Debian Patches
Status for ltt-control/2.16.0-3
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| fix-lttng-health-check-manpage.patch | Fix lintian manpage-has-bad-whatis-entry | no | ||||
| 0001-fix-liblttng-ctl-stub-error_log_time-with-CONFIG_RCU.patch | [PATCH lttng-tools] fix: liblttng-ctl: stub 'error_log_time' with CONFIG_RCU_TLS | Michael Jeanson <mjeanson@efficios.com> | no | 2025-07-10 | ||
| fix_test_notification_ust_buffer_usage_1.patch | Tests: fix: skip buffer usage notification channel test without gdb The last test case of the buffer usage notification tests pauses the consumer daemon using gdb. The other tests that need gdb skip when it isn't available. |
Jérémie Galarneau <jeremie.galarneau@efficios.com> | no | 2026-09-02 | ||
| fix_test_notification_ust_buffer_usage_2.patch | Tests: fix: buffer usage notification test hangs when gdb fails OBSERVED ISSUE ━━━━━━━━━━━━━━ The fifth test case of test_notification_ust_buffer_usage.py hangs until the test suite's timeout expires. The Debian i386 builders frequently hit when building the 2.16.0 package. While the test hangs, the event generator keeps re-launching the gen-ust-events application in a loop. CAUSE ━━━━━ test_buffer_usage_notification_channel() pauses the consumer daemon's data consumption by attaching gdb to it and setting `data_consumption_paused` (a global variable that controls whether or not buffers are consumed). The return values of stop_consumer_by_pid() and resume_consumer_by_pid() are ignored, so the error caused by a missing gdb binary (or one that fails to attach) isn't handled. It is then possible for the consumer daemon to drain the buffers as fast as the generator fills them. The "high buffer usage" notification the test waits for is only emitted if the once-per-second usage sample happens to land while a stream is more than 90% full, which depends on a ton of factor (machine load, available I/O, etc.) Essentially, the other builders pass by "luck" and the full buffers are observed at some point (presumably because those machines are more contended). SOLUTION ━━━━━━━━ Check the exit status of the gdb command and fail the test case when the consumer daemon's data consumption could not be paused or resumed instead of waiting forever for a notification that can't reliably be emitted. |
Jérémie Galarneau <jeremie.galarneau@efficios.com> | no | 2026-09-02 | ||
| fix-tests-uniproc.patch | Skip the tests that require multiples CPUS on single-CPU machines. | Michael Jeanson <mjeanson@debian.org> | no | debian |
All known versions for source package 'ltt-control'
- 2.16.0-3 (forky, sid)
- 2.13.15-2 (trixie)
- 2.13.9-1+deb12u1 (bookworm)
