* [PATCH RESEND v4 1/2] perf/kvm: Port perf kvm to powerpc @ 2015-06-15 5:05 Hemant Kumar 2015-06-15 5:05 ` [PATCH RESEND v4 2/2] perf/kvm: Support HCALL events Hemant Kumar 0 siblings, 1 reply; 2+ messages in thread From: Hemant Kumar @ 2015-06-15 5:05 UTC (permalink / raw) To: linuxppc-dev Cc: linux-kernel, acme, mingo, mpe, sukadev, maddy, warrier, srikar, paulus, scottwood, Hemant Kumar From: Srikar Dronamraju <srikar@linux.vnet.ibm.com> perf kvm can be used to analyze guest exit reasons. This support already exists in x86. Hence, porting it to powerpc. - To trace KVM events : perf kvm stat record If many guests are running, we can track for a specific guest by using --pid as in : perf kvm stat record --pid <pid> - To see the results : perf kvm stat report The result shows the number of exits (from the guest context to host/hypervisor context) grouped by their respective exit reasons with their frequency. This patch makes use of the guest exit reasons available in "trace_book3s.h". It records on two already available tracepoints : "kvm_hv:kvm_guest_exit" and "kvm_hv:kvm_guest_enter". Note : This patch has a dependency on the patch "kvm/powerpc: Export kvm exit reasons" which exports the KVM exit reasons through the uapi. Here is a sample o/p: # pgrep qemu 19378 60515 2 Guests are running on the host. # perf kvm stat record -a ^C[ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 4.153 MB perf.data.guest (39624 samples) ] # perf kvm stat report -p 60515 Analyze events for pid(s) 60515, all VCPUs: VM-EXIT Samples Samples% Time% Min Time Max Time Avg time H_DATA_STORAGE 5006 35.30% 0.13% 1.94us 49.46us 12.37us ( +- 0.52% ) HV_DECREMENTER 4457 31.43% 0.02% 0.72us 16.14us 1.91us ( +- 0.96% ) SYSCALL 2690 18.97% 0.10% 2.84us 528.24us 18.29us ( +- 3.75% ) RETURN_TO_HOST 1789 12.61% 99.76% 1.58us 672791.91us 27470.23us ( +- 3.00% ) EXTERNAL 240 1.69% 0.00% 0.69us 10.67us 1.33us ( +- 5.34% ) Total Samples:14182, Total events handled time:49264158.30us. Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Signed-off-by: Hemant Kumar <hemant@linux.vnet.ibm.com> --- Patch Resend : - Added few more people to cc list. - Rebased it to current tip. Changes : - Moved the uapi related changes to arch/powerpc patchset. This patch has a dependency on : http://www.mail-archive.com/linuxppc-dev@lists.ozlabs.org/msg90506.html which exports the kvm exit reasons. tools/perf/arch/powerpc/Makefile | 1 + tools/perf/arch/powerpc/util/Build | 1 + tools/perf/arch/powerpc/util/kvm-stat.c | 33 +++++++++++++++++++++++++++++++++ 3 files changed, 35 insertions(+) create mode 100644 tools/perf/arch/powerpc/util/kvm-stat.c diff --git a/tools/perf/arch/powerpc/Makefile b/tools/perf/arch/powerpc/Makefile index 7fbca17..21322e0 100644 --- a/tools/perf/arch/powerpc/Makefile +++ b/tools/perf/arch/powerpc/Makefile @@ -1,3 +1,4 @@ ifndef NO_DWARF PERF_HAVE_DWARF_REGS := 1 endif +HAVE_KVM_STAT_SUPPORT := 1 diff --git a/tools/perf/arch/powerpc/util/Build b/tools/perf/arch/powerpc/util/Build index 7b8b0d1..c8fe207 100644 --- a/tools/perf/arch/powerpc/util/Build +++ b/tools/perf/arch/powerpc/util/Build @@ -1,5 +1,6 @@ libperf-y += header.o libperf-y += sym-handling.o +libperf-y += kvm-stat.o libperf-$(CONFIG_DWARF) += dwarf-regs.o libperf-$(CONFIG_DWARF) += skip-callchain-idx.o diff --git a/tools/perf/arch/powerpc/util/kvm-stat.c b/tools/perf/arch/powerpc/util/kvm-stat.c new file mode 100644 index 0000000..24e06bf --- /dev/null +++ b/tools/perf/arch/powerpc/util/kvm-stat.c @@ -0,0 +1,33 @@ +#include "../../util/kvm-stat.h" +#include <asm/kvm_perf_book3s.h> + +define_exit_reasons_table(hv_exit_reasons, kvm_trace_symbol_exit); + +static struct kvm_events_ops exit_events = { + .is_begin_event = exit_event_begin, + .is_end_event = exit_event_end, + .decode_key = exit_event_decode_key, + .name = "VM-EXIT" +}; + +const char *const kvm_events_tp[] = { + "kvm_hv:kvm_guest_exit", + "kvm_hv:kvm_guest_enter", + NULL, +}; + +struct kvm_reg_events_ops kvm_reg_events_ops[] = { + { .name = "vmexit", .ops = &exit_events }, + { NULL, NULL }, +}; + +const char * const kvm_skip_events[] = { + NULL, +}; + +int cpu_isa_init(struct perf_kvm_stat *kvm, const char *cpuid __maybe_unused) +{ + kvm->exit_reasons = hv_exit_reasons; + kvm->exit_reasons_isa = "HV"; + return 0; +} -- 1.9.3 ^ permalink raw reply related [flat|nested] 2+ messages in thread
* [PATCH RESEND v4 2/2] perf/kvm: Support HCALL events 2015-06-15 5:05 [PATCH RESEND v4 1/2] perf/kvm: Port perf kvm to powerpc Hemant Kumar @ 2015-06-15 5:05 ` Hemant Kumar 0 siblings, 0 replies; 2+ messages in thread From: Hemant Kumar @ 2015-06-15 5:05 UTC (permalink / raw) To: linuxppc-dev Cc: linux-kernel, acme, mingo, mpe, sukadev, maddy, warrier, srikar, paulus, scottwood, Hemant Kumar powerpc provides hcall events that also provides insights into guest behaviour. Enhance perf kvm to record and analyze hcall events. - To trace hcall events : perf kvm stat record - To show the results : perf kvm stat report --event=hcall The result shows the number of hypervisor calls from the guest grouped by their respective reasons displayed with the frequency. This patch makes use of two additional tracepoints "kvm_hv:kvm_hcall_enter" and "kvm_hv:kvm_hcall_exit". It uses the pSeries hypervisor codes exported through uapi to classify the hcalls into their respective reasons. Note : This patch has a dependency on "kvm/powerpc: Export HCALL reason codes" which exports HCALL reasons through uapi. # pgrep qemu A sample output : 19378 60515 2 VMs running. # perf kvm stat record -a ^C[ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 4.153 MB perf.data.guest (39624 samples) ] # perf kvm stat report -p 60515 --event=hcall Analyze events for pid(s) 60515, all VCPUs: HCALL-EVENT Samples Samples% Time% Min Time Max Time Avg time H_VIO_SIGNAL 1034 38.44% 15.77% 0.36us 1.59us 0.44us ( +- 0.66% ) H_SEND_CRQ 652 24.24% 10.97% 0.39us 1.84us 0.49us ( +- 1.20% ) H_IPI 523 19.44% 62.05% 1.35us 19.70us 3.44us ( +- 2.88% ) H_PUT_TERM_CHAR 411 15.28% 8.03% 0.38us 3.77us 0.57us ( +- 1.61% ) H_GET_TERM_CHAR 50 1.86% 0.99% 0.40us 0.98us 0.57us ( +- 3.37% ) H_EOI 20 0.74% 2.19% 2.22us 4.72us 3.17us ( +- 5.96% ) Total Samples:2690, Total events handled time:2896.94us. Signed-off-by: Hemant Kumar <hemant@linux.vnet.ibm.com> --- Changes: - Moved the uapi related changes to arch/powerpc side patchset. This patch has a dependency on : http://www.mail-archive.com/linuxppc-dev@lists.ozlabs.org/msg90507.html which export hcall reasons through uapi. tools/perf/arch/powerpc/util/kvm-stat.c | 61 +++++++++++++++++++++++++++++++++ 1 file changed, 61 insertions(+) diff --git a/tools/perf/arch/powerpc/util/kvm-stat.c b/tools/perf/arch/powerpc/util/kvm-stat.c index 24e06bf..0d3ea47 100644 --- a/tools/perf/arch/powerpc/util/kvm-stat.c +++ b/tools/perf/arch/powerpc/util/kvm-stat.c @@ -1,7 +1,9 @@ #include "../../util/kvm-stat.h" #include <asm/kvm_perf_book3s.h> +#include "../../util/debug.h" define_exit_reasons_table(hv_exit_reasons, kvm_trace_symbol_exit); +define_exit_reasons_table(hcall_reasons, kvm_trace_symbol_hcall); static struct kvm_events_ops exit_events = { .is_begin_event = exit_event_begin, @@ -10,14 +12,73 @@ static struct kvm_events_ops exit_events = { .name = "VM-EXIT" }; +static void hcall_event_get_key(struct perf_evsel *evsel, + struct perf_sample *sample, + struct event_key *key) +{ + key->info = 0; + key->key = perf_evsel__intval(evsel, sample, KVM_HCALL_REASON); +} + +static const char *get_exit_reason(u64 exit_code) +{ + struct exit_reasons_table *tbl = hcall_reasons; + + while (tbl->reason != NULL) { + if (tbl->exit_code == exit_code) + return tbl->reason; + tbl++; + } + + pr_err("Unknown kvm hcall exit code: %lld\n", + (unsigned long long)exit_code); + return "UNKNOWN"; +} + +static bool hcall_event_end(struct perf_evsel *evsel, + struct perf_sample *sample __maybe_unused, + struct event_key *key __maybe_unused) +{ + return (!strcmp(evsel->name, KVM_HCALL_EXIT_TRACE)); +} + +static bool hcall_event_begin(struct perf_evsel *evsel, + struct perf_sample *sample, struct event_key *key) +{ + if (!strcmp(evsel->name, KVM_HCALL_ENTRY_TRACE)) { + hcall_event_get_key(evsel, sample, key); + return true; + } + + return false; +} +static void hcall_event_decode_key(struct perf_kvm_stat *kvm __maybe_unused, + struct event_key *key, + char *decode) +{ + const char *hcall_reason = get_exit_reason(key->key); + + scnprintf(decode, DECODE_STR_LEN, "%s", hcall_reason); +} + +static struct kvm_events_ops hcall_events = { + .is_begin_event = hcall_event_begin, + .is_end_event = hcall_event_end, + .decode_key = hcall_event_decode_key, + .name = "HCALL-EVENT", +}; + const char *const kvm_events_tp[] = { "kvm_hv:kvm_guest_exit", "kvm_hv:kvm_guest_enter", + "kvm_hv:kvm_hcall_enter", + "kvm_hv:kvm_hcall_exit", NULL, }; struct kvm_reg_events_ops kvm_reg_events_ops[] = { { .name = "vmexit", .ops = &exit_events }, + { .name = "hcall", .ops = &hcall_events }, { NULL, NULL }, }; -- 1.9.3 ^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2015-06-15 5:07 UTC | newest] Thread overview: 2+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2015-06-15 5:05 [PATCH RESEND v4 1/2] perf/kvm: Port perf kvm to powerpc Hemant Kumar 2015-06-15 5:05 ` [PATCH RESEND v4 2/2] perf/kvm: Support HCALL events Hemant Kumar
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).