linux-f2fs-devel.lists.sourceforge.net archive mirror
 help / color / mirror / Atom feed
From: Chao Yu <yuchao0@huawei.com>
To: Jaegeuk Kim <jaegeuk@kernel.org>, Chao Yu <chao@kernel.org>
Cc: linux-kernel@vger.kernel.org, linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH] f2fs: allocate memory in batch in build_sit_info()
Date: Wed, 10 Jul 2019 09:25:26 +0800	[thread overview]
Message-ID: <6a110d61-3dac-2142-d4e5-cda928aa4938@huawei.com> (raw)
In-Reply-To: <20190709172347.GA53646@jaegeuk-macbookpro.roam.corp.google.com>

On 2019/7/10 1:23, Jaegeuk Kim wrote:
> On 07/09, Chao Yu wrote:
>> On 2019-7-9 7:46, Jaegeuk Kim wrote:
>>> On 07/04, Chao Yu wrote:
>>>> build_sit_info() allocate all bitmaps for each segment one by one,
>>>> it's quite low efficiency, this pach changes to allocate large
>>>> continuous memory at a time, and divide it and assign for each bitmaps
>>>
>>> It may give more failure rate?
>>
>> For android, I think there should be no problem, since while startup, memory
>> should be sufficient for f2fs mount.
>> For server or desktop, if there is any failure on memory allocation,
>> f2fs_kzalloc will fallback to vmalloc, so that would not be worse than before,
>> right?
>>
>> Or if you worry about this really, could we add a fast path:
> 
> How much time can we really get with this big alloc?

Total build_sit_info()'s elapsed time reduces from 20ms to 10ms on 128gb device.

