[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference turris::digital_unix

Title:DIGITAL UNIX(FORMERLY KNOWN AS DEC OSF/1)
Notice:Welcome to the Digital UNIX Conference
Moderator:SMURF::DENHAM
Created:Thu Mar 16 1995
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:10068
Total number of notes:35879

8700.0. "Uninterruptable sleep state in ps?" by NETRIX::"[email protected]" (Keith Austin) Tue Feb 04 1997 13:39

When I do a "ps auxw", a few processes have a state of 'U', which according to
the man page is a "Uninterruptible sleeping process."  How can a process get 
into this state?  In this particular case, the process is switching between
U,S,
and R states (and process is running much slower than normal); however, I have

seen some processes get into this U state, never come out and couldn't be 
killed by user or superuser.

TIA,

Keith Austin

PS.  O/S is DU V3.2G

[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name
DateLines
8700.1Some infoRHETT::PARKERThu Feb 06 1997 11:5117
    
    Hi Keith,
    
    When a process blocks waiting for I/O in a device driver, the
    driver often uses sleep/wakeup as a synchronization mechanism. 
    In the call to sleep/mpsleep, one specifies if the sleep can
    be interrupted by OR'ing in a flag called PCATCH. If this is
    not OR'ed in, the sleep cannot be interrupted by a signal. The
    signal will get OR'ed into the mask for later delivery. 
    
    What do the processes you are interested in do? Could be the
    disk/tape driver or pseudo-tty driver or who knows...
    
    Hth,
    
    Lee
    
8700.2SMURF::DENHAMDigital UNIX KernelThu Feb 06 1997 13:243
    For these kinds of issues, kernel stack traces of the stuck
    processes is most helpful. Also knowing who's the parent process
    can help.