Skip to contentSkip to conversation

tools/perf: Add bitfield_swap to handle branch_stack endian issue

PATCH Applied Madhavan Srinivasan 4 participants 10 messages Oct 16, 2021 – Oct 28, 2021 perf events

v2 of 3 Applied Latest is v3 →

lore
Share & tools
lore

Madhavan Srinivasan

author
#0 lore raw

PATCH v2 1/2 tools/perf: Add bitfield_swap to handle branch_stack endian issue

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
author
#1 lore raw

PATCH v2 2/2 tools/perf/test: Add endian test for struct branch_flags

Extend sample-parsing test to include branch_flag endian-reverse test. Patch include "util/trace-event.h" to sample-parsing for importing bigendian() and extends samples_same() to include "needs_swap" to detect/enable check for endian-swap.

Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Updated commit message

 tools/perf/tests/sample-parsing.c | 43 +++++++++++++++++++++++++++----
 1 file changed, 38 insertions(+), 5 deletions(-)
tools/perf/tests/sample-parsing.c +38 -5
index 8fd8a4ef97da..a67158c12773 100644
@@ -13,6 +13,7 @@
 #include "evsel.h"
 #include "debug.h"
 #include "util/synthetic-events.h"
+#include "util/trace-event.h"
 
 #include "tests.h"
 

@@ -30,9 +31,18 @@
 	}						\
 } while (0)
 
+/*
+ * Hardcode the expected values for branch_entry flags.
+ * These are based on the input value (213) specified
+ * in branch_stack variable.
+ */
+#define BS_EXPECTED_BE	0xa00d000000000000
+#define BS_EXPECTED_LE	0xd5000000
+#define FLAG(s)	s->branch_stack->entries[i].flags
+
 static bool samples_same(const struct perf_sample *s1,
 			 const struct perf_sample *s2,
-			 u64 type, u64 read_format)
+			 u64 type, u64 read_format, bool needs_swap)
 {
 	size_t i;
 

@@ -100,8 +110,14 @@ static bool samples_same(const struct perf_sample *s1,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		COMP(branch_stack->nr);
 		COMP(branch_stack->hw_idx);
-		for (i = 0; i < s1->branch_stack->nr; i++)
-			MCOMP(branch_stack->entries[i]);
+		for (i = 0; i < s1->branch_stack->nr; i++) {
+			if (needs_swap)
+				return ((bigendian()) ?
+					(FLAG(s2).value == BS_EXPECTED_BE) :
+					(FLAG(s2).value == BS_EXPECTED_LE));
+			else
+				MCOMP(branch_stack->entries[i]);
+		}
 	}
 
 	if (type & PERF_SAMPLE_REGS_USER) {

@@ -248,7 +264,7 @@ static int do_test(u64 sample_type, u64 sample_regs, u64 read_format)
 		},
 	};
 	struct sample_read_value values[] = {{1, 5}, {9, 3}, {2, 7}, {6, 4},};
-	struct perf_sample sample_out;
+	struct perf_sample sample_out, sample_out_endian;
 	size_t i, sz, bufsz;
 	int err, ret = -1;
 

@@ -313,12 +329,29 @@ static int do_test(u64 sample_type, u64 sample_regs, u64 read_format)
 		goto out_free;
 	}
 
-	if (!samples_same(&sample, &sample_out, sample_type, read_format)) {
+	if (!samples_same(&sample, &sample_out, sample_type, read_format, evsel.needs_swap)) {
 		pr_debug("parsing failed for sample_type %#"PRIx64"\n",
 			 sample_type);
 		goto out_free;
 	}
 
+	if (sample_type == PERF_SAMPLE_BRANCH_STACK) {
+		evsel.needs_swap = true;
+		evsel.sample_size = __evsel__sample_size(sample_type);
+		err = evsel__parse_sample(&evsel, event, &sample_out_endian);
+		if (err) {
+			pr_debug("%s failed for sample_type %#"PRIx64", error %d\n",
+				 "evsel__parse_sample", sample_type, err);
+			goto out_free;
+		}
+
+		if (!samples_same(&sample, &sample_out_endian, sample_type, read_format, evsel.needs_swap)) {
+			pr_debug("parsing failed for sample_type %#"PRIx64"\n",
+				 sample_type);
+			goto out_free;
+		}
+	}
+
 	ret = 0;
 out_free:
 	free(event);
Signature · 1 line
2.31.1
#2 lore raw
In reply to Madhavan Srinivasan
Quoted 92 lines · Madhavan Srinivasan

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +49 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];

tools/perf/util/evsel.c int evsel__parse_sample(struct evsel *evsel, union perf_event *event, · line 2478

