- 17 Jun, 2019 20 commits
-
-
Adrian Hunter authored
Synthesize a PEBS sample using basic information (ip, timestamp) only. Other PEBS information will be added in later patches. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/r/20190610072803.10456-7-adrian.hunter@intel.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Adrian Hunter authored
Factor out common sample preparation for re-use when synthesizing PEBS samples. Signed-off-by:
-
Adrian Hunter authored
Add infrastructure to prepare for synthesizing PEBS samples but leave the actual synthesis to later patches. Signed-off-by:
-
Adrian Hunter authored
PEBS data is encoded in Block Item Packets (BIP). Populate a new structure intel_pt_blk_items with the values and, upon a Block End Packet (BEP), report them as a new Intel PT sample type INTEL_PT_BLK_ITEMS. Signed-off-by:
-
Adrian Hunter authored
Add Intel PT packet decoder test. This test feeds byte sequences to the Intel PT packet decoder and checks the results. Changes to the packet context are also checked. Committer testing: # perf test "Intel PT" 65: Intel PT packet decoder : Ok # perf test -v "Intel PT" 65: Intel PT packet decoder : --- start --- test child forked, pid 6360 Decoded ok: 00 PAD Decoded ok: 04 TNT N (1) Decoded ok: 06 TNT T (1) Decoded ok: 80 TNT NNNNNN (6) Decoded ok: fe TNT TTTTTT (6) Decoded ok: 02 a3 02 00 00 00 00 00 TNT N (1) Decoded ok: 02 a3 03 00 00 00 00 00 TNT T (1) Decoded ok: 02 a3 00 00 00 00 00 80 TNT NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN (47) Decoded ok: 02 a3 ff ff ff ff ff ff TNT TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT (47) Decoded ok: 0d TIP no ip Decoded ok: 2d 01 02 TIP 0x201 Decoded ok: 4d 01 02 03 04 TIP 0x4030201 Decoded ok: 6d 01 02 03 04 05 06 TIP 0x60504030201 Decoded ok: 8d 01 02 03 04 05 06 TIP 0x60504030201 Decoded ok: cd 01 02 03 04 05 06 07 08 TIP 0x807060504030201 Decoded ok: 11 TIP.PGE no ip Decoded ok: 31 01 02 TIP.PGE 0x201 Decoded ok: 51 01 02 03 04 TIP.PGE 0x4030201 Decoded ok: 71 01 02 03 04 05 06 TIP.PGE 0x60504030201 Decoded ok: 91 01 02 03 04 05 06 TIP.PGE 0x60504030201 Decoded ok: d1 01 02 03 04 05 06 07 08 TIP.PGE 0x807060504030201 Decoded ok: 01 TIP.PGD no ip Decoded ok: 21 01 02 TIP.PGD 0x201 Decoded ok: 41 01 02 03 04 TIP.PGD 0x4030201 Decoded ok: 61 01 02 03 04 05 06 TIP.PGD 0x60504030201 Decoded ok: 81 01 02 03 04 05 06 TIP.PGD 0x60504030201 Decoded ok: c1 01 02 03 04 05 06 07 08 TIP.PGD 0x807060504030201 Decoded ok: 1d FUP no ip Decoded ok: 3d 01 02 FUP 0x201 Decoded ok: 5d 01 02 03 04 FUP 0x4030201 Decoded ok: 7d 01 02 03 04 05 06 FUP 0x60504030201 Decoded ok: 9d 01 02 03 04 05 06 FUP 0x60504030201 Decoded ok: dd 01 02 03 04 05 06 07 08 FUP 0x807060504030201 Decoded ok: 02 43 02 04 06 08 0a 0c PIP 0x60504030201 (NR=0) Decoded ok: 02 43 03 04 06 08 0a 0c PIP 0x60504030201 (NR=1) Decoded ok: 99 00 MODE.Exec 16 Decoded ok: 99 01 MODE.Exec 64 Decoded ok: 99 02 MODE.Exec 32 Decoded ok: 99 20 MODE.TSX TXAbort:0 InTX:0 Decoded ok: 99 21 MODE.TSX TXAbort:0 InTX:1 Decoded ok: 99 22 MODE.TSX TXAbort:1 InTX:0 Decoded ok: 02 83 TraceSTOP Decoded ok: 02 03 12 00 CBR 0x12 Decoded ok: 19 01 02 03 04 05 06 07 TSC 0x7060504030201 Decoded ok: 59 12 MTC 0x12 Decoded ok: 02 73 00 00 00 00 00 TMA CTC 0x0 FC 0x0 Decoded ok: 02 73 01 02 00 00 00 TMA CTC 0x201 FC 0x0 Decoded ok: 02 73 00 00 00 ff 01 TMA CTC 0x0 FC 0x1ff Decoded ok: 02 73 80 c0 00 ff 01 TMA CTC 0xc080 FC 0x1ff Decoded ok: 03 CYC 0x0 Decoded ok: 0b CYC 0x1 Decoded ok: fb CYC 0x1f Decoded ok: 07 02 CYC 0x20 Decoded ok: ff fe CYC 0xfff Decoded ok: 07 01 02 CYC 0x1000 Decoded ok: ff ff fe CYC 0x7ffff Decoded ok: 07 01 01 02 CYC 0x80000 Decoded ok: ff ff ff fe CYC 0x3ffffff Decoded ok: 07 01 01 01 02 CYC 0x4000000 Decoded ok: ff ff ff ff fe CYC 0x1ffffffff Decoded ok: 07 01 01 01 01 02 CYC 0x200000000 Decoded ok: ff ff ff ff ff fe CYC 0xffffffffff Decoded ok: 07 01 01 01 01 01 02 CYC 0x10000000000 Decoded ok: ff ff ff ff ff ff fe CYC 0x7fffffffffff Decoded ok: 07 01 01 01 01 01 01 02 CYC 0x800000000000 Decoded ok: ff ff ff ff ff ff ff fe CYC 0x3fffffffffffff Decoded ok: 07 01 01 01 01 01 01 01 02 CYC 0x40000000000000 Decoded