* [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
@ 2024-07-12 7:34 Chao Yu
2024-07-25 16:55 ` Jaegeuk Kim
0 siblings, 1 reply; 6+ messages in thread
From: Chao Yu @ 2024-07-12 7:34 UTC (permalink / raw)
To: jaegeuk; +Cc: linux-kernel, linux-f2fs-devel
All directory blocks are within the scope of i_size, so let's limit
the end_block to just check valid dirent blocks.
Meanwhile, it uses dir_blocks() instead of variable for cleanup in
__f2fs_find_entry().
Signed-off-by: Chao Yu <chao@kernel.org>
---
fs/f2fs/dir.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
index 02c9355176d3..d4591c215f07 100644
--- a/fs/f2fs/dir.c
+++ b/fs/f2fs/dir.c
@@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
int s = GET_DENTRY_SLOTS(fname->disk_name.len);
unsigned int nbucket, nblock;
unsigned int bidx, end_block;
+ unsigned long last_block;
struct page *dentry_page;
struct f2fs_dir_entry *de = NULL;
pgoff_t next_pgofs;
bool room = false;
int max_slots;
+ last_block = dir_blocks(dir);
nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
nblock = bucket_blocks(level);
bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
le32_to_cpu(fname->hash) % nbucket);
end_block = bidx + nblock;
+ end_block = min_t(unsigned int, end_block, last_block);
while (bidx < end_block) {
/* no need to allocate new dentry pages to all the indices */
@@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
const struct f2fs_filename *fname,
struct page **res_page)
{
- unsigned long npages = dir_blocks(dir);
struct f2fs_dir_entry *de = NULL;
unsigned int max_depth;
unsigned int level;
@@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
goto out;
}
- if (npages == 0)
+ if (dir_blocks(dir) == 0)
goto out;
max_depth = F2FS_I(dir)->i_current_depth;
--
2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
2024-07-12 7:34 [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF Chao Yu
@ 2024-07-25 16:55 ` Jaegeuk Kim
2024-07-26 1:06 ` Chao Yu
0 siblings, 1 reply; 6+ messages in thread
From: Jaegeuk Kim @ 2024-07-25 16:55 UTC (permalink / raw)
To: Chao Yu; +Cc: linux-kernel, linux-f2fs-devel
On 07/12, Chao Yu wrote:
> All directory blocks are within the scope of i_size, so let's limit
> the end_block to just check valid dirent blocks.
Do we really need this?
>
> Meanwhile, it uses dir_blocks() instead of variable for cleanup in
> __f2fs_find_entry().
>
> Signed-off-by: Chao Yu <chao@kernel.org>
> ---
> fs/f2fs/dir.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
> index 02c9355176d3..d4591c215f07 100644
> --- a/fs/f2fs/dir.c
> +++ b/fs/f2fs/dir.c
> @@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
> int s = GET_DENTRY_SLOTS(fname->disk_name.len);
> unsigned int nbucket, nblock;
> unsigned int bidx, end_block;
> + unsigned long last_block;
> struct page *dentry_page;
> struct f2fs_dir_entry *de = NULL;
> pgoff_t next_pgofs;
> bool room = false;
> int max_slots;
>
> + last_block = dir_blocks(dir);
> nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
> nblock = bucket_blocks(level);
>
> bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
> le32_to_cpu(fname->hash) % nbucket);
> end_block = bidx + nblock;
> + end_block = min_t(unsigned int, end_block, last_block);
>
> while (bidx < end_block) {
> /* no need to allocate new dentry pages to all the indices */
> @@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> const struct f2fs_filename *fname,
> struct page **res_page)
> {
> - unsigned long npages = dir_blocks(dir);
> struct f2fs_dir_entry *de = NULL;
> unsigned int max_depth;
> unsigned int level;
> @@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> goto out;
> }
>
> - if (npages == 0)
> + if (dir_blocks(dir) == 0)
> goto out;
>
> max_depth = F2FS_I(dir)->i_current_depth;
> --
> 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
2024-07-25 16:55 ` Jaegeuk Kim
@ 2024-07-26 1:06 ` Chao Yu
2024-07-27 3:27 ` Jaegeuk Kim
0 siblings, 1 reply; 6+ messages in thread
From: Chao Yu @ 2024-07-26 1:06 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: linux-kernel, linux-f2fs-devel
On 2024/7/26 0:55, Jaegeuk Kim wrote:
> On 07/12, Chao Yu wrote:
>> All directory blocks are within the scope of i_size, so let's limit
>> the end_block to just check valid dirent blocks.
