2022-11-04 05:46:58

by Jingbo Xu

Subject: [PATCH 0/2] erofs: misc fixes for erofs in fscache mode

There are several minor fixes for the data path of the fscache mode.

Jingbo Xu (2):
erofs: put metabuf in error path in fscache mode
erofs: get correct count for unmapped range in fscache mode

fs/erofs/fscache.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)

--
2.19.1.6.gb485710b

2022-11-04 05:47:10

by Jingbo Xu

Subject: [PATCH 2/2] erofs: get correct count for unmapped range in fscache mode

For unmapped range, the returned map.m_llen is zero, and thus the
calculated count is unexpected zero.

Prior to the refactoring introduced by commit 1ae9470c3e14 ("erofs:
clean up .read_folio() and .readahead() in fscache mode"), only the
readahead routine suffers from this. With the refactoring of making
.read_folio() and .readahead() calling one common routine, both
read_folio and readahead have this issue now.

Fix this by calculating count separately in unmapped condition.

Fixes: c665b394b9e8 ("erofs: implement fscache-based data readahead")
Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
Signed-off-by: Jingbo Xu <[email protected]>
---
fs/erofs/fscache.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
index 83559008bfa8..260fa4737fc0 100644
--- a/fs/erofs/fscache.c
+++ b/fs/erofs/fscache.c
@@ -296,15 +296,16 @@ static int erofs_fscache_data_read(struct address_space *mapping,
return PAGE_SIZE;
}

- count = min_t(size_t, map.m_llen - (pos - map.m_la), len);
- DBG_BUGON(!count || count % PAGE_SIZE);
-
if (!(map.m_flags & EROFS_MAP_MAPPED)) {
+ count = len;
iov_iter_xarray(&iter, READ, &mapping->i_pages, pos, count);
iov_iter_zero(count, &iter);
return count;
}

+ count = min_t(size_t, map.m_llen - (pos - map.m_la), len);
+ DBG_BUGON(!count || count % PAGE_SIZE);
+
mdev = (struct erofs_map_dev) {
.m_deviceid = map.m_deviceid,
.m_pa = map.m_pa,
--
2.19.1.6.gb485710b

2022-11-04 06:21:29

by Jingbo Xu

Subject: [PATCH 1/2] erofs: put metabuf in error path in fscache mode

For tail packing layout, put metabuf when error is encountered.

Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
Signed-off-by: Jingbo Xu <[email protected]>
---
fs/erofs/fscache.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
index fe05bc51f9f2..83559008bfa8 100644
--- a/fs/erofs/fscache.c
+++ b/fs/erofs/fscache.c
@@ -287,8 +287,10 @@ static int erofs_fscache_data_read(struct address_space *mapping,
return PTR_ERR(src);

iov_iter_xarray(&iter, READ, &mapping->i_pages, pos, PAGE_SIZE);
- if (copy_to_iter(src + offset, size, &iter) != size)
+ if (copy_to_iter(src + offset, size, &iter) != size) {
+ erofs_put_metabuf(&buf);
return -EFAULT;
+ }
iov_iter_zero(PAGE_SIZE - size, &iter);
erofs_put_metabuf(&buf);
return PAGE_SIZE;
--
2.19.1.6.gb485710b

2022-11-04 10:22:12

by Gao Xiang

Subject: Re: [PATCH 2/2] erofs: get correct count for unmapped range in fscache mode

On Fri, Nov 04, 2022 at 01:40:28PM +0800, Jingbo Xu wrote:
> For unmapped range, the returned map.m_llen is zero, and thus the
> calculated count is unexpected zero.
>
> Prior to the refactoring introduced by commit 1ae9470c3e14 ("erofs:
> clean up .read_folio() and .readahead() in fscache mode"), only the
> readahead routine suffers from this. With the refactoring of making
> .read_folio() and .readahead() calling one common routine, both
> read_folio and readahead have this issue now.
>
> Fix this by calculating count separately in unmapped condition.
>
> Fixes: c665b394b9e8 ("erofs: implement fscache-based data readahead")
> Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
> Signed-off-by: Jingbo Xu <[email protected]>

