diff mbox series

[04/12] fuse: support large folios for non-writeback writes

Message ID 20241109001258.2216604-5-joannelkoong@gmail.com (mailing list archive)
State New
Headers show
Series fuse: support large folios | expand

Commit Message

Joanne Koong Nov. 9, 2024, 12:12 a.m. UTC 
Add support for folios larger than one page size for non-writeback
writes.

Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
---
 fs/fuse/file.c | 29 ++++++++++++++++++-----------
 1 file changed, 18 insertions(+), 11 deletions(-)

Comments

Joanne Koong Nov. 12, 2024, 5:32 p.m. UTC | #1 
On Fri, Nov 8, 2024 at 4:13 PM Joanne Koong <joannelkoong@gmail.com> wrote:
>
> Add support for folios larger than one page size for non-writeback

I'll change this naming from "non-writeback" to "writethrough"

> writes.
>
> Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
> ---
>  fs/fuse/file.c | 29 ++++++++++++++++++-----------
>  1 file changed, 18 insertions(+), 11 deletions(-)
>
> diff --git a/fs/fuse/file.c b/fs/fuse/file.c
> index a89fdc55a40b..6ee23ab9b7f2 100644
> --- a/fs/fuse/file.c
> +++ b/fs/fuse/file.c
> @@ -1146,19 +1146,15 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
>         num = min(num, max_pages << PAGE_SHIFT);
>
>         ap->args.in_pages = true;
> -       ap->descs[0].offset = offset;
>
>         while (num) {
>                 size_t tmp;
>                 struct folio *folio;
>                 pgoff_t index = pos >> PAGE_SHIFT;
> -               unsigned int bytes = min(PAGE_SIZE - offset, num);
> -
> - again:
> -               err = -EFAULT;
> -               if (fault_in_iov_iter_readable(ii, bytes))
> -                       break;
> +               unsigned int bytes;
> +               unsigned int folio_offset;
>
> +       again:
>                 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN,
>                                             mapping_gfp_mask(mapping));

This __filemap_get_folio() call (and the one in fuse_write_begin() as
well) needs to also set the order on the fgf flag to enable large
folios, else all folios returned will be order 0.

I'll fix this in v2.

>                 if (IS_ERR(folio)) {
> @@ -1166,10 +1162,20 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
>                         break;
>                 }
>
> +               folio_offset = ((index - folio->index) << PAGE_SHIFT) + offset;
> +               bytes = min(folio_size(folio) - folio_offset, num);
> +
> +               err = -EFAULT;
> +               if (fault_in_iov_iter_readable(ii, bytes)) {
> +                       folio_unlock(folio);
> +                       folio_put(folio);
> +                       break;
> +               }
> +
>                 if (mapping_writably_mapped(mapping))
>                         flush_dcache_folio(folio);
>
> -               tmp = copy_folio_from_iter_atomic(folio, offset, bytes, ii);
> +               tmp = copy_folio_from_iter_atomic(folio, folio_offset, bytes, ii);
>                 flush_dcache_folio(folio);
>
>                 if (!tmp) {
> @@ -1180,6 +1186,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
>
>                 err = 0;
>                 ap->folios[ap->num_folios] = folio;
> +               ap->descs[ap->num_folios].offset = folio_offset;
>                 ap->descs[ap->num_folios].length = tmp;
>                 ap->num_folios++;
>
> @@ -1187,11 +1194,11 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
>                 pos += tmp;
>                 num -= tmp;
>                 offset += tmp;
> -               if (offset == PAGE_SIZE)
> +               if (offset == folio_size(folio))
>                         offset = 0;
>
> -               /* If we copied full page, mark it uptodate */
> -               if (tmp == PAGE_SIZE)
> +               /* If we copied full folio, mark it uptodate */
> +               if (tmp == folio_size(folio))
>                         folio_mark_uptodate(folio);
>
>                 if (folio_test_uptodate(folio)) {
> --
> 2.43.5
>
Joanne Koong Nov. 13, 2024, 6:41 p.m. UTC | #2 
On Tue, Nov 12, 2024 at 9:32 AM Joanne Koong <joannelkoong@gmail.com> wrote:
>
> On Fri, Nov 8, 2024 at 4:13 PM Joanne Koong <joannelkoong@gmail.com> wrote:
> >
> > Add support for folios larger than one page size for non-writeback
>
> I'll change this naming from "non-writeback" to "writethrough"
>
> > writes.
> >
> > Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
> > ---
> >  fs/fuse/file.c | 29 ++++++++++++++++++-----------
> >  1 file changed, 18 insertions(+), 11 deletions(-)
> >
> > diff --git a/fs/fuse/file.c b/fs/fuse/file.c
> > index a89fdc55a40b..6ee23ab9b7f2 100644
> > --- a/fs/fuse/file.c
> > +++ b/fs/fuse/file.c
> > @@ -1146,19 +1146,15 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
> >         num = min(num, max_pages << PAGE_SHIFT);
> >
> >         ap->args.in_pages = true;
> > -       ap->descs[0].offset = offset;
> >
> >         while (num) {
> >                 size_t tmp;
> >                 struct folio *folio;
> >                 pgoff_t index = pos >> PAGE_SHIFT;
> > -               unsigned int bytes = min(PAGE_SIZE - offset, num);
> > -
> > - again:
> > -               err = -EFAULT;
> > -               if (fault_in_iov_iter_readable(ii, bytes))
> > -                       break;
> > +               unsigned int bytes;
> > +               unsigned int folio_offset;
> >
> > +       again:
> >                 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN,
> >                                             mapping_gfp_mask(mapping));
>
> This __filemap_get_folio() call (and the one in fuse_write_begin() as
> well) needs to also set the order on the fgf flag to enable large
> folios, else all folios returned will be order 0.
>
> I'll fix this in v2.

