]> git.dujemihanovic.xyz Git - u-boot.git/log
u-boot.git
14 months agoMakefile: make u-boot-initial-env target depend explicitly on scripts_basic
Rasmus Villemoes [Tue, 3 Oct 2023 10:02:17 +0000 (12:02 +0200)]
Makefile: make u-boot-initial-env target depend explicitly on scripts_basic

We're seeing sporadic errors like

  ENVC    include/generated/env.txt
  HOSTCC  scripts/basic/fixdep
  ENVP    include/generated/env.in
  ENVT    include/generated/environment.h
  HOSTCC  tools/printinitialenv
/bin/sh: 1: scripts/basic/fixdep: not found
make[1]: *** [scripts/Makefile.host:95: tools/printinitialenv] Error 127
make[1]: *** Deleting file 'tools/printinitialenv'
make: *** [Makefile:2446: u-boot-initial-env] Error 2
make: *** Waiting for unfinished jobs....

where sometimes the "fixdep: not found" is instead "fixdep: Permission
denied" and the Error 127 becomes 126.

This smells like a race condition, and indeed it is: Currently,
u-boot-initial-env is a prerequisite of the envtools target, which
also lists scripts_basic as a prerequisite:

envtools: u-boot-initial-env scripts_basic $(version_h) $(timestamp_h) tools/version.h
$(Q)$(MAKE) $(build)=tools/env

However, the u-boot-initial-env rule involves building the
printinitialenv helper, which in turn is built using an if_changed_dep
rule. That means we must ensure scripts/basic/fixdep is built and
ready before trying to build printinitialenv, i.e. the
u-boot-initial-env rule itself must depend on the phony scripts_basic
target.

Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoboard: xen: introduce virtio-blk support
Andrii Chepurnyi [Tue, 3 Oct 2023 08:58:28 +0000 (08:58 +0000)]
board: xen: introduce virtio-blk support

Added new xenguest_arm64_virtio_defconfig which
enables support for virtio-blk using various types
of transport like virtio-pci, vrtio-mmio. Currently
supported: up to 2 PCI host bridges and 10 MMIO devices.
Note: DT parsing code was partly taken from pci-uclass.c
Limitation: All memory regions should be
below 4GB address space.

Signed-off-by: Andrii Chepurnyi <andrii_chepurnyi@epam.com>
14 months agoboard: synquacer: set actual gd->ram_top and gd->ram_size
Masahisa Kojima [Tue, 3 Oct 2023 02:29:57 +0000 (11:29 +0900)]
board: synquacer: set actual gd->ram_top and gd->ram_size

Current gd->ram_size and gd->ram_top reflect only the
first DRAM bank even if the SynQuacer Developerbox could
have up to three DRAM banks.
With the commit 06d514d77c37 ("lmb: consider EFI memory map"),
the first DRAM bank indicates <4GB address, so whole >4GB memory
is marked as EFI_BOOT_SERVICES_DATA and it results that
U-Boot can not access >4GB memory.

Since 64-bits DRAM address is fully available on the SynQuacer
Developerbox, let's set the installed DIMM information to
gd->ram_top and gd->ram_size.

Signed-off-by: Masahisa Kojima <masahisa.kojima@linaro.org>
Acked-by: Jassi Brar <jaswinder.singh@linaro.org>
14 months agoinput: avoid NULL dereference
Heinrich Schuchardt [Tue, 3 Oct 2023 01:09:01 +0000 (03:09 +0200)]
input: avoid NULL dereference

Before using the result of env_get("stdin") we must check if it is NULL.

Avoid #if. This resolves the -Wunused-but-set-variable issue and we don't
need a dummy assignment in the else branch. Anyway this warning is
disabled in the Makefile.

For sake of readability use an early return after the configuration check.

Checking CONFIG_SPL_BUILD is incorrect as env_get() is only defined if
CONFIG_$(SPL_TPL)ENV_SUPPORT=y.

Fixes: 985ca3945fa3 ("spl: input: Allow input in SPL and TPL")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
14 months agomkimage: allow internalization of data-position
Lars Feyaerts [Mon, 2 Oct 2023 08:00:14 +0000 (10:00 +0200)]
mkimage: allow internalization of data-position

Make it possible for data that was externalized using a static external
position (-p) to be internalized. Enables the ability to convert
existing FIT images built with -p to be converted to a FIT image where the
data is internal, to be converted to a FIT image where the data is
external relative to the end of the FIT (-E) or change the initial
static external position to a different static external position (-p).

Removing the original external-data-related properties ensures that
they're not present after conversion. Without this, they would still be
present in the resulting FIT even if the FIT has been, for example,
internalized.

Signed-off-by: Lars Feyaerts <lars@bitbiz.be>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agocheckpatch: skip fdtdec_* check for tools
Lars Feyaerts [Mon, 2 Oct 2023 08:00:13 +0000 (10:00 +0200)]
checkpatch: skip fdtdec_* check for tools

Have checkpatch.pl skip warnings for use of fdtdec_* functions in
ooling; livetree isn't used there.

Signed-off-by: Lars Feyaerts <lars@bitbiz.be>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoMerge tag 'fsl-qoirq-2023-10-10' of https://source.denx.de/u-boot/custodians/u-boot...
Tom Rini [Wed, 11 Oct 2023 14:21:33 +0000 (10:21 -0400)]
Merge tag 'fsl-qoirq-2023-10-10' of https://source.denx.de/u-boot/custodians/u-boot-fsl-qoriq

Drop legacy PPA secure FW support
support for MC reserved memory
reset the FLSHxCR1 registers for nxp_fspi

14 months agoMerge branch '2023-10-10-blk-sandbox-support-binding-a-device-with-a-given-logical...
Tom Rini [Wed, 11 Oct 2023 01:47:50 +0000 (21:47 -0400)]
Merge branch '2023-10-10-blk-sandbox-support-binding-a-device-with-a-given-logical-block-size'

To quote the author:
At present on Sandbox when binding to a host backing file, the host
block device is created with a hard-coded 512 bytes block size.

