add error checks to offline probe post-processing Add error check codes on post processing and improve it for offline probe events as: - post processing fails if no matched symbol found in map(-ENOENT) or strdup() failed(-ENOMEM). - Even if the symbol name is the same, it updates symbol address and offset. Signed-off-by: Masami Hiramatsu Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/148411443738.9978.4617979132625405545.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-event.c | 50 +++++++++++++++++++++++------------ 1 file changed, 33 insertions(+), 17 deletions(-) diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 4a57c8a60bd9..aa8a9227080a 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -610,6 +610,33 @@ error: return ret ? : -ENOENT; } +/* Adjust symbol name and address */ +static int post_process_probe_trace_point(struct probe_trace_point *tp, + struct map *map, unsigned long offs) +{ + struct symbol *sym; + u64 addr = tp->address + tp->offset - offs; + + sym = map__find_symbol(map, addr); + if (!sym) + return -ENOENT; + + if (strcmp(sym->name, tp->symbol)) { + /* If we have no realname, use symbol for it */ + if (!tp->realname) + tp->realname = tp->symbol; + else + free(tp->symbol); + tp->symbol = strdup(sym->name); + if (!tp->symbol) + return -ENOMEM; + } + tp->offset = addr - sym->start; + tp->address -= offs; + + return 0; +} + /* * Rename DWARF symbols to ELF symbols -- gcc sometimes optimizes functions * and generate new symbols with suffixes such as .constprop.N or .isra.N @@ -622,11 +649,9 @@ static int post_process_offline_probe_trace_events(struct probe_trace_event *tevs, int ntevs, const char *pathname) { - struct symbol *sym; struct map *map; unsigned long stext = 0; - u64 addr; - int i; + int i, ret = 0; /* Prepare a map for offline binary */ map = dso__new_map(pathname); @@ -636,23 +661,14 @@ post_process_offline_probe_trace_events(struct probe_trace_event *tevs, } for (i = 0; i < ntevs; i++) { - addr = tevs[i].point.address + tevs[i].point.offset - stext; - sym = map__find_symbol(map, addr); - if (!sym) - continue; - if (!strcmp(sym->name, tevs[i].point.symbol)) - continue; - /* If we have no realname, use symbol for it */ - if (!tevs[i].point.realname) - tevs[i].point.realname = tevs[i].point.symbol; - else - free(tevs[i].point.symbol); - tevs[i].point.symbol = strdup(sym->name); - tevs[i].point.offset = addr - sym->start; + ret = post_process_probe_trace_point(&tevs[i].point, + map, stext); + if (ret < 0) + break; } map__put(map); - return 0; + return ret; } static int add_exec_to_probe_trace_events(struct probe_trace_event *tevs, From 613f050d68a8ed3c0b18b9568698908ef7bbc1f7 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 11 Jan 2017 15:01:57 +0900 Subject: [PATCH 3/4] perf probe: Fix to probe on gcc generated functions in modules Fix to probe on gcc generated functions on modules. Since probing on a module is based on its symbol name, it should be adjusted on actual symbols. E.g. without this fix, perf probe shows probe definition on non-exist symbol as below. $ perf probe -m build-x86_64/net/netfilter/nf_nat.ko -F in_range* in_range.isra.12 $ perf probe -m build-x86_64/net/netfilter/nf_nat.ko -D in_range p:probe/in_range nf_nat:in_range+0 With this fix, perf probe correctly shows a probe on gcc-generated symbol. $ perf probe -m build-x86_64/net/netfilter/nf_nat.ko -D in_range p:probe/in_range nf_nat:in_range.isra.12+0 This also fixes same problem on online module as below. $ perf probe -m i915 -D assert_plane p:probe/assert_plane i915:assert_plane.constprop.134+0 Signed-off-by: Masami Hiramatsu Tested-by: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/148411450673.9978.14905987549651656075.