Have experienced today strange issue with my fresh RELENG_7 ( sources from yesterday April 13, userland and kernel in sync ) - sometimes some processes stop running without any visible reason. Have seen it twice today during KDE compilation - make process just waiting for something while nothing else compiles or some other way prevents this process from running ( no SIGSTOP performed, no Ctrl-S performed on console). Well it possible related to ( or triggered by ) new ports compilation feature for SMP machines ( most of KDE port marked as MAKE_JOBS_SAFE=yes ) but anyway this stuck process behavior is strange. kill -SIGCONT not helps, but it killable ( Ctrl-C helps at least ) Well some info related to this process: procstat -t 63472 output: PID TID COMM TDNAME CPU PRI STATE WCHAN 63472 100059 make - 1 92 sleep wait procstat -kk 63472 output: PID TID COMM TDNAME KSTACK 63472 100059 make - mi_switch+0x2c8 sleepq_switch+0xd9 sleepq_catch_signals+0x239 sleepq_wait_sig+0x14 _sleep+0x307 kern_wait+0xa36 wait4+0x3b syscall+0x2b3 Xint0x80_syscall+0x20 Some related sysctls output: kern.smp.cpus: 2 kern.smp.disabled: 0 kern.smp.active: 1 uname -msr output: FreeBSD 7.2-PRERELEASE i386
On Tue, 14 Apr 2009 15:04:11 +0200, Oleg V. Nauman <oleg@opentransfer.com> wrote:> > Have experienced today strange issue with my fresh RELENG_7 ( sources > from yesterday April 13, userland and kernel in sync ) - sometimes some > processes stop running without any visible reason. Have seen it twice > today during KDE compilation - make process just waiting for something > while nothing else compiles or some other way prevents this process from > running ( no SIGSTOP performed, no Ctrl-S performed on console). Well it > possible related to ( or triggered by ) new ports compilation feature > for SMP machines ( most of KDE port marked as MAKE_JOBS_SAFE=yes ) but > anyway this stuck process behavior is strange. kill -SIGCONT not helps, > but it killable ( Ctrl-C helps at least ) > Well some info related to this process: > procstat -t 63472 output: > > PID TID COMM TDNAME CPU PRI STATE WCHAN > 63472 100059 make - 1 92 sleep wait > > procstat -kk 63472 output: > > PID TID COMM TDNAME KSTACK > 63472 100059 make - mi_switch+0x2c8 > sleepq_switch+0xd9 sleepq_catch_signals+0x239 sleepq_wait_sig+0x14 > _sleep+0x307 kern_wait+0xa36 wait4+0x3b syscall+0x2b3 > Xint0x80_syscall+0x20 > > Some related sysctls output: > > kern.smp.cpus: 2 > kern.smp.disabled: 0 > kern.smp.active: 1 > > uname -msr output: > FreeBSD 7.2-PRERELEASE i386It sounds like http://www.mail-archive.com/freebsd-stable@freebsd.org/msg102628.html . There are some tips in that mailthread about debugging it. Ronald.
Quoting Ronald Klop <ronald-freebsd8@klop.yi.org>:> On Tue, 14 Apr 2009 15:04:11 +0200, Oleg V. Nauman > <oleg@opentransfer.com> wrote: > >> >> Have experienced today strange issue with my fresh RELENG_7 ( >> sources from yesterday April 13, userland and kernel in sync ) - >> sometimes some processes stop running without any visible reason. >> Have seen it twice today during KDE compilation - make process just >> waiting for something while nothing else compiles or some other >> way prevents this process from running ( no SIGSTOP performed, no >> Ctrl-S performed on console). Well it possible related to ( or >> triggered by ) new ports compilation feature for SMP machines ( >> most of KDE port marked as MAKE_JOBS_SAFE=yes ) but anyway this >> stuck process behavior is strange. kill -SIGCONT not helps, but it >> killable ( Ctrl-C helps at least ) >> Well some info related to this process: >> procstat -t 63472 output: >> >> PID TID COMM TDNAME CPU PRI STATE WCHAN >> 63472 100059 make - 1 92 sleep wait >> >> procstat -kk 63472 output: >> >> PID TID COMM TDNAME KSTACK >> 63472 100059 make - mi_switch+0x2c8 >> sleepq_switch+0xd9 sleepq_catch_signals+0x239 sleepq_wait_sig+0x14 >> _sleep+0x307 kern_wait+0xa36 wait4+0x3b syscall+0x2b3 >> Xint0x80_syscall+0x20 >> >> Some related sysctls output: >> >> kern.smp.cpus: 2 >> kern.smp.disabled: 0 >> kern.smp.active: 1 >> >> uname -msr output: >> FreeBSD 7.2-PRERELEASE i386 > > It sounds like > http://www.mail-archive.com/freebsd-stable@freebsd.org/msg102628.html . > There are some tips in that mailthread about debugging it.Thank you. Looks like it was the same issue.> > Ronald.