Such assumption works for most cases, but for situation that with a raw
image file dump from a pre-formatted GPT partitioned disk image from a
4KiB block size device, when binding this file to a host device and mapping
this device to a blkmap, "blkmap" command like "blkmap part" won't work
correctly, due to block size mismatch during parsing the partition table.

This series updates Sandbox block driver, as well as the blkmap driver,
to get rid of the hard-coded 512 bytes block size assumption.

This series is available at u-boot-x86/blk for testing.

Test log (512 block size):

  => host bind 0 test.img
  => host info
  dev       blocks  blksz label           path
    0       262144    512 0               test.img
  => blkmap create 0
  Created "0"
  => blkmap map 0 0 40000 linear host 0 0
  Block 0x0+0x40000 mapped to block 0x0 of "host 0"
  => blkmap info
  Device 0: Vendor: U-Boot Rev: 1.0 Prod: blkmap
              Type: Hard Disk
              Capacity: 128.0 MB = 0.1 GB (262144 x 512)
  => blkmap part

  Partition Map for BLKMAP device 0  --   Partition Type: EFI

  Part    Start LBA       End LBA         Name
          Attributes
          Type GUID
          Partition GUID
    1     0x00000022      0x000000bd      "u-boot-spl"
          attrs:  0x0000000000000000
          type:   5b193300-fc78-40cd-8002-e86c45580b47
                  (5b193300-fc78-40cd-8002-e86c45580b47)
          guid:   0bb6bb6e-4aac-4c27-be03-016b01e7b941
    2     0x00000822      0x00000c84      "u-boot"
          attrs:  0x0000000000000000
          type:   2e54b353-1271-4842-806f-e436d6af6985
                  (2e54b353-1271-4842-806f-e436d6af6985)
          guid:   91d50814-8e31-4cc0-97dc-779e1dc59056
    3     0x00000c85      0x0000cc84      "rootfs"
          attrs:  0x0000000000000004
          type:   0fc63daf-8483-4772-8e79-3d69d8477de4
                  (linux)
          guid:   42799722-6e55-46e6-afa9-529e7af3f03b

Test log (4096 block size):

  => host bind 0 test.img 4096
  => host info
  dev       blocks  blksz label           path
    0        32768   4096 0               test.img
  => blkmap create 0
  Created "0"
  => blkmap map 0 0 8000 linear host 0 0
  Block 0x0+0x8000 mapped to block 0x0 of "host 0"
  => blkmap info
  Device 0: Vendor: U-Boot Rev: 1.0 Prod: blkmap
              Type: Hard Disk
              Capacity: 128.0 MB = 0.1 GB (32768 x 4096)
  => blkmap part

  Partition Map for BLKMAP device 0  --   Partition Type: EFI

  Part    Start LBA       End LBA         Name
          Attributes
          Type GUID
          Partition GUID
    1     0x00000100      0x00001fff      "primary"
          attrs:  0x0000000000000000
          type:   0fc63daf-8483-4772-8e79-3d69d8477de4
                  (linux)
          guid:   eba904d7-72c1-4dbd-bb4e-36be49cba5e3
    2     0x00002000      0x00007ffa      "primary"
          attrs:  0x0000000000000000
          type:   0fc63daf-8483-4772-8e79-3d69d8477de4
                  (linux)
          guid:   c48c360e-db47-46da-ab87-26416fad3cd3

14 months agodisk: part: Handle blkmap device in print_part_header()
Bin Meng [Tue, 26 Sep 2023 08:43:45 +0000 (16:43 +0800)]
disk: part: Handle blkmap device in print_part_header()

Print out the blkmap device type when showing partition header for
a blkmap device.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agodisk: part: Print out the unknown device uclass id
Bin Meng [Tue, 26 Sep 2023 08:43:44 +0000 (16:43 +0800)]
disk: part: Print out the unknown device uclass id

It's helpful to output the device uclass id for unknown devices
during the debugging process.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agodm: blk: Drop blk_{read,write}_devnum()
Bin Meng [Tue, 26 Sep 2023 08:43:43 +0000 (16:43 +0800)]
dm: blk: Drop blk_{read,write}_devnum()

blk_{read,write}_devnum() are no longer used by anywhere in the
source tree. Drop them.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agocmd: blk_common: Stop using hard-coded block size for Sandbox operations
Bin Meng [Tue, 26 Sep 2023 08:43:42 +0000 (16:43 +0800)]
cmd: blk_common: Stop using hard-coded block size for Sandbox operations

commit 3d2fc7971454 ("cmd: blk: Allow generic read/write operations to work in sandbox")
used the hard-coded block size (512) for accessing the sandbox host
device. Now that we have added support for non-512 block size for both
Sandbox host device and blkmap driver, let's stop using the hard-coded
block size.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agodm: blk: Rename get_desc() and make it externally visible
Bin Meng [Tue, 26 Sep 2023 08:43:41 +0000 (16:43 +0800)]
dm: blk: Rename get_desc() and make it externally visible

get_desc() can be useful outside blk-uclass.c. Let's change it to
an API and make it externally visible.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agocmd: blk_common: Use macros for the return values
Bin Meng [Tue, 26 Sep 2023 08:43:40 +0000 (16:43 +0800)]
cmd: blk_common: Use macros for the return values

Avoid using magic number 0/1 for the command result.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
14 months agoblk: blkmap: Support mapping to device of any block size
Bin Meng [Tue, 26 Sep 2023 08:43:39 +0000 (16:43 +0800)]
blk: blkmap: Support mapping to device of any block size

