From 7ac760c2f78ddd8e1bd633767b01becfbbf96720 Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Thu, 2 Dec 2010 06:34:01 +0200 Subject: UBI: fix corrupted PEB detection for NOR flash My new shiny code for corrupted PEB detection has NOR specific bug. We tread PEB as corrupted and preserve it, if 1. EC header is OK. 2. VID header is corrupted. 3. data area is not "all 0xFFs" In case of NOR we have 'nor_erase_prepare()' quirk, which invalidates the headers before erasing the PEB. And we invalidate first the VID header, and then the EC header. So if a power cut happens after we have invalidated the VID header, but before we have invalidated the EC header, we end up with a PEB which satisfies the above 3 conditions, and the scanning code will treat it as corrupted, and will print scary warnings, wrongly. This patch fixes the issue by firt invalidating the EC header, then invalidating the VID header. In case of power cut inbetween, we still just lose the EC header, and UBI can deal with this situation gracefully. Thanks to Anatolij Gustschin for tracking this down. Signed-off-by: Artem Bityutskiy Reported-by: Anatolij Gustschin Tested-by: Anatolij Gustschin --- drivers/mtd/ubi/io.c | 37 +++++++++++++++++++++++++------------ drivers/mtd/ubi/scan.c | 4 ++++ 2 files changed, 29 insertions(+), 12 deletions(-) diff --git a/drivers/mtd/ubi/io.c b/drivers/mtd/ubi/io.c index c2960ac9f39..811775aa8ee 100644 --- a/drivers/mtd/ubi/io.c +++ b/drivers/mtd/ubi/io.c @@ -482,10 +482,17 @@ static int nor_erase_prepare(struct ubi_device *ubi, int pnum) uint32_t data = 0; struct ubi_vid_hdr vid_hdr; - addr = (loff_t)pnum * ubi->peb_size + ubi->vid_hdr_aloffset; + /* + * It is important to first invalidate the EC header, and then the VID + * header. Otherwise a power cut may lead to valid EC header and + * invalid VID header, in which case UBI will treat this PEB as + * corrupted and will try to preserve it, and print scary warnings (see + * the header comment in scan.c for more information). + */ + addr = (loff_t)pnum * ubi->peb_size; err = ubi->mtd->write(ubi->mtd, addr, 4, &written, (void *)&data); if (!err) { - addr -= ubi->vid_hdr_aloffset; + addr += ubi->vid_hdr_aloffset; err = ubi->mtd->write(ubi->mtd, addr, 4, &written, (void *)&data); if (!err) @@ -494,18 +501,24 @@ static int nor_erase_prepare(struct ubi_device *ubi, int pnum) /* * We failed to write to the media. This was observed with Spansion - * S29GL512N NOR flash. Most probably the eraseblock erasure was - * interrupted at a very inappropriate moment, so it became unwritable. - * In this case we probably anyway have garbage in this PEB. + * S29GL512N NOR flash. Most probably the previously eraseblock erasure + * was interrupted at a very inappropriate moment, so it became + * unwritable. In this case we probably anyway have garbage in this + * PEB. */ err1 = ubi_io_read_vid_hdr(ubi, pnum, &vid_hdr, 0); - if (err1 == UBI_IO_BAD_HDR_EBADMSG || err1 == UBI_IO_BAD_HDR) - /* - * The VID header is corrupted, so we can safely erase this - * PEB and not afraid that it will be treated as a valid PEB in - * case of an unclean reboot. - */ - return 0; + if (err1 == UBI_IO_BAD_HDR_EBADMSG || err1 == UBI_IO_BAD_HDR) { + struct ubi_ec_hdr ec_hdr; + + err1 = ubi_io_read_ec_hdr(ubi, pnum, &ec_hdr, 0); + if (err1 == UBI_IO_BAD_HDR_EBADMSG || err1 == UBI_IO_BAD_HDR) + /* + * Both VID and EC headers are corrupted, so we can + * safely erase this PEB and not afraid that it will be + * treated as a valid PEB in case of an unclean reboot. + */ + return 0; + } /* * The PEB contains a valid VID header, but we cannot invalidate it. diff --git a/drivers/mtd/ubi/scan.c b/drivers/mtd/ubi/scan.c index 204345be8e6..79ca304fc4d 100644 --- a/drivers/mtd/ubi/scan.c +++ b/drivers/mtd/ubi/scan.c @@ -953,6 +953,10 @@ static int process_eb(struct ubi_device *ubi, struct ubi_scan_info *si, * impossible to distinguish it from a PEB which just * contains garbage because of a power cut during erase * operation. So we just schedule this PEB for erasure. + * + * Besides, in case of NOR flash, we deliberatly + * corrupt both headers because NOR flash erasure is + * slow and can start from the end. */ err = 0; else -- cgit v1.2.3-70-g09d2