lock_page/LockPage/UnlockPage

lock_page/LockPage/UnlockPage

[Jim Mostek]

Just a minor nit, but it seems to me that if UnlockPage wakes up
sleepers, LockPage should go to sleep.

The interface:

	LockPage
	UnlockPage
	TryLockPage(page)

should all be bit operators and then:

	lock_page
	unlock_page
	trylock_page

should be the ones that actually sleep/wakeup.


Not a big deal, but ....

Jim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/

Re: lock_page/LockPage/UnlockPage

[Andrea Arcangeli]

On Mon, 10 Apr 2000, Jim Mostek wrote:

>Just a minor nit, but it seems to me that if UnlockPage wakes up
>sleepers, LockPage should go to sleep.

LockPage can be executed only in places where you know the page to be
unlocked. You could use it for example to set the PG_locked bitflag before
adding the page to the hashtable.

LockPage could be implemented with:

	if (test_and_set_bit(PG_locked, &page->flags)
		BUG();

And TryLockPage() should really not be changed since by design it's only a
bitflag operator and it's right it to remains so.

Only UnlockPage() should be called unlock_page() if you care about style.

Andrea

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/