perf cs-etm: Always allocate memory for cs_etm_queue::prev_packet
PATCH 1/2 perf cs-etm: Always allocate memory for cs_etm_queue::prev_packet
Robert Walker reported a segmentation fault is observed when process CoreSight trace data; this issue can be easily reproduced by the command 'perf report --itrace=i1000i' for decoding tracing data.
If neither the 'b' flag (synthesize branches events) nor 'l' flag (synthesize last branch entries) are specified to option '--itrace', cs_etm_queue::prev_packet will not been initialised. After merging the code to support exception packets and sample flags, there introduced a number of uses of cs_etm_queue::prev_packet without checking whether it is valid, for these cases any accessing to uninitialised prev_packet will cause crash.
As cs_etm_queue::prev_packet is used more widely now and it's already hard to follow which functions have been called in a context where the validity of cs_etm_queue::prev_packet has been checked, this patch always allocates memory for cs_etm_queue::prev_packet.
--- tools/perf/util/cs-etm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
tools/perf/util/cs-etm.c +3 -5
@@ -422,11 +422,9 @@ static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm)if (!etmq->packet) goto out_free; -if (etm->synth_opts.last_branch || etm->sample_branches) { - etmq->prev_packet =zalloc (szp); -if (!etmq->prev_packet) - goto out_free; - } + etmq->prev_packet =zalloc (szp); +if (!etmq->prev_packet) + goto out_free;if (etm->synth_opts.last_branch) {size_t sz =sizeof (struct branch_stack );
Signature · 1 line
2.17.1
PATCH 2/2 perf cs-etm: Don't check cs_etm_queue::prev_packet validity
Since cs_etm_queue::prev_packet is allocated for all cases, it will never be NULL pointer; now validity checking prev_packet is pointless, remove all of them.
--- tools/perf/util/cs-etm.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-)
tools/perf/util/cs-etm.c +1 -5
@@ -979,7 +979,6 @@ static int cs_etm__sample(struct cs_etm_queue *etmq)* PREV_PACKET is a branch. */ if (etm->synth_opts.last_branch && - etmq->prev_packet && etmq->prev_packet->sample_type ==CS_ETM_RANGE && etmq->prev_packet->last_instr_taken_branch)cs_etm__update_last_branch_rb (etmq);
@@ -1012,7 +1011,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq) etmq->period_instructions = instrs_over; } -if (etm->sample_branches && etmq->prev_packet) { +if (etm->sample_branches) {bool generate_sample = false;/* Generate sample for tracing on packet */
@@ -1069,9 +1068,6 @@ static int cs_etm__flush(struct cs_etm_queue *etmq)struct cs_etm_auxtrace *etm = etmq->etm;struct cs_etm_packet *tmp; -if (!etmq->prev_packet) -return 0 ; -/* Handle start tracing packet */ if (etmq->prev_packet->sample_type ==CS_ETM_EMPTY ) goto swap_packet;
Signature · 1 line
2.17.1
Re: [PATCH v1 2/2] perf cs-etm: Don't check cs_etm_queue::prev_packet validity
Quoted 8 lines · Leo Yan
Since
cs_etm_queue::prev_packetis allocated for all cases, it will never be NULL pointer; now validity checkingprev_packetis pointless, remove all of them.Signed-off-by Leo Yan <[email protected]>--- tools/perf/util/cs-etm.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-)
Regards
Rob
Quoted 31 lines
tools/perf/util/cs-etm.c +1
-5@@ -979,7 +979,6 @@ static int cs_etm__sample(struct cs_etm_queue *etmq)* PREV_PACKET is a branch. */ if (etm->synth_opts.last_branch && - etmq->prev_packet && etmq->prev_packet->sample_type ==CS_ETM_RANGE && etmq->prev_packet->last_instr_taken_branch)cs_etm__update_last_branch_rb (etmq);@@ -1012,7 +1011,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq) etmq->period_instructions = instrs_over; } -if (etm->sample_branches && etmq->prev_packet) { +if (etm->sample_branches) {bool generate_sample = false;/* Generate sample for tracing on packet */ @@ -1069,9 +1068,6 @@ static int cs_etm__flush(struct cs_etm_queue *etmq)struct cs_etm_auxtrace *etm = etmq->etm;struct cs_etm_packet *tmp; -if (!etmq->prev_packet) -return 0 ; -/* Handle start tracing packet */ if (etmq->prev_packet->sample_type ==CS_ETM_EMPTY ) goto swap_packet;
