On 2020/1/8 20:04, Jaegeuk Kim wrote:
> On 01/08, Chao Yu wrote:
>> On 2020/1/7 10:07, Jaegeuk Kim wrote:
>>> Setting 0x40 in /sys/fs/f2fs/dev/ipu_policy gives a way to turn off
>>> bio cache, which is useufl to check whether block layer using hardware
>>> encryption engine merges IOs correctly.
>>>
>>> Signed-off-by: Jaegeuk Kim <[email protected]>
>>> ---
>>> Documentation/filesystems/f2fs.txt | 1 +
>>> fs/f2fs/segment.c | 2 +-
>>> fs/f2fs/segment.h | 1 +
>>> 3 files changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/Documentation/filesystems/f2fs.txt b/Documentation/filesystems/f2fs.txt
>>> index 41b5aa94b30f..cd93bcc34726 100644
>>> --- a/Documentation/filesystems/f2fs.txt
>>> +++ b/Documentation/filesystems/f2fs.txt
>>> @@ -335,6 +335,7 @@ Files in /sys/fs/f2fs/<devname>
>>> 0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
>>> 0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
>>> 0x10: F2FS_IPU_FSYNC.
>>
>> . -> ,
>
> Actually, we can't do it. I revised it a bit instead.
One more question, why skipping 0x20 bit position?
Thanks,
>
>>
>> Reviewed-by: Chao Yu <[email protected]>
>>
>> Thanks,
>>
>>> + 0x40: F2FS_IPU_NOCACHE disables bio caches.
>>>
>>> min_ipu_util This parameter controls the threshold to trigger
>>> in-place-updates. The number indicates percentage
>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>> index a9519532c029..311fe4937f6a 100644
>>> --- a/fs/f2fs/segment.c
>>> +++ b/fs/f2fs/segment.c
>>> @@ -3289,7 +3289,7 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)
>>>
>>> stat_inc_inplace_blocks(fio->sbi);
>>>
>>> - if (fio->bio)
>>> + if (fio->bio && !(SM_I(sbi)->ipu_policy & (1 << F2FS_IPU_NOCACHE)))
>>> err = f2fs_merge_page_bio(fio);
>>> else
>>> err = f2fs_submit_page_bio(fio);
>>> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
>>> index a1b3951367cd..02e620470eef 100644
>>> --- a/fs/f2fs/segment.h
>>> +++ b/fs/f2fs/segment.h
>>> @@ -623,6 +623,7 @@ enum {
>>> F2FS_IPU_SSR_UTIL,
>>> F2FS_IPU_FSYNC,
>>> F2FS_IPU_ASYNC,
>>> + F2FS_IPU_NOCACHE,
>>> };
>>>
>>> static inline unsigned int curseg_segno(struct f2fs_sb_info *sbi,
>>>
> .
>
On 01/08, Chao Yu wrote:
> On 2020/1/8 20:04, Jaegeuk Kim wrote:
> > On 01/08, Chao Yu wrote:
> >> On 2020/1/7 10:07, Jaegeuk Kim wrote:
> >>> Setting 0x40 in /sys/fs/f2fs/dev/ipu_policy gives a way to turn off
> >>> bio cache, which is useufl to check whether block layer using hardware
> >>> encryption engine merges IOs correctly.
> >>>
> >>> Signed-off-by: Jaegeuk Kim <[email protected]>
> >>> ---
> >>> Documentation/filesystems/f2fs.txt | 1 +
> >>> fs/f2fs/segment.c | 2 +-
> >>> fs/f2fs/segment.h | 1 +
> >>> 3 files changed, 3 insertions(+), 1 deletion(-)
> >>>
> >>> diff --git a/Documentation/filesystems/f2fs.txt b/Documentation/filesystems/f2fs.txt
> >>> index 41b5aa94b30f..cd93bcc34726 100644
> >>> --- a/Documentation/filesystems/f2fs.txt
> >>> +++ b/Documentation/filesystems/f2fs.txt
> >>> @@ -335,6 +335,7 @@ Files in /sys/fs/f2fs/<devname>
> >>> 0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
> >>> 0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
> >>> 0x10: F2FS_IPU_FSYNC.
> >>
> >> . -> ,
> >
> > Actually, we can't do it. I revised it a bit instead.
>
> One more question, why skipping 0x20 bit position?
It seems original patch missed to add comment.
From f9447095de55a3cda1023a37a5e1cb6dd2f54ebb Mon Sep 17 00:00:00 2001
From: Jaegeuk Kim <[email protected]>
Date: Wed, 8 Jan 2020 15:10:02 -0800
Subject: [PATCH] f2fs: update f2fs document regarding to fsync_mode
This patch adds missing fsync_mode entry in f2fs document.
