From d2d4edbebe07ddb77980656abe7b9bc7a9e0cdf7 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 11 Jan 2017 14:59:38 +0900 Subject: 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(-) (limited to 'tools/perf/util/probe-finder.c') 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; -- cgit v1.2.3 From 613f050d68a8ed3c0b18b9568698908ef7bbc1f7 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 11 Jan 2017 15:01:57 +0900 Subject: 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-finder.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/perf/util/probe-finder.c') 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)) { -- cgit v1.2.3