bcache: bch_extent_bad() stale dirty pointer

[Eric Wheeler <bcache@xxxxxxxxxxxxxxxxxx>]

Hi Coly and Mingzhe,

Have you ever seen this dmesg notice before?  It only printed two lines 
into dmesg:

[Apr11 19:28] bcache: bch_extent_bad() stale dirty pointer, stale 1, key: 0:15956311168 len 24 -> [0:32419560 gen 88] dirty
[  +0.014297] bcache: bch_extent_bad() stale dirty pointer, stale 1, key: 0:15956311168 len 24 -> [0:32419560 gen 88] dirty

This doesn't look like its a major concern, after all just this is a 
pr_info, and the "stale" value of "1" is well below BUCKET_GC_GEN_MAX (96) 
so I'm not at risk of hitting btree_bug_on().

Is this something that should be addressed?

Near extents.c:537:

static bool bch_extent_bad(struct btree_keys *bk, const struct bkey *k)
{
        struct btree *b = container_of(bk, struct btree, keys);
        unsigned int i, stale;
        char buf[80];

        if (!KEY_PTRS(k) ||
            bch_extent_invalid(bk, k))
                return true;

        for (i = 0; i < KEY_PTRS(k); i++)
                if (!ptr_available(b->c, k, i))
                        return true;

        for (i = 0; i < KEY_PTRS(k); i++) {
                stale = ptr_stale(b->c, k, i);

                if (stale && KEY_DIRTY(k)) {
                        bch_extent_to_text(buf, sizeof(buf), k);
                        pr_info("stale dirty pointer, stale %u, key: %s\n",  <<<<<<<
                                stale, buf);
                }

                btree_bug_on(stale > BUCKET_GC_GEN_MAX, b,
                             "key too stale: %i, need_gc %u",
                             stale, b->c->need_gc);

                if (stale)
                        return true;

                if (expensive_debug_checks(b->c) &&
                    bch_extent_bad_expensive(b, k, i))
                        return true;
        }

        return false;
}



--
Eric Wheeler