Trying to kill a process that just won't die...

[don]

Felipe Alfaro Solana said:
> On Tue, 2004-01-20 at 05:15, Gregory Gulik wrote:
>> F S   UID   PID  PPID  C PRI  NI ADDR    SZ WCHAN  TTY          TIME CMD
>> 1 D     0 11952     1  0  80   2    -  1341 wait_o ?        00:00:00
>> gtar
>
> This process is blocked waiting for I/O, so I can't be killed until the
> I/O request is completed. This is something usual for processes which
> use NFS: when the remote NFS dies or stops responding, the client
> program stays at D and can't be killed.

I find the above very interesting, and it reminds me of the purpose of the
"MIH routine" in mainframe systems... MIH = Missing Interrupt Handler. In
mainframe parlance, if the "device end interrupt" got "lost", the
associated IO device would appear hung and no further IO to/from it would
occur. The MIH routine detects a "stalled device queue" and re-initiates
the IO. (Which may fail with an IO error, but that's OK, the error can be
processed.

Maybe Linuc needs an "MIH routine, especially for network devices.