Nouveau - Apr 2020 - [PATCH 2/2] mm/thp: Rename pmd_mknotpresent() as pmd_mknotvalid()

On Fri, Mar 20, 2020 at 10:24:17AM +0530, Anshuman Khandual
wrote:
> pmd_present() is expected to test positive after pmdp_mknotpresent() as the
> PMD entry still points to a valid huge page in memory. pmdp_mknotpresent()
> implies that given PMD entry is just invalidated from MMU perspective while
> still holding on to pmd_page() referred valid huge page thus also clearing
> pmd_present() test. This creates the following situation which is counter
> intuitive.
> 
> [pmd_present(pmd_mknotpresent(pmd)) = true]
> 
> This renames pmd_mknotpresent() as pmd_mknotvalid() reflecting the
helper's
> functionality more accurately while changing the above mentioned situation
> as follows. This does not create any functional change.
> 
> [pmd_present(pmd_mknotvalid(pmd)) = true]
> 
> This is not applicable for platforms that define own pmdp_invalidate() via
> __HAVE_ARCH_PMDP_INVALIDATE. Suggestion for renaming came during a previous
> discussion here.
Bikeshed alert: maybe pmd_mkinvalid() would be better, given that this is
a one-trick pony for pmdp_invalidate()?

Will

On 04/21/2020 02:33 AM, Will Deacon wrote:
> On Fri, Mar 20, 2020 at 10:24:17AM +0530, Anshuman Khandual wrote:
>> pmd_present() is expected to test positive after pmdp_mknotpresent() as
the
>> PMD entry still points to a valid huge page in memory.
pmdp_mknotpresent()
>> implies that given PMD entry is just invalidated from MMU perspective
while
>> still holding on to pmd_page() referred valid huge page thus also
clearing
>> pmd_present() test. This creates the following situation which is
counter
>> intuitive.
>>
>> [pmd_present(pmd_mknotpresent(pmd)) = true]
>>
>> This renames pmd_mknotpresent() as pmd_mknotvalid() reflecting the
helper's
>> functionality more accurately while changing the above mentioned
situation
>> as follows. This does not create any functional change.
>>
>> [pmd_present(pmd_mknotvalid(pmd)) = true]
>>
>> This is not applicable for platforms that define own pmdp_invalidate()
via
>> __HAVE_ARCH_PMDP_INVALIDATE. Suggestion for renaming came during a
previous
>> discussion here.
> 
> Bikeshed alert: maybe pmd_mkinvalid() would be better, given that this is
> a one-trick pony for pmdp_invalidate()?
I had thought about making it pmd_mkinvalid() earlier. But as we were replacing
pmd_mknotpresent(), hence went with similar pattern pmd_mknotvalid() which was
originally suggested by Catalin. There is an existing pte_mknotpresent() in arc
platform as well. I dont have a very strong opinion either way, will be happy
to rename. But then still wondering if we really need to.
> 
> Will
>

On Tue, Apr 21, 2020 at 04:57:26AM +0530, Anshuman Khandual
wrote:
> 
> 
> On 04/21/2020 02:33 AM, Will Deacon wrote:
> > On Fri, Mar 20, 2020 at 10:24:17AM +0530, Anshuman Khandual wrote:
> >> pmd_present() is expected to test positive after
pmdp_mknotpresent() as the
> >> PMD entry still points to a valid huge page in memory.
pmdp_mknotpresent()
> >> implies that given PMD entry is just invalidated from MMU
perspective while
> >> still holding on to pmd_page() referred valid huge page thus also
clearing
> >> pmd_present() test. This creates the following situation which is
counter
> >> intuitive.
> >>
> >> [pmd_present(pmd_mknotpresent(pmd)) = true]
> >>
> >> This renames pmd_mknotpresent() as pmd_mknotvalid() reflecting the
helper's
> >> functionality more accurately while changing the above mentioned
situation
> >> as follows. This does not create any functional change.
> >>
> >> [pmd_present(pmd_mknotvalid(pmd)) = true]
> >>
> >> This is not applicable for platforms that define own
pmdp_invalidate() via
> >> __HAVE_ARCH_PMDP_INVALIDATE. Suggestion for renaming came during a
previous
> >> discussion here.
> > 
> > Bikeshed alert: maybe pmd_mkinvalid() would be better, given that this
is
> > a one-trick pony for pmdp_invalidate()?
> 
> I had thought about making it pmd_mkinvalid() earlier. But as we were
replacing
> pmd_mknotpresent(), hence went with similar pattern pmd_mknotvalid() which
was
> originally suggested by Catalin. There is an existing pte_mknotpresent() in
arc
> platform as well. I dont have a very strong opinion either way, will be
happy
> to rename. But then still wondering if we really need to.
I just think that having pmdp_invalidate() call pmd_mkinvalid() makes a lot
of sense and, since this is a pure renaming patch, then that's worth taking
into consideration.

If you go with pmd_mkinvalid(), then:

Acked-by: Will Deacon <will at kernel.org>

Will