| Age | Commit message (Collapse) | Author |
|---|
|
Recent cleanups made some variables redundant. Both: "blocklen" and
"erase_len" ended up being used as temporary variables used locally in
some short code paths.
Signed-off-by: Rafał Miłecki <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
Calculate "erasesize" in the same place where other offsets and lengths
/ sizes are calculated. It makes code more consistent and will allow
further cleanups.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
In both code paths "blockstart" is calculated the same way. Unify it.
Signed-off-by: Rafał Miłecki <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
We already have DIV_ROUND_UP() to simplify code so add ROUND_UP() as
well. This makes code in flash_write_buf() easier to follow.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
When dealing with env data not aligned to flash blocks flash_write_buf()
has to use an extra buffer. It reads existing flash content to it,
modifies required part and writes it back.
While reading and writing a size stored in "write_total" is used. It's
what should be used when allocating the buffer too.
In some cases allocating memory of "erase_len" size could result in
allocating too big buffer. That wouldn't break anything but it was
making code less intuitive.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Use "erasesize" instead of "blocklen" in flash_write_buf()'s loop. This
change touches code executed for bad NAND blocks so it doesn't affect
any behaviour (for NAND flashes "blocklen" and "erasesize" are equal).
This just makes code a bit more consistent as "erasesize" is what is
used all around inside the writing loop.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
In v2024.10, "make envtools" is broken for at least these defconfigs:
am335x_evm_defconfig
rpi_3_defconfig
rpi_4_defconfig
mx7dsabresd_defconfig
wandboard_defconfig
imx8mp_evk_defconfig
The only defconfig we use for which it is not broken is
stm32mp13_defconfig. They all work just fine in v2024.07.
The symptoms are slightly different, but all related to the fact that
some transitively included header uses IS_ENABLED or CONFIG_IS_ENABLED
without linux/kconfig.h having already been included.
A simple git bisect doesn't produce anything sensible, it ends up at
3a9f642ca94 (crypto: nuvoton: npcm_sha: Support SHA 384/512) which
clearly has nothing to do with this. But digging deeper, one
eventually finds 0f92fa45603 ("env: Remove <common.h> and add needed
includes").
So at first I tried adding "#include <linux/kconfig.h>" in
include/env_default.h and include/env_flags.h. That fixes it for some,
but not all, of the above. For example rpi_3_defconfig still fails,
then in log.h complaining about BIT() and u8 not being defined. At
least BIT() is should have gotten from bitops.h, except that that's
behind ifdef __KERNEL__, so not set for the envtools build.
It turns out that the envtools source code in fw_env_private.h already
has some hackery to deal with all this, in the form of the __ASSEMBLY__
games it plays before including config.h. It seems that if we just
make sure to do that include early enough, so that config.h is indeed
parsed with that __ASSEMBLY__ hackery in place, everything builds
fine.
Fixes: 0f92fa45603 ("env: Remove <common.h> and add needed includes")
Signed-off-by: Rasmus Villemoes <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
|
|
As already done for NOR chips, if device ESIZE and ENVSECTORS static
configurations are both zero, then autodetect them at runtime.
Cc: Joe Hershberger <[email protected]>
cc: Stefan Agner <[email protected]>
cc: Rasmus Villemoes <[email protected]>
Signed-off-by: Anthony Loiseau <[email protected]>
|
|
It's totally valid for read() to provide less bytes than requested
maximum. It may happen if there is no more data available yet or source
pushes data in small chunks.
This actually happens when trying to read env data from NVMEM device.
Kernel may provide NVMEM content in page size parts (like 4096 B).
This fixes warnings like:
Warning on /sys/bus/nvmem/devices/u-boot-env0/nvmem: Attempted to read 16384 bytes but got 4096
Warning on /sys/bus/nvmem/devices/u-boot-env0/nvmem: Attempted to read 12288 bytes but got 4096
Warning on /sys/bus/nvmem/devices/u-boot-env0/nvmem: Attempted to read 8192 bytes but got 4096
Since the main loop in flash_read_buf() is used to read blocks this
patch adds a new nested one.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Rename "addr0" and "addr1" to "buf0" and "buf1" accordingly. Name "addr"
suggests that variable contains a numeric value being some kind of
address. Name "buf" is de facto a standard name for pointer to allocated
memory for reading data to.
While at it drop redundant checks for NULL before calling free().
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Call to fread() was changed to check for return value. The problem is it
can't be checked for returning 1 (as it is) to determine success.
We call fread() with buffer size as "size" argument. Reading any
"compatible" value shorter than buffer size will result in returning 0
even on success.
