rpi2 head -r327485 (e.g.): rpi2 leaves one "CPU n" always idle for some boots
Allan Jude
allanjude at freebsd.org
Tue Jan 2 20:01:24 UTC 2018
On 2018-01-02 14:48, Mark Millard wrote:
> I've seen this over many versions of head for months
> but have never managed to find a way to force it to
> happen. It just shows up once and a while.
>
> Thus, I'm just dumping out some top and kernel information
> here for reference. I've used:
>
> openssl speed 1>/dev/null 2>&1 &
> openssl speed 1>/dev/null 2>&1 &
> openssl speed 1>/dev/null 2>&1 &
> openssl speed 1>/dev/null 2>&1 &
>
> to give the rpi2 4 active processes. Various outputs
> are from different times without a reboot between.
>
> top -CaePores shows the likes of:
>
> PID USERNAME THR PRI NICE SIZE RES SWAP STATE C TIME CPU COMMAND
> 614 root 1 20 0 10452K 10480K 0K select 1 0:00 0.03% /usr/sbin/ntpd -g -c /etc/ntp.conf -p /var/run/ntpd.pid -f /var/db/ntpd.drift
> 661 root 1 52 0 9984K 6132K 0K select 1 0:00 0.00% /usr/sbin/sshd
> 751 root 1 101 0 7256K 4276K 0K RUN 1 0:28 99.57% openssl speed
> 750 root 1 100 0 7256K 4276K 0K CPU0 0 0:32 94.83% openssl speed
> 753 root 1 86 0 7256K 4276K 0K RUN 3 0:13 52.36% openssl speed
> 752 root 1 86 0 7256K 4276K 0K CPU3 3 0:14 46.54% openssl speed
> 363 root 1 20 0 6428K 3840K 0K select 3 0:00 0.00% /sbin/devd
> . . .
>
> and:
>
> last pid: 754; load averages: 3.70, 2.38, 1.58 up 0+00:16:50 01:59:37
> 21 processes: 5 running, 16 sleeping
> CPU 0: 94.9% user, 0.0% nice, 0.0% system, 5.1% interrupt, 0.0% idle
> CPU 1: 99.6% user, 0.0% nice, 0.0% system, 0.4% interrupt, 0.0% idle
> CPU 2: 0.0% user, 0.0% nice, 0.0% system, 0.0% interrupt, 100% idle
> CPU 3: 100% user, 0.0% nice, 0.0% system, 0.0% interrupt, 0.0% idle
> Mem: 12M Active, 1136K Inact, 56M Wired, 30M Buf, 722M Free
> Swap: 1536M Total, 6M Free
>
> From problem boot to problem boot, the CPU that stays
> idle has varied but usually has been CPU 2. I've never
> seen 2 or more stuck in idle.
>
> show allpcpu shows the likes of:
>
> db> show allpcpu
> Current CPU: 0
>
> cpuid = 0
> dynamic pcpu = 0x3d2540
> curthread = 0xd8478ae0: pid 2032 tid 100150 "openssl"
> curpcb = 0xd852ae98
> fpcurthread = 0xd8478ae0: pid 2032 "openssl"
> idlethread = 0xc376fae0: tid 100002 "idle: cpu0"
> curpmap = 0xd8e43bf4
> curvnet = 0
>
> cpuid = 1
> dynamic pcpu = 0x3998540
> curthread = 0xd7e5b3a0: pid 2031 tid 100173 "openssl"
> curpcb = 0xda7e0e98
> fpcurthread = 0xd7e5b3a0: pid 2031 "openssl"
> idlethread = 0xc376f740: tid 100003 "idle: cpu1"
> curpmap = 0xd8e43ec4
> curvnet = 0
>
> cpuid = 2
> dynamic pcpu = 0x3999540
> curthread = 0xc376f3a0: pid 10 tid 100004 "idle: cpu2"
> curpcb = 0xc378ae98
> fpcurthread = none
> idlethread = 0xc376f3a0: tid 100004 "idle: cpu2"
> curpmap = 0
> curvnet = 0
>
> cpuid = 3
> dynamic pcpu = 0x399a540
> curthread = 0xd8477000: pid 2034 tid 100167 "openssl"
> curpcb = 0xd876de98
> fpcurthread = 0xd8477000: pid 2034 "openssl"
> idlethread = 0xc376f000: tid 100005 "idle: cpu3"
> curpmap = 0xc377ab04
> curvnet = 0
>
> In other words: it appears that the cpuN (here cpu2) is
> left with idle scheduled all the time for some reason.
>
> ps from db> shows things like:
>
>
> db> ps
> pid ppid pgrp uid state wmesg wchan cmd
> 2034 714 2034 0 R+ openssl
> 2033 714 2033 0 R+ CPU 3 openssl
> 2032 714 2032 0 R+ CPU 0 openssl
> 2031 714 2031 0 R+ CPU 1 openssl
>
> (then later:)
>
> db> ps
> pid ppid pgrp uid state wmesg wchan cmd
> 2034 714 2034 0 R+ CPU 3 openssl
> 2033 714 2033 0 R+ openssl
> 2032 714 2032 0 R+ CPU 0 openssl
> 2031 714 2031 0 R+ CPU 1 openssl
>
> There is also:
>
> 10 0 0 0 RL (threaded) [idle]
> 100002 CanRun [idle: cpu0]
> 100003 CanRun [idle: cpu1]
> 100004 CanRun [idle: cpu2]
> 100005 CanRun [idle: cpu3]
>
>
> These are from:
>
> # uname -apKU
> FreeBSD rpi2 12.0-CURRENT FreeBSD 12.0-CURRENT r327485M arm armv7 1200054 1200054
>
>
> ===
> Mark Millard
> markmi at dsl-only.net
>
> _______________________________________________
> freebsd-hackers at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-hackers
> To unsubscribe, send any mail to "freebsd-hackers-unsubscribe at freebsd.org"
>
Have you tried using 'cpuset' to force each of the 4 to run on a
specific CPU? What happens if you force it to run on cpu #2?
--
Allan Jude
More information about the freebsd-hackers
mailing list