No, it is never the same block failing... But the good news is that I finally succeed using ubifs on kirkwood: I had to upgrade uboot.... the old uboot was doing some bad init stuff i guess...
I will do more tests in the coming days, and keep you in touch. Regards, Romary Le 24/08/2011 18:43, Daniel Golle a écrit : > Is it always the same blocks failing? > If so, it's more likely to be some error on NAND Level, otherwise it > could also be SPI communication somehow failing or weird bugs in mtd > or ubi... > What NAND chip is used on your board? > I had a SheevaPlug for testing, but the internal power supply died a > week ago and probably killed something else as well when that happend > and the thing appears to be dead now. I still got some other Kirkwood > boards for testing, those are equipped with > NAND device: Manufacturer ID: 0xec, Chip ID: 0xd3 (Samsung NAND 1GiB > 3,3V 8-bit) > which also come with a few factory bad-blocks... > I currently use one of those a my local office server, so it's running > almost 24/7 for about 2 weeks now and survived all > filesystem-stress-tests (including installing and using a > debian-chroot on ubifs using debootstrap)... > >> ubiformat: mtd2 (nand), size 532676608 bytes (508.0 MiB), 4064 >> eraseblocks of 131072 bytes (128.0 KiB), min. I/O size 2048 bytes > > looks like it matches the parameters we are using now, so block > alignment is unlikely to be the cause. > What is also strange about the output is that it detects wrong ECCs on > PEB 25,26,27,32 and then keeps on torturing PEB 58 instead of having a > closer look at those... > > Maybe I was just lucky with the NANDs I had for now...? > > On 08/24/2011 03:48 AM, Romary Sonrier wrote: >> Hello >> >> I have just tried >> LINUX_VERSION:=2.6.39.2 >> and >> LINUX_VERSION:=3.0 >> >> Both gave the same result: once you do intensiv i/o on ubifs (like >> find /<mountpoint>/), i get strange message: >> >>> UBI: run torture test for PEB 58 UBI: PEB 58 passed torture test, do >>> not mark it as bad UBI error: ubi_io_read: error -74 (ECC error) >>> while reading 129024 bytes from PEB 25:2048, read 129024 bytes UBI >>> warning: ubi_eba_copy_leb: error -74 while reading data from PEB 25 >>> UBI error: ubi_io_read: error -74 (ECC error) while reading 129024 >>> bytes from PEB 26:2048, read 129024 bytes UBI warning: >>> ubi_eba_copy_leb: error -74 while reading data from PEB 26 UBI >>> error: ubi_io_read: error -74 (ECC error) while reading 129024 bytes >>> from PEB 27:2048, read 129024 bytes UBI warning: ubi_eba_copy_leb: >>> error -74 while reading data from PEB 27 UBI error: ubi_io_read: >>> error -74 (ECC error) while reading 129024 bytes from PEB 32:2048, >>> read 129024 bytes UBI warning: ubi_eba_copy_leb: error -74 while >>> reading data from PEB 32 UBI: run torture test for PEB 58 UBI: PEB >>> 58 passed torture test, do not mark it as bad UBI: ruUBI: PEB 58 >>> passed torture test, do not mark it as bad >> >> Could it comme from marvell mtd driver or some nand block alignement >> issue ? >> >> Best regards, >> >> Romary Sonrier >> > > _______________________________________________ > openwrt-devel mailing list > openwrt-devel@lists.openwrt.org > https://lists.openwrt.org/mailman/listinfo/openwrt-devel _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