Ran some benchmarks and trying to get the largest folios possible from
__filemap_get_folio() is an over-optimization and slows down writes
significantly. I'll leave this as is for v2, and we could look into
optimizing this in the future.

>
> >                 if (IS_ERR(folio)) {
> > @@ -1166,10 +1162,20 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
> >                         break;
> >                 }
> >
> > +               folio_offset = ((index - folio->index) << PAGE_SHIFT) + offset;
> > +               bytes = min(folio_size(folio) - folio_offset, num);
> > +
> > +               err = -EFAULT;
> > +               if (fault_in_iov_iter_readable(ii, bytes)) {
> > +                       folio_unlock(folio);
> > +                       folio_put(folio);
> > +                       break;
> > +               }
> > +
> >                 if (mapping_writably_mapped(mapping))
> >                         flush_dcache_folio(folio);
> >
> > -               tmp = copy_folio_from_iter_atomic(folio, offset, bytes, ii);
> > +               tmp = copy_folio_from_iter_atomic(folio, folio_offset, bytes, ii);
> >                 flush_dcache_folio(folio);
> >
> >                 if (!tmp) {
> > @@ -1180,6 +1186,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
> >
> >                 err = 0;
> >                 ap->folios[ap->num_folios] = folio;
> > +               ap->descs[ap->num_folios].offset = folio_offset;
> >                 ap->descs[ap->num_folios].length = tmp;
> >                 ap->num_folios++;
> >
> > @@ -1187,11 +1194,11 @@ static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
> >                 pos += tmp;
> >                 num -= tmp;
> >                 offset += tmp;
> > -               if (offset == PAGE_SIZE)
> > +               if (offset == folio_size(folio))
> >                         offset = 0;
> >
> > -               /* If we copied full page, mark it uptodate */
> > -               if (tmp == PAGE_SIZE)
> > +               /* If we copied full folio, mark it uptodate */
> > +               if (tmp == folio_size(folio))
> >                         folio_mark_uptodate(folio);
> >
> >                 if (folio_test_uptodate(folio)) {
> > --
> > 2.43.5
> >
diff mbox series

Patch

diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index a89fdc55a40b..6ee23ab9b7f2 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1146,19 +1146,15 @@  static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
 	num = min(num, max_pages << PAGE_SHIFT);
 
 	ap->args.in_pages = true;
-	ap->descs[0].offset = offset;
 
 	while (num) {
 		size_t tmp;
 		struct folio *folio;
 		pgoff_t index = pos >> PAGE_SHIFT;
-		unsigned int bytes = min(PAGE_SIZE - offset, num);
-
- again:
-		err = -EFAULT;
-		if (fault_in_iov_iter_readable(ii, bytes))
-			break;
+		unsigned int bytes;
+		unsigned int folio_offset;
 
+	again:
 		folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN,
 					    mapping_gfp_mask(mapping));
 		if (IS_ERR(folio)) {
@@ -1166,10 +1162,20 @@  static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
 			break;
 		}
 
+		folio_offset = ((index - folio->index) << PAGE_SHIFT) + offset;
+		bytes = min(folio_size(folio) - folio_offset, num);
+
+		err = -EFAULT;
+		if (fault_in_iov_iter_readable(ii, bytes)) {
+			folio_unlock(folio);
+			folio_put(folio);
+			break;
+		}
+
 		if (mapping_writably_mapped(mapping))
 			flush_dcache_folio(folio);
 
-		tmp = copy_folio_from_iter_atomic(folio, offset, bytes, ii);
+		tmp = copy_folio_from_iter_atomic(folio, folio_offset, bytes, ii);
 		flush_dcache_folio(folio);
 
 		if (!tmp) {
@@ -1180,6 +1186,7 @@  static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
 
 		err = 0;
 		ap->folios[ap->num_folios] = folio;
+		ap->descs[ap->num_folios].offset = folio_offset;
 		ap->descs[ap->num_folios].length = tmp;
 		ap->num_folios++;
 
@@ -1187,11 +1194,11 @@  static ssize_t fuse_fill_write_pages(struct fuse_io_args *ia,
 		pos += tmp;
 		num -= tmp;
 		offset += tmp;
-		if (offset == PAGE_SIZE)
+		if (offset == folio_size(folio))
 			offset = 0;
 
-		/* If we copied full page, mark it uptodate */
-		if (tmp == PAGE_SIZE)
+		/* If we copied full folio, mark it uptodate */
+		if (tmp == folio_size(folio))
 			folio_mark_uptodate(folio);
 
 		if (folio_test_uptodate(folio)) {