Flushing kernel image after decompression was taking 113 milliseconds
with U-boot 2022.10. With U-boot 2023.01 and 2023.04, flushing
the same amount of memory takes approx 1.5 seconds. With
U-boot 2023.07-rc6, it takes 6.5 seconds.
powerpc flush_cache() function used to call WATCHDOG_RESET() after
flushing every cacheline. At that time WATCHDOG_RESET() was light
so the operation was almost seamless.
But commit 29caf9305b ("cyclic: Use schedule() instead of
WATCHDOG_RESET()") replaced WATCHDOG_RESET() by schedule() and that
started to hurt with U-boot 2022.10.
And in U-boot 2023.07-rc6 that's even worse after
commit 26e8ebcd7c ("watchdog: mpc8xxx: Make it generic").
In the meantime commit 729c1fe656 ("powerpc: introduce
CONFIG_CACHE_FLUSH_WATCHDOG_THRESHOLD") gives us the opportinity to
only call schedule() every given chunk of data instead of every
cacheline. As explained in that commit there is no point in pinging
the watchdog after every cacheline flush, so lets define a sensible
default chunk size of 4k which matches to size of a page on most
powerpc platforms.
With that new default threshold, the culprit flushing performed after
kernel image decompression now takes 85 milliseconds on a powerpc 8xx.
Fixes: 29caf9305b ("cyclic: Use schedule() instead of WATCHDOG_RESET()")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Uncompressing a 1.7Mbytes FIT image on U-boot 2023.04 takes
approx 7s on a powerpc 8xx.
The same on U-boot 2023.07-rc6 takes approx 28s unless watchdog
is disabled.
During that decompression, LzmaDec_DecodeReal() calls schedule
1.6 million times, that is every 4µs in average.
In the past it used to be a call to WATCHDOG_RESET() which was
just calling hw_watchdog_reset().
But the combination of commit 29caf9305b ("cyclic: Use schedule()
instead of WATCHDOG_RESET()") and commit 26e8ebcd7c ("watchdog:
mpc8xxx: Make it generic") results in an heavier processing.
However, there is absolutely no point in calling schedule() that
often.
By moving and keeping only one call to schedule() in the main
loop the number of calls is reduced to 1.2 million which is still
too much. So add logic to only call schedule every 1024 times.
That leads to a call to schedule approx every 6ms which is still
far enough to entertain the watchdog which has a 1s timeout on
powerpc 8xx.
powerpc 8xx being one of the slowest targets we have today in
U-boot, and most other watchdogs having a timeout of one minutes
instead of one second like the 8xx, this fix should not have
negative impact on other targets.
Fixes: 29caf9305b ("cyclic: Use schedule() instead of WATCHDOG_RESET()")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Simon Glass <sjg@chromium.org>
Instead of using the hard-coded value of 0x1f, use 'TASK_TAG'
macro instead to construct the ucd_req_ptr->header.dword_0
This is in sync with what the Linux UFS driver does, i.e.
set the byte0 equal to TASK_TAG (see [1]).
Setting it to a fixed value of 0x1f is wrong as we define
TASK_TAG as 0 inside u-boot ufs framework. So, instead we
should use the macro value directly.
[1]. https://github.com/torvalds/linux/blob/master/drivers/ufs/core/ufshcd.c#L2705
Signed-off-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
This driver considers that a node with an empty label is the top.
But the led class has changed, if a label is not provided for a led,
the label is filed with the node name. So we update this driver
to use a wrapper to manage the top led node.
Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
In case of OF_SEPARATE (!OF_EMBED), the devicetree blob is placed
after _end, and fdt_find_separate() always returns _end. There is
a .bss section after _end and the section is cleared before relocation.
When GD_FLG_SKIP_RELOC is set, relocation is skipped, so the blob is
still in .bss section, but will be cleared. As a result, the devicetree
become invalid.
To avoid this issue, should relocate it regardless of GD_FLG_SKIP_RELOC
in reloc_fdt().
Signed-off-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This fixes CVE-2022-37434 [1] and bases on 2 commits from Mark
Adler's zlib master repo - the original fix of CVE bug [2] and
the fix for the fix [3].
[1]
https://github.com/advisories/GHSA-cfmr-vrgj-vqwv
[2]
eff308af42
[3]
1eb7682f84
Fixes: e89516f031 ("zlib: split up to match original source tree")
Signed-off-by: Oleksandr Suvorov <oleksandr.suvorov@foundries.io>
The different CPSW sub-system Ethernet ports have different PHY mode
control registers. In order to allow the modes to get configured
independently only the register for the port in question must be
accessed, otherwise we would just be re-configuring the mode for port 1,
while leaving all others at their power-on defaults. Fix this issue by
adding a port-number based offset to the mode control base register
address based on the fact that the control registers for the different
ports are spaced exactly 0x4 bytes apart.
