Update the zstd implementation to match Linux zstd 1.5.2 from commit
2aa14b1ab2.
This was motivated by running into decompression corruption issues when
trying to uncompress files compressed with newer versions of zstd. zstd
users also claim significantly improved decompression times with newer
zstd versions which is a side benefit.
Original zstd code was copied from Linux commit 2aa14b1ab2 which is a
custom-built implementation based on zstd 1.3.1. Linux switched to an
implementation that is a copy of the upstream zstd code in Linux commit
e0c1b49f5b, this results in a large code diff. However this should make
future updates easier along with other benefits[1].
This commit is a straight mirror of the Linux zstd code, except to:
- update a few #include that do not translate cleanly
- linux/swab.h -> asm/byteorder.h
- linux/limits.h -> linux/kernel.h
- linux/module.h -> linux/compat.h
- remove assert() from debug.h so it doesn't conflict with u-boot's
assert()
- strip out the compressor code as was done in the previous u-boot zstd
- update existing zstd users to the new Linux zstd API
- change the #define for MEM_STATIC to use INLINE_KEYWORD for codesize
- add a new KConfig option that sets zstd build options to minify code
based on zstd's ZSTD_LIB_MINIFY[2].
These changes were tested by booting a zstd 1.5.2 compressed kernel inside a
FIT. And the squashfs changes by loading a file from zstd compressed squashfs
with sqfsload. buildman was used to compile test other boards and check for
binary bloat, as follows:
> $ buildman -b zstd2 --boards dh_imx6,m53menlo,mvebu_espressobin-88f3720,sandbox,sandbox64,stm32mp15_dhcom_basic,stm32mp15_dhcor_basic,turris_mox,turris_omnia -sS
> Summary of 6 commits for 9 boards (8 threads, 1 job per thread)
> 01: Merge branch '2023-01-10-platform-updates'
> arm: w+ m53menlo dh_imx6
> 02: lib: zstd: update to latest Linux zstd 1.5.2
> aarch64: (for 2/2 boards) all -3186.0 rodata +920.0 text -4106.0
> arm: (for 5/5 boards) all +1254.4 rodata +940.0 text +314.4
> sandbox: (for 2/2 boards) all -4452.0 data -16.0 rodata +640.0 text -5076.0
[1] e0c1b49f5b
[2] f302ad8811/lib/libzstd.mk (L31)
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
[trini: Set ret to -EINVAL for the error of "failed to detect
compressed" to fix warning, drop ZSTD_SRCSIZEHINT_MAX for non-Linux host
tool builds]
Signed-off-by: Tom Rini <trini@konsulko.com>
This makes it possible to run the patman test suite simply by invoking
'pytest' from the patman sub-directory:
$ cd tools/patman
$ pytest
Otherwise, the top level .checkpatch.conf would be ignored and
multiple test_checkpatch.py tests would fail.
Signed-off-by: Maxim Cournoyer <maxim.cournoyer@savoirfairelinux.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This reverts commit 648d8186dd, because
it broke usage of patman on Linux, whose check script doesn't know
about '--strict' or '--u-boot'.
Reported-by: Sjoerd Simons <sjoerd@collabora.com>
Signed-off-by: Maxim Cournoyer <maxim.cournoyer@savoirfairelinux.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Sjoerd Simons <sjoerd@collabora.com>
In some cases it is useful to position an entry over the top of a symbol
in an ELF file. For example, if the symbol holds a version string then it
allows the string to be accessed from the fdtmap.
Add support for this.
Suggested-by: Pali Rohár <pali@kernel.org>
Suggested-by: Keith Short <keithshort@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
The current support for updating variables in a binary is hard-coded to
work with U-Boot:
- It assumes the image starts at __image_copy_start
- It uses the existing U-Boot-specific entry types
It is useful for other projects to use these feature.
Add properties to enable writing symbols for any blob, a way of specifying
the base symbol and a way of providing the ELF filename to allow symbol
lookup to take place.
