possible bug
Im spannish user of linux, I have detected an error on my distro and I ask to local average users and told me that prob. is a kernel bug. Sometimes when I do a kill -9 pid the command dont tells any error but no pid die, the task still freeze but the os works perfect and I can continue using my linux box with any problem, but I read that a kill -) command MUST finnish the running task, Im quite newbie with linux but I hope is a real bug and dont make you losing time.Normaly the freeze task is edonkey or overnet.
My system specs:
amd k7-900 448 ram
hd ide ibm 40 gb 7200 rpm
voodoo3
mainboard msi ms-6330
distro: MDK 9.0 kernel: 2.4.19-16 mdk
I can make an ssh account on my machine when the problem appear to let you enter on my linux and watch the problem or I can send you whatever you want.
see you
Eduardo Galan de Andres
Spain, Madrid
01-10-2002
Hi Eduardo :)
Quoted 3 lines
Im spannish user of linux, I have detected an error on my distro and I ask to local average users and told me that prob. is a kernel bug.
I'm spanish, too ;)) But let's go to the matter: probably the PID
you're trying to 'kill -9' is stuck in 'D' state (or any other uninterruptible state), so it's not a kernel bug ;) If this is not the case, you may have hit a kernel bug.
Quoted 2 lines
but I read that a kill -)
command MUST finnish the running task
AFAIK this is not exact ;) The 'SIGKILL' signal (that is, 9),
cannot be trapped and so its action cannot be changed. Moreover, it is unblockable, so it's always 'fatal' ;))) it is always *sent*. This doesn't mean that the process will die. If the process is in any uninterruptible state, it won't be interrupted!. You will have to wait until the process is woken up and then it will die.
The more probable state, in my experience, is the 'D' state,
which if I remember well, is something like 'disk sleep', waiting for disk i/o, etc...
Quoted 2 lines
Normaly the freeze task
is edonkey or overnet.
Oh... The eDonkey client... AFAIK, this is closed source, so any
problem related to this client is difficult (if not impossible) to catch...
Raúl
Quoted 3 lines · DervishD
I'm spanish, too ;)) But let's go to the matter: probably the PID
you're trying to 'kill -9' is stuck in 'D' state (or any other uninterruptible state), so it's not a kernel bug ;) If this is not
How come? Process stuck in 'D' state *is* a kernel bug (well, the only legitimate reason for it is a hard,nointr NFS mount and remote server down but I still think that this mount option for NFS should be never used).
--
vda
Hi Denis :)
Quoted 3 lines
Quoted 2 lines
you're trying to 'kill -9' is stuck in 'D' state (or any other uninterruptible state), so it's not a kernel bug ;) If this is not
How come? Process stuck in 'D' state *is* a kernel bug
Yes if it is stuck in that state forever, but I've suffered
processes stuck in 'D' state due to, for example, my CD recorder frozen... It can last more than 600 seconds, and it's not a kernel bug, just a 'long-lasting' transient state ;)
Anyway, last time I had a process stuck in 'D' state was back in
2.2 kernel, more than a year ago.
Raúl