Re: [f2fs-dev] [PATCH v2] f2fs: Add bggc_block_io to adjust the priority of BG_GC when issuing IO

[Chao Yu via Linux-f2fs-devel <linux-f2fs-devel@lists.sourceforge.net>]

On 2025/8/7 17:12, Liao Yuanhong wrote:
> 
> On 8/7/2025 4:38 PM, Chao Yu wrote:
>> On 8/6/25 15:09, Liao Yuanhong wrote:
>>> Currently, we have encountered some issues while testing ZUFS. In
>>> situations near the storage limit (e.g., 50GB remaining), and after
>>> simulating fragmentation by repeatedly writing and deleting data, we found
>>> that application installation and startup tests conducted after idling for
>>> a few minutes take significantly longer several times that of traditional
>>> UFS. Tracing the operations revealed that the majority of I/Os were issued
>>> by background GC, which blocks normal I/O operations.
>>>
>>> Under normal circumstances, ZUFS indeed requires more background GC and
>>> employs a more aggressive GC strategy. However, I aim to find a way to
>>> minimize the impact on regular I/O operations under these near-limit
>>> conditions. To address this, I have introduced a bggc_block_io feature,
>>> which controls the prioritization of background GC in the presence of I/Os.
>>> This switch can be adjusted at the framework level to implement different
>>> strategies. If set to ALL_IO_PRIOR, all background GC operations will be
>>> skipped during active I/O issuance. The default option remains consistent
>>> with the current strategy, ensuring no change in behavior.
>>>
>>> Signed-off-by: Liao Yuanhong <liaoyuanhong@vivo.com>
>>> ---
>>> Changes in v2:
>>>     Non ZUFS can also be adjusted through this option.
>>> ---
>>>   Documentation/ABI/testing/sysfs-fs-f2fs | 13 +++++++++++++
>>>   fs/f2fs/f2fs.h                          | 18 +++++++++++-------
>>>   fs/f2fs/super.c                         |  2 ++
>>>   fs/f2fs/sysfs.c                         |  9 +++++++++
>>>   4 files changed, 35 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/Documentation/ABI/testing/sysfs-fs-f2fs b/Documentation/ABI/testing/sysfs-fs-f2fs
>>> index bc0e7fefc39d..12fda11d4da5 100644
>>> --- a/Documentation/ABI/testing/sysfs-fs-f2fs
>>> +++ b/Documentation/ABI/testing/sysfs-fs-f2fs
>>> @@ -883,3 +883,16 @@ Date:        June 2025
>>>   Contact:    "Daeho Jeong" <daehojeong@google.com>
>>>   Description:    Control GC algorithm for boost GC. 0: cost benefit, 1: greedy
>>>           Default: 1
>>> +
>>> +What:        /sys/fs/f2fs/<disk>/bggc_block_io
>>> +Date:        August 2025
>>> +Contact:    "Liao Yuanhong" <liaoyuanhong@vivo.com>
>>> +Description:    Used to adjust the BG_GC priority when issuing IO, with a default value
>>> +        of 1.
>> Default value is 2 if CONFIG_BLK_DEV_ZONED is off?
>>
>> Thanks,
> 
> Sorry, I missed updating the description. Is it more reasonable to set the default value to 0 for versions that disable CONFIG_BLK_DEV_ZONED? If so, I will submit a new version to correct the description and initial value settings here. Thanks, Liao

Before this patch, the default behavior is something like ALL_IO_PRIOR?

What do you think of defining ALL_IO_PRIOR w/ 0, and use it by default?

