[PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Thomas Richter]

In linux-next
commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
causes the perf tests 100 126 to fail on s390:

Output before:
 # ./perf test 100
 100: perf trace BTF general tests         : FAILED!
 #

The root cause is the change from int to int16_t for the
cpu maps. The size of the CPU key value pair changes from
four bytes to two bytes. However a two byte key size is
not supported for bpf_map__update_elem().
Note: validate_map_op() in libbpf.c emits warning
 libbpf: map '__augmented_syscalls__': \
	 unexpected key size 2 provided, expected 4
when key size is set to int16_t.

Therefore change to variable size back to 4 bytes for
invocation of bpf_map__update_elem().

Output after:
 # ./perf test 100
 100: perf trace BTF general tests         : Ok
 #

Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: James Clark <james.clark@linaro.org>
---
 tools/perf/builtin-trace.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index 092c5f6404ba..464c97a11852 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 		 * CPU the bpf-output event's file descriptor.
 		 */
 		perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
+			int mycpu = cpu.cpu;
+
 			bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
-					&cpu.cpu, sizeof(int),
+					&mycpu, sizeof(int),
 					xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
-						       cpu.cpu, 0),
+						       mycpu, 0),
 					sizeof(__u32), BPF_ANY);
 		}
 	}
-- 
2.48.1

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Ian Rogers]

On Mon, Mar 24, 2025 at 8:28 AM Thomas Richter <tmricht@linux.ibm.com> wrote:
>
> In linux-next
> commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> causes the perf tests 100 126 to fail on s390:
>
> Output before:
>  # ./perf test 100
>  100: perf trace BTF general tests         : FAILED!
>  #
>
> The root cause is the change from int to int16_t for the
> cpu maps. The size of the CPU key value pair changes from
> four bytes to two bytes. However a two byte key size is
> not supported for bpf_map__update_elem().

Nice catch!

> Note: validate_map_op() in libbpf.c emits warning
>  libbpf: map '__augmented_syscalls__': \
>          unexpected key size 2 provided, expected 4
> when key size is set to int16_t.

Wow, weird.

> Therefore change to variable size back to 4 bytes for
> invocation of bpf_map__update_elem().
>
> Output after:
>  # ./perf test 100
>  100: perf trace BTF general tests         : Ok
>  #
>
> Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
> Cc: Ian Rogers <irogers@google.com>
> Cc: James Clark <james.clark@linaro.org>
> ---
>  tools/perf/builtin-trace.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> index 092c5f6404ba..464c97a11852 100644
> --- a/tools/perf/builtin-trace.c
> +++ b/tools/perf/builtin-trace.c
> @@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
>                  * CPU the bpf-output event's file descriptor.
>                  */
>                 perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
> +                       int mycpu = cpu.cpu;
> +
>                         bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
> -                                       &cpu.cpu, sizeof(int),
> +                                       &mycpu, sizeof(int),

nit: It is usually preferred to do "sizeof(mycpu)" to avoid the
problems like the one this fix is fixing. And I'm blamed for the bad
code in:
5e6da6be3082 perf trace: Migrate BPF augmentation to use a skeleton

Reviewed-by: Ian Rogers <irogers@google.com>

Thanks for the fixes!
Ian

>                                         xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
> -                                                      cpu.cpu, 0),
> +                                                      mycpu, 0),
>                                         sizeof(__u32), BPF_ANY);
>                 }
>         }
> --
> 2.48.1
>

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Ian Rogers]

On Mon, Mar 24, 2025 at 8:38 AM Ian Rogers <irogers@google.com> wrote:
>
> On Mon, Mar 24, 2025 at 8:28 AM Thomas Richter <tmricht@linux.ibm.com> wrote:
> >
> > In linux-next
> > commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > causes the perf tests 100 126 to fail on s390:
> >
> > Output before:
> >  # ./perf test 100
> >  100: perf trace BTF general tests         : FAILED!
> >  #
> >
> > The root cause is the change from int to int16_t for the
> > cpu maps. The size of the CPU key value pair changes from
> > four bytes to two bytes. However a two byte key size is
> > not supported for bpf_map__update_elem().
>
> Nice catch!
>
> > Note: validate_map_op() in libbpf.c emits warning
> >  libbpf: map '__augmented_syscalls__': \
> >          unexpected key size 2 provided, expected 4
> > when key size is set to int16_t.
>
> Wow, weird.