tip-bot for Leo Yan
merge bot[tip:perf/urgent] perf cs-etm: Don't check cs_etm_queue::prev_packet validity
In reply to Leo Yanperf cs-etm: Don't check cs_etm_queue::prev_packet validity
Commit-ID: cf0c37b6dbf74fb71bea07b516612d29e00dcbc4 Gitweb: https://git.kernel.org/tip/cf0c37b6dbf74fb71bea07b516612d29e00dcbc4 Author: Leo Yan <[email protected]> AuthorDate: Sun, 28 Apr 2019 16:32:28 +0800 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Thu, 2 May 2019 16:00:20 -0400
perf cs-etm: Don't check cs_etm_queue::prev_packet validity
Since cs_etm_queue::prev_packet is allocated for all cases, it will never be NULL pointer; now validity checking prev_packet is pointless, remove all of them.
--- tools/perf/util/cs-etm.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-)
tools/perf/util/cs-etm.c +1 -5
@@ -981,7 +981,6 @@ static int cs_etm__sample(struct cs_etm_queue *etmq)* PREV_PACKET is a branch. */ if (etm->synth_opts.last_branch && - etmq->prev_packet && etmq->prev_packet->sample_type ==CS_ETM_RANGE && etmq->prev_packet->last_instr_taken_branch)cs_etm__update_last_branch_rb (etmq);
@@ -1014,7 +1013,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq) etmq->period_instructions = instrs_over; } -if (etm->sample_branches && etmq->prev_packet) { +if (etm->sample_branches) {bool generate_sample = false;/* Generate sample for tracing on packet */
@@ -1071,9 +1070,6 @@ static int cs_etm__flush(struct cs_etm_queue *etmq)struct cs_etm_auxtrace *etm = etmq->etm;struct cs_etm_packet *tmp; -if (!etmq->prev_packet) -return 0 ; -/* Handle start tracing packet */ if (etmq->prev_packet->sample_type ==CS_ETM_EMPTY ) goto swap_packet;
Hi,
Quoted 17 lines · Leo Yan
Robert Walker reported a segmentation fault is observed when process CoreSight trace data; this issue can be easily reproduced by the command 'perf report --itrace=i1000i' for decoding tracing data.
If neither the 'b' flag (synthesize branches events) nor 'l' flag (synthesize last branch entries) are specified to option '--itrace',
cs_etm_queue::prev_packetwill not been initialised. After merging the code to support exception packets and sample flags, there introduced a number of uses ofcs_etm_queue::prev_packetwithout checking whether it is valid, for these cases any accessing to uninitialisedprev_packetwill cause crash.As
cs_etm_queue::prev_packetis used more widely now and it's already hard to follow which functions have been called in a context where the validity ofcs_etm_queue::prev_packethas been checked, this patch always allocates memory forcs_etm_queue::prev_packet.
Quoted 8 lines
Reported-by Robert Walker <[email protected]>Suggested-by Robert Walker <[email protected]>Fixes 7100b12cf474 ("perf cs-etm: Generate branch sample for exception packet")Fixes 24fff5eb2b93 ("perf cs-etm: Avoid stale branch samples when flush packet")Signed-off-by Leo Yan <[email protected]>--- tools/perf/util/cs-etm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
I've tested these with the trace from the HiKey960 and the segfault no longer occurs
Regards
Rob
Quoted 20 lines
tools/perf/util/cs-etm.c +3
-5@@ -422,11 +422,9 @@ static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm)if (!etmq->packet) goto out_free; -if (etm->synth_opts.last_branch || etm->sample_branches) { - etmq->prev_packet =zalloc (szp); -if (!etmq->prev_packet) - goto out_free; - } + etmq->prev_packet =zalloc (szp); +if (!etmq->prev_packet) + goto out_free;if (etm->synth_opts.last_branch) {size_t sz =sizeof (struct branch_stack );
Em Sun, Apr 28, 2019 at 04:32:27PM +0800, Leo Yan escreveu:
Quoted 12 lines
Robert Walker reported a segmentation fault is observed when process CoreSight trace data; this issue can be easily reproduced by the command 'perf report --itrace=i1000i' for decoding tracing data.