Modify code to use fstat() to determine expected read length.
This fixes regression that broke using fw_env with NVMEM devices.
Fixes: c059a22b7776 ("tools: env: fw_env: Fix unused-result warning")
Cc: Jaehoon Chung <[email protected]>
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Fix unused-result warning about fread.
tools/env/fw_env.c: In function ‘find_nvmem_device’:
tools/env/fw_env.c:1751:3: warning: ignoring return value of ‘fread’, declared with attribute warn_unused_result [-Wunused-result]
1751 | fread(buf, sizeof(buf), 1, fp);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Jaehoon Chung <[email protected]>
|
|
The function calls opendir() but missing the corresponding
closedir() before exit the function.
Add missing closedir() to fix it.
Signed-off-by: Miaoqian Lin <[email protected]>
|
|
A new DT binding for describing environment data block has been added in
Linux's commit 5db1c2dbc04c ("dt-bindings: nvmem: add U-Boot environment
variables binding"). Once we get a proper Linux NVMEM driver it'll be
possible to use Linux's binary interface for user-space as documented
in the:
https://www.kernel.org/doc/html/latest/driver-api/nvmem.html
This commits makes fw_env fallback to looking for a compatible NVMEM
device in case config file isn't present. In a long term this may make
config files redundant and avoid code (info) duplication.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Environment variables can be stored in two formats:
1. Single entry with header containing CRC32
2. Two entries with extra flags field in each entry header
For that reason fw_env_open() has two main code paths and there are
pointers for CRC32/flags/data.
Previous implementation was a bit hard to follow:
1. It was checking for used format twice (in reversed order each time)
2. It was setting "environment" global struct fields to some temporary
values that required extra comments explaining it
This change simplifies that code:
1. It introduces two clear code paths
2. It sets "environment" global struct fields values only once it really
knows them
To be fair there are *two* crc32() calls now and an extra pointer
variable but that should be cheap enough and worth it.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
It's usually easier to understand code & follow it if all arguments are
passed explicitly. Many coding styles also discourage using global
variables.
Behaviour of flash_io() was a bit unintuitive as it was writing to a
buffer referenced in a global struct. That required developers to
remember how it works and sometimes required hacking "environment"
global struct variable to read data into a proper buffer.
Signed-off-by: Rafał Miłecki <[email protected]>
|
|
Commit "fw_setenv: lock the flash only if it was locked before"
checks for Locked status with uninitialized erase data.
Address by moving the test for MEMISLOCKED.
Fixes: 8a726b852502 ("fw_setenv: lock the flash only if it was locked before")
Signed-off-by: Joakim Tjernlund <[email protected]>
|
|
On some cases, the actual number of bytes read can be shorter
than what was requested. This can be handled gracefully by
taking this difference into account instead of exiting.
Signed-off-by: Thibault Ferrante <[email protected]>
|
|
The UBI_IOCVOLUP ioctl can fail if exclusive access to the volume isn't
obtained. If this happens, the flush operation doesn't return error,
leaving the caller without knowledge of missing flush.
Fix this by forwarding the error (-1) from ubi_update_start().
Fixes: 34255b92e6e ("tools: env: Add support for direct read/write UBI volumes")
Signed-off-by: Martin Hundebøll <[email protected]>
|
|
Set this variable to 0 to avoid a warning about an unused variable. This
happens on gcc 7.5.0 for me.
Signed-off-by: Simon Glass <[email protected]>
|
|
The printf() string produces a warning about %d not matching size_t. Fix
it and put the format string on one line to avoid a checkpatch warning.
Signed-off-by: Simon Glass <[email protected]>
|
|
With current implementation of fw_setenv, it is always locks u-boot-env
region if lock interface is implemented for such mtd device. You can
not control lock of this region with fw_setenv, there is no option for
it in config or in application itself. Because of this situation may
happen problems like in this thread on xilinx forum:
https://forums.xilinx.com/t5/Embedded-Linux/Flash-be-locked-after-use-fw-setenv-from-user-space
/td-p/1027851
A short summary of that link is: some person has issue with some spi
chip which has lock interface but doesn't locks properly which leads to
lock of whole flash memory on lock of u-boot-env region. As resulted
solution hack was added into spi-nor.c driver for this chip with lock
disablement.
Instead fix this problem by adding logic to fw_setenv only lock the
flash if it was already locked when we attempted to use it.
