| Message ID | 20220513090237.10444-4-adrian.hunter@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | perf intel-pt: Add support for tracing KVM test programs | expand |
Hi Adrian, On Fri, May 13, 2022 at 2:03 AM Adrian Hunter <adrian.hunter@intel.com> wrote: > > A common case for KVM test programs is that the guest object code can be > found in the hypervisor process (i.e. the test program running on the > host). To support that, copy the host thread's maps to the guest thread's > maps. Note, we do not discover the guest until we encounter a guest event, > which works well because it is not until then that we know that the host > thread's maps have been set up. > > Typically the main function for the guest object code is called > "guest_code", hence the name chosen for this feature. Ok, so that's just a convention and there's no hard-coded support for the "guest_code" function in this code, right? > > This is primarily aimed at supporting Intel PT, or similar, where trace > data can be recorded for a guest. Refer to the final patch in this series > "perf intel-pt: Add guest_code support" for an example. > > Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> > --- > tools/perf/util/event.c | 7 +++- > tools/perf/util/machine.c | 70 +++++++++++++++++++++++++++++++++++ > tools/perf/util/machine.h | 2 + > tools/perf/util/session.c | 7 ++++ > tools/perf/util/symbol_conf.h | 3 +- > 5 files changed, 86 insertions(+), 3 deletions(-) > > diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c > index 6439c888ae38..0476bb3a4188 100644 > --- a/tools/perf/util/event.c > +++ b/tools/perf/util/event.c > @@ -683,9 +683,12 @@ static bool check_address_range(struct intlist *addr_list, int addr_range, > int machine__resolve(struct machine *machine, struct addr_location *al, > struct perf_sample *sample) > { > - struct thread *thread = machine__findnew_thread(machine, sample->pid, > - sample->tid); > + struct thread *thread; > > + if (symbol_conf.guest_code && !machine__is_host(machine)) > + thread = machine__findnew_guest_code(machine, sample->pid); > + else > + thread = machine__findnew_thread(machine, sample->pid, sample->tid); > if (thread == NULL) > return -1; > > diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c > index e67b5a7670f3..ae2e1fb422e2 100644 > --- a/tools/perf/util/machine.c > +++ b/tools/perf/util/machine.c > @@ -392,6 +392,76 @@ struct machine *machines__find_guest(struct machines *machines, pid_t pid) > return machine; > } > > +/* > + * A common case for KVM test programs is that the guest object code can be > + * found in the hypervisor process (i.e. the test program running on the host). > + * To support that, copy the host thread's maps to the guest thread's maps. > + * Note, we do not discover the guest until we encounter a guest event, > + * which works well because it is not until then that we know that the host > + * thread's maps have been set up. > + */ > +static struct thread *findnew_guest_code(struct machine *machine, But this function returns a thread and IIUC that's the task which does the host to guest transition. Then why not calling it just findnew__hypervisor() ? Thanks, Namhyung > + struct machine *host_machine, > + pid_t pid) > +{ > + struct thread *host_thread; > + struct thread *thread; > + int err; > + > + if (!machine) > + return NULL; > + > + thread = machine__findnew_thread(machine, -1, pid); > + if (!thread) > + return NULL; > + > + /* Assume maps are set up if there are any */ > + if (thread->maps->nr_maps) > + return thread; > + > + host_thread = machine__find_thread(host_machine, -1, pid); > + if (!host_thread) > + goto out_err; > + > + thread__set_guest_comm(thread, pid); > + > + /* > + * Guest code can be found in hypervisor process at the same address > + * so copy host maps. > + */ > + err = maps__clone(thread, host_thread->maps); > + thread__put(host_thread); > + if (err) > + goto out_err; > + > + return thread; > + > +out_err: > + thread__zput(thread); > + return NULL; > +} > +
