Matthew Wilcox
2020-Feb-19 21:00 UTC
[Ocfs2-devel] [PATCH v7 09/24] mm: Put readahead pages in cache earlier
From: "Matthew Wilcox (Oracle)" <willy at infradead.org>
When populating the page cache for readahead, mappings that use
->readpages must populate the page cache themselves as the pages are
passed on a linked list which would normally be used for the page cache's
LRU. For mappings that use ->readpage or the upcoming ->readahead method,
we can put the pages into the page cache as soon as they're allocated,
which solves a race between readahead and direct IO. It also lets us
remove the gfp argument from read_pages().
Use the new readahead_page() API to implement the repeated calls to
->readpage(), just like most filesystems will. This iterator also
supports huge pages, even though none of the filesystems have been
converted to use them yet.
Signed-off-by: Matthew Wilcox (Oracle) <willy at infradead.org>
---
include/linux/pagemap.h | 20 +++++++++++++++++
mm/readahead.c | 48 +++++++++++++++++++++++++----------------
2 files changed, 49 insertions(+), 19 deletions(-)
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index 55fcea0249e6..4989d330fada 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -647,8 +647,28 @@ struct readahead_control {
/* private: use the readahead_* accessors instead */
pgoff_t _index;
unsigned int _nr_pages;
+ unsigned int _batch_count;
};
+static inline struct page *readahead_page(struct readahead_control *rac)
+{
+ struct page *page;
+
+ BUG_ON(rac->_batch_count > rac->_nr_pages);
+ rac->_nr_pages -= rac->_batch_count;
+ rac->_index += rac->_batch_count;
+ rac->_batch_count = 0;
+
+ if (!rac->_nr_pages)
+ return NULL;
+
+ page = xa_load(&rac->mapping->i_pages, rac->_index);
+ VM_BUG_ON_PAGE(!PageLocked(page), page);
+ rac->_batch_count = hpage_nr_pages(page);
+
+ return page;
+}
+
/* The number of pages in this readahead block */
static inline unsigned int readahead_count(struct readahead_control *rac)
{
diff --git a/mm/readahead.c b/mm/readahead.c
index 83df5c061d33..aaa209559ba2 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -113,15 +113,14 @@ int read_cache_pages(struct address_space *mapping, struct
list_head *pages,
EXPORT_SYMBOL(read_cache_pages);
-static void read_pages(struct readahead_control *rac, struct list_head *pages,
- gfp_t gfp)
+static void read_pages(struct readahead_control *rac, struct list_head *pages)
{
const struct address_space_operations *aops = rac->mapping->a_ops;
+ struct page *page;
struct blk_plug plug;
- unsigned page_idx;
if (!readahead_count(rac))
- return;
+ goto out;
blk_start_plug(&plug);
@@ -130,23 +129,23 @@ static void read_pages(struct readahead_control *rac,
struct list_head *pages,
readahead_count(rac));
/* Clean up the remaining pages */
put_pages_list(pages);
- goto out;
- }
-
- for (page_idx = 0; page_idx < readahead_count(rac); page_idx++) {
- struct page *page = lru_to_page(pages);
- list_del(&page->lru);
- if (!add_to_page_cache_lru(page, rac->mapping, page->index,
- gfp))
+ rac->_index += rac->_nr_pages;
+ rac->_nr_pages = 0;
+ } else {
+ while ((page = readahead_page(rac))) {
aops->readpage(rac->file, page);
- put_page(page);
+ put_page(page);
+ }
}
-out:
blk_finish_plug(&plug);
BUG_ON(!list_empty(pages));
- rac->_nr_pages = 0;
+ BUG_ON(readahead_count(rac));
+
+out:
+ /* If we were called due to a conflicting page, skip over it */
+ rac->_index++;
}
/*
@@ -165,9 +164,11 @@ void __do_page_cache_readahead(struct address_space
*mapping,
LIST_HEAD(page_pool);
loff_t isize = i_size_read(inode);
gfp_t gfp_mask = readahead_gfp_mask(mapping);
+ bool use_list = mapping->a_ops->readpages;
struct readahead_control rac = {
.mapping = mapping,
.file = filp,
+ ._index = index,
._nr_pages = 0,
};
unsigned long i;
@@ -184,6 +185,8 @@ void __do_page_cache_readahead(struct address_space
*mapping,
if (index + i > end_index)
break;
+ BUG_ON(index + i != rac._index + rac._nr_pages);
+
page = xa_load(&mapping->i_pages, index + i);
if (page && !xa_is_value(page)) {
/*
@@ -191,15 +194,22 @@ void __do_page_cache_readahead(struct address_space
*mapping,
* contiguous pages before continuing with the next
* batch.
