| Message ID | 20240214202055.77776-2-sidhartha.kumar@oracle.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [1/2] mm/migrate_device: further convert migrate_device_unmap() to folios | expand |
Sidhartha Kumar <sidhartha.kumar@oracle.com> writes: > Use folio api functions from the already defined src and dst folio > variables. > > Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> > --- > mm/migrate_device.c | 14 +++++++------- > 1 file changed, 7 insertions(+), 7 deletions(-) > > diff --git a/mm/migrate_device.c b/mm/migrate_device.c > index 9152a329b0a68..a48d5cdb28553 100644 > --- a/mm/migrate_device.c > +++ b/mm/migrate_device.c > @@ -843,17 +843,17 @@ void migrate_device_finalize(unsigned long *src_pfns, > remove_migration_ptes(src, dst, false); > folio_unlock(src); > > - if (is_zone_device_page(page)) > - put_page(page); > + if (folio_is_zone_device(src)) > + folio_put(src); > else > - putback_lru_page(page); > + folio_putback_lru(src); > > if (newpage != page) { > - unlock_page(newpage); > - if (is_zone_device_page(newpage)) > - put_page(newpage); Defining migrate_pfn_to_folio() would also allow the removal of the newpage and page variables entirely which I think would make this clearer. As an aside is there any motivation for making these changes other than as a general cleanup? I ask only because I have been looking at allowing device pages with order > 0 so have some of these clean-ups in a local tree as they're a pre-requisite for that. - Alistair > + folio_unlock(dst); > + if (folio_is_zone_device(dst)) > + folio_put(dst); > else > - putback_lru_page(newpage); > + folio_putback_lru(dst); > } > } > }
On 2/14/24 2:45 PM, Alistair Popple wrote: > > Sidhartha Kumar <sidhartha.kumar@oracle.com> writes: > >> Use folio api functions from the already defined src and dst folio >> variables. >> >> Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> >> --- >> mm/migrate_device.c | 14 +++++++------- >> 1 file changed, 7 insertions(+), 7 deletions(-) >> >> diff --git a/mm/migrate_device.c b/mm/migrate_device.c >> index 9152a329b0a68..a48d5cdb28553 100644 >> --- a/mm/migrate_device.c >> +++ b/mm/migrate_device.c >> @@ -843,17 +843,17 @@ void migrate_device_finalize(unsigned long *src_pfns, >> remove_migration_ptes(src, dst, false); >> folio_unlock(src); >> >> - if (is_zone_device_page(page)) >> - put_page(page); >> + if (folio_is_zone_device(src)) >> + folio_put(src); >> else >> - putback_lru_page(page); >> + folio_putback_lru(src); >> >> if (newpage != page) { >> - unlock_page(newpage); >> - if (is_zone_device_page(newpage)) >> - put_page(newpage); > > Defining migrate_pfn_to_folio() would also allow the removal of the > newpage and page variables entirely which I think would make this > clearer. > > As an aside is there any motivation for making these changes other than > as a general cleanup? I ask only because I have been looking at allowing > device pages with order > 0 so have some of these clean-ups in a local > tree as they're a pre-requisite for that. > > - Alistair > Hello, The motivation is just general cleanup. In folio-compat.c I saw that putback_lru_page() does not have much users left so I could convert them and then just get rid of putback_lru_page(). Should I still continue with a v2 that will include defining a migrate_pfn_to_folio() or wait for your clean-ups? Thanks, Sid >> + folio_unlock(dst); >> + if (folio_is_zone_device(dst)) >> + folio_put(dst); >> else >> - putback_lru_page(newpage); >> + folio_putback_lru(dst); >> } >> } >> } >
Sidhartha Kumar <sidhartha.kumar@oracle.com> writes: > On 2/14/24 2:45 PM, Alistair Popple wrote: >> Sidhartha Kumar <sidhartha.kumar@oracle.com> writes: >> >>> Use folio api functions from the already defined src and dst folio >>> variables. >>> >>> Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> >>> --- >>> mm/migrate_device.c | 14 +++++++------- >>> 1 file changed, 7 insertions(+), 7 deletions(-) >>> >>> diff --git a/mm/migrate_device.c b/mm/migrate_device.c >>> index 9152a329b0a68..a48d5cdb28553 100644 >>> --- a/mm/migrate_device.c >>> +++ b/mm/migrate_device.c >>> @@ -843,17 +843,17 @@ void migrate_device_finalize(unsigned long *src_pfns, >>> remove_migration_ptes(src, dst, false); >>> folio_unlock(src); >>> - if (is_zone_device_page(page)) >>> - put_page(page); >>> + if (folio_is_zone_device(src)) >>> + folio_put(src); >>> else >>> - putback_lru_page(page); >>> + folio_putback_lru(src); >>> if (newpage != page) { >>> - unlock_page(newpage); >>> - if (is_zone_device_page(newpage)) >>> - put_page(newpage); >> Defining migrate_pfn_to_folio() would also allow the removal of the >> newpage and page variables entirely which I think would make this >> clearer. >> As an aside is there any motivation for making these changes other >> than >> as a general cleanup? I ask only because I have been looking at allowing >> device pages with order > 0 so have some of these clean-ups in a local >> tree as they're a pre-requisite for that. >> - Alistair >> > > Hello, > > The motivation is just general cleanup. In folio-compat.c I saw that > putback_lru_page() does not have much users left so I could convert > them and then just get rid of putback_lru_page(). Should I still > continue with a v2 that will include defining a migrate_pfn_to_folio() > or wait for your clean-ups? No, don't wait for mine. Please continue with defining migrate_pfn_to_folio() for v2, I was just curious in case you had some larger goal in mind. Thanks. - Alistair > Thanks, > Sid > >>> + folio_unlock(dst); >>> + if (folio_is_zone_device(dst)) >>> + folio_put(dst); >>> else >>> - putback_lru_page(newpage); >>> + folio_putback_lru(dst); >>> } >>> } >>> } >>
diff --git a/mm/migrate_device.c b/mm/migrate_device.c index 9152a329b0a68..a48d5cdb28553 100644 --- a/mm/migrate_device.c +++ b/mm/migrate_device.c @@ -843,17 +843,17 @@ void migrate_device_finalize(unsigned long *src_pfns, remove_migration_ptes(src, dst, false); folio_unlock(src); - if (is_zone_device_page(page)) - put_page(page); + if (folio_is_zone_device(src)) + folio_put(src); else - putback_lru_page(page); + folio_putback_lru(src); if (newpage != page) { - unlock_page(newpage); - if (is_zone_device_page(newpage)) - put_page(newpage); + folio_unlock(dst); + if (folio_is_zone_device(dst)) + folio_put(dst); else - putback_lru_page(newpage); + folio_putback_lru(dst); } } }
Use folio api functions from the already defined src and dst folio variables. Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> --- mm/migrate_device.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)