perf record: why we used type casting of (uint64_t *) instead of int

PATCH In review Anshuman Khandual 3 participants 7 messages May 24, 2012 – May 25, 2012 perf events

Share & tools

Anshuman Khandual

author

2012-05-25 05:28 UTC #1 lore raw

This code is breaking in powerpc systems.

(1) 'opt->value' gets updated inside the function parse_branch_stack via dereferencing a (uint64_t *) type casted pointer.

(2) But the value is not accessible when we again use opt->value via dereferencing a (int *) type casted pointer.

(3) As a result record.opts.branch_stack remains 0 and unchanged by parse_branch_stack

This is caused by bit representation of 'uint64_t' and 'int' in powerpc systems. Bytes update for the data (when accessed trough (uint64_t *) casting) is no longer available to the data when accessed through (int *) type casting. Verified this from bit representation of the data (accessed through both type casting methods).

However this problem does not seem to be present on an Intel box. Integer dereferencing of the opt->value still gives the value which was updated as (uint64_t).

All this problem would not have been there if we had used (int *) instead of (uint64_t *) in the first place inside parse_branch_stack function.

Quoted 47 lines · Anshuman Khandual

Hey Stephane,
Just wondering why we used the type casting of (uint64_t *) on a data which is defined as "int" in the structure of "perf_record_opts".
struct perf_record_opts {
        struct perf_target target;
        bool         call_graph;
        bool         group;
        bool         inherit_stat;
        bool         no_delay;
        bool         no_inherit;
        bool         no_samples;
        bool         pipe_output;
        bool         raw_samples;
        bool         sample_address;
        bool         sample_time;
        bool         sample_id_all_missing;
        bool         exclude_guest_missing;
        bool         period;
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int user_freq;
        int          branch_stack;
        u64          default_interval;
        u64          user_interval;
};
static int
parse_branch_stack(const struct option *opt, const char *str, int unset) {
#define ONLY_PLM \
        (PERF_SAMPLE_BRANCH_USER        |\
         PERF_SAMPLE_BRANCH_KERNEL      |\
         PERF_SAMPLE_BRANCH_HV)

        uint64_t *mode = (uint64_t *)opt->value;
--
Regards
Anshuman Khandual
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Stephane Eranian

2012-05-25 08:20 UTC #2 lore raw

Quoted 15 lines · Anshuman Khandual

This code is breaking in powerpc systems.
(1) 'opt->value' gets updated inside the function parse_branch_stack via dereferencing a (uint64_t *) type casted pointer.
(2) But the value is not accessible when we again use opt->value via dereferencing a (int *) type casted pointer.
(3) As a result record.opts.branch_stack remains 0 and unchanged by parse_branch_stack
This is caused by bit representation of 'uint64_t' and 'int' in powerpc systems. Bytes update for the data (when accessed trough (uint64_t *) casting) is no longer available to the data when accessed through (int *) type casting. Verified this from bit representation of the data (accessed through both type casting methods).

Quoted 6 lines

However this problem does not seem to be present on an Intel box. Integer dereferencing of the opt->value still gives the value which was updated as (uint64_t).
All this problem would not have been there if we had used (int *) instead of (uint64_t *) in the first place inside parse_branch_stack function.

The bug is that in struct record_opts, branch_stack is declared int instead of u64.
I can post a patch to fix that. The value is eventually passed to struct perf_event_attr.branch_sample_type which is defined as u64.

I can post a patch to fix that.

Thanks for catching this.

Quoted 51 lines · Anshuman Khandual

Quoted 48 lines
Hey Stephane,
Just wondering why we used the type casting of (uint64_t *) on a data which is defined as "int" in the structure of "perf_record_opts".
struct perf_record_opts {
struct perf_target target;
bool call_graph;
bool group;
bool inherit_stat;
bool no_delay;
bool no_inherit;
bool no_samples;
bool pipe_output;
bool raw_samples;
bool sample_address;
bool sample_time;
bool sample_id_all_missing;
bool exclude_guest_missing;
bool period;
unsigned int freq;
unsigned int mmap_pages;
unsigned int user_freq;
int branch_stack;
u64 default_interval;
u64 user_interval;
};
static int
parse_branch_stack(const struct option *opt, const char *str, int unset) {
#define ONLY_PLM \
(PERF_SAMPLE_BRANCH_USER |\
PERF_SAMPLE_BRANCH_KERNEL |\
PERF_SAMPLE_BRANCH_HV)
uint64_t *mode = (uint64_t *)opt->value;
--
Regards
Anshuman Khandual
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Stephane Eranian

