| Message ID | 20230414110821.21548-3-p.raghav@samsung.com (mailing list archive) |
|---|---|
| State | Mainlined, archived |
| Headers | show |
| Series | convert create_page_buffers to create_folio_buffers | expand |
On Fri, Apr 14, 2023 at 01:08:19PM +0200, Pankaj Raghav wrote: > Folio version of alloc_page_buffers() helper. This is required to convert > create_page_buffers() to create_folio_buffers() later in the series. > > It removes one call to compound_head() compared to alloc_page_buffers(). I would convert alloc_page_buffers() to folio_alloc_buffers() and add static struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size, bool retry) { return folio_alloc_buffers(page_folio(page), size, retry); } in buffer_head.h (there are only five callers, so this feels like a better tradeoff than creating a new function) > Signed-off-by: Pankaj Raghav <p.raghav@samsung.com> > --- > fs/buffer.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 59 insertions(+) > > diff --git a/fs/buffer.c b/fs/buffer.c > index 44380ff3a31f..0f9c2127543d 100644 > --- a/fs/buffer.c > +++ b/fs/buffer.c > @@ -900,6 +900,65 @@ struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size, > } > EXPORT_SYMBOL_GPL(alloc_page_buffers); > > +/* > + * Create the appropriate buffers when given a folio for data area and > + * the size of each buffer.. Use the bh->b_this_page linked list to > + * follow the buffers created. Return NULL if unable to create more > + * buffers. > + * > + * The retry flag is used to differentiate async IO (paging, swapping) > + * which may not fail from ordinary buffer allocations. > + */ > +struct buffer_head *alloc_folio_buffers(struct folio *folio, unsigned long size, > + bool retry) > +{ > + struct buffer_head *bh, *head; > + gfp_t gfp = GFP_NOFS | __GFP_ACCOUNT; > + long offset; > + struct mem_cgroup *memcg, *old_memcg; > + > + if (retry) > + gfp |= __GFP_NOFAIL; > + > + /* The folio lock pins the memcg */ > + memcg = folio_memcg(folio); > + old_memcg = set_active_memcg(memcg); > + > + head = NULL; > + offset = folio_size(folio); > + while ((offset -= size) >= 0) { > + bh = alloc_buffer_head(gfp); > + if (!bh) > + goto no_grow; > + > + bh->b_this_page = head; > + bh->b_blocknr = -1; > + head = bh; > + > + bh->b_size = size; > + > + /* Link the buffer to its folio */ > + set_bh_folio(bh, folio, offset); > + } > +out: > + set_active_memcg(old_memcg); > + return head; > +/* > + * In case anything failed, we just free everything we got. > + */ > +no_grow: > + if (head) { > + do { > + bh = head; > + head = head->b_this_page; > + free_buffer_head(bh); > + } while (head); > + } > + > + goto out; > +} > +EXPORT_SYMBOL_GPL(alloc_folio_buffers); > + > static inline void > link_dev_buffers(struct page *page, struct buffer_head *head) > { > -- > 2.34.1 >
On 2023-04-14 15:06, Matthew Wilcox wrote: > On Fri, Apr 14, 2023 at 01:08:19PM +0200, Pankaj Raghav wrote: >> Folio version of alloc_page_buffers() helper. This is required to convert >> create_page_buffers() to create_folio_buffers() later in the series. >> >> It removes one call to compound_head() compared to alloc_page_buffers(). > > I would convert alloc_page_buffers() to folio_alloc_buffers() and > add > > static struct buffer_head *alloc_page_buffers(struct page *page, > unsigned long size, bool retry) > { > return folio_alloc_buffers(page_folio(page), size, retry); > } > > in buffer_head.h > > (there are only five callers, so this feels like a better tradeoff > than creating a new function) > That is a good idea and follows the usual pattern for folio conversion. I will send a new version soon with your other comments as well. -- Pankaj
diff --git a/fs/buffer.c b/fs/buffer.c index 44380ff3a31f..0f9c2127543d 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -900,6 +900,65 @@ struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size, } EXPORT_SYMBOL_GPL(alloc_page_buffers); +/* + * Create the appropriate buffers when given a folio for data area and + * the size of each buffer.. Use the bh->b_this_page linked list to + * follow the buffers created. Return NULL if unable to create more + * buffers. + * + * The retry flag is used to differentiate async IO (paging, swapping) + * which may not fail from ordinary buffer allocations. + */ +struct buffer_head *alloc_folio_buffers(struct folio *folio, unsigned long size, + bool retry) +{ + struct buffer_head *bh, *head; + gfp_t gfp = GFP_NOFS | __GFP_ACCOUNT; + long offset; + struct mem_cgroup *memcg, *old_memcg; + + if (retry) + gfp |= __GFP_NOFAIL; + + /* The folio lock pins the memcg */ + memcg = folio_memcg(folio); + old_memcg = set_active_memcg(memcg); + + head = NULL; + offset = folio_size(folio); + while ((offset -= size) >= 0) { + bh = alloc_buffer_head(gfp); + if (!bh) + goto no_grow; + + bh->b_this_page = head; + bh->b_blocknr = -1; + head = bh; + + bh->b_size = size; + + /* Link the buffer to its folio */ + set_bh_folio(bh, folio, offset); + } +out: + set_active_memcg(old_memcg); + return head; +/* + * In case anything failed, we just free everything we got. + */ +no_grow: + if (head) { + do { + bh = head; + head = head->b_this_page; + free_buffer_head(bh); + } while (head); + } + + goto out; +} +EXPORT_SYMBOL_GPL(alloc_folio_buffers); + static inline void link_dev_buffers(struct page *page, struct buffer_head *head) {
Folio version of alloc_page_buffers() helper. This is required to convert create_page_buffers() to create_folio_buffers() later in the series. It removes one call to compound_head() compared to alloc_page_buffers(). Signed-off-by: Pankaj Raghav <p.raghav@samsung.com> --- fs/buffer.c | 59 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+)