| Age | Commit message (Collapse) | Author |
|---|
|
Add support to clk_gate ops to reference the clk ID at an offset by
using the just introduced gates_offs value from the unified muxes +
gates implementation.
Gate clock that doesn't have gates_offs set won't be affected as the
offset will simply be 0 and won't be offset of any value.
Signed-off-by: Christian Marangi <[email protected]>
|
|
Add support for gates in clk_tree for infrasys ops.
Infracfg clks can have a sum of gates and muxes, and current solution
handle this by duplicating the driver and split clks for mux and clks
for gates. Upstream linux kernel handle this differently and doesn't
have this distinction.
To be closer to the upstream kernel clock definition, implement
additional logic to have gates defined in the clk_tree and implement
variant for the infrasys ops to handle gates defined in the tree.
Similar to how it's done with factor and mux, we introduce gates_offs.
Upstream kernel follow the similar logic with all the ID defined as
FDIVS, MUXES and finally GATES.
Signed-off-by: Christian Marangi <[email protected]>
|
|
We currently return 0 if XTAL rate is requested in get_mux_rate. This
deviates from what is done in get_factor_rate and is totally wrong as it
can cause unwanted results (division by 0 crash)
For infrasys that makes use of CLK_XTAL, assume xtal_rate to be defined
in clk_tree and return the rate when BYPASS_XTAL is not enabled with
clk ID 0 index parents.
Signed-off-by: Christian Marangi <[email protected]>
|
|
There is currently a massive bug that makes any gate clk that have
CLK_XTAL as parent to return the wrong clock.
Following the code, with CLK_XTAL defined as TOPCKGEN parent, the
topckgen get_rate is called. The clk ID (0) is parsed and only in some
corner case (scenario where fixed clock are not defined) the correct
XTAL rate will be returned as get_factor or get_mux is called (that have
correct handling for CLK_XTAL). With fixed clock defined, the rate that
will be returned will always be the FIRST ELEMENT of the fixed clock
table instead of the hardcoded XTAL rate.
To handle this, add additional logic and if the flag is set to
PARENT_XTAL for the gate, return the XTAL rate directly.
We assume the clk_tree to have xtal_rate defined with clk gates that
have XTAL as parents.
Signed-off-by: Christian Marangi <[email protected]>
|
|
Bevor this commit, only clause 22 access was possible. After this commit,
clause 45 direct access will available as well.
Note that there is a slight change of behavior: Before this commit, the
C45E bit was set to whatever value was left in the register from the
previous access. After this commit, we adopt the common practice of
discerning C45 from C22 using the devad argument.
Signed-off-by: Philip Oberfichtner <[email protected]>
Reviewed-by: Marek Vasut <[email protected]>
|
|
Option Acknowledgment (OACK) is an extension of TFTP protocol (see rfc2347).
Not all tftp servers implements it. For example it does not supported by
tftpd server from debian-11 (https://packages.debian.org/bullseye/tftpd).
Starting the "tftpput $loadaddr $size out_file" command with such server
will results in the following packets flow:
192.168.27.3 192.168.27.1 TFTP Write Request, ...
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 0
192.168.27.3 192.168.27.1 TFTP Write Request, ...
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 0
192.168.27.3 192.168.27.1 TFTP Write Request, ...
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 0
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 0
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 0
...
so, no data transfer happening.
Here is a packets flow for tftp-server with OACK support
(tftpd-hpa: https://packages.debian.org/stable/tftpd-hpa)
192.168.27.3 192.168.27.1 TFTP Write Request, ...
192.168.27.1 192.168.27.3 TFTP Option Acknowledgement, ...
192.168.27.3 192.168.27.1 TFTP Data Packet, Block: 1
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 1
192.168.27.3 192.168.27.1 TFTP Data Packet, Block: 2
192.168.27.1 192.168.27.3 TFTP Acknowledgement, Block: 2
and this time data transfer starts normally.
