On Wed, 26 Sep 2012 09:39:08 +0800 Fengguang Wu <fengguang.wu@xxxxxxxxx>
wrote:
> Hi Alexander,
>
> FYI, there are new compile warnings show up in
>
> tree: git://neil.brown.name/md for-next
> head: 0179d361662367ab0f80b5822b162a1c0ba9b57d
> commit: eadb9fb1d4da32ed1b143a6fc1997c4c42ea0422 [32/33] md: When RAID5 is dirty, force reconstruct-write instead of read-modify-write.
> config: m68k-allyesconfig
>
> All warnings:
>
> drivers/md/raid5.c: In function 'handle_stripe_dirtying':
> drivers/md/raid5.c:2866:3: warning: format '%lu' expects argument of type 'long unsigned int', but argument 4 has type 'sector_t' [-Wformat]
> drivers/md/raid5.c:2866:3: warning: format '%lu' expects argument of type 'long unsigned int', but argument 5 has type 'sector_t' [-Wformat]
>
> vim +2866 drivers/md/raid5.c
> 2850 sector_t recovery_cp = conf->mddev->recovery_cp;
> 2851
> 2852 /* RAID6 requires 'rcw' in current implementation.
> 2853 * Otherwise, check whether resync is now happening or should start.
> 2854 * If yes, then the array is dirty (after unclean shutdown or
> 2855 * initial creation), so parity in some stripes might be inconsistent.
> 2856 * In this case, we need to always do reconstruct-write, to ensure
> 2857 * that in case of drive failure or read-error correction, we
> 2858 * generate correct data from the parity.
> 2859 */
> 2860 if (conf->max_degraded == 2 ||
> 2861 (recovery_cp < MaxSector && sh->sector >= recovery_cp)) {
> 2862 /* Calculate the real rcw later - for now make it
> 2863 * look like rcw is cheaper
> 2864 */
> 2865 rcw = 1; rmw = 2;
> > 2866 pr_debug("force RCW max_degraded=%u, recovery_cp=%lu sh->sector=%lu\n",
> 2867 conf->max_degraded, recovery_cp, sh->sector);
> 2868 } else for (i = disks; i--; ) {
> 2869 /* would I have to read this buffer for read_modify_write */
> 2870 struct r5dev *dev = &sh->dev[i];
> 2871 if ((dev->towrite || i == sh->pd_idx) &&
> 2872 !test_bit(R5_LOCKED, &dev->flags) &&
> 2873 !(test_bit(R5_UPTODATE, &dev->flags) ||
> 2874 test_bit(R5_Wantcompute, &dev->flags))) {
>
> ---
> 0-DAY kernel build testing backend Open Source Technology Centre
> Fengguang Wu, Yuanhan Liu Intel Corporation
Thanks again!
I've fixed this up.
NeilBrown
Attachment:
signature.asc
Description: PGP signature
