Revert "gfs2: Don't reject a supposedly full bitmap if we have blocks reserved"
authorAndreas Gruenbacher <agruenba@redhat.com>
Tue, 6 Oct 2020 12:29:04 +0000 (14:29 +0200)
committerAndreas Gruenbacher <agruenba@redhat.com>
Mon, 18 Jan 2021 15:56:37 +0000 (16:56 +0100)
This reverts commit e79e0e1428188b24c3b57309ffa54a33c4ae40c4.

It turns out that we're only setting the GBF_FULL flag of a bitmap if we've
been scanning from the beginning of the bitmap until the end and we haven't
found a single free block, and we're not skipping reservations in that process,
either.  This means that in gfs2_rbm_find, we can always skip bitmaps with the
GBF_FULL flag set.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
fs/gfs2/rgrp.c

index f759f79d2af76939012a45fa6f9282cdf1dcf3f7..faf9f21f3e98b1c5e2d5bf2bbf558e13f6e3ed68 100644 (file)
@@ -1754,8 +1754,7 @@ static int gfs2_rbm_find(struct gfs2_rbm *rbm, u8 state, u32 *minext,
 
        while(1) {
                bi = rbm_bi(rbm);
-               if ((ip == NULL || !gfs2_rs_active(&ip->i_res)) &&
-                   test_bit(GBF_FULL, &bi->bi_flags) &&
+               if (test_bit(GBF_FULL, &bi->bi_flags) &&
                    (state == GFS2_BLKST_FREE))
                        goto next_bitmap;