[PATCH] perf inject: correct event attribute sizes

[PATCH] perf inject: correct event attribute sizes

[Al Grant]

When perf inject reads a perf.data file from an older version of perf,
it writes event attributes into the output with the original size field,
but lays them out as if they had the size currently used. Readers see
a corrupt file. Update the size field to match the layout.

From: Denis Nikitin <denik@google.com>
Signed-off-by: Al Grant <al.grant@foss.arm.com>

  tools/perf/util/header.c | 8 ++++++++
  1 file changed, 8 insertions(+)

diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 9cf4efdcbbbd..762eb94bd532 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -3326,6 +3326,14 @@ int perf_session__write_header(struct
perf_session *session,
         attr_offset = lseek(ff.fd, 0, SEEK_CUR);

         evlist__for_each_entry(evlist, evsel) {
+               if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
+                       /*
+                        * We are likely in "perf inject" and have read 
+                        * from an older file. Update attr size so that
+                        * reader gets the right offset to the ids.
+                        */
+                       evsel->core.attr.size = sizeof(evsel->core.attr);
+               }
                 f_attr = (struct perf_file_attr){
                         .attr = evsel->core.attr,
                         .ids  = {

Re: [PATCH] perf inject: correct event attribute sizes

[Mark Rutland]

On Wed, Aug 26, 2020 at 10:42:04AM +0100, Al Grant wrote:
> When perf inject reads a perf.data file from an older version of perf,
> it writes event attributes into the output with the original size field,
> but lays them out as if they had the size currently used. Readers see
> a corrupt file. Update the size field to match the layout.
> 
> From: Denis Nikitin <denik@google.com>
> Signed-off-by: Al Grant <al.grant@foss.arm.com>

Did Denis write this patch?

If so, we need an S-o-B line from them.

Mark.

>  tools/perf/util/header.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index 9cf4efdcbbbd..762eb94bd532 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -3326,6 +3326,14 @@ int perf_session__write_header(struct
> perf_session *session,
>         attr_offset = lseek(ff.fd, 0, SEEK_CUR);
> 
>         evlist__for_each_entry(evlist, evsel) {
> +               if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
> +                       /*
> +                        * We are likely in "perf inject" and have read +
> * from an older file. Update attr size so that
> +                        * reader gets the right offset to the ids.
> +                        */
> +                       evsel->core.attr.size = sizeof(evsel->core.attr);
> +               }
>                 f_attr = (struct perf_file_attr){
>                         .attr = evsel->core.attr,
>                         .ids  = {

Re: [PATCH] perf inject: correct event attribute sizes

[Al Grant]

On 26/08/2020 12:27, Mark Rutland wrote:
> On Wed, Aug 26, 2020 at 10:42:04AM +0100, Al Grant wrote:
>> When perf inject reads a perf.data file from an older version of perf,
>> it writes event attributes into the output with the original size field,
>> but lays them out as if they had the size currently used. Readers see
>> a corrupt file. Update the size field to match the layout.
>>
>> From: Denis Nikitin <denik@google.com>
>> Signed-off-by: Al Grant <al.grant@foss.arm.com>
> 
> Did Denis write this patch?

Joint work - I fixed the sizes in the event attributes section,
but Denis noticed that the copies in HEADER_EVENT_DESC also need
fixing, so the final text is his.

> If so, we need an S-o-B line from them.

I have his approval to upstream this, so:

Signed-off-by: Denis Nikitin <denik@google.com>

Al


> 
> Mark.
> 
>>   tools/perf/util/header.c | 8 ++++++++
>>   1 file changed, 8 insertions(+)
>>
>> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
>> index 9cf4efdcbbbd..762eb94bd532 100644
>> --- a/tools/perf/util/header.c
>> +++ b/tools/perf/util/header.c
>> @@ -3326,6 +3326,14 @@ int perf_session__write_header(struct
>> perf_session *session,
>>          attr_offset = lseek(ff.fd, 0, SEEK_CUR);
>>
>>          evlist__for_each_entry(evlist, evsel) {
>> +               if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
>> +                       /*
>> +                        * We are likely in "perf inject" and have read +
>> * from an older file. Update attr size so that
>> +                        * reader gets the right offset to the ids.
>> +                        */
>> +                       evsel->core.attr.size = sizeof(evsel->core.attr);
>> +               }
>>                  f_attr = (struct perf_file_attr){
>>                          .attr = evsel->core.attr,
>>                          .ids  = {

Re: [PATCH] perf inject: correct event attribute sizes

[Arnaldo Carvalho de Melo]

Em Wed, Aug 26, 2020 at 10:42:04AM +0100, Al Grant escreveu:
> When perf inject reads a perf.data file from an older version of perf,
> it writes event attributes into the output with the original size field,
> but lays them out as if they had the size currently used. Readers see
> a corrupt file. Update the size field to match the layout.
> 
> From: Denis Nikitin <denik@google.com>
> Signed-off-by: Al Grant <al.grant@foss.arm.com>

Ok, so the author is Denis, has he provided his Signed-off-by?

- Arnaldo
 
>  tools/perf/util/header.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index 9cf4efdcbbbd..762eb94bd532 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -3326,6 +3326,14 @@ int perf_session__write_header(struct
> perf_session *session,
>         attr_offset = lseek(ff.fd, 0, SEEK_CUR);
> 
>         evlist__for_each_entry(evlist, evsel) {
> +               if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
> +                       /*
> +                        * We are likely in "perf inject" and have read +
> * from an older file. Update attr size so that
> +                        * reader gets the right offset to the ids.
> +                        */
> +                       evsel->core.attr.size = sizeof(evsel->core.attr);
> +               }
>                 f_attr = (struct perf_file_attr){
>                         .attr = evsel->core.attr,
>                         .ids  = {

-- 

- Arnaldo

[PATCH] perf inject: correct event attribute sizes

[Al Grant]

When perf inject reads a perf.data file from an older version of perf,
it writes event attributes into the output with the original size field,
but lays them out as if they had the size currently used. Readers see
a corrupt file. Update the size field to match the layout.

