Skip to content

rec: add watchdog to detect tlog_rec_transfer spin loop after child exit #378

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
uncaney wants to merge 1 commit into
base: main
Choose a base branch
from
Open

rec: add watchdog to detect tlog_rec_transfer spin loop after child exit #378

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 21 additions & 0 deletions lib/tlog/rec.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@
#include <config.h>
#include <tlog/rec.h>
#include <tlog/rec_item.h>
#include <time.h>
#include <tlog/json_sink.h>
#include <tlog/syslog_json_writer.h>
#ifdef TLOG_JOURNAL_ENABLED
Expand Down Expand Up @@ -834,6 +835,12 @@ tlog_rec_transfer(struct tlog_errs **perrs,
struct tlog_pkt pkt = TLOG_PKT_VOID;
struct tlog_pkt_pos tty_pos = TLOG_PKT_POS_VOID;
struct tlog_pkt_pos log_pos = TLOG_PKT_POS_VOID;
/*
* Watchdog: detect tlog_pkt_pos_cmp / tlog_sink_write spin loop
* that occurs after the wrapped child exits with buffered data
* that the position state machine never finishes flushing.
*/
time_t watchdog_last_progress = time(NULL);

tlog_rec_exit_signum = 0;
tlog_rec_alarm_set = false;
Expand Down Expand Up @@ -894,6 +901,17 @@ tlog_rec_transfer(struct tlog_errs **perrs,
* Transfer I/O and window changes
*/
while (tlog_rec_exit_signum == 0) {
/*
* Watchdog: child has exited but we've made no progress in 30s.
* That's a bug: tlog_pkt_pos_cmp / tlog_sink_write is in a spin
* loop that never reaches a terminal state. Force exit to avoid
* burning a CPU forever (cf armoirjaune incident 2026-05-06).
*/
if (tlog_rec_child_exited && (time(NULL) - watchdog_last_progress > 30)) {
tlog_errs_pushs(perrs, "Watchdog: child exited and no flush progress in 30s, forcing exit");
return_grc = TLOG_GRC_FROM(errno, ETIMEDOUT);
break;
}
/* Expected exit conditions */
if (tlog_rec_child_exited) {
if (grc == TLOG_GRC_FROM(errno, EIO) || (tlog_pkt_is_eof(&pkt))) {
Expand Down Expand Up @@ -932,6 +950,7 @@ tlog_rec_transfer(struct tlog_errs **perrs,
}
break;
}
watchdog_last_progress = time(NULL);
continue;
}

Expand All @@ -942,6 +961,7 @@ tlog_rec_transfer(struct tlog_errs **perrs,
grc = tlog_sink_write(log_sink, &pkt, &log_pos, NULL);
if (grc == TLOG_RC_OK) {
log_pending = true;
watchdog_last_progress = time(NULL);
} else if (grc != TLOG_GRC_FROM(errno, EINTR)) {
return_grc = grc;
TLOG_ERRS_RAISECS(grc, "Failed logging terminal data");
Expand All @@ -967,6 +987,7 @@ tlog_rec_transfer(struct tlog_errs **perrs,
return_grc = grc;
}
} else if (tlog_pkt_is_eof(&pkt)) {
watchdog_last_progress = time(NULL);
tlog_sink_io_close(tty_sink, pkt.data.io.output);
/* Continue if only input was closed */
if (pkt.data.io.output) {
Expand Down
Loading