2012-05-25 08:44 UTC #4 lore raw

Re: [PATCH] perf record: Fixing record option data type in parse_branch_stack

In reply to Anshuman Khandual

Hi,

It should be something like that instead:

tools/perf/perf.h +1 -1

index 8a9687e..c9ca7c4 100644
@@ -228,7 +228,7 @@ struct perf_record_opts {
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int user_freq;
-       int          branch_stack;
+       u64          branch_stack;
        u64          default_interval;
        u64          user_interval;
 };

Quoted 113 lines · Anshuman Khandual

perf record: Fixing record option data type in parse_branch_stack
Currently parse_branch_stack does not update record.opts.branch_stack value in powerpc architecture. opt->value is declared as int in struct perf_record_opts. But is worked on as uint64_t isnide the function. This breaks functionality in poweprc due to bit representation of uint64_t which is inaccessible as int.
Signed-off-by Anshuman Khandual <[email protected]>
---
tools/perf/builtin-record.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/builtin-record.c +1 -2
index e5cb084..161c0f1 100644
@@ -677,7 +677,7 @@ parse_branch_stack(const struct option *opt, const char *str, int unset)
         PERF_SAMPLE_BRANCH_KERNEL      |\
         PERF_SAMPLE_BRANCH_HV)
 
-       uint64_t *mode = (uint64_t *)opt->value;
+       int *mode = (int *)opt->value;
        const struct branch_mode *br;
        char *s, *os = NULL, *p;
        int ret = -1;
--
1.7.9.5
Quoted 78 lines · Anshuman Khandual
This code is breaking in powerpc systems.
(1) 'opt->value' gets updated inside the function parse_branch_stack via dereferencing a (uint64_t *) type casted pointer.
(2) But the value is not accessible when we again use opt->value via dereferencing a (int *) type casted pointer.
(3) As a result record.opts.branch_stack remains 0 and unchanged by parse_branch_stack
This is caused by bit representation of 'uint64_t' and 'int' in powerpc systems. Bytes update for the data (when accessed trough (uint64_t *) casting) is no longer available to the data when accessed through (int *) type casting. Verified this from bit representation of the data (accessed through both type casting methods).
However this problem does not seem to be present on an Intel box. Integer dereferencing of the opt->value still gives the value which was updated as (uint64_t).
All this problem would not have been there if we had used (int *) instead of (uint64_t *) in the first place inside parse_branch_stack function.
Quoted 41 lines · Anshuman Khandual
Hey Stephane,
Just wondering why we used the type casting of (uint64_t *) on a data which is defined as "int" in the structure of "perf_record_opts".
struct perf_record_opts {
struct perf_target target;
bool call_graph;
bool group;
bool inherit_stat;
bool no_delay;
bool no_inherit;
bool no_samples;
bool pipe_output;
bool raw_samples;
bool sample_address;
bool sample_time;
bool sample_id_all_missing;
bool exclude_guest_missing;
bool period;
unsigned int freq;
unsigned int mmap_pages;
unsigned int user_freq;
int branch_stack;
u64 default_interval;
u64 user_interval;
};
static int
parse_branch_stack(const struct option *opt, const char *str, int unset) {
#define ONLY_PLM \
(PERF_SAMPLE_BRANCH_USER |\
PERF_SAMPLE_BRANCH_KERNEL |\
PERF_SAMPLE_BRANCH_HV)
uint64_t *mode = (uint64_t *)opt->value;
--
Regards
Anshuman Khandual
Quoted 6 lines
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Anshuman Khandual

author

2012-05-25 08:55 UTC #3 lore raw

PATCH 1/1 perf record: Fixing record option data type in parse_branch_stack

In reply to Anshuman Khandual

perf record: Fixing record option data type in parse_branch_stack

Currently parse_branch_stack does not update record.opts.branch_stack value in powerpc architecture. opt->value is declared as int in struct perf_record_opts. But is worked on as uint64_t isnide the function. This breaks functionality in poweprc due to bit representation of uint64_t which is inaccessible as int.