Thanks,

> 
>>> +
>>> +        ==================  =============================================
>>> +        value                description
>>> +        bggc_block_io = 0   Prioritize background GC
>>> +        bggc_block_io = 1   Stop background GC only when issuing read I/O
>>> +        bggc_block_io = 2   Stop background GC when issuing I/O
>>> +        ==================  =============================================
>>> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
>>> index 46be7560548c..fe41b733fbc2 100644
>>> --- a/fs/f2fs/f2fs.h
>>> +++ b/fs/f2fs/f2fs.h
>>> @@ -155,6 +155,12 @@ enum blkzone_allocation_policy {
>>>       BLKZONE_ALLOC_PRIOR_CONV,    /* Prioritize writing to conventional zones */
>>>   };
>>> +enum bggc_block_io_policy {
>>> +    BGGC_PRIOR,
>>> +    READ_IO_PRIOR,
>>> +    ALL_IO_PRIOR,
>>> +};
>>> +
>>>   /*
>>>    * An implementation of an rwsem that is explicitly unfair to readers. This
>>>    * prevents priority inversion when a low-priority reader acquires the read lock
>>> @@ -1804,6 +1810,7 @@ struct f2fs_sb_info {
>>>       spinlock_t dev_lock;            /* protect dirty_device */
>>>       bool aligned_blksize;            /* all devices has the same logical blksize */
>>>       unsigned int first_seq_zone_segno;    /* first segno in sequential zone */
>>> +    unsigned int bggc_block_io;        /* For adjust the BG_GC priority when issuing IO */
>>>       /* For write statistics */
>>>       u64 sectors_written_start;
>>> @@ -2998,13 +3005,10 @@ static inline bool is_idle(struct f2fs_sb_info *sbi, int type)
>>>       if (sbi->gc_mode == GC_URGENT_HIGH)
>>>           return true;
>>> -    if (zoned_gc) {
>>> -        if (is_inflight_read_io(sbi))
>>> -            return false;
>>> -    } else {
>>> -        if (is_inflight_io(sbi, type))
>>> -            return false;
>>> -    }
>>> +    if (sbi->bggc_block_io == READ_IO_PRIOR && is_inflight_read_io(sbi))
>>> +        return false;
>>> +    if (sbi->bggc_block_io == ALL_IO_PRIOR && is_inflight_io(sbi, type))
>>> +        return false;
>>>       if (sbi->gc_mode == GC_URGENT_MID)
>>>           return true;
>>> diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
>>> index e16c4e2830c2..a21cecc5424e 100644
>>> --- a/fs/f2fs/super.c
>>> +++ b/fs/f2fs/super.c
>>> @@ -4629,9 +4629,11 @@ static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
>>>       logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
>>>       sbi->aligned_blksize = true;
>>> +    sbi->bggc_block_io = ALL_IO_PRIOR;
>>>   #ifdef CONFIG_BLK_DEV_ZONED
>>>       sbi->max_open_zones = UINT_MAX;
>>>       sbi->blkzone_alloc_policy = BLKZONE_ALLOC_PRIOR_SEQ;
>>> +    sbi->bggc_block_io = READ_IO_PRIOR;
>>>   #endif
>>>       for (i = 0; i < max_devices; i++) {
>>> diff --git a/fs/f2fs/sysfs.c b/fs/f2fs/sysfs.c
>>> index f736052dea50..efea15209788 100644
>>> --- a/fs/f2fs/sysfs.c
>>> +++ b/fs/f2fs/sysfs.c
>>> @@ -674,6 +674,13 @@ static ssize_t __sbi_store(struct f2fs_attr *a,
>>>           sbi->blkzone_alloc_policy = t;
>>>           return count;
>>>       }
>>> +
>>> +    if (!strcmp(a->attr.name, "bggc_block_io")) {
>>> +        if (t < BGGC_PRIOR || t > ALL_IO_PRIOR)
>>> +            return -EINVAL;
>>> +        sbi->bggc_block_io = t;
>>> +        return count;
>>> +    }
>>>   #endif
>>>   #ifdef CONFIG_F2FS_FS_COMPRESSION
>>> @@ -1172,6 +1179,7 @@ F2FS_SBI_GENERAL_RW_ATTR(max_read_extent_count);
>>>   #ifdef CONFIG_BLK_DEV_ZONED
>>>   F2FS_SBI_GENERAL_RO_ATTR(unusable_blocks_per_sec);
>>>   F2FS_SBI_GENERAL_RW_ATTR(blkzone_alloc_policy);
>>> +F2FS_SBI_GENERAL_RW_ATTR(bggc_block_io);
>>>   #endif
>>>   F2FS_SBI_GENERAL_RW_ATTR(carve_out);
>>>   F2FS_SBI_GENERAL_RW_ATTR(reserved_pin_section);
>>> @@ -1342,6 +1350,7 @@ static struct attribute *f2fs_attrs[] = {
>>>   #ifdef CONFIG_BLK_DEV_ZONED
>>>       ATTR_LIST(unusable_blocks_per_sec),
>>>       ATTR_LIST(blkzone_alloc_policy),
>>> +    ATTR_LIST(bggc_block_io),
>>>   #endif
>>>   #ifdef CONFIG_F2FS_FS_COMPRESSION
>>>       ATTR_LIST(compr_written_block),



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