At present if a device to map has a block size other than 512,
the blkmap map process just fails. There is no reason why we
can't just use the block size of the mapped device.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agocmd: blkmap: Make map_handlers[] and its .fn static
Bin Meng [Tue, 26 Sep 2023 08:43:38 +0000 (16:43 +0800)]
cmd: blkmap: Make map_handlers[] and its .fn static

These are only used in cmd/blkmap.c.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoblk: blkmap: Make bind/unbind routines static
Bin Meng [Tue, 26 Sep 2023 08:43:37 +0000 (16:43 +0800)]
blk: blkmap: Make bind/unbind routines static

These 2 are only used in drivers/block/blkmap.c.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agocmd: host: Print out the block size of the host device
Bin Meng [Tue, 26 Sep 2023 08:43:36 +0000 (16:43 +0800)]
cmd: host: Print out the block size of the host device

It's useful if we can print out the block size of the host device
in the "host info" command.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoblk: host_dev: Sanity check on the size of host backing file
Bin Meng [Tue, 26 Sep 2023 08:43:35 +0000 (16:43 +0800)]
blk: host_dev: Sanity check on the size of host backing file

Since we are emulating a block device, its size should be multiple
of the configured block size.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoblk: host_dev: Make host_sb_detach_file() and host_sb_ops static
Bin Meng [Tue, 26 Sep 2023 08:43:34 +0000 (16:43 +0800)]
blk: host_dev: Make host_sb_detach_file() and host_sb_ops static

They are only used in drivers/block/host_dev.c.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoblk: sandbox: Support binding a device with a given logical block size
Bin Meng [Tue, 26 Sep 2023 08:43:33 +0000 (16:43 +0800)]
blk: sandbox: Support binding a device with a given logical block size

Allow optionally set the logical block size of the host device to
bind in the "host bind" command. If not given, defaults to 512.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
14 months agocmd: host: Mandate the filename parameter in the 'bind' command
Bin Meng [Tue, 26 Sep 2023 08:43:32 +0000 (16:43 +0800)]
cmd: host: Mandate the filename parameter in the 'bind' command

At present the host bind command does not require filename to be
provided. When it is not given NULL is passed to the host device
driver, which ends up failure afterwards.

Change to mandate the filename so that it is useful.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoblk: Use a macro for the typical block size
Bin Meng [Tue, 26 Sep 2023 08:43:31 +0000 (16:43 +0800)]
blk: Use a macro for the typical block size

Avoid using the magic number 512 directly.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoMerge branch '2023-10-09-assorted-fixes'
Tom Rini [Tue, 10 Oct 2023 12:54:17 +0000 (08:54 -0400)]
Merge branch '2023-10-09-assorted-fixes'

- Cleanup how we pick what to launch in SPL, a few test changes, some TI
  K3 platform updates, top-level Makefile fixes and related cleanup,
  correct a problem with LMB overlap, other assorted fixes.

14 months agospi: nxp_fspi: reset the FLSHxCR1 registers
Han Xu [Wed, 13 Sep 2023 21:15:35 +0000 (16:15 -0500)]
spi: nxp_fspi: reset the FLSHxCR1 registers

Reset the FLSHxCR1 registers to default value. ROM may set the register
value and it affects the SPI NAND normal functions.

Signed-off-by: Han Xu <han.xu@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agoboard: freescale: ls1088a: declare MC reserved regions
Laurentiu Tudor [Wed, 27 Sep 2023 15:30:49 +0000 (18:30 +0300)]
board: freescale: ls1088a: declare MC reserved regions

Populate the device tree with the MC reserved memory regions.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agoboard: freescale: ls2080a: declare MC reserved regions
Laurentiu Tudor [Wed, 27 Sep 2023 15:30:48 +0000 (18:30 +0300)]
board: freescale: ls2080a: declare MC reserved regions

Populate the device tree with the MC reserved memory regions.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agodrivers: net: fsl-mc: add support for MC reserved memory
Laurentiu Tudor [Wed, 27 Sep 2023 15:30:47 +0000 (18:30 +0300)]
drivers: net: fsl-mc: add support for MC reserved memory

Add support for declaring in device tree the reserved memory ranges
required for MC. Since the MC firmware acts as any DMA master present
in the SoC, the reserved memory ranges need also be identity mapped
in the SMMU, so create the required 'iommu-addresses' property in
the reserved memory nodes.
For now this support is used only on LX2160A SoCs.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agoarmv8: fsl-layerscape: make some functions static
Laurentiu Tudor [Wed, 27 Sep 2023 15:30:46 +0000 (18:30 +0300)]
armv8: fsl-layerscape: make some functions static

Some functions are not used outside this file, so make them static.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agofsl-layerscape: drop obsolete PPA secure firmware support
Laurentiu Tudor [Wed, 23 Aug 2023 13:25:46 +0000 (16:25 +0300)]
fsl-layerscape: drop obsolete PPA secure firmware support

PPA was a secure firmware developed in-house which is no longer
supported and replaced by TF-A quite some years ago. Drop support
for it.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agoconfigs: layerscape: delete defconfigs using legacy PPA secure FW
Laurentiu Tudor [Tue, 10 Oct 2023 02:08:36 +0000 (10:08 +0800)]
configs: layerscape: delete defconfigs using legacy PPA secure FW

PPA was a secure firmware developed in-house which is no longer
supported and replaced by TF-A quite some years ago. This makes
the defconfigs that make use of PPA obsolete, so remove them.

Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
 [Merged part 1 and part 2]
Signed-off-by: Peng Fan <peng.fan@nxp.com>
14 months agotest: Fix SPL tests not being run
Sean Anderson [Fri, 29 Sep 2023 16:06:54 +0000 (12:06 -0400)]
test: Fix SPL tests not being run

SPL doesn't have OF_LIVE enabled, so we can only run tests with a flat
tree. Don't skip them even if they don't use the devicetree.

