Hi Sebastian and Steve, This patch fixes a problem [1] I saw when backporting 'net: Properly annotate the try-lock for the seqlock" from 5.4 to 4.19 which also looks like it should apply to -devel. This was tested on 5.14 but applies cleanly to rt-devel as well. Thanks, Tom [1] https://lore.kernel.org/linux-rt-users/20201110153853.463368981@xxxxxxxxxxx/T/#m0be3a1290486f3493828dedcf7f06d5fbba092a8 v2: Mike Galbraith pointed out that write_sequnlock() already contains __raw_write_seqcount_end(). --- The patch ('net: Properly annotate the try-lock for the seqlock") adds a seqcount_acquire() in qdisc_run_begin() but omits the corresponding seqcount_release() in qdisc_run_end(). Add it unconditionally, since qdisc_run_end() is never called unless qdisc_run_begin() succeeds, and if it succeeds, seqcount_acquire() will have been called. Signed-off-by: Tom Zanussi <zanussi@xxxxxxxxxx> --- include/net/sch_generic.h | 3 +++ 1 file changed, 3 insertions(+) diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index 112d2dca8b08..221470e38031 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -192,7 +192,10 @@ static inline bool qdisc_run_begin(struct Qdisc *qdisc) static inline void qdisc_run_end(struct Qdisc *qdisc) { #ifdef CONFIG_PREEMPT_RT + seqcount_t *s = &qdisc->running.seqcount; + write_sequnlock(&qdisc->running); + seqcount_release(&s->dep_map, 1, _RET_IP_); #else write_seqcount_end(&qdisc->running); #endif -- 2.17.1
