[Bug 278826] [hpet] cdev->si_refcount leakage when enable hpet as timecounter hardware
Date: Tue, 07 May 2024 11:47:34 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=278826 --- Comment #4 from Austin Zhang <austin.zhang@dell.com> --- Normally, when a process exits, the si_refcount associated with /dev/hpet0 should return to its expected value, assuming it’s incremented and decremented correctly during cdev ops. However, we can inflate the si_refcount using a simple bash script below: while true; do uname >/dev/null;done This unusually high count suggests that the si_refcount isn't being managed properly ``` [root@freebsd-main ~]# while true; do uname >/dev/null;done ^C [root@freebsd-main ~]# dtrace -n 'fbt::dev_ref:entry {printf("[%s]: invoke dev_ref: %s, refcount:%d", execname, args[0]->si_name, args[0]->si_refcount)}' dtrace: description 'fbt::dev_ref:entry ' matched 1 probe CPU ID FUNCTION:NAME 1 43845 dev_ref:entry [uname]: invoke dev_ref: hpet0, refcount:42439 ``` -- You are receiving this mail because: You are the assignee for the bug.