| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <74b01cee-12dc-2a95-9817-4f89a3fbd441@linux.intel.com>
Date: Thu, 5 Mar 2020 16:02:10 -0500
From: "Liang, Kan" <kan.liang@...ux.intel.com>
To: Arnaldo Carvalho de Melo <arnaldo.melo@...il.com>
Cc: jolsa@...hat.com, peterz@...radead.org, mingo@...hat.com,
linux-kernel@...r.kernel.org, namhyung@...nel.org,
adrian.hunter@...el.com, mathieu.poirier@...aro.org,
ravi.bangoria@...ux.ibm.com, alexey.budankov@...ux.intel.com,
vitaly.slobodskoy@...el.com, pavel.gerasimov@...el.com,
mpe@...erman.id.au, eranian@...gle.com, ak@...ux.intel.com
Subject: Re: [PATCH 02/12] perf tools: Support PERF_SAMPLE_BRANCH_HW_INDEX
On 3/5/2020 3:25 PM, Arnaldo Carvalho de Melo wrote:
> Em Fri, Feb 28, 2020 at 08:30:01AM -0800, kan.liang@...ux.intel.com escreveu:
>> From: Kan Liang <kan.liang@...ux.intel.com>
>>
>> A new branch sample type PERF_SAMPLE_BRANCH_HW_INDEX has been introduced
>> in latest kernel.
>>
>> Enable HW_INDEX by default in LBR call stack mode.
>> If kernel doesn't support the sample type, switching it off.
>>
>> Add HW_INDEX in attr_fprintf as well. User can check whether the branch
>> sample type is set via debug information or header.
>
> So while this works with a kernel where PERF_SAMPLE_BRANCH_HW_INDEX is
> present and we get, from the committer notes I was putting together
> while testing/applying this cset:
>
> First collect some samples with LBR callchains, system wide, for a few
> seconds:
>
> # perf record --call-graph lbr -a sleep 5
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.625 MB perf.data (224 samples) ]
> #
>
> Now lets use 'perf evlist -v' to look at the branch_sample_type:
>
> # perf evlist -v
> cycles: size: 120, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
> #
>
> So the machine has the kernel feature, and it was correctly added to
> perf_event_attr.branch_sample_type, for the default 'cycles' event.
>
> Cool, and look at that 'attr.precise_ip: 3' part, the kernel is OK with
> having that together with attr.branch_sample_type with HW_INDEX set.
>
> The problem happens when I go test this in an older kernel, where the
> kernel doesn't know about HW_INDEX, we get it disabled but then
> precise_ip is set to zero in its detection , even if at the end we get
> it to 3, as expected, which got me a bit confused, I'll investigate this
> a bit more to try and avoid these extra probes for the max precise level
> that fails in older kernels due to branch_sample_type having HW_INDEX
> :-\
It looks like this is an expected behavior for the event with maximum
precise config for current perf tool.
The related commits are as below:
commit ID: 4e8a5c155137 ("perf evsel: Fix max perf_event_attr.precise_ip
detection")
commit ID: cd136189370c ("perf evsel: Do not rely on errno values for
precise_ip fallback")
Before handling any standard fallback (not just HW_INDEX), perf tool
will try all the precise_ip value first.