Ah, I guess it is a mismatch with the declaration in
tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c?h=perf-tools-next#n31
```
/* bpf-output associated map */
struct __augmented_syscalls__ {
       __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
       __type(key, int);
       __type(value, __u32);
       __uint(max_entries, MAX_CPUS);
} __augmented_syscalls__ SEC(".maps");
```
but this looks wrong. The values are file descriptors, so should be
ints. I think it should be:
```
struct __augmented_syscalls__ {
       __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
       __type(key, int16_t);
       __type(value, int);
       __uint(max_entries, MAX_CPUS);
} __augmented_syscalls__ SEC(".maps");
```
I'm not sure if max_entries can be a uint16_t too. I suspect this may
well interfere with BPF_MAP_TYPE_PERF_EVENT_ARRAY and its use by
bpf_perf_event_output. Probably best to keep things in the BPF code as
they are and do your builtin-trace.c fix. Perhaps someone else can
clean this up.

Thanks,
Ian

> > Therefore change to variable size back to 4 bytes for
> > invocation of bpf_map__update_elem().
> >
> > Output after:
> >  # ./perf test 100
> >  100: perf trace BTF general tests         : Ok
> >  #
> >
> > Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
> > Cc: Ian Rogers <irogers@google.com>
> > Cc: James Clark <james.clark@linaro.org>
> > ---
> >  tools/perf/builtin-trace.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> > index 092c5f6404ba..464c97a11852 100644
> > --- a/tools/perf/builtin-trace.c
> > +++ b/tools/perf/builtin-trace.c
> > @@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
> >                  * CPU the bpf-output event's file descriptor.
> >                  */
> >                 perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
> > +                       int mycpu = cpu.cpu;
> > +
> >                         bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
> > -                                       &cpu.cpu, sizeof(int),
> > +                                       &mycpu, sizeof(int),
>
> nit: It is usually preferred to do "sizeof(mycpu)" to avoid the
> problems like the one this fix is fixing. And I'm blamed for the bad
> code in:
> 5e6da6be3082 perf trace: Migrate BPF augmentation to use a skeleton
>
> Reviewed-by: Ian Rogers <irogers@google.com>
>
> Thanks for the fixes!
> Ian
>
> >                                         xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
> > -                                                      cpu.cpu, 0),
> > +                                                      mycpu, 0),
> >                                         sizeof(__u32), BPF_ANY);
> >                 }
> >         }
> > --
> > 2.48.1
> >

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Namhyung Kim]

Hello,

On Mon, Mar 24, 2025 at 08:50:36AM -0700, Ian Rogers wrote:
> On Mon, Mar 24, 2025 at 8:38 AM Ian Rogers <irogers@google.com> wrote:
> >
> > On Mon, Mar 24, 2025 at 8:28 AM Thomas Richter <tmricht@linux.ibm.com> wrote:
> > >
> > > In linux-next
> > > commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > > causes the perf tests 100 126 to fail on s390:
> > >
> > > Output before:
> > >  # ./perf test 100
> > >  100: perf trace BTF general tests         : FAILED!
> > >  #
> > >
> > > The root cause is the change from int to int16_t for the
> > > cpu maps. The size of the CPU key value pair changes from
> > > four bytes to two bytes. However a two byte key size is
> > > not supported for bpf_map__update_elem().
> >
> > Nice catch!
> >
> > > Note: validate_map_op() in libbpf.c emits warning
> > >  libbpf: map '__augmented_syscalls__': \
> > >          unexpected key size 2 provided, expected 4
> > > when key size is set to int16_t.
> >
> > Wow, weird.
> 
> Ah, I guess it is a mismatch with the declaration in
> tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c:
> https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c?h=perf-tools-next#n31
> ```
> /* bpf-output associated map */
> struct __augmented_syscalls__ {
>        __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
>        __type(key, int);
>        __type(value, __u32);
>        __uint(max_entries, MAX_CPUS);
> } __augmented_syscalls__ SEC(".maps");
> ```
> but this looks wrong. The values are file descriptors, so should be
> ints. I think it should be:
> ```
> struct __augmented_syscalls__ {
>        __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
>        __type(key, int16_t);
>        __type(value, int);
>        __uint(max_entries, MAX_CPUS);
> } __augmented_syscalls__ SEC(".maps");
> ```
> I'm not sure if max_entries can be a uint16_t too. I suspect this may
> well interfere with BPF_MAP_TYPE_PERF_EVENT_ARRAY and its use by
> bpf_perf_event_output. Probably best to keep things in the BPF code as
> they are and do your builtin-trace.c fix. Perhaps someone else can
> clean this up.

