perf/s390: Regression: Move uid filtering to BPF filters
PATCH v2 1/1 perf/s390: Regression: Move uid filtering to BPF filters
V1 --> V2: Added Jiri Olsa's suggestion and introduced
member bpf_perf_event_opts::no_ioctl_enable.
On linux-next
commit b4c658d4d63d61 ("perf target: Remove uid from target") introduces a regression on s390. In fact the regression exists on all platforms when the event supports auxiliary data gathering.
Command
# ./perf record -u 0 -aB --synth=no -- ./perf test -w thloop [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.011 MB perf.data ] # ./perf report --stats | grep SAMPLE #
does not generate samples in the perf.data file.
On x86 command
# sudo perf record -e intel_pt// -u 0 ls is broken too.
Looking at the sequence of calls in 'perf record' reveals this behavior:
1. The event 'cycles' is created and enabled:
record__open()
+-> evlist__apply_filters()
+-> perf_bpf_filter__prepare()
+-> bpf_program.attach_perf_event()
+-> bpf_program.attach_perf_event_opts()
+-> __GI___ioctl(..., PERF_EVENT_IOC_ENABLE, ...)The event 'cycles' is enabled and active now. However the event's ring-buffer to store the samples generated by hardware is not allocated yet. This happens now after enabling the event:
2. The event's fd is mmap() to create the ring buffer:
record__open()
+-> record__mmap()
+-> record__mmap_evlist()
+-> evlist__mmap_ex()
+-> perf_evlist__mmap_ops()
+-> mmap_per_cpu()
+-> mmap_per_evsel()
+-> mmap__mmap()
+-> perf_mmap__mmap()
+-> mmap() This allocates the ring-buffer for the event 'cycles'. With mmap() the kernel creates the ring buffer:
perf_mmap(): kernel function to create the event's ring
| buffer to save the sampled data.
|
+-> ring_buffer_attach(): Allocates memory for ring buffer.
| The PMU has auxiliary data setup function. The
| has_aux(event) condition is true and the PMU's
| stop() is called to stop sampling. It is not
| restarted:
| if (has_aux(event))
| perf_event_stop(event, 0);
|
+-> cpumsf_pmu_stop():Hardware sampling is stopped. No samples are generated and saved anymore.
3. After the event 'cycles' has been mapped, the event is enabled a second time in:
__cmd_record()
+-> evlist__enable()
+-> __evlist__enable()
+-> evsel__enable_cpu()
+-> perf_evsel__enable_cpu()
+-> perf_evsel__run_ioctl()
+-> perf_evsel__ioctl()
+-> __GI___ioctl(., PERF_EVENT_IOC_ENABLE, .)
The second
ioctl(fd, PERF_EVENT_IOC_ENABLE, 0);
is just a NOP in this case. The first invocation in (1.) sets the
event::state to PERF_EVENT_STATE_ACTIVE. The kernel functions
perf_ioctl()
+-> _perf_ioctl()
+-> _perf_event_enable()
+-> __perf_event_enable() returns immediately because event::state is already set to PERF_EVENT_STATE_ACTIVE.
This happens on s390, because the event 'cycles' offers the possibility to save auxilary data. The PMU call backs setup_aux() and free_aux() are defined. Without both call back functions, cpumsf_pmu_stop() is not invoked and sampling continues.
To remedy this, remove the first invocation of
ioctl(..., PERF_EVENT_IOC_ENABLE, ...).
in step (1.) Create the event in step (1.) and enable it in step (3.) after the ring buffer has been mapped.
Make the change backward compatible and introduce a new structure member bpf_perf_event_opts::no_ioctl_enable. It defaults to false and only bpf_program__attach_perf_event() sets it to true. This way only perf tool invocation do not enable the sampling event.
