[PATCH bpf] libbpf: eBPF fails on events with auxiliary data

[Thomas Richter <tmricht@linux.ibm.com>]

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.

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 BPF program to perf event")
To: Andrii Nakryiko <andriin@fb.com>
To: Ian Rogers <irogers@google.com>
To: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Tested-by: Ilya Leoshkevich <iii@linux.ibm.com>
---
 tools/lib/bpf/libbpf.c | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 162ebd16a59f..5973412a1031 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -10960,12 +10960,6 @@ 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;
-	}
 
 	return &link->link;
 err_out:
-- 
2.50.0