ok: ff ff ff ff ff ff ff ff fe CYC 0x1fffffffffffffff Decoded ok: 07 01 01 01 01 01 01 01 01 02 CYC 0x2000000000000000 Decoded ok: ff ff ff ff ff ff ff ff ff 0e CYC 0xffffffffffffffff Decoded ok: 02 c8 01 02 03 04 05 VMCS 0x504030201 Decoded ok: 02 f3 OVF Decoded ok: 02 f3 OVF Decoded ok: 02 f3 OVF Decoded ok: 02 82 02 82 02 82 02 82 02 82 02 82 02 82 02 82 PSB Decoded ok: 02 82 02 82 02 82 02 82 02 82 02 82 02 82 02 82 PSB Decoded ok: 02 82 02 82 02 82 02 82 02 82 02 82 02 82 02 82 PSB Decoded ok: 02 23 PSBEND Decoded ok: 02 c3 88 01 02 03 04 05 06 07 00 MNT 0x7060504030201 Decoded ok: 02 12 01 02 03 04 PTWRITE 0x4030201 IP:0 Decoded ok: 02 32 01 02 03 04 05 06 07 08 PTWRITE 0x807060504030201 IP:0 Decoded ok: 02 92 01 02 03 04 PTWRITE 0x4030201 IP:1 Decoded ok: 02 b2 01 02 03 04 05 06 07 08 PTWRITE 0x807060504030201 IP:1 Decoded ok: 02 62 EXSTOP IP:0 Decoded ok: 02 e2 EXSTOP IP:1 Decoded ok: 02 c2 00 00 00 00 00 00 00 00 MWAIT 0x0 Hints 0x0 Extensions 0x0 Decoded ok: 02 c2 01 02 03 04 05 06 07 08 MWAIT 0x807060504030201 Hints 0x1 Extensions 0x1 Decoded ok: 02 c2 ff 02 03 04 07 06 07 08 MWAIT 0x8070607040302ff Hints 0xff Extensions 0x3 Decoded ok: 02 22 00 00 PWRE 0x0 HW:0 CState:0 Sub-CState:0 Decoded ok: 02 22 01 02 PWRE 0x201 HW:0 CState:0 Sub-CState:2 Decoded ok: 02 22 80 34 PWRE 0x3480 HW:1 CState:3 Sub-CState:4 Decoded ok: 02 22 00 56 PWRE 0x5600 HW:0 CState:5 Sub-CState:6 Decoded ok: 02 a2 00 00 00 00 00 PWRX 0x0 Last CState:0 Deepest CState:0 Wake Reason 0x0 Decoded ok: 02 a2 01 02 03 04 05 PWRX 0x504030201 Last CState:0 Deepest CState:1 Wake Reason 0x2 Decoded ok: 02 a2 ff ff ff ff ff PWRX 0xffffffffff Last CState:15 Deepest CState:15 Wake Reason 0xf Decoded ok: 02 63 00 BBP SZ 8-byte Type 0x0 Decoded ok: 02 63 80 BBP SZ 4-byte Type 0x0 Decoded ok: 02 63 1f BBP SZ 8-byte Type 0x1f Decoded ok: 02 63 9f BBP SZ 4-byte Type 0x1f Decoded ok: 04 00 00 00 00 BIP ID 0x00 Value 0x0 Decoded ok: fc 00 00 00 00 BIP ID 0x1f Value 0x0 Decoded ok: 04 01 02 03 04 BIP ID 0x00 Value 0x4030201 Decoded ok: fc 01 02 03 04 BIP ID 0x1f Value 0x4030201 Decoded ok: 04 00 00 00 00 00 00 00 00 BIP ID 0x00 Value 0x0 Decoded ok: fc 00 00 00 00 00 00 00 00 BIP ID 0x1f Value 0x0 Decoded ok: 04 01 02 03 04 05 06 07 08 BIP ID 0x00 Value 0x807060504030201 Decoded ok: fc 01 02 03 04 05 06 07 08 BIP ID 0x1f Value 0x807060504030201 Decoded ok: 02 33 BEP IP:0 Decoded ok: 02 b3 BEP IP:1 Decoded ok: 02 33 BEP IP:0 Decoded ok: 02 b3 BEP IP:1 test child finished with 0 ---- end ---- Intel PT packet decoder: Ok # Signed-off-by:
-
Adrian Hunter authored
Add 3 new packets to supports PEBS via PT, namely Block Begin Packet (BBP), Block Item Packet (BIP) and Block End Packet (BEP). PEBS data is encoded into multiple BIP packets that come between BBP and BEP. The BEP packet might be associated with a FUP packet. That is indicated by using a separate packet type (INTEL_PT_BEP_IP) similar to other packets types with the _IP suffix. Refer to the Intel SDM for more information about PEBS via PT: https://software.intel.com/en-us/articles/intel-sdm May 2019 version: Vol. 3B 18.5.5.2 PEBS output to Intel
® Processor Trace Decoding of BIP packets conflicts with single-byte TNT packets. Since BIP packets only occur in the context of a block (i.e. between BBP and BEP), that context must be recorded and passed to the packet decoder. Signed-off-by: -
Mathieu Poirier authored
Call function cs_etm_set_option() once with all relevant options set rather than multiple times to avoid going through the list of CPU more than once. Suggested-by:
Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by:
Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: linux-arm-kernel@lists.infradead.org Link: http://lkml.kernel.org/r/20190611204528.20093-1-mathieu.poirier@linaro.orgSigned-off-by:
-
Raphael Gault authored
