Very Computer

by **Fritz** » Wed, 11 Jun 2003 00:10:44

> If my process call a system call and already in kernal mode.
> then I send a kill -9 to my process.
> The system will change to user mode and recive the signal?

Signal delivery occurs by setting a bit in the uarea structure associated
with the process. Typically, checks for posted signals (i.e. looking at
the signal flag in uarea) occur at return from a system call, return from a
trap, and wakeup from sleep.

If your process is in the kernel during a system call, kill -9 will not
interrrupt the system call unless there is code which specifically tests
for signals (i.e. while the system call is waiting for an event).

RFM
--
To reply, translate domain from l33+ 2p33|< to alpha.
4=a 0=o 3=e +=t

by **Amri** » Wed, 11 Jun 2003 13:00:36

> > If my process call a system call and already in kernal mode.
> > then I send a kill -9 to my process.
> > The system will change to user mode and recive the signal?

> Signal delivery occurs by setting a bit in the uarea structure associated
> with the process. Typically, checks for posted signals (i.e. looking at
> the signal flag in uarea) occur at return from a system call, return from a
> trap, and wakeup from sleep.

> If your process is in the kernel during a system call, kill -9 will not
> interrrupt the system call unless there is code which specifically tests
> for signals (i.e. while the system call is waiting for an event).

> RFM

One doubt:-

i think, if the current system call getting executed in the kernel
mode, itself generates the -9 signal (bad argument). Then this process
will be able to handle, this new signal

Amrith

by **Fritz** » Thu, 12 Jun 2003 08:12:00

> i think, if the current system call getting executed in the kernel
> mode, itself generates the -9 signal (bad argument). Then this process
> will be able to handle, this new signal

I don't see that; my strong suspicion is that the signal flag is set in the
process uarea.

RFM
--
To reply, translate domain from l33+ 2p33|< to alpha.
4=a 0=o 3=e +=t

by **Fritz** » Wed, 18 Jun 2003 07:58:04

> i think, if the current system call getting executed in the kernel
> mode, itself generates the -9 signal (bad argument).

Amrith,

It just occurred to me that you might have meant "bad argument to a
function."

If you do something in the kernel that results in a fault (e.g. NULL
pointer dereference or reference to non-existent memory location), the
result is a system panic.

RFM

by **Amri** » Wed, 18 Jun 2003 12:12:15

Fritz.

System panic is rare case, since OS will validate the arguments in the
registers before executing the system call handler.

Amrith