Output after:
# ./perf record -aB --synth=no -u 0 -- ./perf test -w thloop 2
[ perf record: Woken up 3 times to write data ]
[ perf record: Captured and wrote 0.876 MB perf.data ]
# ./perf report --stats | grep SAMPLE
SAMPLE events: 16200 (99.5%)
SAMPLE events: 16200
#The software event succeeded before and after the patch:
# ./perf record -e cpu-clock -aB --synth=no -u 0 -- \
./perf test -w thloop 2
[ perf record: Woken up 7 times to write data ]
[ perf record: Captured and wrote 2.870 MB perf.data ]
# ./perf report --stats | grep SAMPLE
SAMPLE events: 53506 (99.8%)
SAMPLE events: 53506
#To: Andrii Nakryiko <[email protected]>
To: Ian Rogers <[email protected]>
To: Ilya Leoshkevich <[email protected]>
--- tools/lib/bpf/libbpf.c | 19 +++++++++++++------ tools/lib/bpf/libbpf.h | 3 ++- 2 files changed, 15 insertions(+), 7 deletions(-)
tools/lib/bpf/libbpf.c +13 -6
@@ -10911,6 +10911,7 @@ struct bpf_link *bpf_program__attach_perf_event_opts(const struct bpf_program *pstruct bpf_link_perf *link;int prog_fd, link_fd =-1 , err;bool force_ioctl_attach; +bool no_ioctl_enable;if (!OPTS_VALID (opts, bpf_perf_event_opts))return libbpf_err_ptr (-EINVAL );
@@ -10965,11 +10966,14 @@ struct bpf_link *bpf_program__attach_perf_event_opts(const struct bpf_program *p } link->link.fd = pfd; } -if (ioctl (pfd,PERF_EVENT_IOC_ENABLE ,0 ) <0 ) { - err = -errno; -pr_warn ("prog '%s': failed to enable perf_event FD %d: %s\n" , - prog->name, pfd,errstr (err)); - goto err_out; + no_ioctl_enable =OPTS_GET (opts, no_ioctl_enable, false); +if (!no_ioctl_enable) { +if (ioctl (pfd,PERF_EVENT_IOC_ENABLE ,0 ) <0 ) { + err = -errno; +pr_warn ("prog '%s': failed to enable perf_event FD %d: %s\n" , + prog->name, pfd,errstr (err)); + goto err_out; + } }return &link->link;
@@ -10982,7 +10986,10 @@ struct bpf_link *bpf_program__attach_perf_event_opts(const struct bpf_program *pstruct bpf_link *bpf_program__attach_perf_event (const struct bpf_program *prog,int pfd) { -return bpf_program__attach_perf_event_opts (prog, pfd,NULL ); +DECLARE_LIBBPF_OPTS (bpf_perf_event_opts, pe_opts); + + pe_opts.no_ioctl_enable = true; +return bpf_program__attach_perf_event_opts (prog, pfd, &pe_opts); }/*
tools/lib/bpf/libbpf.h +2 -1
@@ -499,9 +499,10 @@ struct bpf_perf_event_opts {__u64 bpf_cookie;/* don't use BPF link when attach BPF program */ bool force_ioctl_attach; +bool no_ioctl_enable; size_t :0; }; -#define bpf_perf_event_opts__last_field force_ioctl_attach +#define bpf_perf_event_opts__last_field no_ioctl_enable LIBBPF_API struct bpf_link *bpf_program__attach_perf_event (const struct bpf_program *prog,int pfd);
Signature · 1 line
2.50.1
Quoted 127 lines · Thomas Richter
V1 --> V2: Added Jiri Olsa's suggestion and introduced member
bpf_perf_event_opts::no_ioctl_enable.On linux-next
commitb4c658d4d63d61("perf target: Remove uid from target") introduces a regression on s390. In fact the regression exists on all platforms when the event supports auxiliary data gathering.Command
# ./perf record -u 0 -aB --synth=no -- ./perf test -w thloop [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.011 MB perf.data ] # ./perf report --stats | grep SAMPLE #does not generate samples in the perf.data file.
On x86 command# sudo perf record -e intel_pt// -u 0 ls is broken too.Looking at the sequence of calls in 'perf record' reveals this behavior:
1. The event 'cycles' is created and enabled:
record__open()+-> evlist__apply_filters() +-> perf_bpf_filter__prepare() +-> bpf_program.attach_perf_event() +-> bpf_program.attach_perf_event_opts() +-> __GI___ioctl(..., PERF_EVENT_IOC_ENABLE, ...)The event 'cycles' is enabled and active now. However the event's ring-buffer to store the samples generated by hardware is not allocated yet. This happens now after enabling the event:
2. The event's fd is
mmap()to create the ring buffer:
record__open()+-> record__mmap() +-> record__mmap_evlist() +-> evlist__mmap_ex() +-> perf_evlist__mmap_ops() +-> mmap_per_cpu() +-> mmap_per_evsel() +-> mmap__mmap() +-> perf_mmap__mmap() +-> mmap()This allocates the ring-buffer for the event 'cycles'. With
mmap()
the kernel creates the ring buffer:
perf_mmap(): kernel function to create the event's ring | buffer to save the sampled data. |+-> ring_buffer_attach(): Allocates memory for ring buffer. | The PMU has auxiliary data setup function. The | has_aux(event) condition is true and the PMU's | stop() is called to stop sampling. It is not | restarted: | if (has_aux(event)) | perf_event_stop(event, 0); | +-> cpumsf_pmu_stop():Hardware sampling is stopped. No samples are generated and saved anymore.