If neither the 'b' flag (synthesize branches events) nor 'l' flag (synthesize last branch entries) are specified to option '--itrace',
cs_etm_queue::prev_packetwill not been initialised. After merging the code to support exception packets and sample flags, there introduced a number of uses ofcs_etm_queue::prev_packetwithout checking whether it is valid, for these cases any accessing to uninitialisedprev_packetwill cause crash.
Quoted 8 lines
As
cs_etm_queue::prev_packetis used more widely now and it's already hard to follow which functions have been called in a context where the validity ofcs_etm_queue::prev_packethas been checked, this patch always allocates memory forcs_etm_queue::prev_packet.Reported-by Robert Walker <[email protected]>Suggested-by Robert Walker <[email protected]>Fixes 7100b12cf474 ("perf cs-etm: Generate branch sample for exception packet")
Thanks, applied both to perf/urgent, testing them now in the containers.
- Arnaldo
Quoted 27 lines
Fixes 24fff5eb2b93 ("perf cs-etm: Avoid stale branch samples when flush packet")Signed-off-by Leo Yan <[email protected]>--- tools/perf/util/cs-etm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)tools/perf/util/cs-etm.c +3
-5@@ -422,11 +422,9 @@ static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm)if (!etmq->packet) goto out_free; -if (etm->synth_opts.last_branch || etm->sample_branches) { - etmq->prev_packet =zalloc (szp); -if (!etmq->prev_packet) - goto out_free; - } + etmq->prev_packet =zalloc (szp); +if (!etmq->prev_packet) + goto out_free;if (etm->synth_opts.last_branch) {size_t sz =sizeof (struct branch_stack );Signature · 1 line
2.17.1
Signature · 1 line
- Arnaldo
tip-bot for Leo Yan
merge botperf cs-etm: Always allocate memory for cs_etm_queue::prev_packet
Commit-ID: 35bb59c10a6d0578806dd500477dae9cb4be344e Gitweb: https://git.kernel.org/tip/35bb59c10a6d0578806dd500477dae9cb4be344e Author: Leo Yan <[email protected]> AuthorDate: Sun, 28 Apr 2019 16:32:27 +0800 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Thu, 2 May 2019 16:00:20 -0400
perf cs-etm: Always allocate memory for cs_etm_queue::prev_packet
Robert Walker reported a segmentation fault is observed when process CoreSight trace data; this issue can be easily reproduced by the command 'perf report --itrace=i1000i' for decoding tracing data.
If neither the 'b' flag (synthesize branches events) nor 'l' flag (synthesize last branch entries) are specified to option '--itrace', cs_etm_queue::prev_packet will not been initialised. After merging the code to support exception packets and sample flags, there introduced a number of uses of cs_etm_queue::prev_packet without checking whether it is valid, for these cases any accessing to uninitialised prev_packet will cause crash.
As cs_etm_queue::prev_packet is used more widely now and it's already hard to follow which functions have been called in a context where the validity of cs_etm_queue::prev_packet has been checked, this patch always allocates memory for cs_etm_queue::prev_packet.
--- tools/perf/util/cs-etm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
tools/perf/util/cs-etm.c +3 -5
@@ -422,11 +422,9 @@ static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm)if (!etmq->packet) goto out_free; -if (etm->synth_opts.last_branch || etm->sample_branches) { - etmq->prev_packet =zalloc (szp); -if (!etmq->prev_packet) - goto out_free; - } + etmq->prev_packet =zalloc (szp); +if (!etmq->prev_packet) + goto out_free;if (etm->synth_opts.last_branch) {size_t sz =sizeof (struct branch_stack );