In order to subsequently add more tests for the arm64 architecture we compile the tests target for arm64 systematically. Further explanation provided by Mark Rutland: Given prior questions regarding this commit, it's probably worth spelling things out more explicitly, e.g. Currently we only build the arm64/tests directory if CONFIG_DWARF_UNWIND is selected, which is fine as the only test we have is arm64/tests/dwarf-unwind.o. So that we can add more tests to the test directory, let's unconditionally build the directory, but conditionally build dwarf-unwind.o depending on CONFIG_DWARF_UNWIND. There should be no functional change as a result of this patch. Signed-off-by:
Raphael Gault <raphael.gault@arm.com> Acked-by:
Mark Rutland <mark.rutland@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Will Deacon <will.deacon@arm.com> Cc: linux-arm-kernel@lists.infradead.org Link: http://lkml.kernel.org/r/20190611125315.18736-2-raphael.gault@arm.comSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-5.3-20190611' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: perf record: Alexey Budankov: - Allow mixing --user-regs with --call-graph=dwarf, making sure that the minimal set of registers for DWARF unwinding is present in the set of user registers requested to be present in each sample, while warning the user that this may make callchains unreliable if more that the minimal set of registers is needed to unwind. yuzhoujian: - Add support to collect callchains from kernel or user space only, IOW allow setting the perf_event_attr.exclude_callchain_{kernel,user} bits from the command line. perf trace: Arnaldo Carvalho de Melo: - Remove x86_64 specific syscall numbers from the augmented_raw_syscalls BPF in-kernel collector of augmented raw_syscalls:sys_{enter,exit} payloads, use instead the syscall numbers obtainer either by the arch specific syscalltbl generators or from audit-libs. - Allow 'perf trace' to ask for the number of bytes to collect for string arguments, for now ask for PATH_MAX, i.e. the whole pathnames, which ends up being just a way to speficy which syscall args are pathnames and thus should be read using bpf_probe_read_str(). - Skip unknown syscalls when expanding strace like syscall groups. This helps using the 'string' group of syscalls to work in arm64, where some of the syscalls present in x86_64 that deal with strings, for instance 'access', are deprecated and this should not be asked for tracing. Leo Yan: - Exit when failing to build eBPF program. perf config: Arnaldo Carvalho de Melo: - Bail out when a handler returns failure for a key-value pair. This helps with cases where processing a key-value pair is not just a matter of setting some tool specific knob, involving, for instance building a BPF program to then attach to the list of events 'perf trace' will use, e.g. augmented_raw_syscalls.c. perf.data: Kan Liang: - Read and store die ID information available in new Intel processors in CPUID.1F in the CPU topology written in the perf.data header. perf stat: Kan Liang: - Support per-die aggregation. Documentation: Arnaldo Carvalho de Melo: - Update perf.data documentation about the CPU_TOPOLOGY, MEM_TOPOLOGY, CLOCKID and DIR_FORMAT headers. Song Liu: - Add description of headers HEADER_BPF_PROG_INFO and HEADER_BPF_BTF. Leo Yan: - Update default value for llvm.clang-bpf-cmd-template in 'man perf-config'. JVMTI: Jiri Olsa: - Address gcc string overflow warning for strncpy() core: - Remove superfluous nthreads system_wide setup in perf_evsel__alloc_fd(). Intel PT: Adrian Hunter: - Add support for samples to contain IPC ratio, collecting cycles information from CYC packets, showing the IPC info periodically, because Intel PT does not update the cycle count on every branch or instruction, the incremental values will often be zero. When there are values, they will be the number of instructions and number of cycles since the last update, and thus represent the average IPC since the last IPC value. E.g.: # perf record --cpu 1 -m200000 -a -e intel_pt/cyc/u sleep 0.0001 rounding mmap pages size to 1024M (262144 pages) [ perf record: Woken up 0 times to write data ] [ perf record: Captured and wrote 2.208 MB perf.data ] # perf script --insn-trace --xed -F+ipc,-dso,-cpu,-tid # <SNIP + add line numbering to make sense of IPC counts e.g.: (18/3)> 1 cc1 63501.650479626: 7f5219ac27bf _int_free+0x3f jnz 0x7f5219ac2af0 IPC: 0.81 (36/44) 2 cc1 63501.650479626: 7f5219ac27c5 _int_free+0x45 cmp $0x1f, %rbp 3 cc1 