Hi Jaegeuk, On 2017/7/22 4:54, Jaegeuk Kim wrote: > Hi Chao, > > On 07/21, Chao Yu wrote: >> When ->freeze_fs is called from lvm for doing snapshot, it needs to >> make sure there will be no more changes in filesystem's data, however, >> previously, background GC wasn't aware of freezing, so in environment >> with active background GC thread, data of snapshot becomes unstable. > > What about flush/discard threads?
Updated. Thanks, > > Thanks, > >> >> This patch fixes this issue by adding sb_{start,end}_intwrite in GC >> flow. >> >> Signed-off-by: Chao Yu <yuch...@huawei.com> >> --- >> fs/f2fs/gc.c | 8 ++++++-- >> 1 file changed, 6 insertions(+), 2 deletions(-) >> >> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c >> index 76ad2c3d88db..1c0117f60083 100644 >> --- a/fs/f2fs/gc.c >> +++ b/fs/f2fs/gc.c >> @@ -55,6 +55,8 @@ static int gc_thread_func(void *data) >> } >> #endif >> >> + sb_start_intwrite(sbi->sb); >> + >> /* >> * [GC triggering condition] >> * 0. GC is not conducted currently. >> @@ -69,12 +71,12 @@ static int gc_thread_func(void *data) >> * So, I'd like to wait some time to collect dirty segments. >> */ >> if (!mutex_trylock(&sbi->gc_mutex)) >> - continue; >> + goto next; >> >> if (!is_idle(sbi)) { >> increase_sleep_time(gc_th, &wait_ms); >> mutex_unlock(&sbi->gc_mutex); >> - continue; >> + goto next; >> } >> >> if (has_enough_invalid_blocks(sbi)) >> @@ -93,6 +95,8 @@ static int gc_thread_func(void *data) >> >> /* balancing f2fs's metadata periodically */ >> f2fs_balance_fs_bg(sbi); >> +next: >> + sb_end_intwrite(sbi->sb); >> >> } while (!kthread_should_stop()); >> return 0; >> -- >> 2.13.1.388.g69e6b9b4f4a9