As we can see there is no OACK packet in the first case. Investigating
an issue we'll find out:
1) tftp_start() sets
tftp_state = STATE_SEND_WRQ;
2) on OACK tftp_handler() sets
tftp_state = STATE_DATA;
and send a first DATA packet.
3) on ACK tftp_handler() will call a tftp_send() function.
tftp_send() will
* tftpd with OACK support:
Current state is STATE_DATA, so transmittion of data packet will
happen.
* tftpd without OACK support
Current state is STATE_SEND_WRQ, so retransmission of WRQ packet
will happen. Thus tftpd-server will retransmit an ACK.
This will repeats until timeout happens.
According to RFC1350 this is wrong. We should start data transfer
instead of WRQ retransmission.
This patch fix an issue, so tftpput works fine with both types of servers.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
|
eth_env_set_enetaddr_by_index() declaration is duplicated in eth_internal.h
and net.h, but all units including eth_internal.h already include net.h.
Remove the superfluous declaration.
Signed-off-by: Alexander Sverdlin <[email protected]>
|
|
https://source.denx.de/u-boot/custodians/u-boot-video
- improve video sync performance with background syncing (cyclic)
- fix dropping characters when pasting commands over the UART
- enable background syncing by default for boards using VIDEO
- make sandbox video more responsive
|
|
USB is stopped using driver model now, in dm_remove_devices_flags() in
announce_and_cleanup() at the top of this file.
The usb_stop() call actually unbinds devices.
When a USB device is unbound, it causes any bootflows attached to it to
be removed, via a call to bootdev_clear_bootflows() from
bootdev_pre_unbind(). This obviously makes it impossible to boot the
bootflow.
However, when booting a bootflow that relies on USB, usb_stop() is
called, which unbinds the device. At that point any information
attached to the bootflow is dropped.
This is quite risky since the contents of freed memory are not
guaranteed to remain unchanged. Depending on what other options are
done before boot, a hard-to-find bug may crop up.
Drop the call to this old function.
Leave the netconsole call there, since this needs conversion to
driver model.
Signed-off-by: Simon Glass <[email protected]>
Suggested-by: Shantur Rathore <[email protected]>
Reviewed-by: Mattijs Korpershoek <[email protected]>
|
|
Recently we are introducing multiple git subtree projects and
it is the right time to have a universal script to update
various subtrees and replace the dts/update-dts-subtree.sh.
update-subtree.sh is a wrapper of git subtree commands.
Usage: From U-Boot top directory,
run
$ ./tools/update-subtree.sh pull <subtree-name> <release-tag>
for pulling a tag from the upstream.
Or run
$ ./tools/update-subtree.sh pick <subtree-name> <commit-id>
for cherry-pick a commit from the upstream.
Currently <subtree-name> supports dts, mbedtls and lwip.
Signed-off-by: Raymond Mao <[email protected]>
|
|
When CONFIG_SYS_MMC_ENV_PART=2, the environment is loaded from
the USER partition (hwpart=0) instead of from the
BOOT1 partition (hwpart=2).
IS_ENABLED() cannot be used for non-boolean KConfig options.
Its documentation states:
> * IS_ENABLED(CONFIG_FOO) evaluates to 1 if CONFIG_FOO is set to 'y',
> * 0 otherwise.
So in our case, IS_ENABLED(CONFIG_SYS_MMC_ENV_PART) evaluates to 0.
Because of this, the hwpart variable is never assigned and mmc_offset()
ends up switching back to the USER hwpart (0) instead of BOOT1 (2).
Fix it by using #define instead.
Tested with:
# have CONFIG_SYS_MMC_ENV_PART=2 in defconfig
# 1. Erase mmc0boot1
=> mmc dev 0 2
=> mmc erase 0 400
=> mmc read ${loadaddr} 0 400
=> md ${loadaddr} 4
82000000: 00000000 00000000 00000000 00000000 ................