With this it is possible to update a Zephyr image, such as zephyr.bin
after it has been built.
Signed-off-by: Simon Glass <sjg@chromium.org>
In some cases it is useful to have an entry overlap with another in a
section, either to update the contents within a blob, or to add an entry
to the fdtmap that covers only part of the blob.
Add support for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
It is sometimes useful to define an entry which does not have its own
contents but does appear in the image. The contents are set by the section
which contains it, even though it appears as an entry in the fdtmap.
Add support for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Instead of the bash script, use binman to generate the FIT for arm64.
For 32-bit boards, use binman for all images, dropping the intermediate
files.
With this change, only Zynq is now using SPL_FIT_GENERATOR so update the
Kconfig rule accordingly.
Clean up the Makefile to the extent possible. Unfortunately, two boards
do not use SPL_FRAMEWORK so don't enable the u-boot.img rule:
evb-rk3036
kylin-rk3036
So a small remnant remains.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add the required binman images to replace the Makefile rules which are
currently used. This includes subsuming:
- tpl/u-boot-tpl-rockchip.bin if TPL is enabled
- idbloader.img if either or both of SPL and TPL are enabled
- u-boot.itb if SPL_FIT is enabled
- u-boot-rockchip.bin if SPL is used, either using u-boot.itb when
SPL_FIT is enabled or u-boot.img when it isn't
Note that the intermediate files are dropped with binman, since it
producing everything in one pass. This means that
tpl/u-boot-tpl-rockchip.bin is not created, for example.
Note that for some 32-bit rk3288 boards, rockchip-optee.dtsi is included.
Signed-off-by: Simon Glass <sjg@chromium.org>
Enable multiple-images so we can generate more than one image. Also
add a comment for the end of the #if block.
Signed-off-by: Simon Glass <sjg@chromium.org>
This boards uses SPL_FIT so does not need to support loading a raw image.
Drop it to avoid binman trying to insert a symbol which has no value.
Signed-off-by: Simon Glass <sjg@chromium.org>
Some blobs are actually not necessary for the board to work correctly. Add
a property to allow this to be indicated. Missing optional blobs do not
cause a build failure.
Signed-off-by: Simon Glass <sjg@chromium.org>
OP-TEE has a format with a binary header that can be used instead of the
ELF file. With newer versions of OP-TEE this may be required on some
platforms.
Add support for this in binman. First, add a method to obtain the ELF
sections from an entry, then use that in the FIT support. We then end up
with the ability to support both types of OP-TEE files, depending on which
one is passed in with the entry argument (TEE=xxx in the U-Boot build).
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a function which checks whether data is in ELF format or not. This
will be used by binman to check this for entries.
Signed-off-by: Simon Glass <sjg@chromium.org>
Support entries which can be optional depending on their contents. This
allows special entry types which appear in the image only when needed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Several etypes have this reference in their documentation. Now that we are
using rST, link to the section directly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
At present only the image (which is a section) has a filename. Move this
implementation to the entry_Section class so that any section can have a
filename. With this, the section data is written to a file.
This allows parts of an image to be written, along with the entire image.
Make a note that this can be used to include the contents of a section in
one image in another (later) image.
Signed-off-by: Simon Glass <sjg@chromium.org>
This breaks chromebook_coral and it is also not how things should work. If
a board needs to bind GPIOs as part of a pinctrl driver this can be done
during the bind step, if needed.
We cannot probe pinctrl devices when binding as a rule, since it cannot be
supported on some platforms.
The bind and probe steps are separate in U-Boot and they should remain
separate.
This reverts commit f9ec791b5e.
Signed-off-by: Simon Glass <sjg@chromium.org>
The fdt_path_offset() function is slow since it must scan the tree.
This substantial overhead now applies to all boards.
The original code may not be ideal but it is fit for purpose and is only
needed on a few boards.
Reverting this reduces time to set up driver model by about 30ms.
