Re: [PATCH 0/2 v2] remove PF_MEMALLOC_NORECLAIM

[Theodore Ts'o]

On Thu, Sep 05, 2024 at 10:05:15AM -0400, Kent Overstreet wrote:
> > That may be the currrent state of affiars; but is it
> > ****guaranteed**** forever and ever, amen, that GFP_KERNEL will never
> > fail if the amount of memory allocated was lower than a particular
> > multiple of the page size?  If so, what is that size?  I've checked,
> > and this is not documented in the formal interface.
> 
> Yeah, and I think we really need to make that happen, in order to head
> off a lot more sillyness in the future.

I don't think there's any "sillyness"; I hear that you believe that
it's silly, but I think what we have is totally fine.

I've done a quick check of ext4, and we do check the error returns in
most if not all of the calls where we pass in __GFP_NOFAIL and/or are
small allocations less than the block size.  We won't crash if someone
sends a patch which violates the documented guarantee of __GFP_NOFAIL.

So what's the sillynesss?

In any case, Michal has said ix-nay on making GFP_KERNEL == GFP_NOFAIL
for small allocations as documented guarantee, as opposed to the way
things work today, so as far as I'm concerned, the matter is closed.

	       	    	      	     	- Ted