3. After the event 'cycles' has been mapped, the event is enabled a second time in:
__cmd_record()+-> evlist__enable() +-> __evlist__enable() +-> evsel__enable_cpu() +-> perf_evsel__enable_cpu() +-> perf_evsel__run_ioctl() +-> perf_evsel__ioctl() +-> __GI___ioctl(.,
PERF_EVENT_IOC_ENABLE, .)
The second
ioctl(fd, PERF_EVENT_IOC_ENABLE, 0);
is just a NOP in this case. The first invocation in (1.) sets theevent::statetoPERF_EVENT_STATE_ACTIVE. The kernel functionsperf_ioctl()+-> _perf_ioctl() +-> _perf_event_enable() +-> __perf_event_enable() returns immediately because
event::stateis already set to
PERF_EVENT_STATE_ACTIVE.This happens on s390, because the event 'cycles' offers the possibility
to save auxilary data. The PMU call backssetup_aux()andfree_aux()are defined. Without both call back functions,cpumsf_pmu_stop()is not invoked and sampling continues.To remedy this, remove the first invocation of
ioctl(..., PERF_EVENT_IOC_ENABLE, ...).
in step (1.) Create the event in step (1.) and enable it in step (3.) after the ring buffer has been mapped.
Make the change backward compatible and introduce a new structure memberbpf_perf_event_opts::no_ioctl_enable. It defaults to false and onlybpf_program__attach_perf_event()sets it to true. This way only perf tool invocation do not enable the sampling event.Output after:
# ./perf record -aB --synth=no -u 0 -- ./perf test -w thloop 2 [ perf record: Woken up 3 times to write data ] [ perf record: Captured and wrote 0.876 MB perf.data ] # ./perf report --stats | grep SAMPLE SAMPLE events: 16200 (99.5%) SAMPLE events: 16200 #The software event succeeded before and after the patch:
# ./perf record -e cpu-clock -aB --synth=no -u 0 -- \ ./perf test -w thloop 2 [ perf record: Woken up 7 times to write data ] [ perf record: Captured and wrote 2.870 MB perf.data ] # ./perf report --stats | grep SAMPLE SAMPLE events: 53506 (99.8%) SAMPLE events: 53506 #Fixes 63f2f5ee856ba ("libbpf: add ability to attach/detach BPFprogram to perf event")
To: Andrii Nakryiko <[email protected]>
To: Ian Rogers <[email protected]>
To: Ilya Leoshkevich <[email protected]>
Quoted 6 lines
Signed-off-by Thomas Richter <[email protected]>Suggested-by Jiri Olsa <[email protected]>---
tools/lib/bpf/libbpf.c | 19 +++++++++++++------
tools/lib/bpf/libbpf.h | 3 ++-2 files changed, 15 insertions(+), 7 deletions(-)
What do you think about rather calling the new field ioctl_enable? So that we don't get double negations in the API users and implementation - they are sometimes unnecessarily confusing.
I also think enablement should be the default in bpf_program__attach_perf_event(), and perf should now call bpf_program__attach_perf_event_opts() instead.
Based on your request in v1, I can offer to take over the patch and send a v3 with the changes I suggested above.
Quoted 139 lines · Ilya Leoshkevich
Quoted 127 lines · Thomas Richter
V1 --> V2: Added Jiri Olsa's suggestion and introduced member
bpf_perf_event_opts::no_ioctl_enable.On linux-next
commitb4c658d4d63d61("perf target: Remove uid from target") introduces a regression on s390. In fact the regression exists on all platforms when the event supports auxiliary data gathering.Command
# ./perf record -u 0 -aB --synth=no -- ./perf test -w thloop [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.011 MB perf.data ] # ./perf report --stats | grep SAMPLE #does not generate samples in the perf.data file.
