On 06.10.21 17:28, Al Viro wrote:
On Wed, Oct 06, 2021 at 05:23:49PM +0200, David Hildenbrand wrote:
On 06.10.21 17:22, Al Viro wrote:
On Wed, Oct 06, 2021 at 03:58:14PM +0100, Matthew Wilcox wrote:
David expressed some unease about the lack of typesafety in patches
1 & 2 of the page->slab conversion [1], and I'll admit to not being
particularly a fan of passing around an unsigned long. That crystallised
in a discussion with Kent [2] about how to lock a page when you don't know
its type (solution: every memory descriptor type starts with a
pgflags_t)
Why bother making it a struct? What's wrong with __bitwise and letting
sparse catch conversions?
As I raised in my reply, we store all kinds of different things in
page->flags ... not sure if that could be worked around somehow.
What of that? Inline helpers with force-casts for accessing those and
that's it...
It feels to me like using __bitwise for access checks and then still
modifying the __bitwise fields randomly via a backdoor. But sure, if it
works, I'll be happy if we can use that.
--
Thanks,
David / dhildenb