Before revert:
Accumulated time:
47,170 dm_r
53,237 dm_spl
572,986 dm_f
Accumulated time:
44,598 dm_r
50,347 dm_spl
549,133 dm_f
This reverts commit 26f981f295.
Signed-off-by: Simon Glass <sjg@chromium.org>
This board is useful for benchmarking overall U-Boot performance. Enable
the bootstage feature so we get a report.
Since this returns to the boot rom before finishing executing
board_init_r() in SPL, add a few bootstage calls so that we can collect
timing from TPL.
For the stash region, use a portion of SRAM, 64KB below the stack top.
This allows the TPL image to be up to nearly 120KB (it is typically about
64KB). SPL normally runs from SDRAM at 0, so can use the same stash
region.
Signed-off-by: Simon Glass <sjg@chromium.org>
Some platforms cannot honour this and don't need trace before relocation.
Use 'imply' instead, so boards can disable this.
Signed-off-by: Simon Glass <sjg@chromium.org>
This now includes flags and the layout has changed slightly in recent
versions of Linux. Update the generator accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
The flags in this tool don't match the comments or help. Also the variable
names are quite confusing. Update them for consistency.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add messages to make it clearer which part of the FIT creation is failing.
This can happen when an invalid 'algo' property is provided in the .its
file.
Signed-off-by: Simon Glass <sjg@chromium.org>
This can be written before relocation. Move it to the data section, since
accessing BSS before relocation is not permitted.
Signed-off-by: Simon Glass <sjg@chromium.org>
If the FDT overlaps with the data region of the image, or with the stack,
it can become corrupted before relocation. Add a check for this, behind a
debug flag, as it can be very confusing and time-consuming to debug.
Signed-off-by: Simon Glass <sjg@chromium.org>
This function uses gd->fdt_blob a lot and cannot be used to check any
other device tree. Use a parameter instead.
Signed-off-by: Simon Glass <sjg@chromium.org>
This function is a bit messy with several #ifdefs. Convert them to use C
for the conditions.
Rewrite the function comment since most of it is stale.
Signed-off-by: Simon Glass <sjg@chromium.org>
The current approach is a bit of a hack and only works for the tpm
subsystem. Add a Kconfig so that crc8 can be enabled in SPL for other
purposes.
Signed-off-by: Simon Glass <sjg@chromium.org>
Use the notrace attribute so that timer functions can be used when
tracing. This is required to avoid infinite loops when recording a trace.
Signed-off-by: Simon Glass <sjg@chromium.org>
Current default size of 0x100000 is not capable of getting the FIT
buffer during boot when transitioning to using binman generated boot
images for certain K3 devices, so increase it to 0x400000. Since A72 SPL
is coming after relocation to DDR this should not be an issue for any K3
device, so make it default for all.
Signed-off-by: Neha Malcom Francis <n-francis@ti.com>
Acked-by: Andrew Davis <afd@ti.com>
Add sandbox test for the gpt partition type command, the test uses the
persistent data test_gpt_disk_image.bin to check that the first
partition type GUID that identifies the type of the partition has the
"Linux filesystem data" type ( 0FC63DAF-8483-4772-8E79-3D69D8477DE4 ).
Signed-off-by: Enric Balletbo i Serra <eballetbo@redhat.com>
This implements the following command:
part type mmc 0:1
-> print partition type UUID
part type mmc 0:1 uuid
-> set environment variable to partition type UUID
"part type" can be useful when writing a bootcmd which searches for a
specific partition type to enable automatic discovery of partitions and
their intended usage or mount point.
Signed-off-by: Enric Balletbo i Serra <eballetbo@redhat.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
[trini: Fix when CONFIG_PARTITION_TYPE_GUID is disabled and have the
command check for "types" before "type"]
Signed-off-by: Tom Rini <trini@konsulko.com>
Change the string 'a optionnal' to 'an optional'.
Signed-off-by: Enric Balletbo i Serra <eballetbo@redhat.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