On x86 command# sudo perf record -e intel_pt// -u 0 ls is broken too.Looking at the sequence of calls in 'perf record' reveals this behavior:
1. The event 'cycles' is created and enabled:
record__open()+-> evlist__apply_filters() +-> perf_bpf_filter__prepare() +-> bpf_program.attach_perf_event() +-> bpf_program.attach_perf_event_opts() +-> __GI___ioctl(..., PERF_EVENT_IOC_ENABLE, ...)The event 'cycles' is enabled and active now. However the event's ring-buffer to store the samples generated by hardware is not allocated yet. This happens now after enabling the event:
2. The event's fd is
mmap()to create the ring buffer:
record__open()+-> record__mmap() +-> record__mmap_evlist() +-> evlist__mmap_ex() +-> perf_evlist__mmap_ops() +-> mmap_per_cpu() +-> mmap_per_evsel() +-> mmap__mmap() +-> perf_mmap__mmap() +-> mmap()This allocates the ring-buffer for the event 'cycles'. With
mmap()
the kernel creates the ring buffer:
perf_mmap(): kernel function to create the event's ring | buffer to save the sampled data. |+-> ring_buffer_attach(): Allocates memory for ring buffer. | The PMU has auxiliary data setup function. The | has_aux(event) condition is true and the PMU's | stop() is called to stop sampling. It is not | restarted: | if (has_aux(event)) | perf_event_stop(event, 0); | +-> cpumsf_pmu_stop():Hardware sampling is stopped. No samples are generated and saved anymore.
3. After the event 'cycles' has been mapped, the event is enabled a second time in:
__cmd_record()+-> evlist__enable() +-> __evlist__enable() +-> evsel__enable_cpu() +-> perf_evsel__enable_cpu() +-> perf_evsel__run_ioctl() +-> perf_evsel__ioctl() +-> __GI___ioctl(.,
PERF_EVENT_IOC_ENABLE, .)
The second
ioctl(fd, PERF_EVENT_IOC_ENABLE, 0);
is just a NOP in this case. The first invocation in (1.) sets theevent::statetoPERF_EVENT_STATE_ACTIVE. The kernel functionsperf_ioctl()+-> _perf_ioctl() +-> _perf_event_enable() +-> __perf_event_enable() returns immediately because
event::stateis already set to
PERF_EVENT_STATE_ACTIVE.This happens on s390, because the event 'cycles' offers the possibility
to save auxilary data. The PMU call backssetup_aux()andfree_aux()are defined. Without both call back functions,cpumsf_pmu_stop()is not invoked and sampling continues.To remedy this, remove the first invocation of
ioctl(..., PERF_EVENT_IOC_ENABLE, ...).
in step (1.) Create the event in step (1.) and enable it in step (3.) after the ring buffer has been mapped.
Make the change backward compatible and introduce a new structure memberbpf_perf_event_opts::no_ioctl_enable. It defaults to false and onlybpf_program__attach_perf_event()sets it to true. This way only perf tool invocation do not enable the sampling event.Output after:
# ./perf record -aB --synth=no -u 0 -- ./perf test -w thloop 2 [ perf record: Woken up 3 times to write data ] [ perf record: Captured and wrote 0.876 MB perf.data ] # ./perf report --stats | grep SAMPLE SAMPLE events: 16200 (99.5%) SAMPLE events: 16200 #The software event succeeded before and after the patch:
# ./perf record -e cpu-clock -aB --synth=no -u 0 -- \ ./perf test -w thloop 2 [ perf record: Woken up 7 times to write data ] [ perf record: Captured and wrote 2.870 MB perf.data ] # ./perf report --stats | grep SAMPLE SAMPLE events: 53506 (99.8%) SAMPLE events: 53506 #Fixes 63f2f5ee856ba ("libbpf: add ability to attach/detach BPFprogram to perf event")
To: Andrii Nakryiko <[email protected]>
To: Ian Rogers <[email protected]>
To: Ilya Leoshkevich <[email protected]>Quoted 6 lines
Signed-off-by Thomas Richter <[email protected]>Suggested-by Jiri Olsa <[email protected]>---
tools/lib/bpf/libbpf.c | 19 +++++++++++++------
tools/lib/bpf/libbpf.h | 3 ++-2 files changed, 15 insertions(+), 7 deletions(-)What do you think about rather calling the new field
ioctl_enable? So that we don't get double negations in the API users and implementation - they are sometimes unnecessarily confusing.
Quoted 6 lines
I also think enablement should be the default in
bpf_program__attach_perf_event(), and perf should now callbpf_program__attach_perf_event_opts()instead.Based on your request in v1, I can offer to take over the patch and send a v3 with the changes I suggested above.
Yes Ilya, please go ahead.
Thanks very much for your support.
Signature · 9 lines
Thomas Richter, Dept 3303, IBM s390 Linux Development, Boeblingen, Germany -- IBM Deutschland Research & Development GmbH Vorsitzender des Aufsichtsrats: Wolfgang Wendt Geschäftsführung: David Faller Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294