*/
- read_pages(&rac, &page_pool, gfp_mask);
+ read_pages(&rac, &page_pool);
continue;
}
page = __page_cache_alloc(gfp_mask);
if (!page)
break;
- page->index = index + i;
- list_add(&page->lru, &page_pool);
+ if (use_list) {
+ page->index = index + i;
+ list_add(&page->lru, &page_pool);
+ } else if (add_to_page_cache_lru(page, mapping, index + i,
+ gfp_mask) < 0) {
+ put_page(page);
+ read_pages(&rac, &page_pool);
+ continue;
+ }
if (i == nr_to_read - lookahead_size)
SetPageReadahead(page);
rac._nr_pages++;
@@ -210,7 +220,7 @@ void __do_page_cache_readahead(struct address_space
*mapping,
* uptodate then the caller will launch readpage again, and
* will then handle the error.
*/
- read_pages(&rac, &page_pool, gfp_mask);
+ read_pages(&rac, &page_pool);
}
/*
--
2.25.0
John Hubbard
2020-Feb-21 03:19 UTC
[Ocfs2-devel] [PATCH v7 09/24] mm: Put readahead pages in cache earlier
On 2/19/20 1:00 PM, Matthew Wilcox wrote:> From: "Matthew Wilcox (Oracle)" <willy at infradead.org> > > When populating the page cache for readahead, mappings that use > ->readpages must populate the page cache themselves as the pages are > passed on a linked list which would normally be used for the page cache's > LRU. For mappings that use ->readpage or the upcoming ->readahead method, > we can put the pages into the page cache as soon as they're allocated, > which solves a race between readahead and direct IO. It also lets us > remove the gfp argument from read_pages(). > > Use the new readahead_page() API to implement the repeated calls to > ->readpage(), just like most filesystems will. This iterator also > supports huge pages, even though none of the filesystems have been > converted to use them yet. > > Signed-off-by: Matthew Wilcox (Oracle) <willy at infradead.org> > --- > include/linux/pagemap.h | 20 +++++++++++++++++ > mm/readahead.c | 48 +++++++++++++++++++++++++---------------- > 2 files changed, 49 insertions(+), 19 deletions(-) > > diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h > index 55fcea0249e6..4989d330fada 100644 > --- a/include/linux/pagemap.h > +++ b/include/linux/pagemap.h > @@ -647,8 +647,28 @@ struct readahead_control { > /* private: use the readahead_* accessors instead */ > pgoff_t _index; > unsigned int _nr_pages; > + unsigned int _batch_count; > }; > > +static inline struct page *readahead_page(struct readahead_control *rac) > +{ > + struct page *page; > + > + BUG_ON(rac->_batch_count > rac->_nr_pages); > + rac->_nr_pages -= rac->_batch_count; > + rac->_index += rac->_batch_count; > + rac->_batch_count = 0;Is it intentional, to set rac->_batch_count twice (here, and below)? The only reason I can see is if a caller needs to use ->_batch_count in the "return NULL" case, which doesn't seem to come up...> + > + if (!rac->_nr_pages) > + return NULL; > + > + page = xa_load(&rac->mapping->i_pages, rac->_index); > + VM_BUG_ON_PAGE(!PageLocked(page), page); > + rac->_batch_count = hpage_nr_pages(page); > + > + return page; > +} > + > /* The number of pages in this readahead block */ > static inline unsigned int readahead_count(struct readahead_control *rac) > { > diff --git a/mm/readahead.c b/mm/readahead.c > index 83df5c061d33..aaa209559ba2 100644 > --- a/mm/readahead.c > +++ b/mm/readahead.c > @@ -113,15 +113,14 @@ int read_cache_pages(struct address_space *mapping, struct list_head *pages, > > EXPORT_SYMBOL(read_cache_pages); > > -static void read_pages(struct readahead_control *rac, struct list_head *pages, > - gfp_t gfp) > +static void read_pages(struct readahead_control *rac, struct list_head *pages) > { > const struct address_space_operations *aops = rac->mapping->a_ops; > + struct page *page; > struct