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/probe-event.c | 45 ++++++++++++++++++++++------------ tools/perf/util/probe-finder.c | 7 ++++-- tools/perf/util/probe-finder.h | 3 +++ 3 files changed, 37 insertions(+), 18 deletions(-) diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index aa8a9227080a..6a6f44dd594b 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -698,18 +698,31 @@ static int add_exec_to_probe_trace_events(struct probe_trace_event *tevs, return ret; } -static int add_module_to_probe_trace_events(struct probe_trace_event *tevs, - int ntevs, const char *module) +static int +post_process_module_probe_trace_events(struct probe_trace_event *tevs, + int ntevs, const char *module, + struct debuginfo *dinfo) { + Dwarf_Addr text_offs = 0; int i, ret = 0; char *mod_name = NULL; + struct map *map; if (!module) return 0; - mod_name = find_module_name(module); + map = get_target_map(module, false); + if (!map || debuginfo__get_text_offset(dinfo, &text_offs, true) < 0) { + pr_warning("Failed to get ELF symbols for %s\n", module); + return -EINVAL; + } + mod_name = find_module_name(module); for (i = 0; i < ntevs; i++) { + ret = post_process_probe_trace_point(&tevs[i].point, + map, (unsigned long)text_offs); + if (ret < 0) + break; tevs[i].point.module = strdup(mod_name ? mod_name : module); if (!tevs[i].point.module) { @@ -719,6 +732,8 @@ static int add_module_to_probe_trace_events(struct probe_trace_event *tevs, } free(mod_name); + map__put(map); + return ret; } @@ -776,7 +791,7 @@ arch__post_process_probe_trace_events(struct perf_probe_event *pev __maybe_unuse static int post_process_probe_trace_events(struct perf_probe_event *pev, struct probe_trace_event *tevs, int ntevs, const char *module, - bool uprobe) + bool uprobe, struct debuginfo *dinfo) { int ret; @@ -784,7 +799,8 @@ static int post_process_probe_trace_events(struct perf_probe_event *pev, ret = add_exec_to_probe_trace_events(tevs, ntevs, module); else if (module) /* Currently ref_reloc_sym based probe is not for drivers */ - ret = add_module_to_probe_trace_events(tevs, ntevs, module); + ret = post_process_module_probe_trace_events(tevs, ntevs, + module, dinfo); else ret = post_process_kernel_probe_trace_events(tevs, ntevs); @@ -828,30 +844,27 @@ static int try_to_find_probe_trace_events(struct perf_probe_event *pev, } } - debuginfo__delete(dinfo); - if (ntevs > 0) { /* Succeeded to find trace events */ pr_debug("Found %d probe_trace_events.\n", ntevs); ret = post_process_probe_trace_events(pev, *tevs, ntevs, - pev->target, pev->uprobes); + pev->target, pev->uprobes, dinfo); if (ret < 0 || ret == ntevs) { + pr_debug("Post processing failed or all events are skipped. (%d)\n", ret); clear_probe_trace_events(*tevs, ntevs); zfree(tevs); + ntevs = 0; } - if (ret != ntevs) - return ret < 0 ? ret : ntevs; - ntevs = 0; - /* Fall through */ } + debuginfo__delete(dinfo); + if (ntevs == 0) { /* No error but failed to find probe point. */ pr_warning("Probe point '%s' not found.\n", synthesize_perf_probe_point(&pev->point)); return -ENOENT; - } - /* Error path : ntevs < 0 */ - pr_debug("An error occurred in debuginfo analysis (%d).\n", ntevs); - if (ntevs < 0) { + } else if (ntevs < 0) { + /* Error path : ntevs < 0 */ + pr_debug("An error occurred in debuginfo analysis (%d).\n", ntevs); if (ntevs == -EBADF) pr_warning("Warning: No dwarf info found in the vmlinux - " "please rebuild kernel with CONFIG_DEBUG_INFO=y.