2023-07-10 05:04:56

by Chunhai Guo

[permalink] [raw]
Subject: [PATCH] erofs: avoid unnecessary loops in z_erofs_pcluster_readmore() when read page beyond EOF

z_erofs_pcluster_readmore() may take a long time to loop when the page
offset is large enough, which is unnecessary should be prevented.
For example, when the following case is encountered, it will loop 4691368
times, taking about 27 seconds.
- offset = 19217289215
- inode_size = 1442672

Signed-off-by: Chunhai Guo <[email protected]>
---
fs/erofs/zdata.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 5f1890e309c6..d9a0763f4595 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -1841,7 +1841,7 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
}

cur = map->m_la + map->m_llen - 1;
- while (cur >= end) {
+ while ((cur >= end) && (cur < i_size_read(inode))) {
pgoff_t index = cur >> PAGE_SHIFT;
struct page *page;

--
2.25.1



2023-07-10 05:20:16

by Gao Xiang

[permalink] [raw]
Subject: Re: [PATCH] erofs: avoid unnecessary loops in z_erofs_pcluster_readmore() when read page beyond EOF



On 2023/7/10 12:25, Chunhai Guo wrote:
> z_erofs_pcluster_readmore() may take a long time to loop when the page
> offset is large enough, which is unnecessary should be prevented.
> For example, when the following case is encountered, it will loop 4691368
> times, taking about 27 seconds.
> - offset = 19217289215
> - inode_size = 1442672
>
> Signed-off-by: Chunhai Guo <[email protected]>

It looks good to me,

Fixes: 386292919c25 ("erofs: introduce readmore decompression strategy")
Reviewed-by: Gao Xiang <[email protected]>

Thanks,
Gao Xiang

2023-07-10 07:47:13

by Yue Hu

[permalink] [raw]
Subject: Re: [PATCH] erofs: avoid unnecessary loops in z_erofs_pcluster_readmore() when read page beyond EOF

On Mon, 10 Jul 2023 12:25:31 +0800
Chunhai Guo <[email protected]> wrote:

> z_erofs_pcluster_readmore() may take a long time to loop when the page
> offset is large enough, which is unnecessary should be prevented.
> For example, when the following case is encountered, it will loop 4691368
> times, taking about 27 seconds.
> - offset = 19217289215
> - inode_size = 1442672
>
> Signed-off-by: Chunhai Guo <[email protected]>
> ---
> fs/erofs/zdata.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
> index 5f1890e309c6..d9a0763f4595 100644
> --- a/fs/erofs/zdata.c
> +++ b/fs/erofs/zdata.c
> @@ -1841,7 +1841,7 @@ static void z_erofs_pcluster_readmore(struct z_erofs_decompress_frontend *f,
> }
>
> cur = map->m_la + map->m_llen - 1;
> - while (cur >= end) {
> + while ((cur >= end) && (cur < i_size_read(inode))) {
> pgoff_t index = cur >> PAGE_SHIFT;
> struct page *page;
>

Reviewed-by: Yue Hu <[email protected]>

2023-07-10 11:21:57

by Gao Xiang

[permalink] [raw]
Subject: Re: [PATCH] erofs: avoid unnecessary loops in z_erofs_pcluster_readmore() when read page beyond EOF



On 2023/7/10 12:25, Chunhai Guo wrote:
> z_erofs_pcluster_readmore() may take a long time to loop when the page
> offset is large enough, which is unnecessary should be prevented.
> For example, when the following case is encountered, it will loop 4691368
> times, taking about 27 seconds.
> - offset = 19217289215
> - inode_size = 1442672
>
> Signed-off-by: Chunhai Guo <[email protected]>

I will update the subject manually to:
"erofs: avoid useless loops in z_erofs_pcluster_readmore() when reading beyond EOF"

to avoid overly long subject as well...

Thanks,
Gao Xiang

2023-07-11 14:18:13

by Chao Yu

[permalink] [raw]
Subject: Re: [PATCH] erofs: avoid unnecessary loops in z_erofs_pcluster_readmore() when read page beyond EOF

On 2023/7/10 12:25, Chunhai Guo wrote:
> z_erofs_pcluster_readmore() may take a long time to loop when the page
> offset is large enough, which is unnecessary should be prevented.
> For example, when the following case is encountered, it will loop 4691368
> times, taking about 27 seconds.
> - offset = 19217289215
> - inode_size = 1442672
>
> Signed-off-by: Chunhai Guo <[email protected]>

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

Thanks,