On 2019/03/06 1:55, Jaegeuk Kim wrote: > On 03/05, Damien Le Moal wrote: >> Johannes, >> >> On 2019/03/04 18:46, Johannes Thumshirn wrote: >>> Hi Damien, >>> >>> On 04/03/2019 08:04, Damien Le Moal wrote: >>>> @@ -2765,9 +2765,11 @@ static int init_blkz_info(struct f2fs_sb_info *sbi, int devi) >>>> if (nr_sectors & (bdev_zone_sectors(bdev) - 1)) >>>> FDEV(devi).nr_blkz++; >>>> >>>> - FDEV(devi).blkz_type = f2fs_kmalloc(sbi, FDEV(devi).nr_blkz, >>>> - GFP_KERNEL); >>>> - if (!FDEV(devi).blkz_type) >>>> + FDEV(devi).blkz_seq = f2fs_kzalloc(sbi, >>>> + BITS_TO_LONGS(FDEV(devi).nr_blkz) >>>> + * sizeof(unsigned long), >>>> + GFP_KERNEL); >>>> + if (!FDEV(devi).blkz_seq) >>>> return -ENOMEM; >>> >>> Not so sure about F2FS internals, but there is a bitmap_zalloc() in the >>> normal kernel library. >> >> Yes indeed... f2fs_kzalloc uses f2fs_kmalloc(__GFP_ZERO) and f2fs_kmalloc is >> basically kmalloc() or kvmalloc() but with error injection for tests. So I used >> that instead of bitmap_zalloc() to preserve the error injection test. >> >> Jaegeuk, >> >> Which do you prefer ? The patch as it is or switching to bitmap_zalloc() ? >> Whichever is fine with me so I can send a v2 if you prefer bitmap_zalloc(). > > Hi Damien, > > I think f2fs_kmalloc would be fine for fault injection tests. It seems it'd > better to write a clean-up patch which replaces all the bitmap allocations > in f2fs with single f2fs_bitmap_zalloc() at once. Sounds good to me. So will you take the patch as is ? Any comment on it ? This change was tested on a 15TB SMR disks. > > I'll take a look at this. > Thanks, > >> >> Best regards. >> >> -- >> Damien Le Moal >> Western Digital Research > -- Damien Le Moal Western Digital Research