Fixes: 9d0dca1199 ("net: ethernet: ti: Introduce am654 gigabit eth switch subsystem driver")
Signed-off-by: Andreas Dannenberg <dannenberg@ti.com>
Reviewed-by: Siddharth Vadapalli <s-vadapalli@ti.com>
While NFS is widely used in data centres, and private
networks it's quite a nuanced usecase for device firmware.
A lot of devices already disable it.
Various network protocols should really be opt in, not opt
out, because they add extra size and are potential attack
vectors from a security PoV. In the NFS case it doesn't
really make sense for a lot of devices like tables, SBCs etc.
It's also something we don't really want for SystemReady-IR
due to security concerns.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
The new opt-out setting, CONFIG_ENV_MMC_PARTITION, statically sets
the MMC environment partition name. Prior to this patch, the only way
to declare this partition name was by creating a
'u-boot,mmc-env-partition' parameter in the device-tree's /config node.
This setting provides additional flexibility, particularly in cases
where accessing the device-tree is not straightforward (e.g. QEMU).
If undeclared, the device-tree's setting will be used.
Signed-off-by: Emmanuel Di Fede <emmanuel.difede@cysec.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Patrick Delaunay <patrick.delaunay@foss.st.com>
scmi_to_linux_errno() is set to return an appropriate errno
which corresponds to a given SCMI status code.
But the current implementation always returns the same value.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
memalign() returns a pointer which is to be freed by free(). To call
unmap_sysmem() is incorrect, furthermore it was called in a wrong scope.
Also add a check for allocation error.
Fixes: d7af2baa49 ("crypto/fsl: Fix HW accelerated hash commands")
Cc: Breno Lima <breno.lima@nxp.com>
Signed-off-by: Sergei Antonov <saproj@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Without CONFIG_CLK, the build fails with the following message:
LD u-boot
aarch64-none-linux-gnu-ld.bfd: drivers/firmware/scmi/scmi_agent-uclass.o: \
in function `scmi_bind_protocols':
.../drivers/firmware/scmi/scmi_agent-uclass.c:79: undefined reference to \
`_u_boot_list_2_driver_2_scmi_clock'
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Reviewed-by: Sean Anderson <seanga2@gmail.com>
use new memory layout and change uboot load address.
open tpm, tee and more config feature
No need to reserve top memory because the reserved space
is moved to the bottom area of memory.
Signed-off-by: Jim Liu <JJLIU0@nuvoton.com>
To quote the author:
This series provides a means to edit board configuration in U-Boot in a
graphical manner. It supports multiple menu items and allows moving
between them and selecting items. The configuration is defined in a data
format so that code is not needed in most cases. This allows the board
configuration to be provided in the devicetree.
This is still at an early stage, since it only supports menus. Numeric
values are not supported. Most importantly it does not yet support
loading or saving the configuration selected by the user.
To try it out you can use something like:
./tools/expo.py -e test/boot/files/expo_layout.dts \
-l test/boot/files/expo_layout.dts -o cedit.dtb
./u-boot -Tl -c "cedit load hostfs - cedit.dtb; cedit run"
Use the arrow keys to move between menus, enter to open a menu, escape
to exit.
Various minor fixes and improvements are provided in this series:
- Update STB TrueType library to latest
- Support clearing part of the video display
- Support multiple livetrees loaded at runtime
- Support loading and allocating a file
- Support proper measuring of text in expo
- Support simple themes for expo
This was brought in in 2016 and a number of changes have been made since
then. There does not seem to be much change in functionality, but it is
a good idea to update from time to time.
Bring in the latest version:
5736b15 ("re-add perlin noise again")
Add a few necessary functions, with dummies in some cases. Update the tests
as there are subtle changes in rendering, perhaps not for the better.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a new 'cedit' command which allows editing configuration using an
expo. The configuration items appear as menus on the display.
This is extremely basic, only supporting menus and not providing any way
to load or save the configuration.
Signed-off-by: Simon Glass <sjg@chromium.org>
The only way to create an expo at present is by calling the functions to
create each object. It is useful to have more data-driven approach, where
the objects can be specified in a suitable file format and created from
that. This makes testing easier as well.
Add support for describing an expo in a devicetree node. This allows more
complex tests to be set up, as well as providing an easier format for
users. It also provides a better basis for the upcoming configuration
editor.
Signed-off-by: Simon Glass <sjg@chromium.org>
It looks better if menus have a bit of an inset, rather than be drawn hard
up against the background. Also, menu items look better if they have a bit
of spacing between them.
Add theme options for these and implement the required changes.
Signed-off-by: Simon Glass <sjg@chromium.org>
In 'popup' mode, the expo allows moving around the objects in a scene.
When 'enter' is pressed on a menu, it opens and the user can move around
the items in the menu.
Implement this using keypress handles and actions.
Signed-off-by: Simon Glass <sjg@chromium.org>
When a menu is opened, it must be displayed over all other objects in
the scene, so that all its items are visible.
Handle this by drawing the menu object a second time, after all other
objects have been drawn. Draw all of the objects which are dependent
on the menu object.