Signed-off-by: Al Grant <al.grant@foss.arm.com>
---
 tools/perf/util/header.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index be850e9f8852..0d95981df8dd 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -3331,6 +3331,14 @@ int perf_session__write_header(struct perf_session *session,
 	attr_offset = lseek(ff.fd, 0, SEEK_CUR);
 
 	evlist__for_each_entry(evlist, evsel) {
+		if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
+			/*
+			 * We are likely in "perf inject" and have read
+			 * from an older file. Update attr size so that
+			 * reader gets the right offset to the ids.
+			 */
+			evsel->core.attr.size = sizeof(evsel->core.attr);
+		}
 		f_attr = (struct perf_file_attr){
 			.attr = evsel->core.attr,
 			.ids  = {
-- 
2.17.1

[PATCH] perf inject: correct event attribute sizes

[Denis Nikitin]

When perf inject reads a perf.data file from an older version of perf,
it writes event attributes into the output with the original size field,
but lays them out as if they had the size currently used. Readers see
a corrupt file. Update the size field to match the layout.

Signed-off-by: Al Grant <al.grant@foss.arm.com>
Signed-off-by: Denis Nikitin <denik@chromium.org>
---
 tools/perf/util/header.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index be850e9f8852..0d95981df8dd 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -3331,6 +3331,14 @@ int perf_session__write_header(struct perf_session *session,
 	attr_offset = lseek(ff.fd, 0, SEEK_CUR);
 
 	evlist__for_each_entry(evlist, evsel) {
+		if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
+			/*
+			 * We are likely in "perf inject" and have read
+			 * from an older file. Update attr size so that
+			 * reader gets the right offset to the ids.
+			 */
+			evsel->core.attr.size = sizeof(evsel->core.attr);
+		}
 		f_attr = (struct perf_file_attr){
 			.attr = evsel->core.attr,
 			.ids  = {
-- 
2.29.2.684.gfbc64c5ab5-goog

Re: [PATCH] perf inject: correct event attribute sizes

[Jiri Olsa]

On Tue, Dec 15, 2020 at 10:58:16PM -0800, Denis Nikitin wrote:
> When perf inject reads a perf.data file from an older version of perf,
> it writes event attributes into the output with the original size field,
> but lays them out as if they had the size currently used. Readers see
> a corrupt file. Update the size field to match the layout.
> 
> Signed-off-by: Al Grant <al.grant@foss.arm.com>
> Signed-off-by: Denis Nikitin <denik@chromium.org>
> ---
>  tools/perf/util/header.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index be850e9f8852..0d95981df8dd 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -3331,6 +3331,14 @@ int perf_session__write_header(struct perf_session *session,
>  	attr_offset = lseek(ff.fd, 0, SEEK_CUR);
>  
>  	evlist__for_each_entry(evlist, evsel) {
> +		if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
> +			/*
> +			 * We are likely in "perf inject" and have read
> +			 * from an older file. Update attr size so that
> +			 * reader gets the right offset to the ids.
> +			 */
> +			evsel->core.attr.size = sizeof(evsel->core.attr);
> +		}

seems ok, just wondering if it would be better
to fix it in perf_event__process_attr instead

where we know where the data is coming from,
here we could cover for some unexpected case
of having different attr.size?

anyway both would be probably equal, I'm ok
with either way

Acked-by: Jiri Olsa <jolsa@redhat.com>

thanks,
jirka


>  		f_attr = (struct perf_file_attr){
>  			.attr = evsel->core.attr,
>  			.ids  = {
> -- 
> 2.29.2.684.gfbc64c5ab5-goog
> 

Re: [PATCH] perf inject: correct event attribute sizes

[Arnaldo Carvalho de Melo]

Em Wed, Dec 23, 2020 at 11:11:05PM +0100, Jiri Olsa escreveu:
> On Tue, Dec 15, 2020 at 10:58:16PM -0800, Denis Nikitin wrote:
> > When perf inject reads a perf.data file from an older version of perf,
> > it writes event attributes into the output with the original size field,
> > but lays them out as if they had the size currently used. Readers see
> > a corrupt file. Update the size field to match the layout.
> > 
> > Signed-off-by: Al Grant <al.grant@foss.arm.com>
> > Signed-off-by: Denis Nikitin <denik@chromium.org>
> > ---
> >  tools/perf/util/header.c | 8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> > index be850e9f8852..0d95981df8dd 100644
> > --- a/tools/perf/util/header.c
> > +++ b/tools/perf/util/header.c
> > @@ -3331,6 +3331,14 @@ int perf_session__write_header(struct perf_session *session,
> >  	attr_offset = lseek(ff.fd, 0, SEEK_CUR);
> >  
> >  	evlist__for_each_entry(evlist, evsel) {
> > +		if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
> > +			/*
> > +			 * We are likely in "perf inject" and have read
> > +			 * from an older file. Update attr size so that
> > +			 * reader gets the right offset to the ids.
> > +			 */
> > +			evsel->core.attr.size = sizeof(evsel->core.attr);
> > +		}
> 
> seems ok, just wondering if it would be better
> to fix it in perf_event__process_attr instead
> 
> where we know where the data is coming from,
> here we could cover for some unexpected case
> of having different attr.size?
> 
> anyway both would be probably equal, I'm ok
> with either way
> 
> Acked-by: Jiri Olsa <jolsa@redhat.com>

Thanks, applied.

- Arnaldo