While compile checking mtd changes in PR#1359 I've noticed following
compiler warnings and cleaned them up:
fis.c: In function 'fis_remap':
fis.c:143:25: warning: variable 'redboot' set but not used [-Wunused-but-set-variable]
struct fis_image_desc *redboot = NULL;
^~~~~~~
fis.c:142:25: warning: variable 'fisdir' set but not used [-Wunused-but-set-variable]
struct fis_image_desc *fisdir = NULL;
^~~~~~
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The code for calculating the CRC32 signatures for RedBoot FIS partitions
was already included, but for unknown reasons, it was never invoked. Some
bootloaders enforce checking these for loaded kernels, so they should be
written. This patch does so.
Tested-by: Brian Gonyer <bgonyer@gmail.com>
Signed-off-by: Daniel Gimpelevich <daniel@gimpelevich.san-francisco.ca.us>
Using the same method as the D-Link DAP-2695 A1 we use
the "mtd" tool to augment the firmware checkum in flash
on first boot of a new firmware on the D-Link DIR-685.
We need to augment the Makefile for "mtd" to build in
the special WRGG fixup support for Gemini as well.
This works around the problem of the machine not booting
after factory install unless the sysupgrade is applied
immediately.
Based on commit e3875350f3
"ar71xx: add support for D-Link DAP-2695 rev. A1"
Cc: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The D-Link DIR-685 has the same problem as the
D-Link DAP-2695: when flashing the factory image, the
checksum includes the whole flashed image, even the
rootfs_data part with the end of filesystem mark.
Also the whole flashed image is stored in the flash,
so on the first boot, the whole rootfs image is loaded
into memory with the kernel.
This is fixed using the fixwrgg command to mtd, but
for this to work we need to make fixwrgg work with
the Little-Endian ARM DIR-685.
The code tries to be endian agnostic but this fails
because the WRGG image loader doesn't. On ARM, the
file size is stored in little endian format, and on
big-endian systems it is stored in big endian format,
so we can just drop all the friendly htonl() that
will make the shdr->size big endian: this will
actually break the little endian systems, and on
the big endian systems the native endianness will
still be correct.
The magic number is always stored in little endian
format however, so make sure this is always read
in LE32 format. I chose to create a straight-forward
le32_to_cpu() static inline that IMO is simple and
easy to read.
Cc: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Consistently handle boot-count reset and upgrade across
ipq40xx, ipq806x, kirkwood, mvebu
Dual-firmware devices often utilize a specific MTD partition
to record the number of times the boot loader has initiated boot.
Most of these devices are NAND, typically with a 2k erase size.
When this code was ported to the ipq40xx platform, the device in hand
used NOR for this partition, with a 16-byte "record" size. As the
implementation of `mtd resetbc` is by-platform, the hard-coded nature
of this change prevented proper operation of a NAND-based device.
* Unified the "NOR" variant with the rest of the Linksys variants
* Added logging to indicate success and failure
* Provided a meaningful return value for scripting
* "Protected" the use of `mtd resetbc` in start-up scripts so that
failure does not end the boot sequence
* Moved Linksys-specific actions into common `/etc/init.d/bootcount`
For upgrade, these devices need to determine which partition to flash,
as well as set certain U-Boot envirnment variables to change the next
boot to the newly flashed version.
* Moved upgrade-related environment changes out of bootcount
* Combined multiple flashes of environment into single one
* Current-partition detection now handles absence of `boot_part`
Runtime-tested: Linksys EA8300
Signed-off-by: Jeff Kletsky <git-commits@allycomm.com>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
[checkpatch.pl fixes, traded split strings for 80+ chars per line]
This adds an option to set the recovery flag of newer TP-Link MediaTek
boards and remove it after a successful write.
To make use of this feature, add the '-t' option to mtd-write.
The '-t' option takes the mtd partition containing the recovery flag
(usually 'romfile') as an argument. Make sure this partition is not
flagged as read-only!
Example:
> mtd -t romfile write owrt.bin firmware
This command writes the recovery-flag before it begins writing the image
to the firmware partition. After the image-write has been successful,
the recovery flag is removed.
This way, the TP-Link web-recovery is automatically enabled on an
unsucessful flash (e.g. power loss).
This option is only available if the mtd package is compiled for the
ramips target.
Signed-off-by: David Bauer <mail@david-bauer.net>
This commit adds the object 'linksys_bootcount_fix.o' to the ipq40xx
target.
This is needed for the Linksys EA6350v3 device. Without this patch, the
device will switch-back between the current and the last flashed firmware
every 3 (three) reboots. With this patch, the device works as expected.
Signed-off-by: Ryan Pannell <ryan@osukl.com>
Signed-off-by: Oever González <notengobattery@gmail.com>
First of all lengths should be compared after checking all blocks for
being good/bad. It's because requested length may differ from a final
one if there were some bad blocks.
Secondly it makes sense to also compare crc32 since we already have a
new one calculated.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Reading MTD data with (p)read doesn't return any error when accessing
bad block. As the result, with current code, CRC32 covers "data" stored
in bad blocks.