>
> Do we really need this?
f2fs_readdir() and f2fs_empty_dir() uses dir_blocks() for upper boundary,
this patch aligns find_in_level() w/ them.
Also, it can avoid grabbing never used page cache across EOF.
So, we can consider taking this patch?
Thanks,
>
>>
>> Meanwhile, it uses dir_blocks() instead of variable for cleanup in
>> __f2fs_find_entry().
>>
>> Signed-off-by: Chao Yu <chao@kernel.org>
>> ---
>> fs/f2fs/dir.c | 6 ++++--
>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
>> index 02c9355176d3..d4591c215f07 100644
>> --- a/fs/f2fs/dir.c
>> +++ b/fs/f2fs/dir.c
>> @@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
>> int s = GET_DENTRY_SLOTS(fname->disk_name.len);
>> unsigned int nbucket, nblock;
>> unsigned int bidx, end_block;
>> + unsigned long last_block;
>> struct page *dentry_page;
>> struct f2fs_dir_entry *de = NULL;
>> pgoff_t next_pgofs;
>> bool room = false;
>> int max_slots;
>>
>> + last_block = dir_blocks(dir);
>> nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
>> nblock = bucket_blocks(level);
>>
>> bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
>> le32_to_cpu(fname->hash) % nbucket);
>> end_block = bidx + nblock;
>> + end_block = min_t(unsigned int, end_block, last_block);
>>
>> while (bidx < end_block) {
>> /* no need to allocate new dentry pages to all the indices */
>> @@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
>> const struct f2fs_filename *fname,
>> struct page **res_page)
>> {
>> - unsigned long npages = dir_blocks(dir);
>> struct f2fs_dir_entry *de = NULL;
>> unsigned int max_depth;
>> unsigned int level;
>> @@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
>> goto out;
>> }
>>
>> - if (npages == 0)
>> + if (dir_blocks(dir) == 0)
>> goto out;
>>
>> max_depth = F2FS_I(dir)->i_current_depth;
>> --
>> 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
2024-07-26 1:06 ` Chao Yu
@ 2024-07-27 3:27 ` Jaegeuk Kim
2024-07-29 2:54 ` Chao Yu
0 siblings, 1 reply; 6+ messages in thread
From: Jaegeuk Kim @ 2024-07-27 3:27 UTC (permalink / raw)
To: Chao Yu; +Cc: linux-kernel, linux-f2fs-devel
On 07/26, Chao Yu wrote:
> On 2024/7/26 0:55, Jaegeuk Kim wrote:
> > On 07/12, Chao Yu wrote:
> > > All directory blocks are within the scope of i_size, so let's limit
> > > the end_block to just check valid dirent blocks.
> >
> > Do we really need this?
>
> f2fs_readdir() and f2fs_empty_dir() uses dir_blocks() for upper boundary,
> this patch aligns find_in_level() w/ them.
>
> Also, it can avoid grabbing never used page cache across EOF.
>
> So, we can consider taking this patch?
I'm wondering whether the current code has a bug or not.
>
> Thanks,
>
> >
> > >
> > > Meanwhile, it uses dir_blocks() instead of variable for cleanup in
> > > __f2fs_find_entry().
