On Mon, 10 Jun 2013, Sasha Levin wrote:
> > There must be another reason. Lets fix this.
>
> My, I feel silly now.
>
> I was the one who added __GFP_NOFAIL in the first place in
> 2ccd4f4d ("pipe: fail cleanly when root tries F_SETPIPE_SZ
> with big size").
>
> What happens is that root can go ahead and specify any size
> it wants to be used as buffer size - and the kernel will
> attempt to comply by allocation that buffer. Which fails
> if the size is too big.
Could you check that against a boundary? Use vmalloc if larger than a
couple of pages? Maybe PAGE_COSTLY_ORDER or so?
THe higher the order the more likely it is that the allocation will fail.
The PAGE_ORDER_COSTLY (or so) is a reasonable limit as to what size of a
linear contiguous allocation that can be expected to be successful.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>