blk_plug plug; > - unsigned page_idx; > > if (!readahead_count(rac)) > - return; > + goto out; > > blk_start_plug(&plug); > > @@ -130,23 +129,23 @@ static void read_pages(struct readahead_control *rac, struct list_head *pages, > readahead_count(rac)); > /* Clean up the remaining pages */ > put_pages_list(pages); > - goto out; > - } > - > - for (page_idx = 0; page_idx < readahead_count(rac); page_idx++) { > - struct page *page = lru_to_page(pages); > - list_del(&page->lru); > - if (!add_to_page_cache_lru(page, rac->mapping, page->index, > - gfp)) > + rac->_index += rac->_nr_pages; > + rac->_nr_pages = 0; > + } else { > + while ((page = readahead_page(rac))) { > aops->readpage(rac->file, page); > - put_page(page); > + put_page(page); > + } > } > > -out: > blk_finish_plug(&plug); > > BUG_ON(!list_empty(pages)); > - rac->_nr_pages = 0; > + BUG_ON(readahead_count(rac)); > + > +out: > + /* If we were called due to a conflicting page, skip over it */Tiny documentation nit: What if we were *not* called due to a conflicting page? (And what is a "conflicting page", in this context, btw?) The next line unconditionally moves the index ahead, so the "if" part of the comment really confuses me.> + rac->_index++; > } > > /* > @@ -165,9 +164,11 @@ void __do_page_cache_readahead(struct address_space *mapping, > LIST_HEAD(page_pool); > loff_t isize = i_size_read(inode); > gfp_t gfp_mask = readahead_gfp_mask(mapping); > + bool use_list = mapping->a_ops->readpages; > struct readahead_control rac = { > .mapping = mapping, > .file = filp, > + ._index = index, > ._nr_pages = 0, > }; > unsigned long i; > @@ -184,6 +185,8 @@ void __do_page_cache_readahead(struct address_space *mapping, > if (index + i > end_index) > break; > > + BUG_ON(index + i != rac._index + rac._nr_pages); > + > page = xa_load(&mapping->i_pages, index + i); > if (page && !xa_is_value(page)) { > /* > @@ -191,15 +194,22 @@ void __do_page_cache_readahead(struct address_space *mapping, > * contiguous pages before continuing with the next > * batch. > */ > - read_pages(&rac, &page_pool, gfp_mask); > + read_pages(&rac, &page_pool); > continue; > } > > page = __page_cache_alloc(gfp_mask); > if (!page) > break; > - page->index = index + i; > - list_add(&page->lru, &page_pool); > + if (use_list) { > + page->index = index + i; > + list_add(&page->lru, &page_pool); > + } else if (add_to_page_cache_lru(page, mapping, index + i, > + gfp_mask) < 0) {I still think you'll want to compare against !=0, rather than < 0, here.> + put_page(page); > + read_pages(&rac, &page_pool);Doing a read_pages() in the error case is because...actually, I'm not sure yet. Why do we do this? Effectively it's a retry?> + continue; > + } > if (i == nr_to_read - lookahead_size) > SetPageReadahead(page); > rac._nr_pages++; > @@ -210,7 +220,7 @@ void __do_page_cache_readahead(struct address_space *mapping, > * uptodate then the caller will launch readpage again, and > * will then handle the error. > */ > - read_pages(&rac, &page_pool, gfp_mask); > + read_pages(&rac, &page_pool); > } > > /* >Didn't spot any actual errors, just mainly my own questions here. :) thanks, -- John Hubbard NVIDIA
Christoph Hellwig
2020-Feb-24 21:40 UTC
[Ocfs2-devel] [PATCH v7 09/24] mm: Put readahead pages in cache earlier