Quoted 10 lines
tools/perf/util/evsel.c +9 -0
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {

thanks for all the comments, make it easy to review

tools/perf/util/evsel.c int evsel__parse_sample(struct evsel *evsel, union perf_event *event, · line 2490

Quoted 12 lines
tools/perf/util/evsel.c +12 -0
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);

one last thing that confuses me.. the sample is already swapped at this point, right? with perf_event__all64_swap function

should you swap it back first and then do the proper bitfield swap? like we do in PERF_SAMPLE_RAW to get proper u32 values

thanks,
jirka

Quoted 30 lines
tools/perf/util/evsel.c +2 -0
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
author
#3 lore raw
Quoted 120 lines · Jiri Olsa
Quoted 92 lines · Madhavan Srinivasan

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

  tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
  tools/perf/util/evsel.h | 13 ++++++++
  2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +49 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
  	data->weight = *array;
  }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
  int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			struct perf_sample *data)
  {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  	if (type & PERF_SAMPLE_BRANCH_STACK) {
  		const u64 max_branch_nr = UINT64_MAX /
  					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
  		OVERFLOW_CHECK_u64(array);
  		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			return -EFAULT;
 
  		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
  			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];

tools/perf/util/evsel.c int evsel__parse_sample(struct evsel *evsel, union perf_event *event, · line 2478

Quoted 10 lines
tools/perf/util/evsel.c +9 -0
+		} else {
  			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {

thanks for all the comments, make it easy to review

tools/perf/util/evsel.c int evsel__parse_sample(struct evsel *evsel, union perf_event *event, · line 2490

Quoted 12 lines
tools/perf/util/evsel.c +12 -0
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);

one last thing that confuses me.. the sample is already swapped at this point, right? with perf_event__all64_swap function

Quoted 2 lines

should you swap it back first and then do the proper bitfield swap? like we do in PERF_SAMPLE_RAW to get proper u32 values

No, in case of PERF_SAMPLE_RAW type, first word contains the size.

Let me explain with an example (hope will not confuse further :) ), say we have branch_flag.value as 0x4000000000000000 generated in a BE system, which has predict set, now if we undo the swap done by perf_event__all64_swap, value comes out as 0x40 (in a LE system) and if this value fed to bitswap macro, we end up 0x0000000040000000, which is wrong. Instead, by not swapping back, we feed the value in the source-endian to the new macro.

Maddy

Quoted 33 lines

thanks,
jirka

Quoted 30 lines
tools/perf/util/evsel.c +2 -0
+		}
+
  		OVERFLOW_CHECK(array, sz, max_size);
  		array = (void *)array + sz;
  	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
  bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
  bool evsel__is_leader(struct evsel *evsel);
  void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
  #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
Acked-by on v2 1/2
#4 lore raw
In reply to Madhavan Srinivasan
Quoted 9 lines · Madhavan Srinivasan

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Signed-off-by Madhavan Srinivasan <[email protected]>

for both patches

Acked-by Jiri Olsa <[email protected]>

thanks,
jirka

Quoted 135 lines

---
Changelog v1:
- Renamed function and macro
- Added comments in code

 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
maintainer Applied
#5 lore raw

Em Mon, Oct 25, 2021 at 02:07:23PM +0200, Jiri Olsa escreveu:

Quoted 9 lines · Madhavan Srinivasan
Quoted 8 lines

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Quoted 5 lines
Quoted 1 line
Signed-off-by Madhavan Srinivasan <[email protected]>

for both patches

Acked-by Jiri Olsa <[email protected]>

Thanks, applied.

- Arnaldo

Quoted 138 lines

thanks,
jirka

Quoted 135 lines

---
Changelog v1:
- Renamed function and macro
- Added comments in code

 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
Signature · 1 line
- Arnaldo
maintainer
#6 lore raw
In reply to Madhavan Srinivasan

Em Sat, Oct 16, 2021 at 06:20:58PM +0530, Madhavan Srinivasan escreveu:

Quoted 8 lines

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Quoted 5 lines
Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

[acme@quaco perf]$ perf test python
19: 'import perf' in python : FAILED!
[acme@quaco perf]$ perf test -v python
Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc
19: 'import perf' in python :
--- start ---
test child forked, pid 991284
python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: /tmp/build/perf/python/perf.cpython-39-x86_64-linux-gnu.so: undefined symbol: bigendian test child finished with -1
---- end ----
'import perf' in python: FAILED!
[acme@quaco perf]$

Quoted 129 lines
 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
Signature · 1 line
- Arnaldo
#7 lore raw
Quoted 15 lines · Arnaldo Carvalho de Melo

Em Sat, Oct 16, 2021 at 06:20:58PM +0530, Madhavan Srinivasan escreveu:

Quoted 13 lines

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

Please, run 'perf test' before submitting patches, it's not that difficult, please do so.