Fixes: 6ec5178c0ef ("test: Skip flat-tree tests if devicetree is not used")
Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoboard/km/cent2: Fix buffer overflow when fixing MAC address
Francois Berder [Fri, 29 Sep 2023 14:11:00 +0000 (16:11 +0200)]
board/km/cent2: Fix buffer overflow when fixing MAC address

String "/soc/fman/ethernet@e8000" is 25 bytes long
and not 24 due to extra byte for null character at
the end.

Signed-off-by: Francois Berder <fberder@outlook.fr>
14 months agostdio: fix stdio_deregister_dev()
Heinrich Schuchardt [Fri, 29 Sep 2023 00:47:17 +0000 (02:47 +0200)]
stdio: fix stdio_deregister_dev()

When copying the name of a stdio device we must ensure that it is NUL
terminated before passing it to strcmp() to avoid a buffer overrun.

Truncating the name field leads to failure to deregister a stdio device.
When copying we must ensure that the name field sizes match.

Addresses-Coverity-ID: 350462 String not null terminated
Fixes: 5294e97832a6 ("stdio: extend "name" to 32 symbols")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agodm: serial: fix serial_post_probe()
Heinrich Schuchardt [Fri, 29 Sep 2023 00:47:16 +0000 (02:47 +0200)]
dm: serial: fix serial_post_probe()

The size of the name of a udevice is not limited.

When setting the fixed sized name field of a stdio device we must ensure
that the target string is NUL terminated to avoid buffer overflows.

Fixes: 57d92753d4ca ("dm: Add a uclass for serial devices")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agospl: Jump to image at end of board_init_r
Jonas Karlman [Wed, 27 Sep 2023 21:44:13 +0000 (21:44 +0000)]
spl: Jump to image at end of board_init_r

spl_board_prepare_for_boot() is not called before jumping/invoking atf,
optee, opensbi or linux images.

Jump to image at the end of board_init_r() to fix this.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agospl: add __noreturn attribute to spl_invoke_atf function
Chanho Park [Fri, 8 Sep 2023 08:08:56 +0000 (17:08 +0900)]
spl: add __noreturn attribute to spl_invoke_atf function

spl_invoke_atf function will not be returned to SPL. Thus, we need to
set __noreturn function attribute to the function.

Signed-off-by: Chanho Park <chanho61.park@samsung.com>
14 months agotools: iot2050-sign-fw.sh: Make localization of tools dir more robust
Jan Kiszka [Wed, 27 Sep 2023 15:41:09 +0000 (17:41 +0200)]
tools: iot2050-sign-fw.sh: Make localization of tools dir more robust

When building in-tree, there is no source link.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
14 months agoarm: mach-k3: common: fix compile warnings with PHYS_64BIT on 32bit
Matthias Schiffer [Wed, 27 Sep 2023 13:43:14 +0000 (15:43 +0200)]
arm: mach-k3: common: fix compile warnings with PHYS_64BIT on 32bit

Use uintptr_t instead of phys_addr_t where appropriate, so passing the
addresses to writel() doesn't result in compile warnings when PHYS_64BIT
is set for 32bit builds (which is actually a useful configuration, as
the K3 SoC family boots from an R5 SPL, which may pass bank information
based on gd->bd->bi_dram to fdt_fixup_memory_banks() etc., so PHYS_64BIT
is needed for fixing up the upper bank).

Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
14 months agotest/py: sleep: Add a test for the time command
Love Kumar [Wed, 27 Sep 2023 05:03:55 +0000 (10:33 +0530)]
test/py: sleep: Add a test for the time command

Execute "time <sleep cmd>", and validate that it gives the approximately
the correct amount of command execution time.

Signed-off-by: Love Kumar <love.kumar@amd.com>
14 months agomailbox: k3-sec-proxy: fix error handling for missing scfg in FDT
Matthias Schiffer [Tue, 26 Sep 2023 12:42:54 +0000 (14:42 +0200)]
mailbox: k3-sec-proxy: fix error handling for missing scfg in FDT

The wrong field was checked.

Fixes: f9aa41023bd9 ("mailbox: Introduce K3 Secure Proxy Driver")
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Reviewed-by: Nishanth Menon <nm@ti.com>
14 months agotest: lmb: Add test for coalescing and overlap range
Udit Kumar [Tue, 26 Sep 2023 11:24:43 +0000 (16:54 +0530)]
test: lmb: Add test for coalescing and overlap range

Add test case for an address range which is coalescing with one of
range and overlapping with next range

Cc: Simon Glass <sjg@google.com>
Signed-off-by: Udit Kumar <u-kumar1@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agolmb: remove overlapping region with next range
Udit Kumar [Tue, 26 Sep 2023 11:24:42 +0000 (16:54 +0530)]
lmb: remove overlapping region with next range

In case of new memory range to be added is coalesced
with any already added non last lmb region.

And there is possibility that, then region in which new memory
range added is not adjacent to next region. But have some
sections are overlapping.

So along with adjacency check with next lmb region,
check for overlap should be done.

In case overlap  is found, adjust and merge these two lmb
region into one.

Reported-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Udit Kumar <u-kumar1@ti.com>
14 months agoexynos: Cleanup exynos_init
Tom Rini [Thu, 21 Sep 2023 23:32:48 +0000 (19:32 -0400)]
exynos: Cleanup exynos_init

- None of the callers perform error checking and based on the non-empty
  versions of this function, there's no checking to be done, so make
  this a void.
- Add a default weak version of the function.
- Remove the empty versions of exynos_init now that we have a weak
  version.

