Re: page_launder() bug

[Linus Torvalds]

On Mon, 7 May 2001, David S. Miller wrote:
> 
> Marcelo Tosatti writes:
>  > I just thought about this case:
>  >   
>  > We find a dead swap cache page, so dead_swap_page goes to 1.
>  > 
>  > We call swap_writepage(), but in the meantime the swapin readahead code   
>  > got a reference on the swap map for the page.
>  > 
>  > We write the page out because "(swap_count(page) > 1)", and we may
>  > not have __GFP_IO set in the gfp_mask. Boom.

Yes. That looks a lot easier to trigger than my "slow memory
leak" schenario.

> Hmmm, can't this happen without my patch?

No. The old code would never try to write anything if __GFP_IO wasn't set,
because "launder_loop" would never become non-zero.

			Linus