\n"); diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index 0278fe1a4cc6..0d9d6e0803b8 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -1501,7 +1501,8 @@ int debuginfo__find_available_vars_at(struct debuginfo *dbg, } /* For the kernel module, we need a special code to get a DIE */ -static int debuginfo__get_text_offset(struct debuginfo *dbg, Dwarf_Addr *offs) +int debuginfo__get_text_offset(struct debuginfo *dbg, Dwarf_Addr *offs, + bool adjust_offset) { int n, i; Elf32_Word shndx; @@ -1530,6 +1531,8 @@ static int debuginfo__get_text_offset(struct debuginfo *dbg, Dwarf_Addr *offs) if (!shdr) return -ENOENT; *offs = shdr->sh_addr; + if (adjust_offset) + *offs -= shdr->sh_offset; } } return 0; @@ -1545,7 +1548,7 @@ int debuginfo__find_probe_point(struct debuginfo *dbg, unsigned long addr, int baseline = 0, lineno = 0, ret = 0; /* We always need to relocate the address for aranges */ - if (debuginfo__get_text_offset(dbg, &baseaddr) == 0) + if (debuginfo__get_text_offset(dbg, &baseaddr, false) == 0) addr += baseaddr; /* Find cu die */ if (!dwarf_addrdie(dbg->dbg, (Dwarf_Addr)addr, &cudie)) { diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h index f1d8558f498e..2956c5198652 100644 --- a/tools/perf/util/probe-finder.h +++ b/tools/perf/util/probe-finder.h @@ -46,6 +46,9 @@ int debuginfo__find_trace_events(struct debuginfo *dbg, int debuginfo__find_probe_point(struct debuginfo *dbg, unsigned long addr, struct perf_probe_point *ppt); +int debuginfo__get_text_offset(struct debuginfo *dbg, Dwarf_Addr *offs, + bool adjust_offset); + /* Find a line range */ int debuginfo__find_line_range(struct debuginfo *dbg, struct line_range *lr); From 4e71de7986386d5fd3765458f27d612931f27f5e Mon Sep 17 00:00:00 2001 From: Zhou Chengming Date: Mon, 16 Jan 2017 11:21:11 +0800 Subject: [PATCH 4/4] perf/x86/intel: Handle exclusive threadid correctly on CPU hotplug The CPU hotplug function intel_pmu_cpu_starting() sets cpu_hw_events.excl_thread_id unconditionally to 1 when the shared exclusive counters data structure is already availabe for the sibling thread. This works during the boot process because the first sibling gets threadid 0 assigned and the second sibling which shares the data structure gets 1. But when the first thread of the core is offlined and onlined again it shares the data structure with the second thread and gets exclusive thread id 1 assigned as well. Prevent this by checking the threadid of the already online thread. [ tglx: Rewrote changelog ] Signed-off-by: Zhou Chengming Cc: NuoHan Qiao Cc: ak@linux.intel.com Cc: peterz@infradead.org Cc: kan.liang@intel.com Cc: dave.hansen@linux.intel.com Cc: eranian@google.com Cc: qiaonuohan@huawei.com Cc: davidcc@google.com Cc: guohanjun@huawei.com Link: http://lkml.kernel.org/r/1484536871-3131-1-git-send-email-zhouchengming1@huawei.com Signed-off-by: Thomas Gleixner --- --- arch/x86/events/intel/core.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) --- arch/x86/events/intel/core.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index d611cab214a6..eb1484c86bb4 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -3176,13 +3176,16 @@ static void intel_pmu_cpu_starting(int cpu) if (x86_pmu.flags & PMU_FL_EXCL_CNTRS) { for_each_cpu(i, topology_sibling_cpumask(cpu)) { + struct cpu_hw_events *sibling; struct intel_excl_cntrs *c; - c = per_cpu(cpu_hw_events, i).excl_cntrs; + sibling = &per_cpu(cpu_hw_events, i); + c = sibling->excl_cntrs; if (c && c->core_id == core_id) { cpuc->kfree_on_online[1] = cpuc->excl_cntrs; cpuc->excl_cntrs = c; - cpuc->excl_thread_id = 1; + if (!sibling->excl_thread_id) + cpuc->excl_thread_id = 1; break; } }