Re: instability in -current ?

I am still seeing instability in 9.99.20.

What I do is cvs checkout src ; ./build.sh -j 9 ; then I install the resulting kernel and do it all over again, by first deleting /usr/src/ This takes a day or two, so I get these sort-of 48-hour snapshots.

Trying to build most recently yielded:

Dec 9 11:00:03 S syslogd[190]: restart
Dec 9 13:35:22 S syslogd[228]: restart
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] panic: kernel diagnostic assertion "!cpu_intr_p()" failed: file "/c/usr/src/sys/uvm/pmap/pmap_tlb.c", line 979
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] cpu0: Begin traceback...
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] 0x9de71a44: netbsd:vpanic+0x16c
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] Bad frame pointer: 0x80acbb3c
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] cpu0: End traceback...
Dec 9 13:35:22 S /netbsd:
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] dump to dev 92,1 not possible
Dec 9 13:35:22 S /netbsd: [ 55977.4574372] rebooting...

From another tty was running 'top -S -s1':

load averages: 8.36, 6.89, 5.83; up 0+15:33:50 13:34:19 ( THIS IS IN UTC )
69 processes: 6 runnable, 59 sleeping, 4 on CPU
CPU states: 77.7% user, 0.0% nice, 21.3% system, 1.0% interrupt, 0.0% idle
Memory: 335M Act, 11M Inact, 9952K Wired, 30M Exec, 230M File, 271M Free
Swap: 6723M Total, 115M Used, 6608M Free

PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND
14280 root 25 0 40M 28M CPU/2 0:02 59.20% 10.74% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
2531 root 25 0 40M 27M CPU/0 0:01 36.38% 8.06% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
9996 root 25 0 35M 26M RUN/3 0:01 36.86% 6.69% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
18279 root 25 0 34M 26M CPU/3 0:01 36.06% 6.54% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
18312 root 25 0 35M 25M RUN/0 0:01 35.37% 4.93% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
1848 root 26 0 33M 17M RUN/1 0:00 16.91% 1.61% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
1701 root 77 0 12M 7272K poll/0 0:02 1.69% 1.56% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/bin/nbmake realall
22317 root 25 0 33M 16M RUN/1 0:00 12.00% 0.59% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
9448 root 25 0 32M 16M RUN/1 0:00 5.00% 0.24% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses
7210 root 85 0 9368K 1912K poll/0 0:12 0.15% 0.15% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/bin/nbmake _THISDIR_ build_install
27043 root 85 0 10M 4480K poll/0 0:01 0.10% 0.10% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/bin/nbmake _THISDIR_ dependall
74 root 42 0 8832K 2488K CPU/1 7:50 0.00% 0.00% top -S -s1
25314 root 25 0 33M 16M RUN/0 0:00 0.00% 0.00% /c/usr/src/obj/tooldir.NetBSD-9.99.20-evbarm/libexec/gcc/armv7--netbsdelf-eabihf/8.3.0/cc1 -quiet -I /c/usr/src/lib/libcurses

On Sat, Dec 7, 2019 at 12:20 AM Michael Cheponis <michael.cheponis%gmail.com@localhost> wrote:

Dec 7 07:03:40 S /netbsd: [ 33652.4520831] panic: kernel diagnostic assertion "!cpu_intr_p()" failed: file "/c/usr/src/sys/uvm/pmap/pmap_tlb.c", line 979
Dec 7 07:03:40 S /netbsd: [ 33652.4520831] cpu0: Begin traceback...
Dec 7 07:03:40 S /netbsd: [ 33652.4520831] 0xa2121a44: netbsd:vpanic+0x16c
Dec 7 07:03:40 S /netbsd: [ 33652.4520831] Bad frame pointer: 0x80acbbbc
Dec 7 07:03:40 S /netbsd: [ 33652.4520831] cpu0: End traceback...

Dec 7 07:03:40 S /netbsd: [ 1.0000000] NetBSD 9.99.19 (GENERIC) #0: Fri Dec 6 16:44:02 UTC 2019
Dec 7 07:03:40 S /netbsd: [ 1.0000000] mac%S.Culver.Net@localhost:/c/usr/src/sys/arch/evbarm/compile/obj/GENERIC

System was doing ./build.sh -j 9

This kernel is as of a day or two ago. I've been noticing some stability regression compared with 9.99.17

My local build. This is a 9.99.19 kernel as built on a stable kernel 9.99.17 I use as a reference (because it has been very stable).

RPi 3B+