# 2. Restore default environment and save to MMC
=> env default -a -f
=> saveenv
# 3. Read back mmc0boot1 and confirm the env is present
=> mmc read ${loadaddr} 0 400
=> md ${loadaddr} 4
82000000: 13e0632e 72646461 7469665f 3978303d .c..addr_fit=0x9
Fixes: 5b4acb0ff79d ("env: mmc: Apply GPT only on eMMC user HW partition")
Signed-off-by: Mattijs Korpershoek <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
Jiaxun Yang <[email protected]> says:
This is a subset of my previous arm64_be work.
I wish this could be merged first so it would be easier to work
against xtensa and arm64 be support.
|
|
So CONFIG_SYS_BIG_ENDIAN is our cross architecture option for
selecting machine endian, while the old CONFIG_CPU_BIG_ENDIAN
is defined by Arc only.
Use it whenever possible to ensure big endian code path is enabled
for all possible big endian machines.
Signed-off-by: Jiaxun Yang <[email protected]>
|
|
Move SUPPORT_BIG_ENDIAN, SUPPORT_LITTLE_ENDIAN to top-level
arch Kconfig and let architectures select them as necessary.
Remove if guard for Endianness selection choice so we can
have one of SYS_BIG_ENDIAN, SYS_LITTLE_ENDIAN config symbol
defined even on single endian system.
Default endian to SYS_BIG_ENDIAN for MIPS || MICROBLAZE
and LITTLE_ENDIAN for the rest to retain old config
behaviour.
Note: PPC, SH, Xtensa are technically bi-endian, but I
checked compiled u-boot image with readelf, U-Boot currently
only support little endian for SH and Xtensa, Big Endian for
PPC.
Signed-off-by: Jiaxun Yang <[email protected]>
|
|
With sandbox, when U-Boot is waiting for input it syncs the video
display, since presumably the user has finished typing.
Now that cyclic is used for video syncing, we can drop this. Cyclic
will automatically call the video_idle() function when idle.
Signed-off-by: Simon Glass <[email protected]>
|
|
Now that sandbox is using cyclic for video, it trips the 1us time
limit. Updating the sandbox display often takes 20ms or more.
Increase the limit to 100ms to avoid a warning.
Signed-off-by: Simon Glass <[email protected]>
|
|
At present U-Boot flushes the cache after every character written to
the display. This makes the command-line slower, to the point that
pasting in long strings can fail.
Add a cyclic function to sync the display every 10ms. Enable this by
default.
Allow much longer times for sandbox, since the SDL display is quite
slow.
Avoid size growth if the feature is disabled by making the new init and
destroy functions dependent on CYCLIC being enabled.
Signed-off-by: Simon Glass <[email protected]>
|
|
Rather than using a static variable, use the video device's private
data to remember when the last video sync was completed. This allows
each display to have its own sync and avoids using static data in SPL.
Signed-off-by: Simon Glass <[email protected]>
|
|
The cyclic subsystem is currently enabled either in all build phases
or none. For tools this should not be enabled, but since lib/shc256.c
and other files include watchdog.h in the host build, we must make
sure that it is not enabled there.
Add an SPL symbol so that there is more control of this.
Add an include into cyclic.h so that tools can include this file.
Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Devarsh Thakkar <[email protected]>
Reviewed-by: Stefan Roese <[email protected]>
|
|
https://source.denx.de/u-boot/custodians/u-boot-efi
Pull request efi-2024-10-rc2
Documentation:
* Add a description for bootmeth_android
UEFI:
* Provide and use function list_count_nodes()
* Require EFI boot manager for EBBR compliance
* Correct check in efi_load_option_dp_join()
* Adjust config options for capsule updates
SMBIOS:
* Add extended Extended BIOS ROM Size
|
|
The linux kernel has the list_count_nodes() API functions which is
used for counting nodes of a list. This has now been imported in
U-Boot as part of an earlier commit. Use this function and drop the
list_count_items().
Signed-off-by: Sughosh Ganu <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
Use the list_for_each_entry() API to get the efi_mem_list node
directly, instead of making an additional call to list_entry().
