From 553de6e1157df63fc6cdfe4573e04c8edcbe68f2 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 1 Jul 2021 13:39:15 -0300 Subject: [PATCH 1/5] tools headers cpufeatures: Sync with the kernel sources To pick the changes from: 28a99e95f55c6185 ("x86/amd: Use IBPB for firmware calls") This only causes these perf files to be rebuilt: CC /tmp/build/perf/bench/mem-memcpy-x86-64-asm.o CC /tmp/build/perf/bench/mem-memset-x86-64-asm.o And addresses this perf build warning: Warning: Kernel ABI header at 'tools/arch/x86/include/asm/cpufeatures.h' differs from latest version at 'arch/x86/include/asm/cpufeatures.h' diff -u tools/arch/x86/include/asm/cpufeatures.h arch/x86/include/asm/cpufeatures.h Cc: Adrian Hunter Cc: Borislav Petkov Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra --- tools/arch/x86/include/asm/cpufeatures.h | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/arch/x86/include/asm/cpufeatures.h b/tools/arch/x86/include/asm/cpufeatures.h index 00f5227c8459..a77b915d36a8 100644 --- a/tools/arch/x86/include/asm/cpufeatures.h +++ b/tools/arch/x86/include/asm/cpufeatures.h @@ -302,6 +302,7 @@ #define X86_FEATURE_RETPOLINE_LFENCE (11*32+13) /* "" Use LFENCE for Spectre variant 2 */ #define X86_FEATURE_RETHUNK (11*32+14) /* "" Use REturn THUNK */ #define X86_FEATURE_UNRET (11*32+15) /* "" AMD BTB untrain return */ +#define X86_FEATURE_USE_IBPB_FW (11*32+16) /* "" Use IBPB during runtime firmware calls */ /* Intel-defined CPU features, CPUID level 0x00000007:1 (EAX), word 12 */ #define X86_FEATURE_AVX_VNNI (12*32+ 4) /* AVX VNNI instructions */ From b226521923aee7051f4b24df9be5bf07d53f0a2b Mon Sep 17 00:00:00 2001 From: Leo Yan Date: Mon, 25 Jul 2022 18:42:20 +0800 Subject: [PATCH 2/5] perf scripts python: Let script to be python2 compliant The mainline kernel can be used for relative old distros, e.g. RHEL 7. The distro doesn't upgrade from python2 to python3, this causes the building error that the python script is not python2 compliant. To fix the building failure, this patch changes from the python f-string format to traditional string format. Fixes: 12fdd6c009da0d02 ("perf scripts python: Support Arm CoreSight trace data disassembly") Reported-by: Akemi Yagi Signed-off-by: Leo Yan Cc: Alexander Shishkin Cc: ElRepo Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Leo Yan Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20220725104220.1106663-1-leo.yan@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- .../scripts/python/arm-cs-trace-disasm.py | 34 ++++++++++--------- 1 file changed, 18 insertions(+), 16 deletions(-) diff --git a/tools/perf/scripts/python/arm-cs-trace-disasm.py b/tools/perf/scripts/python/arm-cs-trace-disasm.py index 5f57d9829956..4339692a8d0b 100755 --- a/tools/perf/scripts/python/arm-cs-trace-disasm.py +++ b/tools/perf/scripts/python/arm-cs-trace-disasm.py @@ -61,7 +61,7 @@ def get_optional(perf_dict, field): def get_offset(perf_dict, field): if field in perf_dict: - return f"+0x{perf_dict[field]:x}" + return "+%#x" % perf_dict[field] return "" def get_dso_file_path(dso_name, dso_build_id): @@ -76,7 +76,7 @@ def get_dso_file_path(dso_name, dso_build_id): else: append = "/elf" - dso_path = f"{os.environ['PERF_BUILDID_DIR']}/{dso_name}/{dso_build_id}{append}" + dso_path = os.environ['PERF_BUILDID_DIR'] + "/" + dso_name + "/" + dso_build_id + append; # Replace duplicate slash chars to single slash char dso_path = dso_path.replace('//', '/', 1) return dso_path @@ -94,8 +94,8 @@ def read_disam(dso_fname, dso_start, start_addr, stop_addr): start_addr = start_addr - dso_start; stop_addr = stop_addr - dso_start; disasm = [ options.objdump_name, "-d", "-z", - f"--start-address=0x{start_addr:x}", - f"--stop-address=0x{stop_addr:x}" ] + "--start-address="+format(start_addr,"#x"), + "--stop-address="+format(stop_addr,"#x") ] disasm += [ dso_fname ] disasm_output = check_output(disasm).decode('utf-8').split('\n') disasm_cache[addr_range] = disasm_output @@ -109,12 +109,14 @@ def print_disam(dso_fname, dso_start, start_addr, stop_addr): m = disasm_re.search(line) if m is None: continue - print(f"\t{line}") + print("\t" + line) def print_sample(sample): - print(f"Sample = {{ cpu: {sample['cpu']:04} addr: 0x{sample['addr']:016x} " \ - f"phys_addr: 0x{sample['phys_addr']:016x} ip: 0x{sample['ip']:016x} " \ - f"pid: {sample['pid']} tid: {sample['tid']} period: {sample['period']} time: {sample['time']} }}") + print("Sample = { cpu: %04d addr: 0x%016x phys_addr: 0x%016x ip: 0x%016x " \ + "pid: %d tid: %d period: %d time: %d }" % \ + (sample['cpu'], sample['addr'], sample['phys_addr'], \ + sample['ip'], sample['pid'], sample['tid'], \ + sample['period'], sample['time'])) def trace_begin(): print('ARM CoreSight Trace Data Assembler Dump') @@ -131,7 +133,7 @@ def common_start_str(comm, sample): cpu = sample["cpu"] pid = sample["pid"] tid = sample["tid"] - return f"{comm:>16} {pid:>5}/{tid:<5} [{cpu:04}] {sec:9}.{ns:09} " + return "%16s %5u/%-5u [%04u] %9u.%09u " % (comm, pid, tid, cpu, sec, ns) # This code is copied from intel-pt-events.py for printing source code # line and symbols. @@ -171,7 +173,7 @@ def print_srccode(comm, param_dict, sample, symbol, dso): glb_line_number = line_number glb_source_file_name = source_file_name - print(f"{start_str}{src_str}") + print(start_str, src_str) def process_event(param_dict): global cache_size @@ -188,7 +190,7 @@ def process_event(param_dict): symbol = get_optional(param_dict, "symbol") if (options.verbose == True): - print(f"Event type: {name}") + print("Event type: %s" % name) print_sample(sample) # If cannot find dso so cannot dump assembler, bail out @@ -197,7 +199,7 @@ def process_event(param_dict): # Validate dso start and end addresses if ((dso_start == '[unknown]') or (dso_end == '[unknown]')): - print(f"Failed to find valid dso map for dso {dso}") + print("Failed to find valid dso map for dso %s" % dso) return if (name[0:12] == "instructions"): @@ -244,15 +246,15 @@ def process_event(param_dict): # Handle CS_ETM_TRACE_ON packet if start_addr=0 and stop_addr=4 if (start_addr == 0 and stop_addr == 4): - print(f"CPU{cpu}: CS_ETM_TRACE_ON packet is inserted") + print("CPU%d: CS_ETM_TRACE_ON packet is inserted" % cpu) return if (start_addr < int(dso_start) or start_addr > int(dso_end)): - print(f"Start address 0x{start_addr:x} is out of range [ 0x{dso_start:x} .. 0x{dso_end:x} ] for dso {dso}") + print("Start address 0x%x is out of range [ 0x%x .. 0x%x ] for dso %s" % (start_addr, int(dso_start), int(dso_end), dso)) return if (stop_addr < int(dso_start) or stop_addr > int(dso_end)): - print(f"Stop address 0x{stop_addr:x} is out of range [ 0x{dso_start:x} .. 0x{dso_end:x} ] for dso {dso}") + print("Stop address 0x%x is out of range [ 0x%x .. 0x%x ] for dso %s" % (stop_addr, int(dso_start), int(dso_end), dso)) return if (options.objdump_name != None): @@ -267,6 +269,6 @@ def process_event(param_dict): if path.exists(dso_fname): print_disam(dso_fname, dso_vm_start, start_addr, stop_addr) else: - print(f"Failed to find dso {dso} for address range [ 0x{start_addr:x} .. 0x{stop_addr:x} ]") + print("Failed to find dso %s for address range [ 0x%x .. 0x%x ]" % (dso, start_addr, stop_addr)) print_srccode(comm, param_dict, sample, symbol, dso) From 2d86612aacb7805f72873691a2644d7279ed0630 Mon Sep 17 00:00:00 2001 From: Leo Yan Date: Sun, 24 Jul 2022 14:00:12 +0800 Subject: [PATCH 3/5] perf symbol: Correct address for bss symbols When using 'perf mem' and 'perf c2c', an issue is observed that tool reports the wrong offset for global data symbols. This is a common issue on both x86 and Arm64 platforms. Let's see an example, for a test program, below is the disassembly for its .bss section which is dumped with objdump: ... Disassembly of section .bss: 0000000000004040 : ... 0000000000004080 : ... 00000000000040c0 : ... 