perf powerpc-vpadtl: Fix off-by-one in auxtrace_info minimum size check
Subsystems (2)
PATCH 1/1 perf powerpc-vpadtl: Fix off-by-one in auxtrace_info minimum size check
min_sz is set to sizeof(u64) * POWERPC_VPADTL_TYPE, but the code reads auxtrace_info->priv[POWERPC_VPADTL_TYPE], which needs at least POWERPC_VPADTL_TYPE + 1 elements. POWERPC_VPADTL_TYPE is the first enumerator of the priv index enum (0), so min_sz evaluates to 0 and the check validates only the perf_record_auxtrace_info header itself. A PERF_RECORD_AUXTRACE_INFO event carrying a zero-length priv array then passes the size check, and the subsequent priv[POWERPC_VPADTL_TYPE] read runs one u64 past the validated region.
This is the same off-by-one fixed for Intel PT by commit c4362d5e1a5e ("perf intel-pt: Fix off-by-one in auxtrace_info minimum size check") and for Intel BTS by commit b9fb8225951c ("perf intel-bts: Fix off-by-one in auxtrace_info minimum size check").
Use sizeof(u64) * (POWERPC_VPADTL_TYPE + 1) so the highest accessed priv index is covered by the minimum-size validation.
--- tools/perf/util/powerpc-vpadtl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/powerpc-vpadtl.c +1 -1
@@ -683,7 +683,7 @@ int powerpc_vpadtl_process_auxtrace_info(union perf_event *event, struct perf_session *session) {struct perf_record_auxtrace_info *auxtrace_info = &event->auxtrace_info; -size_t min_sz =sizeof (u64) *POWERPC_VPADTL_TYPE ; +size_t min_sz =sizeof (u64) * (POWERPC_VPADTL_TYPE +1 );struct powerpc_vpadtl *vpa;int err;
Signature · 1 line
2.25.1
[email protected]
AI reviewThank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Pre-existing issues:
- [High] Out-of-bounds memory read when indexing preempt_reasons and dispatch_reasons string arrays with unvalidated payload data.
- [High] Missing minimum size validation for auxtrace buffers before parsing boottb_freq.
- [High] Invalid sizeof(record) usage truncates sample.raw_size to 8 bytes instead of the full struct size, causing downstream out-of-bounds reads. --
--- Patch [1]: [PATCH] perf powerpc-vpadtl: Fix off-by-one in auxtrace_info minimum size check --- commit aa49174859e1464fdc810c66395f70e08682bd05
Author: Wang Yan <[email protected]>
perf powerpc-vpadtl: Fix off-by-one in auxtrace_info minimum size check
This commit fixes an off-by-one error in auxtrace_info minimum size validation. It correctly sizes min_sz to include the highest accessed priv index, preventing an out-of-bounds read when validating the struct size.
tools/perf/util/powerpc-vpadtl.c int powerpc_vpadtl_process_auxtrace_info(union perf_event *event, · line 689
Quoted 13 lines
tools/perf/util/powerpc-vpadtl.c +1
-1@@ -683,7 +683,7 @@ int powerpc_vpadtl_process_auxtrace_info(union perf_event *event, struct perf_session *session) {struct perf_record_auxtrace_info *auxtrace_info = &event->auxtrace_info; -size_t min_sz =sizeof (u64) *POWERPC_VPADTL_TYPE ; +size_t min_sz =sizeof (u64) * (POWERPC_VPADTL_TYPE +1 );struct powerpc_vpadtl *vpa;int err;
tools/perf/util/powerpc-vpadtl.c:powerpc_vpadtl_dump() {
...
if (dtl->timebase != 0) {
printf("dispatch_reason:%s, preempt_reason:%s, "
"enqueue_to_dispatch_time:%d, ready_to_enqueue_time:%d, "
"waiting_to_ready_time:%d\n",
dispatch_reasons[dtl->dispatch_reason],
preempt_reasons[dtl->preempt_reason],
be32_to_cpu(dtl->enqueue_to_dispatch_time),
be32_to_cpu(dtl->ready_to_enqueue_time),
be32_to_cpu(dtl->waiting_to_ready_time));
} else {
...
}[Severity: High]
This isn't a bug introduced by this patch, but can we read out-of-bounds memory here?
The dispatch_reasons and preempt_reasons arrays have sizes 11 and 10 respectively. If the untrusted payload contains byte values that exceed these bounds, it looks like they are used as array indices without validation.
tools/perf/util/powerpc-vpadtl.c:powerpc_vpadtl_decode() {
...
ret = powerpc_vpadtl_get_buffer(vpaq);
if (ret <= 0)
return ret;
boottb = (struct boottb_freq *)vpaq->buffer->data;
if (boottb->timebase == 0) {
vpaq->boot_tb = boottb->boot_tb;
...