Signed-off-by: Tom Rini <trini@konsulko.com>
14 months agoboard: Remove essentially empty board files and Makefiles
Tom Rini [Thu, 21 Sep 2023 23:32:47 +0000 (19:32 -0400)]
board: Remove essentially empty board files and Makefiles

As part of reviewing a new platform, Daniel Schwierzeck noted that we
can have an empty Makefile in the board directory and don't need an
empty board.c file as well.  Further with further cleanup in the
Makefile we can now omit the Makefile entirely. Remove a number of now
unnecessary board.c and Makefiles.

Signed-off-by: Tom Rini <trini@konsulko.com>
14 months agoMakefile: Allow for board directories to not have a Makefile
Tom Rini [Thu, 21 Sep 2023 23:32:46 +0000 (19:32 -0400)]
Makefile: Allow for board directories to not have a Makefile

It is entirely possible at this point to have platforms in U-Boot that
do not have board-specific C code (just Kconfig or environment) and so
make it optional to have to descend in to and then build in the board
directory.

Signed-off-by: Tom Rini <trini@konsulko.com>
14 months agoARM: vexpress_ca9x4: Add missing flash width config option
Patryk Biel [Wed, 20 Sep 2023 07:41:20 +0000 (09:41 +0200)]
ARM: vexpress_ca9x4: Add missing flash width config option

Allow for a proper configuration of CFI flash banks avaialble on the vexpress_ca9x4
board. Without this option, the CFI flash incorrectly detects that the board has two
banks of 32MB flash devices, while in reality, the board provides
two flash banks, each with 64MB size. As a result, it becomes impossible to e.g. to
save u-boot env in flash. According to device tree for this board and
its implementation in QEMU, the CFI width should be set to 32 bits.

After applying this fix, CFI flash will correctly detect both flash
banks each with a size of 64MB. As as result the functionality of e.g. saving u-boot
env will work correctly.

Tested on QEMU 6.2.0.

Cc: Kristian Amlie <kristian.amlie@northern.tech>
Signed-off-by: Patryk Biel <pbiel7@gmail.com>
Reviewed-by: Kristian Amlie <kristian.amlie@northern.tech>
14 months agobootstd: use ARCH_DMA_MINALIGN in memalign() when allocating memory
Tony Dinh [Tue, 19 Sep 2023 21:27:21 +0000 (14:27 -0700)]
bootstd: use ARCH_DMA_MINALIGN in memalign() when allocating memory

Use ARCH_DMA_MINALIGN in memalign() when allocating memory to read the script from the media.

Ref: https://lore.kernel.org/u-boot/CAJaLiFy05F3Cr4X4G2mVkppXnBEFZrHQ+5CngYN8eJPg8ENWkg@mail.gmail.com/T/#m26daadc2463fe653b814a94e6309e5e6bb6be1d1

Note: this patch depends on the previous patch
https://patchwork.ozlabs.org/project/uboot/patch/20230917230649.30357-1-mibodhi@gmail.com/

Signed-off-by: Tony Dinh <mibodhi@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoMakefile: Force regeneration of env.txt
Andrew Davis [Fri, 15 Sep 2023 14:43:23 +0000 (09:43 -0500)]
Makefile: Force regeneration of env.txt

If the source .env file changes to one that is also older than the
generated env.txt file then the .env file is not regenerated. This
means when switching board configs we do not regenerate the env.

This can be tested with:

$ make j721e_evm_a72_defconfig
$ make # this may fail to complete but that is okay for this test
$ make am64x_evm_a53_defconfig
$ make
$ vim include/generated/env.txt

Note this is still the J721e env not the AM64 config as expected.

As ENV_FILE is set based on configuration, regenerate anytime
autoconf.h changes.

Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
14 months agoMerge tag 'u-boot-rockchip-20231007' of https://source.denx.de/u-boot/custodians...
Tom Rini [Sun, 8 Oct 2023 13:58:55 +0000 (09:58 -0400)]
Merge tag 'u-boot-rockchip-20231007' of https://source.denx.de/u-boot/custodians/u-boot-rockchip

- Add Board: rk3568 Bananapi R2Pro;
- Update pcie bifurcation support;
- dwc_eth_qos controller support for rk3568 and rk3588;
- Compressed binary support for U-Boot on rockchip platform;
- dts and config updates for different board and soc;

[ trini: Fix conflict on include/spl.h ]
Signed-off-by: Tom Rini <trini@konsulko.com>
14 months agorockchip: rk356x-u-boot: Add bootph-all to i2c0_xfer pinctrl node
Jonas Karlman [Thu, 3 Aug 2023 21:11:54 +0000 (21:11 +0000)]
rockchip: rk356x-u-boot: Add bootph-all to i2c0_xfer pinctrl node

A RK8XX PMIC is typically using i2c0 on RK356x devices. Add bootph-all
to required pinctrl nodes to simplify use of the prevent booting on
power plug-in option in SPL.

With the following Kconfig options and nodes in u-boot.dtsi the prevent
booting on power plug-in option can work in SPL.

  CONFIG_ROCKCHIP_RK8XX_DISABLE_BOOT_ON_POWERON=y
  CONFIG_SPL_I2C=y
  CONFIG_SPL_POWER=y
  CONFIG_SPL_PINCTRL=y
  CONFIG_SPL_PMIC_RK8XX=y

  &i2c0 {
   bootph-pre-ram;
  };

  &rk817 {
   bootph-pre-ram;

   regulators {
   bootph-pre-ram;
   };
  };

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agopower: pmic: rk8xx: Fix power-on source check in SPL
Jonas Karlman [Thu, 3 Aug 2023 21:02:42 +0000 (21:02 +0000)]
power: pmic: rk8xx: Fix power-on source check in SPL

The commit 30975fb73d51 ("rockchip: Add option to prevent booting on
power plug-in") introduce an option to prevent booting a device when the
device was powered on due to power plug-in instead of pressing a power
button.

