Encountered this problem in syslog
kernel:BUG: soft lockup - CPU#2 stuck for 22s! [process name:PID]
According to this link : http://askubuntu.com/questions/401736/what-does-cpu0-stuck-mean
The Linux kernel has a process which monitors each CPU on the system.
There are special interrupt(s) in the kernel. This interrupt(s)
function calls a soft-lockup counter, it will compare the current time
stamp with the specific kernel CPU data structure time information. If
it looks like the current time stamp is greater than the defined
threshold (in seconds) later as compared to the stored time stamp, it is
assumed that the monitoring process or watchdog thread(s) have not
executed in a respectable amount of time.
Why or how can a CPU soft lock occur? How can a CPU get locked if the
kernel is carefully scheduling CPU access? Basically any poorly written
code that loops a lot or infinitely, would own a CPU and get some
priority. It can be a programming problem or 3rd party software.
Locking issues in drivers. Even kernel bugs in important drivers or
the scheduler. A scheduler could tell schedule a driver routine to run
and if that driver has problems and doesn’t check on it, that driver
routine could own or hog that CPU for a longtime. By definition as
described above, the watchdog would catch this and issue a soft lockup
alert.
No comments:
Post a Comment