Thanks,
Kan
>
>
> # perf record -vv --call-graph lbr -a sleep 5
> <SNIP>
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> task 1
> precise_ip 3
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> branch_sample_type USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8
> sys_perf_event_open failed, error -95
> decreasing precise_ip by one (2)
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> task 1
> precise_ip 2
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> branch_sample_type USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8
> sys_perf_event_open failed, error -95
> decreasing precise_ip by one (1)
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> task 1
> precise_ip 1
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> branch_sample_type USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8
> sys_perf_event_open failed, error -95
> decreasing precise_ip by one (0)
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> task 1
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> branch_sample_type USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8
> sys_perf_event_open failed, error -22
> switching off branch HW index support
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> task 1
> precise_ip 3
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> branch_sample_type USER|CALL_STACK|NO_FLAGS|NO_CYCLES
> ------------------------------------------------------------
>
>
>
>> Signed-off-by: Kan Liang <kan.liang@...ux.intel.com>
>> ---
>> tools/perf/util/evsel.c | 15 ++++++++++++---
>> tools/perf/util/evsel.h | 1 +
>> tools/perf/util/perf_event_attr_fprintf.c | 1 +
>> 3 files changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
>> index 05883a45de5b..816d930d774e 100644
>> --- a/tools/perf/util/evsel.c
>> +++ b/tools/perf/util/evsel.c
>> @@ -712,7 +712,8 @@ static void __perf_evsel__config_callchain(struct evsel *evsel,
>> attr->branch_sample_type = PERF_SAMPLE_BRANCH_USER |
>> PERF_SAMPLE_BRANCH_CALL_STACK |
>> PERF_SAMPLE_BRANCH_NO_CYCLES |
>> - PERF_SAMPLE_BRANCH_NO_FLAGS;
>> + PERF_SAMPLE_BRANCH_NO_FLAGS |
>> + PERF_SAMPLE_BRANCH_HW_INDEX;
>> }
>> } else
>> pr_warning("Cannot use LBR callstack with branch stack. "
>> @@ -763,7 +764,8 @@ perf_evsel__reset_callgraph(struct evsel *evsel,
>> if (param->record_mode == CALLCHAIN_LBR) {
>> perf_evsel__reset_sample_bit(evsel, BRANCH_STACK);
>> attr->branch_sample_type &= ~(PERF_SAMPLE_BRANCH_USER |
>> - PERF_SAMPLE_BRANCH_CALL_STACK);
>> + PERF_SAMPLE_BRANCH_CALL_STACK |
>> + PERF_SAMPLE_BRANCH_HW_INDEX);
>> }
>> if (param->record_mode == CALLCHAIN_DWARF) {
>> perf_evsel__reset_sample_bit(evsel, REGS_USER);
>> @@ -1673,6 +1675,8 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
>> evsel->core.attr.ksymbol = 0;
>> if (perf_missing_features.bpf)
>> evsel->core.attr.bpf_event = 0;
>> + if (perf_missing_features.branch_hw_idx)
>> + evsel->core.attr.branch_sample_type &= ~PERF_SAMPLE_BRANCH_HW_INDEX;
>> retry_sample_id:
>> if (perf_missing_features.sample_id_all)
>> evsel->core.attr.sample_id_all = 0;
>> @@ -1784,7 +1788,12 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
>> * Must probe features in the order they were added to the
>> * perf_event_attr interface.
>> */
>> - if (!perf_missing_features.aux_output && evsel->core.attr.aux_output) {
>> + if (!perf_missing_features.branch_hw_idx &&
>> + (evsel->core.attr.branch_sample_type & PERF_SAMPLE_BRANCH_HW_INDEX)) {
>> + perf_missing_features.branch_hw_idx = true;
>> + pr_debug2("switching off branch HW index support\n");
>> + goto fallback_missing_features;
>> + } else if (!perf_missing_features.aux_output && evsel->core.attr.aux_output) {
>> perf_missing_features.aux_output = true;
>> pr_debug2_peo("Kernel has no attr.aux_output support, bailing out\n");
>> goto out_close;
>> diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
>> index 99a0cb60c556..33804740e2ca 100644
>> --- a/tools/perf/util/evsel.h
>> +++ b/tools/perf/util/evsel.h
>> @@ -119,6 +119,7 @@ struct perf_missing_features {
>> bool ksymbol;
>> bool bpf;
>> bool aux_output;
>> + bool branch_hw_idx;
>> };
>>
>> extern struct perf_missing_features perf_missing_features;
>> diff --git a/tools/perf/util/perf_event_attr_fprintf.c b/tools/perf/util/perf_event_attr_fprintf.c
>> index 651203126c71..355d3458d4e6 100644
>> --- a/tools/perf/util/perf_event_attr_fprintf.c
>> +++ b/tools/perf/util/perf_event_attr_fprintf.c
>> @@ -50,6 +50,7 @@ static void __p_branch_sample_type(char *buf, size_t size, u64 value)
>> bit_name(ABORT_TX), bit_name(IN_TX), bit_name(NO_TX),
>> bit_name(COND), bit_name(CALL_STACK), bit_name(IND_JUMP),
>> bit_name(CALL), bit_name(NO_FLAGS), bit_name(NO_CYCLES),
>> + bit_name(HW_INDEX),
>> { .name = NULL, }
>> };
>> #undef bit_name
>> --
>> 2.17.1
>>
>
Powered by blists - more mailing lists