Signed-off-by: Ivan Mikhaylov <[email protected]>
|
|
When using CONFIG_ENV_IS_IN_FAT and the config-file specifies a size
larger than what U-Boot wrote into the env-file, a confusing error
message is shown:
$ fw_printenv
Read error on /boot/uboot.env: Success
Fix this by showing a different error message when read returns too
little data.
Signed-off-by: Harald Seiler <[email protected]>
|
|
We have a board with several revisions. The older ones use a nor flash
with 64k erase size, while the newer have a flash with 4k sectors. The
environment size is 8k.
Currently, we have to put a column containing 0x10000 (64k) in
fw_env.config in order for it to work on the older boards. But that
ends up wasting quite a lot of time on the newer boards that could
just erase the 8k occupied by the environment - strace says the 64k
erase takes 0.405 seconds. With this patch, as expected, that's about
an 8-fold better, at 0.043 seconds.
Having different fw_env.config files for the different revisions is
highly impractical, and the correct information is already available
right at our fingertips. So use the erasesize returned by the
MEMGETINFO ioctl when the fourth and fifth columns (sector size and
#sectors, respectively) are absent or contain 0, a case where the
logic previously used to use the environment size as erase size (and
consequently computed ENVSECTORS(dev) as 1).
As I'm only testing this on a NOR flash, I'm only changing the logic
for that case, though I think it should be possible for the other
types as well.
Signed-off-by: Rasmus Villemoes <[email protected]>
|
|
In the case where one deletes an already-non-existing variable, or sets
a variable to the value it already has, there is no point in writing the
environment back, thus reducing wear on the underlying storage
device.
In the case of redundant environments, if the two environments
differ (e.g. because one is corrupt), make sure that any call of
fw_setenv causes the two to become synchronized, even if the fw_setenv
call does not change anything in the good copy.
Signed-off-by: Rasmus Villemoes <[email protected]>
Acked-by: Joe Hershberger <[email protected]>
|
|
Drop inclusion of crc.h in common.h and use the correct header directly
instead.
With this we can drop the conflicting definition in fw_env.h and rely on
the crc.h header, which is already included.
Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
The following error appears:
tools/env/fw_env.c:1149:25: error: lvalue required as unary ‘&’ operand
rc = write(fd, &ENV_REDUND_OBSOLETE, sizeof(ENV_REDUND_OBSOLETE));
Fixes: d3716dd ("env: Rename the redundancy flags")
Signed-off-by: Pierre-Jean Texier <[email protected]>
Tested-by: Joris Offouga <[email protected]>
Tested-by: Heiko Schocher <[email protected]>
Suggested-by: Heiko Schocher <[email protected]>
Acked-by: Joe Hershberger <[email protected]>
|
|
Since commit d3716dd ("env: Rename the redundancy flags"), the
definitions of ENV_REDUND_OBSOLETE & ENV_REDUND_ACTIVE was moved
to env.h.
Fixes:
tools/env/fw_env.c:122:22: error: ‘ENV_REDUND_ACTIVE’ redeclared as different kind of symbol
static unsigned char ENV_REDUND_ACTIVE = 1;
^~~~~~~~~~~~~~~~~
In file included from tools/env/fw_env.c:13:
include/env.h:63:2: note: previous definition of ‘ENV_REDUND_ACTIVE’ was here
ENV_REDUND_ACTIVE = 1,
^~~~~~~~~~~~~~~~~
tools/env/fw_env.c:127:22: error: ‘ENV_REDUND_OBSOLETE’ redeclared as different kind of symbol
static unsigned char ENV_REDUND_OBSOLETE;
^~~~~~~~~~~~~~~~~~~
In file included from tools/env/fw_env.c:13:
include/env.h:62:2: note: previous definition of ‘ENV_REDUND_OBSOLETE’ was here
ENV_REDUND_OBSOLETE = 0,
Signed-off-by: Pierre-Jean Texier <[email protected]>
Tested-by: Joris Offouga <[email protected]>
Tested-by: Heiko Schocher <[email protected]>
Acked-by: Joe Hershberger <[email protected]>
|
|
Add an ENV prefix to these two flags so that it is clear what they relate
to. Also move them to env.h since they are part of the public API. Use an
enum rather than a #define to tie them together.
Signed-off-by: Simon Glass <[email protected]>
|
|
Move env_set() over to the new header file.
Acked-by: Joe Hershberger <[email protected]>
Signed-off-by: Simon Glass <[email protected]>
|
|
If flash_write fails, whilst we propagate this up to our caller, we need
to avoid swapping in the new file (if we're on a filesystem) in this
case.
