From: Artem Bityutskiy Date: Mon, 23 Jul 2007 13:06:50 +0000 (+0300) Subject: [MTD] [NAND] fix race in nand_base.c X-Git-Tag: v2.6.24-rc1~1364^2~46 X-Git-Url: https://git.openpandora.org/cgi-bin/gitweb.cgi?p=pandora-kernel.git;a=commitdiff_plain;h=c0b8ba7bfe7b4c1f11b3356b65520405b005bd33 [MTD] [NAND] fix race in nand_base.c When we mark block bad we have to get chip because this involves writing to the page's OOB. We hit this bug in UBI - we observed random obscure crashes when it marks block bad from the background thread and there is some parallel task which utilizes flash. This patch also adds a TODO note about BBT table protection which it seems does not exist. Signed-off-by: Artem Bityutskiy Signed-off-by: David Woodhouse --- diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c index 25673eacdd88..24ac6778b1a8 100644 --- a/drivers/mtd/nand/nand_base.c +++ b/drivers/mtd/nand/nand_base.c @@ -24,6 +24,7 @@ * if we have HW ecc support. * The AG-AND chips have nice features for speed improvement, * which are not supported yet. Read / program 4 pages in one go. + * BBT table is not serialized, has to be fixed * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as @@ -360,6 +361,7 @@ static int nand_default_block_markbad(struct mtd_info *mtd, loff_t ofs) /* We write two bytes, so we dont have to mess with 16 bit * access */ + nand_get_device(chip, mtd, FL_WRITING); ofs += mtd->oobsize; chip->ops.len = chip->ops.ooblen = 2; chip->ops.datbuf = NULL; @@ -367,9 +369,11 @@ static int nand_default_block_markbad(struct mtd_info *mtd, loff_t ofs) chip->ops.ooboffs = chip->badblockpos & ~0x01; ret = nand_do_write_oob(mtd, ofs, &chip->ops); + nand_release_device(mtd); } if (!ret) mtd->ecc_stats.badblocks++; + return ret; }