I don't think max_entries can be uint16_t.  Also I'm skeptical to change
the key type to int16_t as you said.  I'll pick Thomas's patch up.

Thanks,
Namhyung

> 
> > > Therefore change to variable size back to 4 bytes for
> > > invocation of bpf_map__update_elem().
> > >
> > > Output after:
> > >  # ./perf test 100
> > >  100: perf trace BTF general tests         : Ok
> > >  #
> > >
> > > Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > > Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
> > > Cc: Ian Rogers <irogers@google.com>
> > > Cc: James Clark <james.clark@linaro.org>
> > > ---
> > >  tools/perf/builtin-trace.c | 6 ++++--
> > >  1 file changed, 4 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> > > index 092c5f6404ba..464c97a11852 100644
> > > --- a/tools/perf/builtin-trace.c
> > > +++ b/tools/perf/builtin-trace.c
> > > @@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
> > >                  * CPU the bpf-output event's file descriptor.
> > >                  */
> > >                 perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
> > > +                       int mycpu = cpu.cpu;
> > > +
> > >                         bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
> > > -                                       &cpu.cpu, sizeof(int),
> > > +                                       &mycpu, sizeof(int),
> >
> > nit: It is usually preferred to do "sizeof(mycpu)" to avoid the
> > problems like the one this fix is fixing. And I'm blamed for the bad
> > code in:
> > 5e6da6be3082 perf trace: Migrate BPF augmentation to use a skeleton
> >
> > Reviewed-by: Ian Rogers <irogers@google.com>
> >
> > Thanks for the fixes!
> > Ian
> >
> > >                                         xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
> > > -                                                      cpu.cpu, 0),
> > > +                                                      mycpu, 0),
> > >                                         sizeof(__u32), BPF_ANY);
> > >                 }
> > >         }
> > > --
> > > 2.48.1
> > >

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Howard Chu]

Hello Ian,

On Mon, Mar 24, 2025 at 08:50:36AM -0700, Ian Rogers wrote:
> On Mon, Mar 24, 2025 at 8:38 AM Ian Rogers <irogers@google.com> wrote:
> >
> > On Mon, Mar 24, 2025 at 8:28 AM Thomas Richter <tmricht@linux.ibm.com> wrote:
> > >
> > > In linux-next
> > > commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > > causes the perf tests 100 126 to fail on s390:
> > >
> > > Output before:
> > >  # ./perf test 100
> > >  100: perf trace BTF general tests         : FAILED!
> > >  #
> > >
> > > The root cause is the change from int to int16_t for the
> > > cpu maps. The size of the CPU key value pair changes from
> > > four bytes to two bytes. However a two byte key size is
> > > not supported for bpf_map__update_elem().
> >
> > Nice catch!
> >
> > > Note: validate_map_op() in libbpf.c emits warning
> > >  libbpf: map '__augmented_syscalls__': \
> > >          unexpected key size 2 provided, expected 4
> > > when key size is set to int16_t.
> >
> > Wow, weird.
> 
> Ah, I guess it is a mismatch with the declaration in
> tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c:
> https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c?h=perf-tools-next#n31
> ```
> /* bpf-output associated map */
> struct __augmented_syscalls__ {
>        __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
>        __type(key, int);
>        __type(value, __u32);
>        __uint(max_entries, MAX_CPUS);
> } __augmented_syscalls__ SEC(".maps");
> ```
> but this looks wrong. The values are file descriptors, so should be
> ints. I think it should be:
> ```
> struct __augmented_syscalls__ {
>        __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
>        __type(key, int16_t);
>        __type(value, int);
>        __uint(max_entries, MAX_CPUS);
> } __augmented_syscalls__ SEC(".maps");
> ```

I agree, regardless of the key, value of fd should be int.

> I'm not sure if max_entries can be a uint16_t too. I suspect this may
> well interfere with BPF_MAP_TYPE_PERF_EVENT_ARRAY and its use by
> bpf_perf_event_output. Probably best to keep things in the BPF code as
> they are and do your builtin-trace.c fix. Perhaps someone else can
> clean this up.

I'll do that since I'm responsible, and I'll based it on Thomas'
changes.