Signed-off-by Anshuman Khandual <[email protected]>

---
 tools/perf/builtin-record.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

tools/perf/builtin-record.c +1 -1

index e5cb084..161c0f1 100644
@@ -677,7 +677,7 @@ parse_branch_stack(const struct option *opt, const char *str, int unset)
 	 PERF_SAMPLE_BRANCH_KERNEL	|\
 	 PERF_SAMPLE_BRANCH_HV)
 
-	uint64_t *mode = (uint64_t *)opt->value;
+	int *mode = (int *)opt->value;
 	const struct branch_mode *br;
 	char *s, *os = NULL, *p;
 	int ret = -1;

Signature · 1 line

1.7.9.5

Quoted 78 lines · Anshuman Khandual

This code is breaking in powerpc systems.
(1) 'opt->value' gets updated inside the function parse_branch_stack via dereferencing a (uint64_t *) type casted pointer.
(2) But the value is not accessible when we again use opt->value via dereferencing a (int *) type casted pointer.
(3) As a result record.opts.branch_stack remains 0 and unchanged by parse_branch_stack
This is caused by bit representation of 'uint64_t' and 'int' in powerpc systems. Bytes update for the data (when accessed trough (uint64_t *) casting) is no longer available to the data when accessed through (int *) type casting. Verified this from bit representation of the data (accessed through both type casting methods).
However this problem does not seem to be present on an Intel box. Integer dereferencing of the opt->value still gives the value which was updated as (uint64_t).
All this problem would not have been there if we had used (int *) instead of (uint64_t *) in the first place inside parse_branch_stack function.
Quoted 41 lines · Anshuman Khandual
Hey Stephane,
Just wondering why we used the type casting of (uint64_t *) on a data which is defined as "int" in the structure of "perf_record_opts".
struct perf_record_opts {
        struct perf_target target;
        bool         call_graph;
        bool         group;
        bool         inherit_stat;
        bool         no_delay;
        bool         no_inherit;
        bool         no_samples;
        bool         pipe_output;
        bool         raw_samples;
        bool         sample_address;
        bool         sample_time;
        bool         sample_id_all_missing;
        bool         exclude_guest_missing;
        bool         period;
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int user_freq;
        int          branch_stack;
        u64          default_interval;
        u64          user_interval;
};
static int
parse_branch_stack(const struct option *opt, const char *str, int unset) {
#define ONLY_PLM \
        (PERF_SAMPLE_BRANCH_USER        |\
         PERF_SAMPLE_BRANCH_KERNEL      |\
         PERF_SAMPLE_BRANCH_HV)

        uint64_t *mode = (uint64_t *)opt->value;
--
Regards
Anshuman Khandual
Quoted 6 lines
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Anshuman Khandual

author

2012-05-25 10:49 UTC #5 lore raw

Re: [PATCH] perf record: Fixing record option data type in parse_branch_stack

In reply to Stephane Eranian

Quoted 10 lines · Stephane Eranian

Hi,
It should be something like that instead:
tools/perf/perf.h +0 -0
index 8a9687e..c9ca7c4 100644

tools/perf/perf.h struct perf_record_opts { · line 234

Quoted 8 lines

tools/perf/perf.h +1 -1

@@ -228,7 +228,7 @@ struct perf_record_opts {
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int user_freq;
-       int          branch_stack;
+       u64          branch_stack;
        u64          default_interval;
        u64          user_interval;
 };

Agreed. I just tried to fix the problem where it was more evident because of the data type mismatch.

Arnaldo Carvalho de Melo

2012-05-25 14:47 UTC #6 lore raw

Re: [PATCH] perf record: Fixing record option data type in parse_branch_stack

Em Fri, May 25, 2012 at 04:02:15PM +0530, Anshuman Khandual escreveu:

Quoted 17 lines · Stephane Eranian

Quoted 16 lines

Hi,

It should be something like that instead:

tools/perf/perf.h +1 -1

index 8a9687e..c9ca7c4 100644
@@ -228,7 +228,7 @@ struct perf_record_opts {
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int user_freq;
-       int          branch_stack;
+       u64          branch_stack;
        u64          default_interval;

Quoted 8 lines

tools/perf/perf.h struct perf_record_opts { · line 234
Quoted 3 lines
tools/perf/perf.h +0 -0
        u64          user_interval;
 };
Agreed. I just tried to fix the problem where it was more evident because of the data type mismatch.

Can you send an updated patch on a message just for that?

- Arnaldo