This feature works by checking the power-on source during PMIC probe
and powers off the device if power-on source was power plug-in.
This check currently runs very late at PMIC probe in U-Boot proper.

Fix so that the power-on source check can work at probe time in SPL.
Also enable probe after bind and remove the PMIC banner in SPL.

With this we can use ROCKCHIP_RK8XX_DISABLE_BOOT_ON_POWERON and
SPL_PMIC_RK8XX to power off the device very quickly after TPL instead
of after TF-A and U-Boot proper has been loaded and run.

  DDR V1.18 f366f69a7d typ 23/07/17-15:48:58
  ln
  LP4/4x derate en, other dram:1x trefi
  ddrconfig:7
  LPDDR4X, 324MHz
  BW=32 Col=10 Bk=8 CS0 Row=17 CS1 Row=17 CS=2 Die BW=16 Size=8192MB

  change to: 324MHz
  clk skew:0x64

  change to: 528MHz
  clk skew:0x58

  change to: 780MHz
  clk skew:0x58

  change to: 1056MHz(final freq)
  clk skew:0x40
  out
  Power Off due to plug-in event

Fixes: 30975fb73d51 ("rockchip: Add option to prevent booting on power plug-in")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
14 months agorockchip: rk356x: Enable poweroff command
Jonas Karlman [Thu, 17 Aug 2023 05:45:04 +0000 (05:45 +0000)]
rockchip: rk356x: Enable poweroff command

With PMIC_RK8XX, SYSRESET and CMD_POWEROFF options enabled it is
possible to power down a board using the poweroff command and turn the
board back on using a power button.

Enable the poweroff command on RK356x boards that have a button wired
to PMIC pwron.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agopower: pmic: rk8xx: Use sysreset implementation of the poweroff command
Jonas Karlman [Thu, 17 Aug 2023 05:45:02 +0000 (05:45 +0000)]
power: pmic: rk8xx: Use sysreset implementation of the poweroff command

Select SYSRESET_CMD_POWEROFF to use the sysreset implementation of the
poweroff command when PMIC_RK8XX is enabled.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoconfigs: rockchip: rk3308: enable CONFIG_OF_LIBFDT_OVERLAY
FUKAUMI Naoki [Mon, 11 Sep 2023 10:01:21 +0000 (19:01 +0900)]
configs: rockchip: rk3308: enable CONFIG_OF_LIBFDT_OVERLAY

enable CONFIG_OF_LIBFDT_OVERLAY and use it on Radxa ROCK Pi S.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
14 months agoconfigs: rockchip: rk3308: use CONFIG_DEFAULT_FDT_FILE
FUKAUMI Naoki [Mon, 11 Sep 2023 10:01:20 +0000 (19:01 +0900)]
configs: rockchip: rk3308: use CONFIG_DEFAULT_FDT_FILE

all rk3308 boards should use their own dtb file.

also, change fdt_addr_r to avoid following error:
 "ERROR: Did not find a cmdline Flattened Device Tree"
it happens on Radxa ROCK Pi S (256MB/512MB) with kernel built from
Radxa BSP.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
14 months agoarm: dts: rockchip: rock-5b: add support for PCIe3 and NVMe
FUKAUMI Naoki [Tue, 5 Sep 2023 11:47:36 +0000 (20:47 +0900)]
arm: dts: rockchip: rock-5b: add support for PCIe3 and NVMe

this patch adds support for PCIe3 (M.2 M key) and enables NVMe.

 => pci
 BusDevFun  VendorId   DeviceId   Device Class       Sub-Class
 _____________________________________________________________
 00.00.00   0x1d87     0x3588     Bridge device           0x04
 01.00.00   0x10ec     0x8125     Network controller      0x00
 02.00.00   0x1d87     0x3588     Bridge device           0x04
 03.00.00   0x1179     0x011a     Mass storage controller 0x08
 => nvme scan
 => nvme info
 Device 0: Vendor: 0x1179 Rev: AGHA4101 Prod: 79CA20WPKRYN
             Type: Hard Disk
             Capacity: 488386.3 MB = 476.9 GB (1000215216 x 512)

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoarm: dts: rockchip: sync DT for RK3588 series with Linux
FUKAUMI Naoki [Tue, 5 Sep 2023 11:47:35 +0000 (20:47 +0900)]
arm: dts: rockchip: sync DT for RK3588 series with Linux

Sync the device tree for RK3588 series with Linux 6.6-rc1.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoconfigs: rockchip: rock-pi-s: use default bootdelay (2s)
FUKAUMI Naoki [Mon, 11 Sep 2023 10:05:08 +0000 (19:05 +0900)]
configs: rockchip: rock-pi-s: use default bootdelay (2s)

align with other boards.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoconfigs: rockchip: add DOS_PARTITION to RK3308 boards defconfig
Massimo Pegorer [Sun, 1 Oct 2023 14:15:29 +0000 (16:15 +0200)]
configs: rockchip: add DOS_PARTITION to RK3308 boards defconfig

Without DOS_PARTITION support U-Boot is not able to boot an OS stored
into an SD card with MBR partitions table. This is still a quite common
case so add DOS_PARTITION (only for U-Boot proper build) to Rockchip
RK3308 EVB, Radxa ROCK Pi S and Firefly roc-rk3308-cc boards: they are
the only RK boards missing of DOS_PARTITION.

Reported-by: Jayantajit Gogoi <jayanta.gogoi525@gmail.com>
Signed-off-by: Massimo Pegorer <massimo.pegorer+oss@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoboard: rockchip: Add Bananapi R2Pro Board
Frank Wunderlich [Wed, 4 Oct 2023 19:04:34 +0000 (21:04 +0200)]
board: rockchip: Add Bananapi R2Pro Board

Add Bananapi R2 Pro board.

tested:
- sdcard
- both front usb-ports
- sata
- wan-port

lan-ports are connected to mt7531 switch where driver needs to be
separated from mtk ethernet-driver.

Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Jonas Karlman <jonas@kwiboo.se>
14 months agoconfigs: rockchip: Enable ethernet driver on RK3588 boards
Jonas Karlman [Sun, 1 Oct 2023 19:17:22 +0000 (19:17 +0000)]
configs: rockchip: Enable ethernet driver on RK3588 boards

Enable DWC_ETH_QOS_ROCKCHIP and related PHY driver on RK3588 boards that
have an enabled gmac node and drop ETH_DESIGNWARE and GMAC_ROCKCHIP for
remaining RK3588 boards.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoconfigs: rockchip: Enable ethernet driver on RK356x boards
Jonas Karlman [Sun, 1 Oct 2023 19:17:21 +0000 (19:17 +0000)]
configs: rockchip: Enable ethernet driver on RK356x boards

Enable DWC_ETH_QOS_ROCKCHIP and related PHY driver on RK356x boards that
have an enabled gmac node.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agonet: dwc_eth_qos_rockchip: Add support for RK3588
Jonas Karlman [Sun, 1 Oct 2023 19:17:20 +0000 (19:17 +0000)]
net: dwc_eth_qos_rockchip: Add support for RK3588

Add rk_gmac_ops and other special handling that is needed for GMAC to
work on RK3588.

rk_gmac_ops was ported from linux commits:
2f2b60a0ec28 ("net: ethernet: stmmac: dwmac-rk: Add gmac support for rk3588")
88619e77b33d ("net: stmmac: rk3588: Allow multiple gmac controller")

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
14 months agonet: dwc_eth_qos: Add glue driver for GMAC on Rockchip RK3568
Jonas Karlman [Sun, 1 Oct 2023 19:17:19 +0000 (19:17 +0000)]
net: dwc_eth_qos: Add glue driver for GMAC on Rockchip RK3568

Add a new glue driver for Rockchip SoCs, i.e RK3568, with a GMAC based
on Synopsys DWC Ethernet QoS IP.

rk_gmac_ops was ported from linux commit:
3bb3d6b1c195 ("net: stmmac: Add RK3566/RK3568 SoC support")

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
14 months agonet: dwc_eth_qos: Stop spam of RX packet not available message
Jonas Karlman [Sun, 1 Oct 2023 19:17:18 +0000 (19:17 +0000)]
net: dwc_eth_qos: Stop spam of RX packet not available message

Remove spam of RX packet not available debug messages when waiting to
receive a packet.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
14 months agonet: dwc_eth_qos: Return error code when start fails
Jonas Karlman [Sun, 1 Oct 2023 19:17:17 +0000 (19:17 +0000)]
net: dwc_eth_qos: Return error code when start fails

Return error code when phy_connect fails or no link can be established.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
14 months agonet: dwc_eth_qos: Drop unused rx_pkt from eqos_priv
Jonas Karlman [Sun, 1 Oct 2023 19:17:16 +0000 (19:17 +0000)]
net: dwc_eth_qos: Drop unused rx_pkt from eqos_priv

rx_pkt is allocated and not used for anything, remove it.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
14 months agorockchip: Add support to generate LZMA compressed U-boot binary
Manoj Sai [Sun, 17 Sep 2023 19:26:28 +0000 (00:56 +0530)]
rockchip: Add support to generate LZMA compressed U-boot binary

Add support for generating a LZMA-compressed U-boot binary with the
help of binman, if CONFIG_SPL_LZMA is selected.

Signed-off-by: Manoj Sai <abbaraju.manojsai@amarulasolutions.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: Add support to generate GZIP compressed U-boot binary
Manoj Sai [Sun, 17 Sep 2023 19:26:27 +0000 (00:56 +0530)]
rockchip: Add support to generate GZIP compressed U-boot binary

Add support for generating a GZIP-compressed U-boot binary with the
help of binman, if CONFIG_SPL_GZIP is selected.

Signed-off-by: Manoj Sai <abbaraju.manojsai@amarulasolutions.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agospl: fit: support for booting a LZMA-compressed U-boot binary
Manoj Sai [Sun, 17 Sep 2023 19:26:26 +0000 (00:56 +0530)]
spl: fit: support for booting a LZMA-compressed U-boot binary

If LZMA Compression support is enabled, LZMA compressed U-Boot
binary will be placed at a specified RAM location which is
defined at CONFIG_SYS_LOAD_ADDR and will be assigned  as the
source address.

image_decomp() function, will decompress the LZMA compressed
U-Boot binary which is placed at source address(CONFIG_SYS_LOAD_ADDR)
to the default CONFIG_SYS_TEXT_BASE location.

spl_load_fit_image function will load the decompressed U-Boot
binary, which is placed at the CONFIG_SYS_TEXT_BASE location.

Signed-off-by: Manoj Sai <abbaraju.manojsai@amarulasolutions.com>
Signed-off-by: Suniel Mahesh <sunil@amarulasolutions.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
14 months agospl: fit: support for booting a GZIP-compressed U-boot binary
Manoj Sai [Sun, 17 Sep 2023 19:26:25 +0000 (00:56 +0530)]
spl: fit: support for booting a GZIP-compressed U-boot binary

If GZIP Compression support is enabled, GZIP compressed U-Boot binary
will be at a specified RAM location which is defined at
CONFIG_SYS_LOAD_ADDR and will be assign it as the source address.

gunzip function in spl_load_fit_image ,will decompress the GZIP
compressed U-Boot binary which is placed at
source address(CONFIG_SYS_LOAD_ADDR)  to the default
CONFIG_SYS_TEXT_BASE location.

spl_load_fit_image function will load the decompressed U-Boot
binary, which is placed at the CONFIG_SYS_TEXT_BASE location.