0000000000004100 : ... First we used 'perf mem record' to run the test program and then used 'perf --debug verbose=4 mem report' to observe what's the symbol info for 'buf1' and 'buf2' structures. # ./perf mem record -e ldlat-loads,ldlat-stores -- false_sharing.exe 8 # ./perf --debug verbose=4 mem report ... dso__load_sym_internal: adjusting symbol: st_value: 0x40c0 sh_addr: 0x4040 sh_offset: 0x3028 symbol__new: buf2 0x30a8-0x30e8 ... dso__load_sym_internal: adjusting symbol: st_value: 0x4080 sh_addr: 0x4040 sh_offset: 0x3028 symbol__new: buf1 0x3068-0x30a8 ... The perf tool relies on libelf to parse symbols, in executable and shared object files, 'st_value' holds a virtual address; 'sh_addr' is the address at which section's first byte should reside in memory, and 'sh_offset' is the byte offset from the beginning of the file to the first byte in the section. The perf tool uses below formula to convert a symbol's memory address to a file address: file_address = st_value - sh_addr + sh_offset ^ ` Memory address We can see the final adjusted address ranges for buf1 and buf2 are [0x30a8-0x30e8) and [0x3068-0x30a8) respectively, apparently this is incorrect, in the code, the structure for 'buf1' and 'buf2' specifies compiler attribute with 64-byte alignment. The problem happens for 'sh_offset', libelf returns it as 0x3028 which is not 64-byte aligned, combining with disassembly, it's likely libelf doesn't respect the alignment for .bss section, therefore, it doesn't return the aligned value for 'sh_offset'. Suggested by Fangrui Song, ELF file contains program header which contains PT_LOAD segments, the fields p_vaddr and p_offset in PT_LOAD segments contain the execution info. A better choice for converting memory address to file address is using the formula: file_address = st_value - p_vaddr + p_offset This patch introduces elf_read_program_header() which returns the program header based on the passed 'st_value', then it uses the formula above to calculate the symbol file address; and the debugging log is updated respectively. After applying the change: # ./perf --debug verbose=4 mem report ... dso__load_sym_internal: adjusting symbol: st_value: 0x40c0 p_vaddr: 0x3d28 p_offset: 0x2d28 symbol__new: buf2 0x30c0-0x3100 ... dso__load_sym_internal: adjusting symbol: st_value: 0x4080 p_vaddr: 0x3d28 p_offset: 0x2d28 symbol__new: buf1 0x3080-0x30c0 ... Fixes: f17e04afaff84b5c ("perf report: Fix ELF symbol parsing") Reported-by: Chang Rui Suggested-by: Fangrui Song Signed-off-by: Leo Yan Acked-by: Namhyung Kim Cc: Alexander Shishkin Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20220724060013.171050-2-leo.yan@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/symbol-elf.c | 45 ++++++++++++++++++++++++++++++++---- 1 file changed, 41 insertions(+), 4 deletions(-) diff --git a/tools/perf/util/symbol-elf.c b/tools/perf/util/symbol-elf.c index ecd377938eea..ef6ced5c5746 100644 --- a/tools/perf/util/symbol-elf.c +++ b/tools/perf/util/symbol-elf.c @@ -233,6 +233,33 @@ Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, return NULL; } +static int elf_read_program_header(Elf *elf, u64 vaddr, GElf_Phdr *phdr) +{ + size_t i, phdrnum; + u64 sz; + + if (elf_getphdrnum(elf, &phdrnum)) + return -1; + + for (i = 0; i < phdrnum; i++) { + if (gelf_getphdr(elf, i, phdr) == NULL) + return -1; + + if (phdr->p_type != PT_LOAD) + continue; + + sz = max(phdr->p_memsz, phdr->p_filesz); + if (!sz) + continue; + + if (vaddr >= phdr->p_vaddr && (vaddr < phdr->p_vaddr + sz)) + return 0; + } + + /* Not found any valid program header */ + return -1; +} + static bool want_demangle(bool is_kernel_sym) { return is_kernel_sym ? symbol_conf.demangle_kernel : symbol_conf.demangle; @@ -1209,6 +1236,7 @@ dso__load_sym_internal(struct dso *dso, struct map *map, struct symsrc *syms_ss, sym.st_value); used_opd = true; } + /* * When loading symbols in a data mapping, ABS symbols (which * has a value of SHN_ABS in its st_shndx) failed at @@ -1262,11 +1290,20 @@ dso__load_sym_internal(struct dso *dso, struct map *map, struct symsrc *syms_ss, goto out_elf_end; } else if ((used_opd && runtime_ss->adjust_symbols) || (!used_opd && syms_ss->adjust_symbols)) { + GElf_Phdr phdr; + + if (elf_read_program_header(syms_ss->elf, + (u64)sym.st_value, &phdr)) { + pr_warning("%s: failed to find program header for " + "symbol: %s st_value: %#" PRIx64 "\n", + __func__, elf_name, (u64)sym.st_value); + continue; + } pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " " - "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__, - (u64)sym.st_value, (u64)shdr.sh_addr, - (u64)shdr.sh_offset); - sym.st_value -= shdr.sh_addr - shdr.sh_offset; + "p_vaddr: %#" PRIx64 " p_offset: %#" PRIx64 "\n", + __func__, (u64)sym.st_value, (u64)phdr.p_vaddr, + (u64)phdr.p_offset); + sym.st_value -= phdr.p_vaddr - phdr.p_offset; } demangled = demangle_sym(dso, kmodule, elf_name); From 882528d2e77687c3ef26abb9c490f77a9c1f6e1a Mon Sep 17 00:00:00 2001 From: Leo Yan Date: Sun, 24 Jul 2022 14:00:13 +0800 Subject: [PATCH 4/5] perf symbol: Skip symbols if SHF_ALLOC flag is not set Some symbols are observed with the 'st_value' field zeroed. E.g. libc.so.6 in Ubuntu contains a symbol '__evoke_link_warning_getwd' which resides in the '.gnu.warning.getwd' section. Unlike normal sections, such kind of sections are used for linker warning when a file calls deprecated functions, but they are not part of memory images, the symbols in these sections should be dropped. This patch checks the section attribute SHF_ALLOC bit, if the bit is not set, it skips symbols to avoid spurious ones. Suggested-by: Fangrui Song Signed-off-by: Leo Yan Acked-by: Namhyung Kim Cc: Alexander Shishkin Cc: Chang Rui Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20220724060013.171050-3-leo.yan@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/symbol-elf.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/tools/perf/util/symbol-elf.c b/tools/perf/util/symbol-elf.c index ef6ced5c5746..b3be5b1d9dbb 100644 --- a/tools/perf/util/symbol-elf.c +++ b/tools/perf/util/symbol-elf.c @@ -1255,6 +1255,17 @@ dso__load_sym_internal(struct dso *dso, struct map *map, struct symsrc *syms_ss, gelf_getshdr(sec, &shdr); + /* + * If the attribute bit SHF_ALLOC is not set, the section + * doesn't occupy memory during process execution. + * E.g. ".gnu.warning.*" section is used by linker to generate + * warnings when calling deprecated functions, the symbols in + * the section aren't loaded to memory during process execution, + * so skip them. + */ + if (!(shdr.sh_flags & SHF_ALLOC)) + continue; + secstrs = secstrs_sym; /* From 9a241805673ec0a826b7ddf84b00f4e03adb0a5e Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 26 Jul 2022 15:09:21 -0700 Subject: [PATCH 5/5] perf bpf: Remove undefined behavior from bpf_perf_object__next() bpf_perf_object__next() folded the last element in the list test with the empty list test. However, this meant that offsets were computed against null and that a struct list_head was compared against a 'struct bpf_perf_object'. Working around this with clang's undefined behavior sanitizer required -fno-sanitize=null and -fno-sanitize=object-size. Remove the undefined behavior by using the regular Linux list APIs and handling the starting case separately from the end testing case. Looking at uses like bpf_perf_object__for_each(), as the constant NULL or non-NULL argument can be constant propagated, the code is no less efficient. Signed-off-by: Ian Rogers Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andrii Nakryiko Cc: Christy Lee Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Miaoqian Lin Cc: Namhyung Kim Cc: Nathan Chancellor Cc: Nick Desaulniers Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Tom Rix Cc: bpf@vger.kernel.org Cc: llvm@lists.linux.dev Link: https://lore.kernel.org/r/20220726220921.2567761-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf-loader.c | 18 +++++++----------- 1 file changed, 7 insertions(+), 11 deletions(-) diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c index f8ad581ea247..cdd6463a5b68 100644 --- a/tools/perf/util/bpf-loader.c +++ b/tools/perf/util/bpf-loader.c @@ -63,20 +63,16 @@ static struct hashmap *bpf_map_hash; static struct bpf_perf_object * bpf_perf_object__next(struct bpf_perf_object *prev) { - struct bpf_perf_object *next; + if (!prev) { + if (list_empty(&bpf_objects_list)) + return NULL; - if (!prev) - next = list_first_entry(&bpf_objects_list, - struct bpf_perf_object, - list); - else - next = list_next_entry(prev, list); - - /* Empty list is noticed here so don't need checking on entry. */ - if (&next->list == &bpf_objects_list) + return list_first_entry(&bpf_objects_list, struct bpf_perf_object, list); + } + if (list_is_last(&prev->list, &bpf_objects_list)) return NULL; - return next; + return list_next_entry(prev, list); } #define bpf_perf_object__for_each(perf_obj, tmp) \