On Mon, Aug 21, 2017 at 03:57:13PM +0800, zhangyu-fnst@xxxxxxxxxxxxxx wrote: > From: Zhang Yu <zhangyu-fnst@xxxxxxxxxxxxxx> > > [TEST/fuzz] case: 004-simple-dump-tree > > Since the wrong key(DATA_RELOC_TREE CHUNK_ITEM 0) in root tree, > error calling print_chunk(), resulting in num_stripes == 0. > > ERROR: > [TEST/fuzz] 004-simple-dump-tree > ctree.h:317: btrfs_chunk_item_size: BUG_ON `num_stripes == 0` > triggered, value 1 > > failed (ignored, ret=134): /myproject/btrfs-progs/btrfs > inspect-internal dump-tree > /myproject/btrfs-progs/tests/fuzz-tests/images/ > bko-155201-wrong-chunk-item-in-root-tree.raw.restored > > test failed for case 004-simple-dump-tree > Makefile:288: recipe for target 'test-fuzz' failed > make: *** [test-fuzz] Error 1 > > So, before printing item, determine the root is valid or not. I don't think this is the right way to fix it. The print-tree function should print everything that's found, possibly doing sanity checks and then only skip the bad data. For debugging or other purposes, we want to get exact state of the trees. The original problem you found is wrong number of stripes, so it should be dealt with in print_chunk. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