- Arnaldo

Quoted 146 lines

[acme@quaco perf]$ perf test python
19: 'import perf' in python : FAILED!
[acme@quaco perf]$ perf test -v python
Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc
19: 'import perf' in python :
--- start ---
test child forked, pid 991284
python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: /tmp/build/perf/python/perf.cpython-39-x86_64-linux-gnu.so: undefined symbol: bigendian test child finished with -1
---- end ----
'import perf' in python: FAILED!
[acme@quaco perf]$

Quoted 129 lines
 tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/evsel.h | 13 ++++++++
 2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
 	data->weight = *array;
 }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
 int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			struct perf_sample *data)
 {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 	if (type & PERF_SAMPLE_BRANCH_STACK) {
 		const u64 max_branch_nr = UINT64_MAX /
 					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
 		OVERFLOW_CHECK_u64(array);
 		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
 			return -EFAULT;
 
 		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
 			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
 			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
 		OVERFLOW_CHECK(array, sz, max_size);
 		array = (void *)array + sz;
 	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
 bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
 bool evsel__is_leader(struct evsel *evsel);
 void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
author
#8 lore raw
Quoted 11 lines · Arnaldo Carvalho de Melo
Quoted 9 lines · Arnaldo Carvalho de Melo

Em Sat, Oct 16, 2021 at 06:20:58PM +0530, Madhavan Srinivasan escreveu:

Quoted 8 lines

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

Please, run 'perf test' before submitting patches, it's not that difficult, please do so.

I did. I ran 'perf test' and did not see any fail. I did not do 'perf test python', will check that out. My bad.

Maddy

Quoted 148 lines

- Arnaldo

Quoted 145 lines

[acme@quaco perf]$ perf test python
19: 'import perf' in python : FAILED!
[acme@quaco perf]$ perf test -v python
Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc
19: 'import perf' in python :
--- start ---
test child forked, pid 991284
python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: /tmp/build/perf/python/perf.cpython-39-x86_64-linux-gnu.so: undefined symbol: bigendian test child finished with -1
---- end ----
'import perf' in python: FAILED!
[acme@quaco perf]$

Quoted 129 lines
  tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
  tools/perf/util/evsel.h | 13 ++++++++
  2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
  	data->weight = *array;
  }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
  int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			struct perf_sample *data)
  {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  	if (type & PERF_SAMPLE_BRANCH_STACK) {
  		const u64 max_branch_nr = UINT64_MAX /
  					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
  		OVERFLOW_CHECK_u64(array);
  		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			return -EFAULT;
 
  		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
  			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
  			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
  		OVERFLOW_CHECK(array, sz, max_size);
  		array = (void *)array + sz;
  	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
  bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
  bool evsel__is_leader(struct evsel *evsel);
  void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
  #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1
author
#9 lore raw
In reply to Arnaldo Carvalho de Melo
Quoted 11 lines · Arnaldo Carvalho de Melo
Quoted 9 lines · Arnaldo Carvalho de Melo

Em Sat, Oct 16, 2021 at 06:20:58PM +0530, Madhavan Srinivasan escreveu:

Quoted 8 lines

branch_stack struct has bit field definition which produces different bit ordering for big/little endian. Because of this, when branch_stack sample is collected in a BE system and viewed/reported in a LE system, bit fields of the branch stack are not presented properly. To address this issue, a evsel__bitfield_swap_branch_stack() is defined and introduced in evsel__parse_sample.

Quoted 21 lines
Quoted 5 lines
Quoted 5 lines
Signed-off-by Madhavan Srinivasan <[email protected]>

---
Changelog v1:
- Renamed function and macro
- Added comments in code

Please, run 'perf test' before submitting patches, it's not that difficult, please do so.

- Arnaldo

Quoted 12 lines

[acme@quaco perf]$ perf test python
19: 'import perf' in python : FAILED!
[acme@quaco perf]$ perf test -v python
Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc
19: 'import perf' in python :
--- start ---
test child forked, pid 991284
python usage test: "echo "import sys ; sys.path.append('/tmp/build/perf/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: /tmp/build/perf/python/perf.cpython-39-x86_64-linux-gnu.so: undefined symbol: bigendian test child finished with -1

using tep_is_bigendian() instead of bigendian() fixes it. Will send a V3 with that changes.