Reviewed-by: Gao Xiang <[email protected]>

Thanks,
Gao Xiang

> ---
> fs/erofs/fscache.c | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
> index 83559008bfa8..260fa4737fc0 100644
> --- a/fs/erofs/fscache.c
> +++ b/fs/erofs/fscache.c
> @@ -296,15 +296,16 @@ static int erofs_fscache_data_read(struct address_space *mapping,
> return PAGE_SIZE;
> }
>
> - count = min_t(size_t, map.m_llen - (pos - map.m_la), len);
> - DBG_BUGON(!count || count % PAGE_SIZE);
> -
> if (!(map.m_flags & EROFS_MAP_MAPPED)) {
> + count = len;
> iov_iter_xarray(&iter, READ, &mapping->i_pages, pos, count);
> iov_iter_zero(count, &iter);
> return count;
> }
>
> + count = min_t(size_t, map.m_llen - (pos - map.m_la), len);
> + DBG_BUGON(!count || count % PAGE_SIZE);
> +
> mdev = (struct erofs_map_dev) {
> .m_deviceid = map.m_deviceid,
> .m_pa = map.m_pa,
> --
> 2.19.1.6.gb485710b

2022-11-04 10:38:48

by Gao Xiang

Subject: Re: [PATCH 1/2] erofs: put metabuf in error path in fscache mode

On Fri, Nov 04, 2022 at 01:40:27PM +0800, Jingbo Xu wrote:
> For tail packing layout, put metabuf when error is encountered.
>
> Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
> Signed-off-by: Jingbo Xu <[email protected]>

Reviewed-by: Gao Xiang <[email protected]>

Thanks,
Gao Xiang

> ---
> fs/erofs/fscache.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
> index fe05bc51f9f2..83559008bfa8 100644
> --- a/fs/erofs/fscache.c
> +++ b/fs/erofs/fscache.c
> @@ -287,8 +287,10 @@ static int erofs_fscache_data_read(struct address_space *mapping,
> return PTR_ERR(src);
>
> iov_iter_xarray(&iter, READ, &mapping->i_pages, pos, PAGE_SIZE);
> - if (copy_to_iter(src + offset, size, &iter) != size)
> + if (copy_to_iter(src + offset, size, &iter) != size) {
> + erofs_put_metabuf(&buf);
> return -EFAULT;
> + }
> iov_iter_zero(PAGE_SIZE - size, &iter);
> erofs_put_metabuf(&buf);
> return PAGE_SIZE;
> --
> 2.19.1.6.gb485710b

2022-11-07 14:53:44

by Jia Zhu

Subject: Re: [External] [PATCH 1/2] erofs: put metabuf in error path in fscache mode

在 2022/11/4 13:40, Jingbo Xu 写道:
> For tail packing layout, put metabuf when error is encountered.
>
> Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
> Signed-off-by: Jingbo Xu <[email protected]>
Reviewed-by: Jia Zhu <[email protected]>
> ---
> fs/erofs/fscache.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
> index fe05bc51f9f2..83559008bfa8 100644
> --- a/fs/erofs/fscache.c
> +++ b/fs/erofs/fscache.c
> @@ -287,8 +287,10 @@ static int erofs_fscache_data_read(struct address_space *mapping,
> return PTR_ERR(src);
>
> iov_iter_xarray(&iter, READ, &mapping->i_pages, pos, PAGE_SIZE);
> - if (copy_to_iter(src + offset, size, &iter) != size)
> + if (copy_to_iter(src + offset, size, &iter) != size) {
> + erofs_put_metabuf(&buf);
> return -EFAULT;
> + }
> iov_iter_zero(PAGE_SIZE - size, &iter);
> erofs_put_metabuf(&buf);
> return PAGE_SIZE;

2022-11-08 03:35:31

by Chao Yu

Subject: Re: [PATCH 1/2] erofs: put metabuf in error path in fscache mode

On 2022/11/4 13:40, Jingbo Xu wrote:
> For tail packing layout, put metabuf when error is encountered.
>
> Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
> Signed-off-by: Jingbo Xu <[email protected]>

Reviewed-by: Chao Yu <[email protected]>

Thanks,

2022-11-08 03:46:27

by Chao Yu

Subject: Re: [PATCH 2/2] erofs: get correct count for unmapped range in fscache mode

On 2022/11/4 13:40, Jingbo Xu wrote:
> For unmapped range, the returned map.m_llen is zero, and thus the
> calculated count is unexpected zero.
>
> Prior to the refactoring introduced by commit 1ae9470c3e14 ("erofs:
> clean up .read_folio() and .readahead() in fscache mode"), only the
> readahead routine suffers from this. With the refactoring of making
> .read_folio() and .readahead() calling one common routine, both
> read_folio and readahead have this issue now.
>
> Fix this by calculating count separately in unmapped condition.
>
> Fixes: c665b394b9e8 ("erofs: implement fscache-based data readahead")
> Fixes: 1ae9470c3e14 ("erofs: clean up .read_folio() and .readahead() in fscache mode")
> Signed-off-by: Jingbo Xu <[email protected]>

Reviewed-by: Chao Yu <[email protected]>

Thanks,