That behavior doesn't match CFE's one (bootloader simply skips bad
blocks) and may result in:
1) Invalid CRC32
2) CFE refusing to boot firmware with a following error:
Boot program checksum is invalid
Fix that problem by checking every block before reading its content.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
The mtd tool is built with different configurations depending on the
target. For example, brcm47xx adds the fixtrx subcommand, without which
an image fails when booting the second time.
Mark the mtd package as nonshared to really fix FS#484.
Signed-off-by: Mirko Parthey <mirko.parthey@web.de>
When a partition offset is given, it is used in an lseek call, which
affects write, but not erase. Add it to the offset for erase calls as
well
Signed-off-by: Felix Fietkau <nbd@nbd.name>
D-Link DAP-1522 is a wireless bridge/access point with 4 LAN
ports and a dual-band wireless chipset.
Specifications:
- Ralink RT2880
- 32 MB of RAM
- 4 MB of Flash
- 4x 10/100/1000 Mbps Ethernet (RTL8366SR)
- 802.11abgn (RT2850)
Flash Instructions:
1. Download lede-ramips-rt288x-dap-1522-a1-squashfs-factory.bin
2. Open the web interface and upload the image
Signed-off-by: George Hopkins <george-hopkins@null.net>
src/linksys_bootcount.c misses to include stdint.h.
Apparently musl doesn't mind and includes this header by default,
but glibc does not and causes the build to fail.
Signed-off-by: Josua Mayer <josua.mayer97@gmail.com>
So far fixtrx was calculating checksum over amount of data matching
partition erase size. It was mostly a workaround of checksum problem
after changing anything in initial TRX content (e.g. formatting JFFS2).
Its main purpose was to make bootloader accept modified TRX. This didn't
provide much protection of flash data against corruption.
This new option lets caller request calculating checksum over a bigger
amount of data. It may be used e.g. to include whole kernel data for
checksum and hopefully make bootloader go info failsafe mode if
something goes wrong.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
We plan to adjust usage of the main buffer to allow reading custom
amount of data for CRC32. This means we need another buffer that will be
always block aligned.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
1) Put sanity checks in one place
2) Respect provided offset
3) Read only as much data as needed for MD5 calculation
Thanks to the last change this is a great speedup and memory saver. On
devices with NAND flash we were allocating & reading about 128 MiB while
something about 8 MiB is enough.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
This avoid long (and unneeded) process of reading all data in case of
running on MTD not containig Seama entity.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
On platforms supporting both: TRX and Seama calling "fixtrx" was
resulting in trying to fix Seama as well.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
The image_check currently fails when it cannot read all magic bytes in a
single chunk. But this can happen when the data are read from a pipe. This
currently breaks the openmesh.sh upgrade script with musl because it uses
dd with a blocksize of 1 to copy the image file to the mtd process.
The read can simply be repeated until enough bytes are read for the magic
byte check. It only stops when either an error was returned or 0 bytes were
read.
Signed-off-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
SVN-Revision: 48891
Seama header has MD5 similarly to TRX and its CRC32. We need to update
it after replacing anything in Seama entity content to make bootloader
accept it.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48688
Recently TRX checking code was changed to detect Seama format and don't
abort whole writing operation because of it. This isn't a good long-term
solution. It's a poor idea to teach every format handler recognizing all
possible formats. Instead it should be handled in a generic code which
should run check depending on the detected format.
This will also allow further improvements like fixing formats other than
TRX after replacing JFFS2.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48639
This is done with existing code from the WRT1900AC port.
It makes sure the "auto_recovery" bootloader option is set,
and resets the s_env boot counter after a successful boot.
This gives users without a serial console connection some
measure of safety.
Signed-off-by: Claudio Leite <leitec@staticky.com>
SVN-Revision: 47433
The u-boot boot counter was never reset after a successful boot,
which sometimes could make some variables become out of sync.
This patch adds support for the boot counter and enables
auto_recovery unconditionally.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
Signed-off-by: Rob Mosher <nyt-openwrt@countercultured.net>
Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 46690
trx.c in mtd would not include endian.h, so on systems that do not have
this header implicitly included from the other headers (like musl), both
__BYTE_ORDER and __BIG_ENDIAN would be undefined and thus 0, leading to
it always presuming a big-endian system. this would lead to issues when
running mtd fixtrx on little-endian systems, as it would never recognize
the TRX magic as result of the broken STORE32_LE() macro.
Signed-off-by: Shiz <hi@shiz.me>
Tested-by: Shiz <hi@shiz.me>
SVN-Revision: 45896
In mtd_verify(), the return value of md5sum() has been
interpreted as error if nonzero, while the function
returns number of processed bytes, which caused
mtd_verify() to always fail.
This patch fixes error checking to interpret only
negative values as errors.
Signed-off-by: Zefir Kurtisi <zefir.kurtisi@neratec.com>
SVN-Revision: 44605
Remove the need for the header file to be exported - we don't need most
of it anyway; all we care about are the offset of the rootfs length and
header crc fields.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 44557
this can be used on nand flashes and will skip bad blocks and run ecc on the
read data before dumping it.
Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 43500
mtd.c:544:7: warning: format ‘%x’ expects argument of type ‘unsigned int’, but argument 3 has type ‘ssize_t’ [-Wformat]
mtd.c:602:1: warning: label ‘done’ defined but not used [-Wunused-label]
Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 43383