Maddy

Quoted 130 lines
Quoted 130 lines

---- end ---- 'import perf' in python: FAILED! [acme@quaco perf]$

Quoted 129 lines
  tools/perf/util/evsel.c | 74 +++++++++++++++++++++++++++++++++++++++--
  tools/perf/util/evsel.h | 13 ++++++++
  2 files changed, 85 insertions(+), 2 deletions(-)
tools/perf/util/evsel.c +72 -2
index dbfeceb2546c..746e642d4d32 100644
@@ -2221,6 +2221,51 @@ void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
  	data->weight = *array;
  }
 
+u64 evsel__bitfield_swap_branch_flags(u64 value)
+{
+	u64 new_val = 0;
+
+	/*
+	 * branch_flags
+	 * union {
+	 * 	u64 values;
+	 * 	struct {
+	 * 		mispred:1	//target mispredicted
+	 * 		predicted:1	//target predicted
+	 * 		in_tx:1		//in transaction
+	 * 		abort:1		//transaction abort
+	 * 		cycles:16	//cycle count to last branch
+	 * 		type:4		//branch type
+	 * 		reserved:40
+	 * 	}
+	 * }
+	 *
+	 * Avoid bswap64() the entire branch_flag.value,
+	 * as it has variable bit-field sizes. Instead the
+	 * macro takes the bit-field position/size,
+	 * swaps it based on the host endianness.
+	 */
+	if (bigendian()) {
+		new_val = bitfield_swap(value, 0, 1);
+		new_val |= bitfield_swap(value, 1, 1);
+		new_val |= bitfield_swap(value, 2, 1);
+		new_val |= bitfield_swap(value, 3, 1);
+		new_val |= bitfield_swap(value, 4, 16);
+		new_val |= bitfield_swap(value, 20, 4);
+		new_val |= bitfield_swap(value, 24, 40);
+	} else {
+		new_val = bitfield_swap(value, 63, 1);
+		new_val |= bitfield_swap(value, 62, 1);
+		new_val |= bitfield_swap(value, 61, 1);
+		new_val |= bitfield_swap(value, 60, 1);
+		new_val |= bitfield_swap(value, 44, 16);
+		new_val |= bitfield_swap(value, 40, 4);
+		new_val |= bitfield_swap(value, 0, 40);
+	}
+
+	return new_val;
+}
+
  int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			struct perf_sample *data)
  {

@@ -2408,6 +2453,8 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  	if (type & PERF_SAMPLE_BRANCH_STACK) {
  		const u64 max_branch_nr = UINT64_MAX /
  					  sizeof(struct branch_entry);
+		struct branch_entry *e;
+		unsigned int i;
 
  		OVERFLOW_CHECK_u64(array);
  		data->branch_stack = (struct branch_stack *)array++;

@@ -2416,10 +2463,33 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
  			return -EFAULT;
 
  		sz = data->branch_stack->nr * sizeof(struct branch_entry);
-		if (evsel__has_branch_hw_idx(evsel))
+		if (evsel__has_branch_hw_idx(evsel)) {
  			sz += sizeof(u64);
-		else
+			e = &data->branch_stack->entries[0];
+		} else {
  			data->no_hw_idx = true;
+			/*
+			 * if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied,
+			 * only nr and entries[] will be output by kernel.
+			 */
+			e = (struct branch_entry *)&data->branch_stack->hw_idx;
+		}
+
+		if (swapped) {
+			/*
+			 * struct branch_flag does not have endian
+			 * specific bit field definition. And bswap
+			 * will not resolve the issue, since these
+			 * are bit fields.
+			 *
+			 * evsel__bitfield_swap_branch_flags() uses a
+			 * bitfield_swap macro to swap the bit position
+			 * based on the host endians.
+			 */
+			for (i = 0; i < data->branch_stack->nr; i++, e++)
+				e->flags.value = evsel__bitfield_swap_branch_flags(e->flags.value);
+		}
+
  		OVERFLOW_CHECK(array, sz, max_size);
  		array = (void *)array + sz;
  	}
tools/perf/util/evsel.h +13 -0
index 1f7edfa8568a..2e82cdbe2c08 100644
@@ -482,4 +482,17 @@ struct evsel *evsel__leader(struct evsel *evsel);
  bool evsel__has_leader(struct evsel *evsel, struct evsel *leader);
  bool evsel__is_leader(struct evsel *evsel);
  void evsel__set_leader(struct evsel *evsel, struct evsel *leader);
+
+/*
+ * Macro to swap the bit-field postition and size.
+ * Used when,
+ * - dont need to swap the entire u64 &&
+ * - when u64 has variable bit-field sizes &&
+ * - when presented in a host endian which is different
+ *   than the source endian of the perf.data file
+ */
+#define bitfield_swap(src, pos, size)	\
+	((((src) >> (pos)) & ((1ull << (size)) - 1)) << (63 - ((pos) + (size) - 1)))
+
+u64 evsel__bitfield_swap_branch_flags(u64 value);
  #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.31.1