> > >
> > > Signed-off-by: Chao Yu <chao@kernel.org>
> > > ---
> > > fs/f2fs/dir.c | 6 ++++--
> > > 1 file changed, 4 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
> > > index 02c9355176d3..d4591c215f07 100644
> > > --- a/fs/f2fs/dir.c
> > > +++ b/fs/f2fs/dir.c
> > > @@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
> > > int s = GET_DENTRY_SLOTS(fname->disk_name.len);
> > > unsigned int nbucket, nblock;
> > > unsigned int bidx, end_block;
> > > + unsigned long last_block;
> > > struct page *dentry_page;
> > > struct f2fs_dir_entry *de = NULL;
> > > pgoff_t next_pgofs;
> > > bool room = false;
> > > int max_slots;
> > > + last_block = dir_blocks(dir);
> > > nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
> > > nblock = bucket_blocks(level);
> > > bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
> > > le32_to_cpu(fname->hash) % nbucket);
> > > end_block = bidx + nblock;
> > > + end_block = min_t(unsigned int, end_block, last_block);
> > > while (bidx < end_block) {
> > > /* no need to allocate new dentry pages to all the indices */
> > > @@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> > > const struct f2fs_filename *fname,
> > > struct page **res_page)
> > > {
> > > - unsigned long npages = dir_blocks(dir);
> > > struct f2fs_dir_entry *de = NULL;
> > > unsigned int max_depth;
> > > unsigned int level;
> > > @@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> > > goto out;
> > > }
> > > - if (npages == 0)
> > > + if (dir_blocks(dir) == 0)
> > > goto out;
> > > max_depth = F2FS_I(dir)->i_current_depth;
> > > --
> > > 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
2024-07-27 3:27 ` Jaegeuk Kim
@ 2024-07-29 2:54 ` Chao Yu
2024-07-29 16:21 ` Jaegeuk Kim
0 siblings, 1 reply; 6+ messages in thread
From: Chao Yu @ 2024-07-29 2:54 UTC (permalink / raw)
To: Jaegeuk Kim; +Cc: linux-kernel, linux-f2fs-devel
On 2024/7/27 11:27, Jaegeuk Kim wrote:
> On 07/26, Chao Yu wrote:
>> On 2024/7/26 0:55, Jaegeuk Kim wrote:
>>> On 07/12, Chao Yu wrote:
>>>> All directory blocks are within the scope of i_size, so let's limit
>>>> the end_block to just check valid dirent blocks.
>>>
>>> Do we really need this?
>>
>> f2fs_readdir() and f2fs_empty_dir() uses dir_blocks() for upper boundary,
>> this patch aligns find_in_level() w/ them.
>>
>> Also, it can avoid grabbing never used page cache across EOF.
>>
>> So, we can consider taking this patch?
>
> I'm wondering whether the current code has a bug or not.
I think it's not a bug, but, in corner case it may waste memory and
cause a bit performance reduction.
Thanks,
>
>>
>> Thanks,
>>
>>>
>>>>
>>>> Meanwhile, it uses dir_blocks() instead of variable for cleanup in
>>>> __f2fs_find_entry().
>>>>
>>>> Signed-off-by: Chao Yu <chao@kernel.org>
>>>> ---
>>>> fs/f2fs/dir.c | 6 ++++--
>>>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
>>>> index 02c9355176d3..d4591c215f07 100644
>>>> --- a/fs/f2fs/dir.c
>>>> +++ b/fs/f2fs/dir.c
>>>> @@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
>>>> int s = GET_DENTRY_SLOTS(fname->disk_name.len);
>>>> unsigned int nbucket, nblock;
>>>> unsigned int bidx, end_block;
>>>> + unsigned long last_block;
>>>> struct page *dentry_page;
>>>> struct f2fs_dir_entry *de = NULL;
>>>> pgoff_t next_pgofs;
>>>> bool room = false;
>>>> int max_slots;
>>>> + last_block = dir_blocks(dir);
>>>> nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
>>>> nblock = bucket_blocks(level);
>>>> bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
>>>> le32_to_cpu(fname->hash) % nbucket);
>>>> end_block = bidx + nblock;
>>>> + end_block = min_t(unsigned int, end_block, last_block);
>>>> while (bidx < end_block) {
>>>> /* no need to allocate new dentry pages to all the indices */
>>>> @@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
>>>> const struct f2fs_filename *fname,
>>>> struct page **res_page)
>>>> {
>>>> - unsigned long npages = dir_blocks(dir);
>>>> struct f2fs_dir_entry *de = NULL;
>>>> unsigned int max_depth;
>>>> unsigned int level;
>>>> @@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
>>>> goto out;
>>>> }
>>>> - if (npages == 0)
>>>> + if (dir_blocks(dir) == 0)
>>>> goto out;
>>>> max_depth = F2FS_I(dir)->i_current_depth;
>>>> --
>>>> 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF
2024-07-29 2:54 ` Chao Yu
@ 2024-07-29 16:21 ` Jaegeuk Kim
0 siblings, 0 replies; 6+ messages in thread
From: Jaegeuk Kim @ 2024-07-29 16:21 UTC (permalink / raw)
To: Chao Yu; +Cc: linux-kernel, linux-f2fs-devel
On 07/29, Chao Yu wrote:
> On 2024/7/27 11:27, Jaegeuk Kim wrote:
> > On 07/26, Chao Yu wrote:
> > > On 2024/7/26 0:55, Jaegeuk Kim wrote:
> > > > On 07/12, Chao Yu wrote:
> > > > > All directory blocks are within the scope of i_size, so let's limit
> > > > > the end_block to just check valid dirent blocks.