Signed-off-by: Simon Glass <sjg@chromium.org>
When a popup menu is closed it shows only the selected item. When it is
open it shows a background and all items, with a highlight that can be
moved between the items.
Add the drawing logic for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present we only support a single menu, so all that can be pointed to
is the current menu item. Rename this action so that we can also add
an action for pointing to an object. This will allow cycling through
the objects in a scene.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present only a single menu is supported. All items are shown and a
pointer object points to the current item.
Add support for multiple menus, one of which is highlighted, indicated
by the highlight_id property in the scene.
The highlighted menu item has a SCENEOF_POINT flag, indicating that it
is currently pointed to.
The popup menu is normally closed, in which case it shows only the
current menu item. When it is opened, it shows all items, allowing the
user to select one.
Rather than requiring the menu item to have a description, require it to
have a label. Use the label (only) for the popup menu.
With this, most of the drawing and layout logic is complete.
Signed-off-by: Simon Glass <sjg@chromium.org>
The current implementation supports a 'pointer' object which points to
the currently highlighted menu item. We want to support highlighting the
label of the menu item instead, e.g. with inverse video. In this case we
will need to 'unhighlight' the old item and highlight the new one.
As a first step, move the pointer logic into a function. This fixes a
bug where the item is hidden when it should not be.
Signed-off-by: Simon Glass <sjg@chromium.org>
It is a pain to manually set the fonts of all objects to be consistent.
Some spacing settings are also better set globally than by manually
positioning each object.
Add a 'theme' to the expo, to hold this information. For now it includes
only the font size.
Signed-off-by: Simon Glass <sjg@chromium.org>
Provide a way to set the full dimensions of objects, i.e. including the
width and height.
For menus, calculate the bounding box of all objects in the menu. Set all
labels to be the same size, so that highlighting works correct, once
implemented.
Signed-off-by: Simon Glass <sjg@chromium.org>
Rather than estimating, measure the text accurately, using the new
vidconsole feature. This allows accurate placement of objects.
Signed-off-by: Simon Glass <sjg@chromium.org>
We currently have just a 'hide' property for each object. In preparation
for adding more properties, convert the struct to use a flags value,
instead of individual booleans. This is more extensible.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present objects only have a position so it is not possible to determine
the amount of space they take up on the display.
Add width and height properties, using a struct to keep all the dimensions
together.
For now this is not used. Future work will set up these new properties.
Signed-off-by: Simon Glass <sjg@chromium.org>
Provide a way to set this value so that it is easy to separate the
statically allocated IDs (generated by the caller) from those
generated dynamically by expo itself.
Signed-off-by: Simon Glass <sjg@chromium.org>
For laying out text accurately it is necessary to know the width and
height of the text. Add a measure() method to the console API, so this
can be supported.
Add an implementation for truetype and a base implementation for the
normal console.
Signed-off-by: Simon Glass <sjg@chromium.org>
It is sometimes necessary to highlight some text in a different colour.
Add an easy way to do this and then restore the original console colours.
Signed-off-by: Simon Glass <sjg@chromium.org>
This uses the private data of the video uclass, not the console uclass
(its child). Update the comment to avoid confusion.
Signed-off-by: Simon Glass <sjg@chromium.org>
Fixes: a032e4b55e ("video: Move console colours to the video uclass")
This feature is different enough from bootstd that it probably deserves
its own log category. It cannot use a uclass since it is not a device.
Add a new category.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present livetree can only be used for the control FDT. It is useful
to be able to use the ofnode API for other FDTs, e.g. those used by
the upcoming configuration editor.
We already have most of the support present, and tests can be marked with
the UT_TESTF_OTHER_FDT flag to use another FDT as a special case. But
with this change, the functionality becomes more generally available.
Plumb in the require support.
Signed-off-by: Simon Glass <sjg@chromium.org>
Ensure that the block of memory used by live tree is aligned according to
the default for structures. This ensures that the root node appears at
the start of the block, so it can be used with free(), rather than being
4 bytes later in some cases.
This corrects a rather obscure bug in unflatten_device_tree().
Fixes: 8b50d526ea ("dm: Add a function to create a 'live' device tree")
Signed-off-by: Simon Glass <sjg@chromium.org>
This functionality current sits in bootstd, but it is more generally
useful. Add a function to load a file into memory, allocating it as
needed. Adjust bootstd to use this version.
Note: Tests are added in the subsequent patch which converts the 'cat'
command to use this function.
Signed-off-by: Simon Glass <sjg@chromium.org>
The putc_xy() method is supposed to return the amount of space used. The
existing implementation erroneously adds the previous sub-pixel position
to the returned value. This spaces out the characters very slightly more
than it should. It is seldom noticeable but it does make accurate
measurement of the text impossible.
Fix this minor but long-standing bug.
Fixes: a29b012037 ("video: Add a console driver that uses TrueType fonts")
Signed-off-by: Simon Glass <sjg@chromium.org>