On Wed, Feb 19, 2020 at 01:00:48PM -0800, Matthew Wilcox wrote:> From: "Matthew Wilcox (Oracle)" <willy at infradead.org> > > When populating the page cache for readahead, mappings that use > ->readpages must populate the page cache themselves as the pages are > passed on a linked list which would normally be used for the page cache's > LRU. For mappings that use ->readpage or the upcoming ->readahead method, > we can put the pages into the page cache as soon as they're allocated, > which solves a race between readahead and direct IO. It also lets us > remove the gfp argument from read_pages(). > > Use the new readahead_page() API to implement the repeated calls to > ->readpage(), just like most filesystems will. This iterator also > supports huge pages, even though none of the filesystems have been > converted to use them yet. > > Signed-off-by: Matthew Wilcox (Oracle) <willy at infradead.org> > --- > include/linux/pagemap.h | 20 +++++++++++++++++ > mm/readahead.c | 48 +++++++++++++++++++++++++---------------- > 2 files changed, 49 insertions(+), 19 deletions(-) > > diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h > index 55fcea0249e6..4989d330fada 100644 > --- a/include/linux/pagemap.h > +++ b/include/linux/pagemap.h > @@ -647,8 +647,28 @@ struct readahead_control { > /* private: use the readahead_* accessors instead */ > pgoff_t _index; > unsigned int _nr_pages; > + unsigned int _batch_count; > }; > > +static inline struct page *readahead_page(struct readahead_control *rac) > +{ > + struct page *page; > + > + BUG_ON(rac->_batch_count > rac->_nr_pages); > + rac->_nr_pages -= rac->_batch_count; > + rac->_index += rac->_batch_count; > + rac->_batch_count = 0; > + > + if (!rac->_nr_pages) > + return NULL; > + > + page = xa_load(&rac->mapping->i_pages, rac->_index); > + VM_BUG_ON_PAGE(!PageLocked(page), page); > + rac->_batch_count = hpage_nr_pages(page); > + > + return page; > +} > + > /* The number of pages in this readahead block */ > static inline unsigned int readahead_count(struct readahead_control *rac) > { > diff --git a/mm/readahead.c b/mm/readahead.c > index 83df5c061d33..aaa209559ba2 100644 > --- a/mm/readahead.c > +++ b/mm/readahead.c > @@ -113,15 +113,14 @@ int read_cache_pages(struct address_space *mapping, struct list_head *pages, > > EXPORT_SYMBOL(read_cache_pages); > > -static void read_pages(struct readahead_control *rac, struct list_head *pages, > - gfp_t gfp) > +static void read_pages(struct readahead_control *rac, struct list_head *pages) > { > const struct address_space_operations *aops = rac->mapping->a_ops; > + struct page *page; > struct blk_plug plug; > - unsigned page_idx; > > if (!readahead_count(rac)) > - return; > + goto out; > > blk_start_plug(&plug); > > @@ -130,23 +129,23 @@ static void read_pages(struct readahead_control *rac, struct list_head *pages, > readahead_count(rac)); > /* Clean up the remaining pages */ > put_pages_list(pages); > - goto out; > - } > - > - for (page_idx = 0; page_idx < readahead_count(rac); page_idx++) { > - struct page *page = lru_to_page(pages); > - list_del(&page->lru); > - if (!add_to_page_cache_lru(page, rac->mapping, page->index, > - gfp)) > + rac->_index += rac->_nr_pages; > + rac->_nr_pages = 0; > + } else { > + while ((page = readahead_page(rac))) { > aops->readpage(rac->file, page); > - put_page(page); > + put_page(page); > + } > } > > -out: > blk_finish_plug(&plug); > > BUG_ON(!list_empty(pages)); > - rac->_nr_pages = 0; > + BUG_ON(readahead_count(rac)); > + > +out: > + /* If we were called due to a conflicting page, skip over it */ > + rac->_index++; > } > > /* > @@ -165,9 +164,11 @@ void __do_page_cache_readahead(struct address_space *mapping, > LIST_HEAD(page_pool); > loff_t isize = i_size_read(inode); > gfp_t gfp_mask = readahead_gfp_mask(mapping); > + bool use_list = mapping->a_ops->readpages;I find this single use variable a little weird. Not a dealbreaker, but just checking the methods would seem a little more obvious to me. Except for this and the other nitpick the patch looks good to me: Reviewed-by: Christoph Hellwig <hch at lst.de>