RE: [RFT PATCH] x86/pat: Fix set_mce_nospec() for pmem

["Luck, Tony" <tony.luck@intel.com>]

>I don't mean from the hw aspect but from the OS one: my simple thinking
> is, *if* a page is marked as HW poison, any further mapping or accessing
> of the page frame is prevented by the mm code.
>
> So you can't access *any* bits there so why do we even bother with whole
> or not whole page? Page is gone...

See the comment above set_mce_nospec() ...

/*
 * Prevent speculative access to the page by either unmapping
 * it (if we do not require access to any part of the page) or
 * marking it uncacheable (if we want to try to retrieve data
 * from non-poisoned lines in the page).
 */
static inline int set_mce_nospec(unsigned long pfn, bool unmap)


It's a choice as to whether the whole page is gone or not. The history for
this is using pmem as storage. The filesystem block size may be less than
the page size. An error in a "block" should only result in that block disappearing
from the file, not the surrounding 4k.

Of course that only works for users of read(2)/write(2) to access the file.
They can't mmap(2) and just get the good bits.

-Tony