Fixes: dbc34323796b ("tools: env: Implement atomic replace for filesystem")
Signed-off-by: Alex Kiernan <[email protected]>
|
|
Signed-off-by: Philip Molloy <[email protected]>
|
|
addr0 is being freed twice, leading to segmentation fault
Signed-off-by: Björn Stenberg <[email protected]>
|
|
As pointed out by Wolfgang Denk, the problem with this fix is that while
interactive users will see that we have found one part of the
environment failed and are using the other, progmatic use will not see
this and can lead to problems.
Signed-off-by: Tom Rini <[email protected]>
|
|
When using a redundant environment a read error should simply mean to
not use that copy instead of giving up completely. The other copy may
be just fine.
Signed-off-by: Joe Hershberger <[email protected]>
Signed-off-by: Ioan-Adrian Ratiu <[email protected]>
|
|
's/environemnt/environment/' and
's/Environemnt/Environment/'
Signed-off-by: Shyam Saini <[email protected]>
|
|
When reading the config file, or a script file, use getline rather than
fgets so line lengths aren't limited by the size of a compiled in buffer
(128 characters for config, 1024 for scripts).
Rename 'dump' to 'line' so it's clear we're working with a line of text.
Signed-off-by: Alex Kiernan <[email protected]>
|
|
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <[email protected]>
|
|
If the U-Boot environment is stored in a regular file and redundant
operation isn't set, then write to a temporary file and perform an
atomic rename.
Signed-off-by: Alex Kiernan <[email protected]>
|
|
Extract write path of flash_io() into a separate function. This patch
should be a functional no-op.
Signed-off-by: Alex Kiernan <[email protected]>
Reviewed-by: Stefano Babic <[email protected]>
|
|
Replace HaveRedundEnv with have_redund_env to fix style violation.
Signed-off-by: Alex Kiernan <[email protected]>
|
|
Pass tools/env/fw_env.c through indent to correct style violations. This
commit consists of only one non-whitespace change:
tools/env/fw_env.c:549: error: do not use assignment in if condition
Signed-off-by: Alex Kiernan <[email protected]>
|
|
Treat the first equal sign as a key/value separation too. This makes
the script files compatible with mkenvimage input file format. It
won't support variables with equal signs anymore, but this seems not
really like a loss.
Signed-off-by: Stefan Agner <[email protected]>
|
|
The single argument `--noheader' is expecting isn't taken from getopt
parsing, but instead from the remaining argv arguments.
Signed-off-by: Alex Kiernan <[email protected]>
|
|
Remove unused ret from fw_env_flush.
Signed-off-by: Jelle van der Waa <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Up to now we were able to read/write environment data from/to UBI
volumes only indirectly by gluebi driver. This driver creates NAND MTD
on top of UBI volumes, which is quite a workaroung for this use case.
Add support for direct read/write UBI volumes in order to not use
obsolete gluebi driver.
Forward-ported from this patch:
http://patchwork.ozlabs.org/patch/619305/
Original patch:
Signed-off-by: Marcin Niestroj <[email protected]>
Forward port:
Signed-off-by: S. Lockwood-Childs <[email protected]>
|
|
This support has been deprecated since v2017.09 due to security issues.
We now remove this support.
Signed-off-by: Tom Rini <[email protected]>
|
|
With the move of environment code from common/ to env/ a number of
changes needed to be made to various make targets. We missed updating
some of the files required for out of tree builds of the tools. Correct
the 'environ' target to know that we need to work under tools/env/ still
(not tools/environ/) and then update the wrappers in env_attr.c and
env_flags.c to point to the new correct file.
Reported-by: Peter Robinson <[email protected]>
Tested-by: Peter Robinson <[email protected]>
Signed-off-by: Tom Rini <[email protected]>
|
|
support fsync
According to fsync specification [1] some special files (e.g., a pipe, FIFO,
or socket) don't support synchronization and return either EROFS or EINVAL.
On the linux side the sys_fsync -> do_fsync() checks if the requested file
has f_op->fsync defined. If not it returns EINVAL [2].
This commit prevents writing error messages for files (devices), which
do not support fsync().
[1] - http://man7.org/linux/man-pages/man2/fsync.2.html
[2] - http://elixir.free-electrons.com/linux/v4.13-rc6/source/fs/sync.c#L183
Signed-off-by: Lukasz Majewski <[email protected]>
Acked-by: Michael Heimpold <[email protected]>
|
|
We are now using an env_ prefix for environment functions. Rename setenv()
for consistency. Also add function comments in common.h.
Suggested-by: Wolfgang Denk <[email protected]>
Signed-off-by: Simon Glass <[email protected]>
|