63501.650479626: 7f5219ac27c9 _int_free+0x49 jbe 0x7f5219ac2b00 4 cc1 63501.650479626: 7f5219ac27cf _int_free+0x4f test $0x8, %al 5 cc1 63501.650479626: 7f5219ac27d1 _int_free+0x51 jnz 0x7f5219ac2b00 6 cc1 63501.650479626: 7f5219ac27d7 _int_free+0x57 movq 0x13c58a(%rip), %rcx 7 cc1 63501.650479626: 7f5219ac27de _int_free+0x5e mov %rdi, %r12 8 cc1 63501.650479626: 7f5219ac27e1 _int_free+0x61 movq %fs:(%rcx), %rax 9 cc1 63501.650479626: 7f5219ac27e5 _int_free+0x65 test %rax, %rax 10 cc1 63501.650479626: 7f5219ac27e8 _int_free+0x68 jz 0x7f5219ac2821 11 cc1 63501.650479626: 7f5219ac27ea _int_free+0x6a leaq -0x11(%rbp), %rdi 12 cc1 63501.650479626: 7f5219ac27ee _int_free+0x6e mov %rdi, %rsi 13 cc1 63501.650479626: 7f5219ac27f1 _int_free+0x71 shr $0x4, %rsi 14 cc1 63501.650479626: 7f5219ac27f5 _int_free+0x75 cmpq %rsi, 0x13caf4(%rip) 15 cc1 63501.650479626: 7f5219ac27fc _int_free+0x7c jbe 0x7f5219ac2821 16 cc1 63501.650479626: 7f5219ac2821 _int_free+0xa1 cmpq 0x13f138(%rip), %rbp 17 cc1 63501.650479626: 7f5219ac2828 _int_free+0xa8 jnbe 0x7f5219ac28d8 18 cc1 63501.650479626: 7f5219ac28d8 _int_free+0x158 testb $0x2, 0x8(%rbx) 19 cc1 63501.650479628: 7f5219ac28dc _int_free+0x15c jnz 0x7f5219ac2ab0 IPC: 6.00 (18/3) <SNIP> - Allow using time ranges with Intel PT, i.e. these features, already present but not optimially usable with Intel PT, should be now: Select the second 10% time slice: $ perf script --time 10%/2 Select from 0% to 10% time slice: $ perf script --time 0%-10% Select the first and second 10% time slices: $ perf script --time 10%/1,10%/2 Select from 0% to 10% and 30% to 40% slices: $ perf script --time 0%-10%,30%-40% cs-etm (ARM): Mathieu Poirier: - Add support for CPU-wide trace scenarios. s390: Thomas Richter: - Fix missing kvm module load for s390. - Fix OOM error in TUI mode on s390 - Support s390 diag event display when doing analysis on !s390 architectures. Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Jiri Olsa authored
Tom Vaden reported false failure of the check_msr() function, because some servers can do POST tracing and enable LBR tracing during bootup. Kan confirmed that check_msr patch was to fix a bug report in guest, so it's ok to disable it for real HW. Reported-by:
Tom Vaden <tom.vaden@hpe.com> Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by:
-
Jiri Olsa authored
It's preffered to use group's ->is_visible callback, so we do not need to use condition attribute assignment. Suggested-by:
-
Kan Liang authored
IMC uncore unit can only be accessed via MMIO on Snow Ridge. The MMIO space of IMC uncore is at the specified offsets from the MEM0_BAR. Add snr_uncore_get_mc_dev() to locate the PCI device with MMIO_BASE and MEM0_BAR register. Add new ops to access the IMC registers via MMIO. Add 3 new free running counters for clocks, read and write bandwidth. Signed-off-by:
Kan Liang <kan.liang@linux.intel.com> Signed-off-by:
-
Kan Liang authored
The client IMC block is accessed by MMIO. Current code uses an informal way to access the block, which is not recommended. Clean up the code by using __iomem annotation and the accessor functions (read[lq]()). Move exit_box() and read_counter() to generic code, which can be shared with the server code later. Signed-off-by:
-
Kan Liang authored
A new MMIO type uncore box is introduced on Snow Ridge server. The counters of MMIO type uncore box can only be accessed by MMIO. Add a new uncore type, uncore_mmio_uncores, for MMIO type uncore blocks. Support MMIO type uncore blocks in CPU hot plug. The MMIO space has to be map/unmap for the first/last CPU. The context also need to be migrated if the bind CPU changes. Add mmio_init() to init and register PMUs for MMIO type uncore blocks. Add a helper to calculate the box_ctl address. The helpers which calculate ctl/ctr can be shared with PCI type uncore blocks. Signed-off-by:
-
Kan Liang authored
For uncore box which can only be accessed by MSR, its reference box->refcnt is updated in CPU hot plug. The uncore boxes need to be initalized and exited accordingly for the first/last CPU of a socket. Starts from Snow Ridge server, a new type of uncore box is introduced, which can only be accessed by MMIO. The driver needs to map/unmap MMIO space for the first/last CPU of a socket. Extract the codes of box ref/unref and init/exit for reuse later. There is no functional change. Signed-off-by:
-
Kan Liang authored
The uncore subsystem on Snow Ridge is similar as previous SKX server. The uncore units on Snow Ridge include Ubox, Chabox, IIO, IRP, M2PCIE, PCU, M2M, PCIE3 and IMC. - The config register encoding and pci device IDs are changed. - For CHA, the umask_ext and filter_tid fields are changed. - For IIO, the ch_mask and fc_mask fields are changed. - For M2M, the mask_ext field is changed. - Add new PCIe3 unit for PCIe3 root port which provides the interface between PCIe devices, plugged into the PCIe port, and the components (in M2IOSF). - IMC can only be accessed via MMIO on Snow Ridge now. Current common code doesn't support it yet. IMC will be supported in following patches. - There are 9 free running counters for IIO CLOCKS and bandwidth In. - Full uncore event list is not published yet. Event constrain is not included in this patch. It will be added later separately. Signed-off-by:
-
Kan Liang authored
Counting with invalid event coding for free-running counter may cause OOPs, e.g. uncore_iio_free_running_0/event=1/. Current code only validate the event with free-running event format, event=0xff,umask=0xXY. Non-free-running event format never be checked for the PMU with free-running counters. Add generic hw_config() to check and reject the invalid event coding for free-running PMU. Signed-off-by:
-
Kan Liang authored
Add new model number for Icelake desktop and server to perf. The data source encoding for Icelake server is the same as Skylake server. Signed-off-by:
-
Kan Liang authored
Add new Icelake desktop CPUID for RAPL, CSTATE and UNCORE. Signed-off-by:
-
Ingo Molnar authored
Signed-off-by:
-
- 14 Jun, 2019 2 commits
-
-
Aaron Lewis authored
Add the CPUID enumeration for Intel's de-feature bits to accommodate passing these de-features through to kvm guests. These de-features are (from SDM vol 1, section 8.1.8): - X86_FEATURE_FDP_EXCPTN_ONLY: If CPUID.(EAX=07H,ECX=0H):EBX[bit 6] = 1, the data pointer (FDP) is updated only for the x87 non-control instructions that incur unmasked x87 exceptions. - X86_FEATURE_ZERO_FCS_FDS: If CPUID.(EAX=07H,ECX=0H):EBX[bit 13] = 1, the processor deprecates FCS and FDS; it saves each as 0000H. Signed-off-by:
Aaron Lewis <aaronlewis@google.com> Signed-off-by:
Borislav Petkov <bp@suse.de> Reviewed-by:
Jim Mattson <jmattson@google.com> Cc: Fenghua Yu <fenghua.yu@intel.com> Cc: Frederic Weisbecker <frederic@kernel.org> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: marcorr@google.com Cc: Peter Feiner <pfeiner@google.com> Cc: pshier@google.com Cc: Robert Hoo <robert.hu@linux.intel.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Thomas Lendacky <Thomas.Lendacky@amd.com> Cc: x86-ml <x86@kernel.org> Link: https://lkml.kernel.org/r/20190605220252.103406-1-aaronlewis@google.com
-
Rajneesh Bhardwaj authored
Intel Ice Lake uncore support already included IMC PCI ID but ICL-NNPI CPUID is missing so add it to fix the probe function. Fixes: e39875d15ad6 ("perf/x86: add Intel Icelake uncore support") Signed-off-by:Rajneesh Bhardwaj <rajneesh.bhardwaj@linux.intel.com> Signed-off-by:
Thomas Gleixner <tglx@linutronix.de> Acked-by:
-
- 13 Jun, 2019 1 commit
-
-
Rajneesh Bhardwaj authored
Add the CPUID model number of Ice Lake Neural Network Processor for Deep Learning Inference (ICL-NNPI) to the Intel family list. Ice Lake NNPI uses model number 0x9D and this will be documented in a future version of Intel Software Development Manual. Signed-off-by:
-
- 10 Jun, 2019 17 commits
-
-
Arnaldo Carvalho de Melo authored