Thanks
Howard
> 
> Thanks,
> Ian
> 
> > > Therefore change to variable size back to 4 bytes for
> > > invocation of bpf_map__update_elem().
> > >
> > > Output after:
> > >  # ./perf test 100
> > >  100: perf trace BTF general tests         : Ok
> > >  #
> > >
> > > Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> > > Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
> > > Cc: Ian Rogers <irogers@google.com>
> > > Cc: James Clark <james.clark@linaro.org>
> > > ---
> > >  tools/perf/builtin-trace.c | 6 ++++--
> > >  1 file changed, 4 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> > > index 092c5f6404ba..464c97a11852 100644
> > > --- a/tools/perf/builtin-trace.c
> > > +++ b/tools/perf/builtin-trace.c
> > > @@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
> > >                  * CPU the bpf-output event's file descriptor.
> > >                  */
> > >                 perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
> > > +                       int mycpu = cpu.cpu;
> > > +
> > >                         bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
> > > -                                       &cpu.cpu, sizeof(int),
> > > +                                       &mycpu, sizeof(int),
> >
> > nit: It is usually preferred to do "sizeof(mycpu)" to avoid the
> > problems like the one this fix is fixing. And I'm blamed for the bad
> > code in:
> > 5e6da6be3082 perf trace: Migrate BPF augmentation to use a skeleton
> >
> > Reviewed-by: Ian Rogers <irogers@google.com>
> >
> > Thanks for the fixes!
> > Ian
> >
> > >                                         xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
> > > -                                                      cpu.cpu, 0),
> > > +                                                      mycpu, 0),
> > >                                         sizeof(__u32), BPF_ANY);
> > >                 }
> > >         }
> > > --
> > > 2.48.1
> > >
> 

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Howard Chu]

On Mon, Mar 24, 2025 at 04:27:56PM +0100, Thomas Richter wrote:
> In linux-next
> commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> causes the perf tests 100 126 to fail on s390:
>
> Output before:
>  # ./perf test 100
>  100: perf trace BTF general tests         : FAILED!

Arnaldo encountered similar failure, but I couldn't reproduce it, so the
fix was stalled.

>  #
>
> The root cause is the change from int to int16_t for the
> cpu maps. The size of the CPU key value pair changes from
> four bytes to two bytes. However a two byte key size is
> not supported for bpf_map__update_elem().
> Note: validate_map_op() in libbpf.c emits warning
>  libbpf: map '__augmented_syscalls__': \
>        unexpected key size 2 provided, expected 4
> when key size is set to int16_t.
>
> Therefore change to variable size back to 4 bytes for
> invocation of bpf_map__update_elem().

Makes sense, thank you so much for digging into this and fixing it.

>
> Output after:
>  # ./perf test 100
>  100: perf trace BTF general tests         : Ok
>  #
>
> Fixes: c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
> Cc: Ian Rogers <irogers@google.com>
> Cc: James Clark <james.clark@linaro.org>
> ---
>  tools/perf/builtin-trace.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> index 092c5f6404ba..464c97a11852 100644
> --- a/tools/perf/builtin-trace.c
> +++ b/tools/perf/builtin-trace.c
> @@ -4375,10 +4375,12 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
>                * CPU the bpf-output event's file descriptor.
>                */
>               perf_cpu_map__for_each_cpu(cpu, i, trace->syscalls.events.bpf_output->core.cpus) {
> +                     int mycpu = cpu.cpu;
> +
>                       bpf_map__update_elem(trace->skel->maps.__augmented_syscalls__,
> -                                     &cpu.cpu, sizeof(int),
> +                                     &mycpu, sizeof(int),
>                                       xyarray__entry(trace->syscalls.events.bpf_output->core.fd,
> -                                                    cpu.cpu, 0),
> +                                                    mycpu, 0),
>                                       sizeof(__u32), BPF_ANY);
>               }
>       }
> --
> 2.48.1
>
>

Acked-by: Howard Chu <howardchu95@gmail.com>

Thanks,
Howard

Re: [PATCH] perf trace: Fix wrong size to bpf_map__update_elem call

[Namhyung Kim]

On Mon, 24 Mar 2025 16:27:56 +0100, Thomas Richter wrote:
> In linux-next
> commit c760174401f6 ("perf cpumap: Reduce cpu size from int to int16_t")
> causes the perf tests 100 126 to fail on s390:
> 
> Output before:
>  # ./perf test 100
>  100: perf trace BTF general tests         : FAILED!
>  #
> 
> [...]
Applied to perf-tools-next, thanks!

Best regards,
Namhyung