On 22-May-01 Matt Dillon wrote:
>:>:
>:>:John Baldwin <[EMAIL PROTECTED]> -- http://www.FreeBSD.org/~jhb/
>:>
>:> Huh? It doesn't look like the same algorithm to me.
>:
>:In exit1() we attempt to free resources early if we can. If we lose the
>:race,
>:we still clean it up in vmspace_free() called from cpu_wait(). If you check
>:the shm pointer and do shmexit() in vmspace_free() just as is done in
>:vmspace_free(), then you will catch any cases that fall through with the shm
>:not being free'd using the same technique currently employed in releasing the
>:vmspace and with a minimal amount of change to the code.
>:
>:--
>:
>:John Baldwin <[EMAIL PROTECTED]> -- http://www.FreeBSD.org/~jhb/
>
> The whole point is to release resources as early as possible. Why would
> you ever want to intentionally introduce a race that will 'sometimes' be
> lost and thus cause a late resource release when you can just as easily
> completely guarentee that the resource will be released early, and thus
> never have to worry about it. That makes no sense at all. From the
> point of view of algorithm design, it's much better to know what *will*
> happen rather then what *might* happen.
Then do you want to fix the race for the vmspace release as well? If so, that
makes sense. Only doing one and not the other makes no sense though. It
seemed you only wanted to close the shmexit() race but not the one for
releasing the user pages. *shrug*
--
John Baldwin <[EMAIL PROTECTED]> -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.Baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!" - http://www.FreeBSD.org/
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
