[Bug 276967] sysctl dev.hwpstate_intel periodically slow

From: <bugzilla-noreply_at_freebsd.org>
Date: Thu, 18 Apr 2024 17:26:02 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=276967

--- Comment #2 from Thomas Steen Rasmussen / Tykling <thomas@gibfest.dk> ---
ok so today it started acting up again:

------------------
[tykling@nuc1 ~]$ time sysctl dev.hwpstate_intel
dev.hwpstate_intel.5.epp: 50
dev.hwpstate_intel.5.%parent: cpu5
dev.hwpstate_intel.5.%pnpinfo: 
dev.hwpstate_intel.5.%location: 
dev.hwpstate_intel.5.%driver: hwpstate_intel
dev.hwpstate_intel.5.%desc: Intel Speed Shift
dev.hwpstate_intel.4.epp: 50
dev.hwpstate_intel.4.%parent: cpu4
dev.hwpstate_intel.4.%pnpinfo: 
dev.hwpstate_intel.4.%location: 
dev.hwpstate_intel.4.%driver: hwpstate_intel
dev.hwpstate_intel.4.%desc: Intel Speed Shift
dev.hwpstate_intel.3.epp: 50
dev.hwpstate_intel.3.%parent: cpu3
dev.hwpstate_intel.3.%pnpinfo: 
dev.hwpstate_intel.3.%location: 
dev.hwpstate_intel.3.%driver: hwpstate_intel
dev.hwpstate_intel.3.%desc: Intel Speed Shift
dev.hwpstate_intel.2.epp: 50
dev.hwpstate_intel.2.%parent: cpu2
dev.hwpstate_intel.2.%pnpinfo: 
dev.hwpstate_intel.2.%location: 
dev.hwpstate_intel.2.%driver: hwpstate_intel
dev.hwpstate_intel.2.%desc: Intel Speed Shift
load: 14.18  cmd: sysctl 38907 [runnable] 17.57r 0.00u 0.00s 0% 1780k
dev.hwpstate_intel.1.epp: 50
dev.hwpstate_intel.1.%parent: cpu1
dev.hwpstate_intel.1.%pnpinfo: 
dev.hwpstate_intel.1.%location: 
dev.hwpstate_intel.1.%driver: hwpstate_intel
dev.hwpstate_intel.1.%desc: Intel Speed Shift
dev.hwpstate_intel.0.epp: 50
dev.hwpstate_intel.0.%parent: cpu0
dev.hwpstate_intel.0.%pnpinfo: 
dev.hwpstate_intel.0.%location: 
dev.hwpstate_intel.0.%driver: hwpstate_intel
dev.hwpstate_intel.0.%desc: Intel Speed Shift
dev.hwpstate_intel.%parent: 

real    0m56.086s
user    0m0.000s
sys     0m0.002s
[tykling@nuc1 ~]$ 
------------------

I ran procstat -kka while it was hanging and grepped for the pid in the output:

------------------
[tykling@nuc1 ~]$ sudo procstat -kka > procstat.txt
[tykling@nuc1 ~]$ grep 38907 procstat.txt 
38907 112055 sysctl              -                   mi_switch+0xbf
sched_bind+0x72 sysctl_epp_select+0xb3 sysctl_root_handler_locked+0x91
sysctl_root+0x244 userland_sysctl+0x176 sys___sysctl+0x60 amd64_syscall+0x110
fast_syscall_common+0xf8 
[tykling@nuc1 ~]$ 
------------------

Was this what you were looking for? I have the rest of procstat.txt still but I
have to reboot the host now because it is affecting operation. Slowly more and
more alerts for weird network related stuff starts triggering until eventually
I cannot reach it anymore over the network.

Thanks!

-- 
You are receiving this mail because:
You are the assignee for the bug.