On 17/05/22 06:13, Namhyung Kim wrote: > Hi Adrian, > > On Fri, May 13, 2022 at 2:03 AM Adrian Hunter <adrian.hunter@intel.com> wrote: >> >> A common case for KVM test programs is that the guest object code can be >> found in the hypervisor process (i.e. the test program running on the >> host). To support that, copy the host thread's maps to the guest thread's >> maps. Note, we do not discover the guest until we encounter a guest event, >> which works well because it is not until then that we know that the host >> thread's maps have been set up. >> >> Typically the main function for the guest object code is called >> "guest_code", hence the name chosen for this feature. > > Ok, so that's just a convention and there's no hard-coded > support for the "guest_code" function in this code, right? That is correct. > >> >> This is primarily aimed at supporting Intel PT, or similar, where trace >> data can be recorded for a guest. Refer to the final patch in this series >> "perf intel-pt: Add guest_code support" for an example. >> >> Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> >> --- >> tools/perf/util/event.c | 7 +++- >> tools/perf/util/machine.c | 70 +++++++++++++++++++++++++++++++++++ >> tools/perf/util/machine.h | 2 + >> tools/perf/util/session.c | 7 ++++ >> tools/perf/util/symbol_conf.h | 3 +- >> 5 files changed, 86 insertions(+), 3 deletions(-) >> >> diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c >> index 6439c888ae38..0476bb3a4188 100644 >> --- a/tools/perf/util/event.c >> +++ b/tools/perf/util/event.c >> @@ -683,9 +683,12 @@ static bool check_address_range(struct intlist *addr_list, int addr_range, >> int machine__resolve(struct machine *machine, struct addr_location *al, >> struct perf_sample *sample) >> { >> - struct thread *thread = machine__findnew_thread(machine, sample->pid, >> - sample->tid); >> + struct thread *thread; >> >> + if (symbol_conf.guest_code && !machine__is_host(machine)) >> + thread = machine__findnew_guest_code(machine, sample->pid); >> + else >> + thread = machine__findnew_thread(machine, sample->pid, sample->tid); >> if (thread == NULL) >> return -1; >> >> diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c >> index e67b5a7670f3..ae2e1fb422e2 100644 >> --- a/tools/perf/util/machine.c >> +++ b/tools/perf/util/machine.c >> @@ -392,6 +392,76 @@ struct machine *machines__find_guest(struct machines *machines, pid_t pid) >> return machine; >> } >> >> +/* >> + * A common case for KVM test programs is that the guest object code can be >> + * found in the hypervisor process (i.e. the test program running on the host). >> + * To support that, copy the host thread's maps to the guest thread's maps. >> + * Note, we do not discover the guest until we encounter a guest event, >> + * which works well because it is not until then that we know that the host >> + * thread's maps have been set up. >> + */ >> +static struct thread *findnew_guest_code(struct machine *machine, > > But this function returns a thread and IIUC that's the task which > does the host to guest transition. Then why not calling it just > findnew__hypervisor() ? The thread returned is in the guest machine. While the code comes from the hypervisor, it is in the guest VM when it runs. From Intel PT point of view, this function allows finding the guest object code by setting up the guest thread and its maps. I will try to improve on the explanation in V2. > > Thanks, > Namhyung > > >> + struct machine *host_machine, >> + pid_t pid) >> +{ >> + struct thread *host_thread; >> + struct thread *thread; >> + int err; >> + >> + if (!machine) >> + return NULL; >> + >> + thread = machine__findnew_thread(machine, -1, pid); >> + if (!thread) >> + return NULL; >> + >> + /* Assume maps are set up if there are any */ >> + if (thread->maps->nr_maps) >> + return thread; >> + >> + host_thread = machine__find_thread(host_machine, -1, pid); >> + if (!host_thread) >> + goto out_err; >> + >> + thread__set_guest_comm(thread, pid); >> + >> + /* >> + * Guest code can be found in hypervisor process at the same address >> + * so copy host maps. >> + */ >> + err = maps__clone(thread, host_thread->maps); >> + thread__put(host_thread); >> + if (err) >> + goto out_err; >> + >> + return thread; >> + >> +out_err: >> + thread__zput(thread); >> + return NULL; >> +} >> +
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 6439c888ae38..0476bb3a4188 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -683,9 +683,12 @@ static bool check_address_range(struct intlist *addr_list, int addr_range, int machine__resolve(struct machine *machine, struct addr_location *al, struct perf_sample *sample) { - struct thread *thread = machine__findnew_thread(machine, sample->pid, - sample->tid); + struct thread *thread; + if (symbol_conf.guest_code && !machine__is_host(machine)) + thread = machine__findnew_guest_code(machine, sample->pid); + else + thread = machine__findnew_thread(machine, sample->pid, sample->tid); if (thread == NULL) return -1; diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index e67b5a7670f3..ae2e1fb422e2 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -392,6 +392,76 @@ struct machine *machines__find_guest(struct machines *machines, pid_t pid) return machine; } +/* + * A