Signed-off-by: Sughosh Ganu <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
Populate the previous memory descriptor node pointer only after it's
parent struct has been initialised. The compiler fixes this logic to
do the right thing, but it is better to have correct code in place.
Signed-off-by: Sughosh Ganu <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
Use the API function list_count_nodes() to count the number of EFI
memory map entries.
Signed-off-by: Sughosh Ganu <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
|
|
Add a function to count the nodes of a list.
Taken from linux 6.11-rc1 tag commit 8400291e289e.
Signed-off-by: Sughosh Ganu <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
After assigning to *dp we must check this value and not dp.
Addresses-Coverity-ID: 501792 Control flow issues (DEADCODE)
Fixes: 58bef195f94e ("cmd: eficonfig: add support for setting fdt")
Signed-off-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
Add initial documentation for the Android bootmeth.
Signed-off-by: Mattijs Korpershoek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
U-Boot claims to create SMBIOS 3.7 tables. The type 0 table has
a field Extended BIOS ROM Size since version 3.1.
BIOS ROM sizes of 16 MiB or above must be written to this field.
Add and fill the missing field.
Signed-off-by: Heinrich Schuchardt <[email protected]>
|
|
Add an additional search path /dtbs, this is where dtbs are installed on
postmarketOS and potentially other distros.
Signed-off-by: Caleb Connolly <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
A system has to support booting via the boot manager to be EBBR compliant.
See the reference to variables Boot#### in the specification.
Signed-off-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Ilias Apalodimas <[email protected]>
|
|
EFI_IGNORE_OSINDICATIONS is used to ignore OsIndications if setvariable
at runtime is not supported and allow the platform to perform capsule
updates on disk. With the recent changes boards can conditionally enable
setvariable at runtime using EFI_RT_VOLATILE_STORE.
Let's make that visible in our Kconfigs and enable EFI_IGNORE_OSINDICATIONS
when set variable at runtime is disabled.
Since EFI_RT_VOLATILE_STORE needs help from the OS to persist the
variables, allow users to ignore OsIndications even if setvariable at
runtime is enabled.
Signed-off-by: Ilias Apalodimas <[email protected]>
Reviewed-by: Heinrich Schuchardt <[email protected]>
|
|
Mikhail Kshevetskiy <[email protected]> says:
v2 changes:
* Drop sw_blink_state structure, move its necessary fields to
led_uc_plat structure.
* Add cyclic_info pointer to led_uc_plat structure. This
simplify code a lot.
* Remove cyclic function search logic. Not needed anymore.
* Fix blinking period. It was twice large.
* Other cleanups.
v3 changes:
* Adapt code to recent cyclic function changes
* Move software blinking functions to separate file
* Other small changes
v4 changes:
* Refactoring of led_set_period() function
v5 changes
* Fix compilation if CONFIG_LED_BLINK is not defined
v6 changes:
* Enable LEDST_BLINK state unconditionally.
* Function led_set_period() becomes available when CONFIG_LED_BLINK
is disabled. This makes led code simpler.
* Software blinking requires about 100 bytes of data for a led. It's
not a good idea to allocate so much memory for each supported led.
Change the code to allocate blinking data only for required leds.
|
|
The standard property
linux,default-trigger = "pattern";
used to get an effect. No blinking parameters can be set yet.
Signed-off-by: Michael Polyntsov <[email protected]>
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
If hardware (or driver) doesn't support leds blinking, it's
now possible to use software implementation of blinking instead.
This relies on cyclic functions.
Signed-off-by: Michael Polyntsov <[email protected]>
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Changes:
* enable LEDST_BLINK state unconditionally
* function led_set_period() becomes available when CONFIG_LED_BLINK
is disabled. This makes led code simpler.
* fix cmd/led.c to work properly when LEDST_BLINK present, but
CONFIG_LED_BLINK is disabled
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Drop all duplicate newlines from the include directory files.
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop all duplicate newlines. No functional change.
Signed-off-by: Marek Vasut <[email protected]>
|