Thanks,

> 
>>
>> build_sit_info()
>>  - try allocate sit_i->bitmap
>>   - success: divide memory
>>  - fail: fallback to old method
>>
>> Thanks,
>>
>>>
>>>> of segment. For large size image, it can expect improving its mount
>>>> speed.
>>>>
>>>> Signed-off-by: Chen Gong <gongchen4@huawei.com>
>>>> Signed-off-by: Chao Yu <yuchao0@huawei.com>
>>>> ---
>>>>  fs/f2fs/segment.c | 51 +++++++++++++++++++++--------------------------
>>>>  fs/f2fs/segment.h |  1 +
>>>>  2 files changed, 24 insertions(+), 28 deletions(-)
>>>>
>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>>> index 402fbbbb2d7c..73c803af1f31 100644
>>>> --- a/fs/f2fs/segment.c
>>>> +++ b/fs/f2fs/segment.c
>>>> @@ -3929,7 +3929,7 @@ static int build_sit_info(struct f2fs_sb_info *sbi)
>>>>  	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
>>>>  	struct sit_info *sit_i;
>>>>  	unsigned int sit_segs, start;
>>>> -	char *src_bitmap;
>>>> +	char *src_bitmap, *bitmap;
>>>>  	unsigned int bitmap_size;
>>>>  
>>>>  	/* allocate memory for SIT information */
>>>> @@ -3950,27 +3950,31 @@ static int build_sit_info(struct f2fs_sb_info *sbi)
>>>>  	if (!sit_i->dirty_sentries_bitmap)
>>>>  		return -ENOMEM;
>>>>  
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>> +	bitmap_size = MAIN_SEGS(sbi) * SIT_VBLOCK_MAP_SIZE * 4;
>>>> +#else
>>>> +	bitmap_size = MAIN_SEGS(sbi) * SIT_VBLOCK_MAP_SIZE * 3;
>>>> +#endif
>>>> +	sit_i->bitmap = f2fs_kzalloc(sbi, bitmap_size, GFP_KERNEL);
>>>> +	if (!sit_i->bitmap)
>>>> +		return -ENOMEM;
>>>> +
>>>> +	bitmap = sit_i->bitmap;
>>>> +
>>>>  	for (start = 0; start < MAIN_SEGS(sbi); start++) {
>>>> -		sit_i->sentries[start].cur_valid_map
>>>> -			= f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE, GFP_KERNEL);
>>>> -		sit_i->sentries[start].ckpt_valid_map
>>>> -			= f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE, GFP_KERNEL);
>>>> -		if (!sit_i->sentries[start].cur_valid_map ||
>>>> -				!sit_i->sentries[start].ckpt_valid_map)
>>>> -			return -ENOMEM;
>>>> +		sit_i->sentries[start].cur_valid_map = bitmap;
>>>> +		bitmap += SIT_VBLOCK_MAP_SIZE;
>>>> +
>>>> +		sit_i->sentries[start].ckpt_valid_map = bitmap;
>>>> +		bitmap += SIT_VBLOCK_MAP_SIZE;
>>>>  
>>>>  #ifdef CONFIG_F2FS_CHECK_FS
>>>> -		sit_i->sentries[start].cur_valid_map_mir
>>>> -			= f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE, GFP_KERNEL);
>>>> -		if (!sit_i->sentries[start].cur_valid_map_mir)
>>>> -			return -ENOMEM;
>>>> +		sit_i->sentries[start].cur_valid_map_mir = bitmap;
>>>> +		bitmap += SIT_VBLOCK_MAP_SIZE;
>>>>  #endif
>>>>  
>>>> -		sit_i->sentries[start].discard_map
>>>> -			= f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE,
>>>> -							GFP_KERNEL);
>>>> -		if (!sit_i->sentries[start].discard_map)
>>>> -			return -ENOMEM;
>>>> +		sit_i->sentries[start].discard_map = bitmap;
>>>> +		bitmap += SIT_VBLOCK_MAP_SIZE;
>>>>  	}
>>>>  
>>>>  	sit_i->tmp_map = f2fs_kzalloc(sbi, SIT_VBLOCK_MAP_SIZE, GFP_KERNEL);
>>>> @@ -4440,21 +4444,12 @@ static void destroy_free_segmap(struct f2fs_sb_info *sbi)
>>>>  static void destroy_sit_info(struct f2fs_sb_info *sbi)
>>>>  {
>>>>  	struct sit_info *sit_i = SIT_I(sbi);
>>>> -	unsigned int start;
>>>>  
>>>>  	if (!sit_i)
>>>>  		return;
>>>>  
>>>> -	if (sit_i->sentries) {
>>>> -		for (start = 0; start < MAIN_SEGS(sbi); start++) {
>>>> -			kvfree(sit_i->sentries[start].cur_valid_map);
>>>> -#ifdef CONFIG_F2FS_CHECK_FS
>>>> -			kvfree(sit_i->sentries[start].cur_valid_map_mir);
>>>> -#endif
>>>> -			kvfree(sit_i->sentries[start].ckpt_valid_map);
>>>> -			kvfree(sit_i->sentries[start].discard_map);
>>>> -		}
>>>> -	}
>>>> +	if (sit_i->sentries)
>>>> +		kvfree(sit_i->bitmap);
>>>>  	kvfree(sit_i->tmp_map);
>>>>  
>>>>  	kvfree(sit_i->sentries);
>>>> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
>>>> index 2fd53462fa27..4d171b489130 100644
>>>> --- a/fs/f2fs/segment.h
>>>> +++ b/fs/f2fs/segment.h
>>>> @@ -226,6 +226,7 @@ struct sit_info {
>>>>  	block_t sit_base_addr;		/* start block address of SIT area */
>>>>  	block_t sit_blocks;		/* # of blocks used by SIT area */
>>>>  	block_t written_valid_blocks;	/* # of valid blocks in main area */
>>>> +	char *bitmap;			/* all bitmaps pointer */
>>>>  	char *sit_bitmap;		/* SIT bitmap pointer */
>>>>  #ifdef CONFIG_F2FS_CHECK_FS
>>>>  	char *sit_bitmap_mir;		/* SIT bitmap mirror */
>>>> -- 
>>>> 2.18.0.rc1
> .
> 


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

      reply	other threads:[~2019-07-10  1:25 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-07-04  8:17 [f2fs-dev] [PATCH] f2fs: allocate memory in batch in build_sit_info() Chao Yu
2019-07-08 23:46 ` Jaegeuk Kim
2019-07-09  5:06   ` Chao Yu
2019-07-09 17:23     ` Jaegeuk Kim
2019-07-10  1:25       ` Chao Yu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6a110d61-3dac-2142-d4e5-cda928aa4938@huawei.com \
    --to=yuchao0@huawei.com \
    --cc=chao@kernel.org \
    --cc=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).