On Thu, Sep 03, 2020 at 12:34:04PM +0100, Robin Murphy wrote: > The attempt to handle huge page allocations was originally added since > the comments around stripping __GFP_COMP in other implementations were > nonsensical, and we naively assumed that split_huge_page() could simply > be called equivalently to split_page(). It turns out that this doesn't > actually work correctly, so just get rid of it - there's little point > going to the effort of allocating huge pages if we're only going to > split them anyway. > > Reported-by: Roman Gushchin <guro@xxxxxx> > Signed-off-by: Robin Murphy <robin.murphy@xxxxxxx> This looks sensibe. We really need to clear it in generic code, but last time I checked there were one or two buggy drivers that assumed __GFP_COMP works and actually gives a compound page (iirc legacy drm stuff). All that stuff really needs fixing.. Reviewed-by: Christoph Hellwig <hch@xxxxxx>