We have $INSTALL_DIR/share/perf-core/strace/groups/string files with syscalls that should be selected when 'string' is used, meaning, in this case, syscalls that receive as one of its arguments a string, like a pathname. But those were first selected and tested on x86_64, and end up failing in architectures where some of those syscalls are not available, like the 'access' syscall on arm64, which makes using 'perf trace -e string' in such archs to fail. Since this the routine doing the validation is used only when reading such files, do not fail when some syscall is not found in the syscalltbl, instead just use pr_debug() to register that in case people are suspicious of problems. Now using 'perf trace -e string' should work on arm64, selecting only the syscalls that have a string and are available on that architecture. Reported-by:
Leo Yan <leo.yan@linaro.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexei Starovoitov <ast@kernel.org> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Martin KaFai Lau <kafai@fb.com> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Song Liu <songliubraving@fb.com> Cc: Suzuki K Poulose <suzuki.poulose@arm.com> Cc: Yonghong Song <yhs@fb.com> Link: https://lkml.kernel.org/r/20190610184754.GU21245@kernel.orgSigned-off-by:
-
Thomas Richter authored
Perf report fails to display s390 specific event numbered bd000 on an x86 platform. For example on s390 this works without error: [root@m35lp76 perf]# uname -m s390x [root@m35lp76 perf]# ./perf record -e rbd000 -- find / >/dev/null [ perf record: Woken up 3 times to write data ] [ perf record: Captured and wrote 0.549 MB perf.data ] [root@m35lp76 perf]# ./perf report -D --stdio > /dev/null [root@m35lp76 perf]# Transfering this perf.data file to an x86 platform and executing the same report command produces: [root@f29 perf]# uname -m x86_64 [root@f29 perf]# ./perf report -i ~/perf.data.m35lp76 --stdio interpreting bpf_prog_info from systems with endianity is not yet supported interpreting btf from systems with endianity is not yet supported 0x8c890 [0x8]: failed to process type: 68 Error: failed to process sample Event bd000 generates auxiliary data which is stored in big endian format in the perf data file. This error is caused by missing endianess handling on the x86 platform when the data is displayed. Fix this by handling s390 auxiliary event data depending on the local platform endianness. Output after on x86: [root@f29 perf]# ./perf report -D -i ~/perf.data.m35lp76 --stdio > /dev/null interpreting bpf_prog_info from systems with endianity is not yet supported interpreting btf from systems with endianity is not yet supported [root@f29 perf]# Committer notes: Fix build breakage on older systems, such as CentOS:6 where using nesting calls to the endian.h macros end up redefining local variables: util/s390-cpumsf.c: In function 's390_cpumsf_trailer_show': util/s390-cpumsf.c:333: error: declaration of '__v' shadows a previous local util/s390-cpumsf.c:333: error: shadowed declaration is here util/s390-cpumsf.c:333: error: declaration of '__x' shadows a previous local util/s390-cpumsf.c:333: error: shadowed declaration is here util/s390-cpumsf.c:334: error: declaration of '__v' shadows a previous local util/s390-cpumsf.c:334: error: shadowed declaration is here util/s390-cpumsf.c:334: error: declaration of '__x' shadows a previous local util/s390-cpumsf.c:334: error: shadowed declaration is here [perfbuilder@455a63ef60dc perf]$ gcc -v |& tail -1 gcc version 4.4.7 20120313 (Red Hat 4.4.7-23) (GCC) [perfbuilder@455a63ef60dc perf]$ Since there are several uses of be64toh(te->flags) Introduce a variable to hold that and then use it, avoiding this case that causes the above problems: - local.bsdes = be16toh((be64toh(te->flags) >> 16 & 0xffff)); + local.bsdes = be16toh((flags >> 16 & 0xffff)); Its the same construct used in s390_cpumsf_diag_show() where we have a 'word' variable that is used just once, s390_cpumsf_basic_show() has lots of uses and also uses a variable to hold the result of be16toh(). Some of those temp variables needed to be converted from 'unsigned long' to 'unsigned long long' so as to build on 32-bit arches such as debian:experimental-x-mipsel, the android NDK ones and fedora:24-x-ARC-uClibc. Signed-off-by:
Thomas Richter <tmricht@linux.ibm.com> Reviewed-by:
Hendrik Brueckner <brueckner@linux.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Link: http://lkml.kernel.org/r/20190522064325.25596-1-tmricht@linux.ibm.comSigned-off-by:
-
Thomas Richter authored