}[Severity: High]
This is a pre-existing issue, but is there a missing minimum size check here?
The auxtrace buffer payload is accessed as a struct boottb_freq and boottb->timebase is checked. If the auxtrace buffer provides fewer bytes than the size of the structure, could this read past the end of the provided payload buffer?
tools/perf/util/powerpc-vpadtl.c:powerpc_vpadtl_sample() {
...
static int powerpc_vpadtl_sample(struct powerpc_vpadtl_entry *record,
struct powerpc_vpadtl *vpa, u64 save, int cpu)
{
...
sample.time = save;
sample.raw_data = record;
sample.raw_size = sizeof(record);
event.sample.header.type = PERF_RECORD_SAMPLE;
...
}[Severity: High]
This isn't a bug introduced by this patch, but does this sizeof() truncate the sample raw size?
Because record is a pointer to struct powerpc_vpadtl_entry, sizeof(record) evaluates to the size of a pointer (8 bytes) rather than the full struct size (48 bytes). Could this cause downstream consumers to read out-of-bounds memory when casting the raw data back to the structure?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Quoted 14 lines · Wang Yan
min_szis set to sizeof(u64) *POWERPC_VPADTL_TYPE, but the code reads auxtrace_info->priv[POWERPC_VPADTL_TYPE], which needs at leastPOWERPC_VPADTL_TYPE+ 1 elements.POWERPC_VPADTL_TYPEis the first enumerator of the priv index enum (0), somin_szevaluates to 0 and the check validates only theperf_record_auxtrace_infoheader itself. APERF_RECORD_AUXTRACE_INFOevent carrying a zero-length priv array then passes the size check, and the subsequent priv[POWERPC_VPADTL_TYPE] read runs one u64 past the validated region.This is the same off-by-one fixed for Intel PT by commit
c4362d5e1a5e("perf intel-pt: Fix off-by-one inauxtrace_infominimum size check") and for Intel BTS by commitb9fb8225951c("perf intel-bts: Fix off-by-one inauxtrace_infominimum size check").
Quoted 6 lines
Use sizeof(u64) * (
POWERPC_VPADTL_TYPE+ 1) so the highest accessed priv index is covered by the minimum-size validation.Fixes c4bbd4ec2e50 ("perf powerpc: Process auxtrace events and display in 'perf report -D'")Signed-off-by Wang Yan <[email protected]>
Quoted 17 lines
--- tools/perf/util/powerpc-vpadtl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)tools/perf/util/powerpc-vpadtl.c +1
-1@@ -683,7 +683,7 @@ int powerpc_vpadtl_process_auxtrace_info(union perf_event *event, struct perf_session *session) {struct perf_record_auxtrace_info *auxtrace_info = &event->auxtrace_info; -size_t min_sz =sizeof (u64) *POWERPC_VPADTL_TYPE ; +size_t min_sz =sizeof (u64) * (POWERPC_VPADTL_TYPE +1 );struct powerpc_vpadtl *vpa;int err;
Quoted 15 lines · Adrian Hunter
Quoted 14 lines · Wang Yan
min_szis set to sizeof(u64) *POWERPC_VPADTL_TYPE, but the code reads auxtrace_info->priv[POWERPC_VPADTL_TYPE], which needs at leastPOWERPC_VPADTL_TYPE+ 1 elements.POWERPC_VPADTL_TYPEis the first enumerator of the priv index enum (0), somin_szevaluates to 0 and the check validates only theperf_record_auxtrace_infoheader itself. APERF_RECORD_AUXTRACE_INFOevent carrying a zero-length priv array then passes the size check, and the subsequent priv[POWERPC_VPADTL_TYPE] read runs one u64 past the validated region.This is the same off-by-one fixed for Intel PT by commit
c4362d5e1a5e("perf intel-pt: Fix off-by-one inauxtrace_infominimum size check") and for Intel BTS by commitb9fb8225951c("perf intel-bts: Fix off-by-one inauxtrace_infominimum size check").
Quoted 8 lines
Quoted 6 lines
Use sizeof(u64) * (
POWERPC_VPADTL_TYPE+ 1) so the highest accessed priv index is covered by the minimum-size validation.Fixes c4bbd4ec2e50 ("perf powerpc: Process auxtrace events and display in 'perf report -D'")Signed-off-by Wang Yan <[email protected]>Reviewed-by Adrian Hunter <[email protected]>
Thanks, applied to perf-tools-next, for v7.4.
- Arnaldo