> > > >
> > > > Do we really need this?
> > >
> > > f2fs_readdir() and f2fs_empty_dir() uses dir_blocks() for upper boundary,
> > > this patch aligns find_in_level() w/ them.
> > >
> > > Also, it can avoid grabbing never used page cache across EOF.
> > >
> > > So, we can consider taking this patch?
> >
> > I'm wondering whether the current code has a bug or not.
>
> I think it's not a bug, but, in corner case it may waste memory and
> cause a bit performance reduction.
Well, I don't think it's really a problem.
>
> Thanks,
>
> >
> > >
> > > Thanks,
> > >
> > > >
> > > > >
> > > > > Meanwhile, it uses dir_blocks() instead of variable for cleanup in
> > > > > __f2fs_find_entry().
> > > > >
> > > > > Signed-off-by: Chao Yu <chao@kernel.org>
> > > > > ---
> > > > > fs/f2fs/dir.c | 6 ++++--
> > > > > 1 file changed, 4 insertions(+), 2 deletions(-)
> > > > >
> > > > > diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
> > > > > index 02c9355176d3..d4591c215f07 100644
> > > > > --- a/fs/f2fs/dir.c
> > > > > +++ b/fs/f2fs/dir.c
> > > > > @@ -305,18 +305,21 @@ static struct f2fs_dir_entry *find_in_level(struct inode *dir,
> > > > > int s = GET_DENTRY_SLOTS(fname->disk_name.len);
> > > > > unsigned int nbucket, nblock;
> > > > > unsigned int bidx, end_block;
> > > > > + unsigned long last_block;
> > > > > struct page *dentry_page;
> > > > > struct f2fs_dir_entry *de = NULL;
> > > > > pgoff_t next_pgofs;
> > > > > bool room = false;
> > > > > int max_slots;
> > > > > + last_block = dir_blocks(dir);
> > > > > nbucket = dir_buckets(level, F2FS_I(dir)->i_dir_level);
> > > > > nblock = bucket_blocks(level);
> > > > > bidx = dir_block_index(level, F2FS_I(dir)->i_dir_level,
> > > > > le32_to_cpu(fname->hash) % nbucket);
> > > > > end_block = bidx + nblock;
> > > > > + end_block = min_t(unsigned int, end_block, last_block);
> > > > > while (bidx < end_block) {
> > > > > /* no need to allocate new dentry pages to all the indices */
> > > > > @@ -361,7 +364,6 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> > > > > const struct f2fs_filename *fname,
> > > > > struct page **res_page)
> > > > > {
> > > > > - unsigned long npages = dir_blocks(dir);
> > > > > struct f2fs_dir_entry *de = NULL;
> > > > > unsigned int max_depth;
> > > > > unsigned int level;
> > > > > @@ -373,7 +375,7 @@ struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir,
> > > > > goto out;
> > > > > }
> > > > > - if (npages == 0)
> > > > > + if (dir_blocks(dir) == 0)
> > > > > goto out;
> > > > > max_depth = F2FS_I(dir)->i_current_depth;
> > > > > --
> > > > > 2.40.1
_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2024-07-29 16:21 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-07-12 7:34 [f2fs-dev] [PATCH] f2fs: don't traverse directory blocks after EOF Chao Yu
2024-07-25 16:55 ` Jaegeuk Kim
2024-07-26 1:06 ` Chao Yu
2024-07-27 3:27 ` Jaegeuk Kim
2024-07-29 2:54 ` Chao Yu
2024-07-29 16:21 ` Jaegeuk Kim
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).