Debugging a OOM error using the TUI interface revealed this issue on s390: [tmricht@m83lp54 perf]$ cat /proc/kallsyms |sort .... 00000001119b7158 B radix_tree_node_cachep 00000001119b8000 B __bss_stop 00000001119b8000 B _end 000003ff80002850 t autofs_mount [autofs4] 000003ff80002868 t autofs_show_options [autofs4] 000003ff80002a98 t autofs_evict_inode [autofs4] .... There is a huge gap between the last kernel symbol __bss_stop/_end and the first kernel module symbol autofs_mount (from autofs4 module). After reading the kernel symbol table via functions: dso__load() +--> dso__load_kernel_sym() +--> dso__load_kallsyms() +--> __dso_load_kallsyms() +--> symbols__fixup_end() the symbol __bss_stop has a start address of 1119b8000 and an end address of 3ff80002850, as can be seen by this debug statement: symbols__fixup_end __bss_stop start:0x1119b8000 end:0x3ff80002850 The size of symbol __bss_stop is 0x3fe6e64a850 bytes! It is the last kernel symbol and fills up the space until the first kernel module symbol. This size kills the TUI interface when executing the following code: process_sample_event() hist_entry_iter__add() hist_iter__report_callback() hist_entry__inc_addr_samples() symbol__inc_addr_samples(symbol = __bss_stop) symbol__cycles_hist() annotated_source__alloc_histograms(..., symbol__size(sym), ...) This function allocates memory to save sample histograms. The symbol_size() marco is defined as sym->end - sym->start, which results in above value of 0x3fe6e64a850 bytes and the call to calloc() in annotated_source__alloc_histograms() fails. The histgram memory allocation might fail, make this failure no-fatal and continue processing. Output before: [tmricht@m83lp54 perf]$ ./perf --debug stderr=1 report -vvvvv \ -i ~/slow.data 2>/tmp/2 [tmricht@m83lp54 perf]$ tail -5 /tmp/2 __symbol__inc_addr_samples(875): ENOMEM! sym->name=__bss_stop, start=0x1119b8000, addr=0x2aa0005eb08, end=0x3ff80002850, func: 0 problem adding hist entry, skipping event 0x938b8 [0x8]: failed to process type: 68 [Cannot allocate memory] [tmricht@m83lp54 perf]$ Output after: [tmricht@m83lp54 perf]$ ./perf --debug stderr=1 report -vvvvv \ -i ~/slow.data 2>/tmp/2 [tmricht@m83lp54 perf]$ tail -5 /tmp/2 symbol__inc_addr_samples map:0x1597830 start:0x110730000 end:0x3ff80002850 symbol__hists notes->src:0x2aa2a70 nr_hists:1 symbol__inc_addr_samples sym:unlink_anon_vmas src:0x2aa2a70 __symbol__inc_addr_samples: addr=0x11094c69e 0x11094c670 unlink_anon_vmas: period++ [addr: 0x11094c69e, 0x2e, evidx=0] => nr_samples: 1, period: 526008 [tmricht@m83lp54 perf]$ There is no error about failed memory allocation and the TUI interface shows all entries. Signed-off-by: -
Thomas Richter authored
Command # perf test -Fv 6 fails with error running test 100 'kvm-s390:kvm_s390_create_vm' failed to parse event 'kvm-s390:kvm_s390_create_vm', err -1, str 'unknown tracepoint' event syntax error: 'kvm-s390:kvm_s390_create_vm' \___ unknown tracepoint when the kvm module is not loaded or not built in. Fix this by adding a valid function which tests if the module is loaded. Loaded modules (or builtin KVM support) have a directory named /sys/kernel/debug/tracing/events/kvm-s390 for this tracepoint. Check for existence of this directory. Signed-off-by:Christian Borntraeger <borntraeger@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Link: http://lkml.kernel.org/r/20190604053504.43073-1-tmricht@linux.ibm.comSigned-off-by:
-
Adrian Hunter authored
Currently only a single explicit time range is accepted. Add support for multiple ranges separated by spaces, which requires the string to be quoted. Update the time utils test accordingly. Signed-off-by:
-
Adrian Hunter authored
