[MTD NAND] Modify check for modules registering NAND devices without ->owner

Make it work even with compilers which lack the wit to notice that
THIS_MODULE is always non-NULL. Use #ifdef MODULE instead. It's only
a temporary debugging check anyway.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>

diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c
index 42cff0a2b93d2d7bd8f690a2a4b44ed2033f8729..08dffb7a9389abbb56d9289663ff8708ebf45c30 100644 (file)
--- a/drivers/mtd/nand/nand_base.c
+++ b/drivers/mtd/nand/nand_base.c
@@ -2312,6 +2312,15 @@ static void nand_resume(struct mtd_info *mtd)
 
 }
 
+/* module_text_address() isn't exported, and it's mostly a pointless
+   test if this is a module _anyway_ -- they'd have to try _really_ hard
+   to call us from in-kernel code if the core NAND support is modular. */
+#ifdef MODULE
+#define caller_is_module() (1)
+#else
+#define caller_is_module() module_text_address((unsigned long)__builtin_return_address(0))
+#endif
+
 /**
  * nand_scan - [NAND Interface] Scan for the NAND device
  * @mtd:       MTD device structure
@@ -2330,12 +2339,8 @@ int nand_scan(struct mtd_info *mtd, int maxchips)
        int i, nand_maf_id, nand_dev_id, busw, maf_id;
        struct nand_chip *this = mtd->priv;
 
-       /* module_text_address() isn't exported. But if _this_ is a module,
-          it's a fairly safe bet that its caller is a module too... and
-          that means the call to module_text_address() gets optimised out
-          without having to resort to ifdefs */
-       if (!mtd->owner && (THIS_MODULE ||
-           module_text_address((unsigned long)__builtin_return_address(0)))) {
+       /* Many callers got this wrong, so check for it for a while... */
+       if (!mtd->owner && caller_is_module()) {
                printk(KERN_CRIT "nand_scan() called with NULL mtd->owner!\n");
                BUG();
        }