staging: lustre: obdclass: don't require lct_owner to be non-NULL.
authorNeilBrown <neilb@suse.com>
Thu, 1 Mar 2018 23:31:25 +0000 (10:31 +1100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 6 Mar 2018 12:17:27 +0000 (04:17 -0800)
Some places in lu_object.c allow lct_owner to be NULL, implying
that the code is built in to the kernel (not a module), but
two places don't.  This prevents us from building lustre into
the kernel.

So remove the requirement and always allow lct_owner to be NULL.

This requires removing an "assert" that the module count is positive,
but this is redundant as module_put() already does the necessary test.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/staging/lustre/lustre/obdclass/lu_object.c

index 964e854..8ddf23b 100644 (file)
@@ -1380,12 +1380,8 @@ static void key_fini(struct lu_context *ctx, int index)
                lu_ref_del(&key->lct_reference, "ctx", ctx);
                atomic_dec(&key->lct_used);
 
-               if ((ctx->lc_tags & LCT_NOREF) == 0) {
-#ifdef CONFIG_MODULE_UNLOAD
-                       LINVRNT(module_refcount(key->lct_owner) > 0);
-#endif
+               if ((ctx->lc_tags & LCT_NOREF) == 0)
                        module_put(key->lct_owner);
-               }
                ctx->lc_value[index] = NULL;
        }
 }
@@ -1619,7 +1615,6 @@ static int keys_fill(struct lu_context *ctx)
                        LINVRNT(key->lct_init);
                        LINVRNT(key->lct_index == i);
 
-                       LASSERT(key->lct_owner);
                        if (!(ctx->lc_tags & LCT_NOREF) &&
                            !try_module_get(key->lct_owner)) {
                                /* module is unloading, skip this key */