Test time ranges work as expected. Committer testing: $ perf test "time utils" 59: time utils : Ok $ perf test -v "time utils" 59: time utils : --- start --- test child forked, pid 31711 parse_nsec_time("0") 0 parse_nsec_time("1") 1000000000 parse_nsec_time("0.000000001") 1 parse_nsec_time("1.000000001") 1000000001 parse_nsec_time("123456.123456") 123456123456000 parse_nsec_time("1234567.123456789") 1234567123456789 parse_nsec_time("18446744073.709551615") 18446744073709551615 perf_time__parse_str("1234567.123456789,1234567.123456789") start time 1234567123456789, end time 1234567123456789 perf_time__parse_str("1234567.123456789,1234567.123456790") start time 1234567123456789, end time 1234567123456790 perf_time__parse_str("1234567.123456789,") start time 1234567123456789, end time 0 perf_time__parse_str(",1234567.123456789") start time 0, end time 1234567123456789 perf_time__parse_str("0,1234567.123456789") start time 0, end time 1234567123456789 perf_time__parse_for_ranges("1234567.123456789,1234567.123456790") start time 1234567123456789, end time 1234567123456790 perf_time__parse_for_ranges("10%/1") first_sample_time 7654321000000000 last_sample_time 7654321000000100 start time 0: 7654321000000000, end time 0: 7654321000000009 perf_time__parse_for_ranges("10%/2") first_sample_time 7654321000000000 last_sample_time 7654321000000100 start time 0: 7654321000000010, end time 0: 7654321000000019 perf_time__parse_for_ranges("10%/1,10%/2") first_sample_time 11223344000000000 last_sample_time 11223344000000100 start time 0: 11223344000000000, end time 0: 11223344000000009 start time 1: 11223344000000010, end time 1: 11223344000000019 perf_time__parse_for_ranges("10%/1,10%/3,10%/10") first_sample_time 11223344000000000 last_sample_time 11223344000000100 start time 0: 11223344000000000, end time 0: 11223344000000009 start time 1: 11223344000000020, end time 1: 11223344000000029 start time 2: 11223344000000090, end time 2: 11223344000000100 test child finished with 0 ---- end ---- time utils: Ok $ Signed-off-by: -
Adrian Hunter authored
Explicit time ranges never contain a percent sign whereas percentage ranges always do, so it is possible to call the correct parser. Signed-off-by:
-
Adrian Hunter authored
Simplify perf_time__parse_for_ranges() error paths slightly. Signed-off-by:
-
Adrian Hunter authored
Correct some punctuation and spelling and correct the format to show that the time resolution is nanoseconds not microseconds. Signed-off-by:
-
Adrian Hunter authored
Prevent percentage time range overlap. This is only a 1 nanosecond change but makes the results more logical e.g. a sample cannot be in both the first 10% and the second 20%. Note, there is a later patch that adds a test for time-utils. Signed-off-by:
-
Adrian Hunter authored
Factor out set_percent_time() so it can be reused. Signed-off-by:
-
Adrian Hunter authored
Currently, options allow only 1 explicit (non-percentage) time range. In preparation for adding support for multiple explicit time ranges, treat time ranges consistently. Instead of treating some time ranges as inclusive and some as excluding the end time, treat all time ranges as inclusive. This is only a 1 nanosecond change but is necessary to treat multiple explicit time ranges in a consistent manner. Note, there is a later patch that adds a test for time-utils. Signed-off-by:
-
Adrian Hunter authored
Set up time ranges for efficient time interval filtering using the new "fast forward" facility. Because decoding is done in time order, intel_pt_time_filter() needs to look only at the next start or end timestamp - refer intel_pt_next_time(). Signed-off-by:
-
Adrian Hunter authored
Implement the lookahead callback to let the decoder access subsequent buffers. intel_pt_lookahead() manages the buffer lifetime and calls the decoder for each buffer until the decoder returns a non-zero value. Signed-off-by:
-
Adrian Hunter authored
Factor out intel_pt_get_buffer() so it can be reused. Signed-off-by:
-
Adrian Hunter authored
Intel PT decoding is done in time order. In order to support efficient time interval filtering, add a facility to "fast forward" towards a particular timestamp. That involves finding the right buffer, stepping to that buffer, and then stepping forward PSBs. Because decoding must begin at a PSB, "fast forward" stops at the last PSB that has a timestamp before the target timestamp. Signed-off-by:
-
Adrian Hunter authored
When the decoder gets the next trace buffer, some state is reset if the buffer is not consecutive to the previous buffer. Add a parameter 'reposition' so that can be done also to support a "fast forward" facility. Signed-off-by:
-