Fixes: 04485987f053 ("f2fs: introduce async IPU policy")
Signed-off-by: Jaegeuk Kim <[email protected]>
---
Documentation/filesystems/f2fs.txt | 3 ++-
fs/f2fs/segment.h | 5 +++--
2 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/Documentation/filesystems/f2fs.txt b/Documentation/filesystems/f2fs.txt
index b80a7b69f210..ee61ace30276 100644
--- a/Documentation/filesystems/f2fs.txt
+++ b/Documentation/filesystems/f2fs.txt
@@ -334,7 +334,8 @@ Files in /sys/fs/f2fs/<devname>
updates in f2fs. User can set:
0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
- 0x10: F2FS_IPU_FSYNC, 0x40: F2FS_IPU_NOCACHE.
+ 0x10: F2FS_IPU_FSYNC, 0x20: F2FS_IPU_ASYNC,
+ 0x40: F2FS_IPU_NOCACHE.
Refer segment.h for details.
min_ipu_util This parameter controls the threshold to trigger
diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
index 5e6cd8d8411d..459dc3901a57 100644
--- a/fs/f2fs/segment.h
+++ b/fs/f2fs/segment.h
@@ -607,9 +607,10 @@ static inline int utilization(struct f2fs_sb_info *sbi)
* threashold,
* F2FS_IPU_FSYNC - activated in fsync path only for high performance flash
* storages. IPU will be triggered only if the # of dirty
- * pages over min_fsync_blocks.
+ * pages over min_fsync_blocks. (=default option)
+ * F2FS_IPU_ASYNC - do IPU given by asynchronous write requests.
* F2FS_IPU_NOCACHE - disable IPU bio cache.
- * F2FS_IPUT_DISABLE - disable IPU. (=default option)
+ * F2FS_IPUT_DISABLE - disable IPU. (=default option in LFS mode)
*/
#define DEF_MIN_IPU_UTIL 70
#define DEF_MIN_FSYNC_BLOCKS 8
--
2.24.0.525.g8f36a354ae-goog
On 2020/1/9 7:18, Jaegeuk Kim wrote:
> On 01/08, Chao Yu wrote:
>> On 2020/1/8 20:04, Jaegeuk Kim wrote:
>>> On 01/08, Chao Yu wrote:
>>>> On 2020/1/7 10:07, Jaegeuk Kim wrote:
>>>>> Setting 0x40 in /sys/fs/f2fs/dev/ipu_policy gives a way to turn off
>>>>> bio cache, which is useufl to check whether block layer using hardware
>>>>> encryption engine merges IOs correctly.
>>>>>
>>>>> Signed-off-by: Jaegeuk Kim <[email protected]>
>>>>> ---
>>>>> Documentation/filesystems/f2fs.txt | 1 +
>>>>> fs/f2fs/segment.c | 2 +-
>>>>> fs/f2fs/segment.h | 1 +
>>>>> 3 files changed, 3 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/Documentation/filesystems/f2fs.txt b/Documentation/filesystems/f2fs.txt
>>>>> index 41b5aa94b30f..cd93bcc34726 100644
>>>>> --- a/Documentation/filesystems/f2fs.txt
>>>>> +++ b/Documentation/filesystems/f2fs.txt
>>>>> @@ -335,6 +335,7 @@ Files in /sys/fs/f2fs/<devname>
>>>>> 0x01: F2FS_IPU_FORCE, 0x02: F2FS_IPU_SSR,
>>>>> 0x04: F2FS_IPU_UTIL, 0x08: F2FS_IPU_SSR_UTIL,
>>>>> 0x10: F2FS_IPU_FSYNC.
>>>>
>>>> . -> ,
>>>
>>> Actually, we can't do it. I revised it a bit instead.
>>
>> One more question, why skipping 0x20 bit position?
>
> It seems original patch missed to add comment.
>
>>From f9447095de55a3cda1023a37a5e1cb6dd2f54ebb Mon Sep 17 00:00:00 2001
> From: Jaegeuk Kim <[email protected]>
> Date: Wed, 8 Jan 2020 15:10:02 -0800
> Subject: [PATCH] f2fs: update f2fs document regarding to fsync_mode
>
> This patch adds missing fsync_mode entry in f2fs document.
>
> Fixes: 04485987f053 ("f2fs: introduce async IPU policy")
> Signed-off-by: Jaegeuk Kim <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Thanks,