2.2.14 feels slow and non-interactive.

[Rik van Riel]

On Sun, 23 Jan 2000, Andrea Arcangeli wrote:
> On Sat, 22 Jan 2000, Rik van Riel wrote:
> 
> >Please answer this question:
> >
> >	Why are there both a __GFP_MED *and* a __GFP_HIGH?
> 
> I'll answer your question once you'll have answered me why there
> are both GFP_USER and GFP_KERNEL in 2.2.15pre4.

Historically these two were different because of the 
GFP flags we passed to try_to_free_pages(), we tried
harder for kernel pages.

And please, NEVER EVER again suggest using the last
few free pages for GFP_KERNEL allocations. We _must_
let PF_MALLOC allocations succeed and GFP_ATOMIC ones
if we can.

GFP_KERNEL can lead to a recursive allocation, that
will give deadlocks when you let them use the last
few free pages. All suggestions to change the semantics
back to the (deadlocking in several instances!) ones
in 2.2.14 will be ignored, perseverance will own you
your own .procmailrc entry.

regards,

Rik
--
The Internet is not a network of computers. It is a network
of people. That is its real strength.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/