Re: [PATCH v5 perf,bpf 07/15] perf, bpf: save bpf_prog_info information as headers to perf.data

From: Jiri Olsa
Date: Mon Mar 04 2019 - 15:37:51 EST


On Mon, Mar 04, 2019 at 07:41:10PM +0000, Song Liu wrote:
>
>
> > On Mar 4, 2019, at 5:52 AM, Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
> >
> > On Wed, Feb 27, 2019 at 09:06:35PM -0800, Song Liu wrote:
> >> This patch enables perf-record to save bpf_prog_info information as
> >> headers to perf.data. A new header type HEADER_BPF_PROG_INFO is
> >> introduced for this data.
> >>
> >> Signed-off-by: Song Liu <songliubraving@xxxxxx>
> >> ---
> >> tools/perf/util/header.c | 143 ++++++++++++++++++++++++++++++++++++++-
> >> tools/perf/util/header.h | 1 +
> >> 2 files changed, 143 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> >> index 4b88de5e9192..16f5bedb0b7d 100644
> >> --- a/tools/perf/util/header.c
> >> +++ b/tools/perf/util/header.c
> >> @@ -18,6 +18,7 @@
> >> #include <sys/utsname.h>
> >> #include <linux/time64.h>
> >> #include <dirent.h>
> >> +#include <bpf/libbpf.h>
> >>
> >> #include "evlist.h"
> >> #include "evsel.h"
> >> @@ -39,6 +40,7 @@
> >> #include "tool.h"
> >> #include "time-utils.h"
> >> #include "units.h"
> >> +#include "bpf-event.h"
> >>
> >> #include "sane_ctype.h"
> >>
> >> @@ -1074,6 +1076,51 @@ static int write_clockid(struct feat_fd *ff,
> >> sizeof(ff->ph->env.clockid_res_ns));
> >> }
> >>
> >> +static int write_bpf_prog_info(struct feat_fd *ff,
> >> + struct perf_evlist *evlist __maybe_unused)
> >> +{
> >> + struct perf_env *env = &ff->ph->env;
> >> + struct rb_root *root;
> >> + struct rb_node *next;
> >> + u32 count = 0;
> >> + int ret;
> >> +
> >> + down_read(&env->bpf_progs.lock);
> >> +
> >> + root = &env->bpf_progs.infos;
> >> + next = rb_first(root);
> >> + while (next) {
> >> + ++count;
> >> + next = rb_next(next);
> >> + }
> >> +
> >> + ret = do_write(ff, &count, sizeof(count));
> >> +
> >> + if (ret < 0)
> >> + goto out;
> >> +
> >> + next = rb_first(root);
> >> + while (next) {
> >> + struct bpf_prog_info_node *node;
> >> + size_t len;
> >> +
> >> + node = rb_entry(next, struct bpf_prog_info_node, rb_node);
> >> + next = rb_next(&node->rb_node);
> >> + len = sizeof(struct bpf_prog_info_linear) +
> >> + node->info_linear->data_len;
> >> +
> >> + /* before writing to file, translate address to offset */
> >> + bpf_program__bpil_addr_to_offs(node->info_linear);
> >> + ret = do_write(ff, node->info_linear, len);
> >> + bpf_program__bpil_offs_to_addr(node->info_linear);
> >
> > what's the reason to call this before the error check?
>
> This will translate the all the pointers to the right address
> (instead of offsets). We only need the offsets when writing to
> files. With this approach, the data is not changed whether the
> function succeeds or not. It is probably not necessary right
> now. But I think this may save us some debugging efforts in
> the future.

ok, please make comment with this

jirka