[coyotos-dev] Sleep for interval

[Charles Landau]

At 12:24 PM -0400 9/28/07, Jonathan S. Shapiro wrote:
>On Tue, 2007-09-25 at 11:43 -0700, Charles Landau wrote:
>>  >In contrast to all other kernel capabilities, which are always
>>  >receiving, the definition of the sleep object is that it will remain in
>>  >a non-receiving state until the invoker's specified interval has run
>>  >out. This has three consequences:
>>  >
>>  >   1. Sleep always happens to the invoker. Process A cannot place process
>  > >      B onto a kernel sleep queue.
>>
>>  So the sleep object cannot be accurately proxied, which is
>>  unfortunate. If you later did decide to implement this outside the
>>  kernel, you could not do so compatibly.
>
>Hmm. This is technically true, but I think that it may not matter.
>
>The difference in behavior is this:
>
>   1. In the kernel case, the invoker appears blocked attempting to
>      invoke.
>   2. In the emulator case, the invoker succeeds, but then remains
>      in the receiving state until the response.

I think it's worse. If you will allow me to describe the scenario in 
KeyKOS/CapROS terms, since I may not be current on the latest 
iteration of Coyotos IPC:

Process A forks the sleep capability, passing as the last key 
parameter a resume key to Process B. In the kernel case, A blocks. In 
the emulator case, A doesn't block (at least not for long, assuming 
the emulator is responsive). I thought this was what you meant when 
you said "Process A cannot place process B onto a kernel sleep queue".

>  > At 1:57 PM -0700 9/14/07, Charles Landau wrote:
>>  >Another notion is system-up time. I can say "sleep until the
>>  >system has run for at least 5 seconds". This might be useful for
>>  >timeouts of other programs.
>>
>>  This request might be difficult to fulfill if it is not supported in
>>  the kernel.
>>
>>  I'm suggesting you give the sleeper the choice of (1) waking up after
>>  x seconds of uptime, or (2) waking up after x seconds or restart
>>  whichever occurs first. Note that uptime gets rolled back on a
>>  restart.
>
>The "sleep until uptime exceeds X" case does not need to be implemented
>by the kernel. It can be implemented by a user-mode agent if the kernel
>will disclose the start tick value.
>
>I am certainly not opposed to this function. In fact, I think it's a
>good idea. I just don't see that the kernel needs to implement it
>directly.

I'm not sure what the "start tick value" is (I thought you started at 
zero), but I agree there are things the kernel can do that will allow 
a user to synthesize this function. It just doesn't seem that hard to 
do in the kernel, and it allows more programs to be oblivious to 
restarts.