Skip to contentSkip to conversation

tools/lib/api: Check nonfilterable in fdarray__filter()

PATCH In review Namhyung Kim 2 participants 3 messages Jul 10

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 tools/lib/api: Check nonfilterable in fdarray__filter()

IIUC fdarray__filter() should check filterable events only and leave non-filterable events to be handled outside. The recent change to set fd to -1 made perf record test stuck forever. Adding the check the flag fixes the problem for me.

Before:

  $ timeout 100 sudo perf test 105
  105: perf record tests                                     : Running (1 active)

Signal (15) while running tests.
Terminating tests with the same signal
Internal test harness failure. Completing any started tests:
105: perf record tests : FAILED!

=== Test Summary ===
Passed main tests : 0

  Passed subtests   : 0
  Skipped tests     : 0
  Failed tests      : 1

List of failed tests:
105: perf record tests

After:

  $ time sudo perf test 105
  105: perf record tests                                     : Ok

=== Test Summary ===
Passed main tests : 1

  Passed subtests   : 0
  Skipped tests     : 0
  Failed tests      : 0

real 0m52.472s
user 0m0.830s
sys 0m0.535s

Fixes f94563fac269 ("perf record: fix poll storm when monitored threads exit")
Cc Jiawei Sun <[email protected]>
---
 tools/lib/api/fd/array.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
tools/lib/api/fd/array.c +3 -2
index ffe8272af59b2da0..b4570e4981492cd7 100644
@@ -117,6 +117,8 @@ int fdarray__filter(struct fdarray *fda, short revents,
 	for (fd = 0; fd < fda->nr; ++fd) {
 		if (!fda->entries[fd].events)
 			continue;
+		if (fda->priv[fd].flags & fdarray_flag__nonfilterable)
+			continue;
 
 		if (fda->entries[fd].revents & revents) {
 			if (entry_destructor)
@@ -132,8 +134,7 @@ int fdarray__filter(struct fdarray *fda, short revents,
 			continue;
 		}
 
-		if (!(fda->priv[fd].flags & fdarray_flag__nonfilterable))
-			++nr;
+		++nr;
 	}
 
 	return nr;
Signature · 1 line
2.55.0.795.g602f6c329a-goog
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] If realloc for fda->priv fails in fdarray__grow(), a dangling pointer is left in fda->entries, leading to a double-free during error cleanup. --

--- Patch [1]: [PATCH] tools/lib/api: Check nonfilterable in fdarray__filter() --- commit f87f1b3c291ca03e96598f96697566a328711de3
Author: Namhyung Kim <[email protected]>

tools/lib/api: Check nonfilterable in fdarray__filter()

IIUC fdarray__filter() should check filterable events only and leave non-filterable events to be handled outside. The recent change to set fd to -1 made perf record test stuck forever. Adding the check the flag fixes the problem for me.

tools/lib/api/fd/array.c int fdarray__filter(struct fdarray *fda, short revents, · line 121

Quoted 10 lines
tools/lib/api/fd/array.c +2 -0
index ffe8272af59b2..b4570e4981492 100644
@@ -117,6 +117,8 @@ int fdarray__filter(struct fdarray *fda, short revents,
 	for (fd = 0; fd < fda->nr; ++fd) {
 		if (!fda->entries[fd].events)
 			continue;
+		if (fda->priv[fd].flags & fdarray_flag__nonfilterable)
+			continue;

[Severity: High]
This is a pre-existing issue, but while looking at this file I noticed a potential double-free in fdarray__grow() during error handling.

If the realloc for fda->entries succeeds but the realloc for fda->priv fails, the error path frees the newly allocated entries:

tools/lib/api/fd/array.c:fdarray__grow() {
    ...
	struct pollfd *entries = realloc(fda->entries, size);

	if (entries == NULL)
		return -ENOMEM;

	priv = realloc(fda->priv, psize);
	if (priv == NULL) {
		free(entries);
		return -ENOMEM;
	}
    ...
}

However, fda->entries is neither updated to point to the new entries block nor cleared to NULL. If realloc moved the allocation to a new block, the old fda->entries pointer is now freed memory.

When fdarray__exit() is later called to clean up, it executes free(fda->entries), which would trigger a double-free on that old pointer.

Would it be better to clear fda->entries to NULL here, or avoid modifying the struct state until both allocations succeed?

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#2 lore raw
Quoted 17 lines · On Fri, Jul 10, 2026 at 08:01:49PM +0000, [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] If realloc for fda->priv fails in fdarray__grow(), a dangling pointer is left in fda->entries, leading to a double-free during error cleanup. --

--- Patch [1]: [PATCH] tools/lib/api: Check nonfilterable in fdarray__filter() --- commit f87f1b3c291ca03e96598f96697566a328711de3
Author: Namhyung Kim <[email protected]>

tools/lib/api: Check nonfilterable in fdarray__filter()

IIUC fdarray__filter() should check filterable events only and leave non-filterable events to be handled outside. The recent change to set fd to -1 made perf record test stuck forever. Adding the check the flag fixes the problem for me.

Quoted 14 lines

tools/lib/api/fd/array.c int fdarray__filter(struct fdarray *fda, short revents, · line 121

Quoted 10 lines
tools/lib/api/fd/array.c +2 -0
index ffe8272af59b2..b4570e4981492 100644
@@ -117,6 +117,8 @@ int fdarray__filter(struct fdarray *fda, short revents,
 	for (fd = 0; fd < fda->nr; ++fd) {
 		if (!fda->entries[fd].events)
 			continue;
+		if (fda->priv[fd].flags & fdarray_flag__nonfilterable)
+			continue;

[Severity: High]
This is a pre-existing issue, but while looking at this file I noticed a potential double-free in fdarray__grow() during error handling.

I'll handle this in a separate patch.

Thanks,
Namhyung

Quoted 33 lines

If the realloc for fda->entries succeeds but the realloc for fda->priv fails, the error path frees the newly allocated entries:

tools/lib/api/fd/array.c:fdarray__grow() {
    ...
	struct pollfd *entries = realloc(fda->entries, size);

	if (entries == NULL)
		return -ENOMEM;

	priv = realloc(fda->priv, psize);
	if (priv == NULL) {
		free(entries);
		return -ENOMEM;
	}
    ...
}

However, fda->entries is neither updated to point to the new entries block nor cleared to NULL. If realloc moved the allocation to a new block, the old fda->entries pointer is now freed memory.

When fdarray__exit() is later called to clean up, it executes free(fda->entries), which would trigger a double-free on that old pointer.

Would it be better to clear fda->entries to NULL here, or avoid modifying the struct state until both allocations succeed?

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1