common case for KVM test programs is that the guest object code can be + * found in the hypervisor process (i.e. the test program running on the host). + * To support that, copy the host thread's maps to the guest thread's maps. + * Note, we do not discover the guest until we encounter a guest event, + * which works well because it is not until then that we know that the host + * thread's maps have been set up. + */ +static struct thread *findnew_guest_code(struct machine *machine, + struct machine *host_machine, + pid_t pid) +{ + struct thread *host_thread; + struct thread *thread; + int err; + + if (!machine) + return NULL; + + thread = machine__findnew_thread(machine, -1, pid); + if (!thread) + return NULL; + + /* Assume maps are set up if there are any */ + if (thread->maps->nr_maps) + return thread; + + host_thread = machine__find_thread(host_machine, -1, pid); + if (!host_thread) + goto out_err; + + thread__set_guest_comm(thread, pid); + + /* + * Guest code can be found in hypervisor process at the same address + * so copy host maps. + */ + err = maps__clone(thread, host_thread->maps); + thread__put(host_thread); + if (err) + goto out_err; + + return thread; + +out_err: + thread__zput(thread); + return NULL; +} + +struct thread *machines__findnew_guest_code(struct machines *machines, pid_t pid) +{ + struct machine *host_machine = machines__find(machines, HOST_KERNEL_ID); + struct machine *machine = machines__findnew(machines, pid); + + return findnew_guest_code(machine, host_machine, pid); +} + +struct thread *machine__findnew_guest_code(struct machine *machine, pid_t pid) +{ + struct machines *machines = machine->machines; + struct machine *host_machine; + + if (!machines) + return NULL; + + host_machine = machines__find(machines, HOST_KERNEL_ID); + + return findnew_guest_code(machine, host_machine, pid); +} + void machines__process_guests(struct machines *machines, machine__process_t process, void *data) { diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h index 0d113771e8c8..01a5fca643b7 100644 --- a/tools/perf/util/machine.h +++ b/tools/perf/util/machine.h @@ -168,6 +168,8 @@ struct machine *machines__find_host(struct machines *machines); struct machine *machines__find(struct machines *machines, pid_t pid); struct machine *machines__findnew(struct machines *machines, pid_t pid); struct machine *machines__find_guest(struct machines *machines, pid_t pid); +struct thread *machines__findnew_guest_code(struct machines *machines, pid_t pid); +struct thread *machine__findnew_guest_code(struct machine *machine, pid_t pid); void machines__set_id_hdr_size(struct machines *machines, u16 id_hdr_size); void machines__set_comm_exec(struct machines *machines, bool comm_exec); diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index f9a320694b85..6577e1227bd5 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1410,6 +1410,13 @@ static struct machine *machines__find_for_cpumode(struct machines *machines, else pid = sample->pid; + /* + * Guest code machine is created as needed and does not use + * DEFAULT_GUEST_KERNEL_ID. + */ + if (symbol_conf.guest_code) + return machines__findnew(machines, pid); + return machines__find_guest(machines, pid); } diff --git a/tools/perf/util/symbol_conf.h b/tools/perf/util/symbol_conf.h index a70b3ec09dac..bc3d046fbb63 100644 --- a/tools/perf/util/symbol_conf.h +++ b/tools/perf/util/symbol_conf.h @@ -43,7 +43,8 @@ struct symbol_conf { report_individual_block, inline_name, disable_add2line_warn, - buildid_mmap2; + buildid_mmap2, + guest_code; const char *vmlinux_name, *kallsyms_name, *source_prefix,
A common case for KVM test programs is that the guest object code can be found in the hypervisor process (i.e. the test program running on the host). To support that, copy the host thread's maps to the guest thread's maps. Note, we do not discover the guest until we encounter a guest event, which works well because it is not until then that we know that the host thread's maps have been set up. Typically the main function for the guest object code is called "guest_code", hence the name chosen for this feature. This is primarily aimed at supporting Intel PT, or similar, where trace data can be recorded for a guest. Refer to the final patch in this series "perf intel-pt: Add guest_code support" for an example. Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> --- tools/perf/util/event.c | 7 +++- tools/perf/util/machine.c | 70 +++++++++++++++++++++++++++++++++++ tools/perf/util/machine.h | 2 + tools/perf/util/session.c | 7 ++++ tools/perf/util/symbol_conf.h | 3 +- 5 files changed, 86 insertions(+), 3 deletions(-)