Signed-off-by: Manoj Sai <abbaraju.manojsai@amarulasolutions.com>
Signed-off-by: Suniel Mahesh <sunil@amarulasolutions.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
14 months agodoc: rockchip: Update and improve info on rk3308, TPL and TF-A
Massimo Pegorer [Sat, 9 Sep 2023 09:33:33 +0000 (11:33 +0200)]
doc: rockchip: Update and improve info on rk3308, TPL and TF-A

Update and improve documentation about build steps for SoCs that
require using TF-A and TPL binaries provided by Rockchip, such as
rk3308. Add rk3308 boards case to rST document. Add ROCK Pi S in
the list of supported boards. Minor page format improvements.

Signed-off-by: Massimo Pegorer <massimo.pegorer+oss@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: Kconfig: Enable external TPL binary for rk3308
Massimo Pegorer [Sat, 9 Sep 2023 09:33:24 +0000 (11:33 +0200)]
rockchip: Kconfig: Enable external TPL binary for rk3308

There is no support to initialize DRAM on rk3308 SoC using U-Boot
TPL or SPL, and therefore an external TPL binary must be used to
package a bootable u-boot-rockchip.bin image.

Default ROCKCHIP_EXTERNAL_TPL to yes if ROCKCHIP_RK3308.
Remove useless TPL_SERIAL.

Signed-off-by: Massimo Pegorer <massimo.pegorer+oss@gmail.com>
Tested-by: FUKAUMI Naoki <naoki@radxa.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: board: Add minimal generic RK3566/RK3568 board
Jonas Karlman [Mon, 21 Aug 2023 22:30:29 +0000 (22:30 +0000)]
rockchip: board: Add minimal generic RK3566/RK3568 board

Add a minimal generic RK3566/RK3568 board that only have eMMC and SDMMC
enabled. This defconfig can be used to boot from eMMC or SD-card on most
RK3566/RK3568 boards that follow reference board design.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: Port IO-domain driver for RK3568 from linux
Jonas Karlman [Mon, 21 Aug 2023 22:30:28 +0000 (22:30 +0000)]
rockchip: Port IO-domain driver for RK3568 from linux

Port the Rockchip IO-domain driver for RK3568 from linux.

The driver auto probe after bind to configure IO-domain based on the
regulator voltage. Compared to the linux driver this driver is not
notified about regulator voltage changes and only configure IO-domain
based on the initial voltage autoset by the regulator.

It is not recommended to enable MMC_IO_VOLTAGE or the mmc signal voltage
and IO-domain may end up out of sync.

Based on the linux commit 28b05a64e47c ("soc: rockchip: io-domain: add
rk3568 support").

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoregulator: rk8xx: Return correct voltage for switchout converters
shengfei Xu [Mon, 21 Aug 2023 22:30:26 +0000 (22:30 +0000)]
regulator: rk8xx: Return correct voltage for switchout converters

The voltage value for switchout converters is always reported as 0 uV.
When the switch is enabled, it's voltage is same as input supply.

Fix this by implementing get_value for switchout converters.

Fixes: ee30068fa574 ("power: pmic: rk809: support rk809 pmic")
Signed-off-by: shengfei Xu <xsf@rock-chips.com>
[jonas@kwiboo.se: fix checkpatch error, update commit message]
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoregulator: rk8xx: Return correct voltage for buck converters
Joseph Chen [Mon, 21 Aug 2023 22:30:25 +0000 (22:30 +0000)]
regulator: rk8xx: Return correct voltage for buck converters

Information from the first range group is always used to calculate the
voltage returned for buck converters. This may result in wrong voltage
reported back to the regulator_get_value caller.

Traverse all the possible BUCK ranges to fix this issue.

Fixes: addd062beacc ("power: pmic: rk816: support rk816 pmic")
Fixes: b62280745e55 ("power: pmic: rk805: support rk805 pmic")
Fixes: b4a35574b38d ("power: pmic: rk817: support rk817 pmic")
Fixes: ee30068fa574 ("power: pmic: rk809: support rk809 pmic")
Signed-off-by: Joseph Chen <chenjh@rock-chips.com>
[jonas@kwiboo.se: fix checkpatch error, simplify buck get_value, update commit message]
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agopower: regulator: Only run autoset once for each regulator
Jonas Karlman [Mon, 21 Aug 2023 22:30:24 +0000 (22:30 +0000)]
power: regulator: Only run autoset once for each regulator

With the commit 4fcba5d556b4 ("regulator: implement basic reference
counter"), keeping regulator enablement in balance become more important.
Calling regulator_autoset multiple times on a fixed regulator increase
the enable count for each call, resulting in an unbalanced enable count.

Introduce a AUTOSET_DONE flag and use it to mark that autoset has run
for the regulator. Return -EALREADY on any subsequent call to autoset.

This fixes so that the enable count is only ever increased by one per
regulator for autoset.

Fixes: 4fcba5d556b4 ("regulator: implement basic reference counter")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: rk3568-nanopi-r5: Enable PCIe on NanoPi R5C and R5S
Jonas Karlman [Wed, 2 Aug 2023 19:59:33 +0000 (19:59 +0000)]
rockchip: rk3568-nanopi-r5: Enable PCIe on NanoPi R5C and R5S

Enable missing PCIe Kconfig options now that PCIe bifurcation is fixed
to make use of the two on-board RTL8125B and the M.2 slot on NanoPi R5C
and NanoPi R5S.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agorockchip: rk3568-nanopi-r5: Update defconfig for NanoPi R5C and R5S
Jonas Karlman [Wed, 2 Aug 2023 19:49:46 +0000 (19:49 +0000)]
rockchip: rk3568-nanopi-r5: Update defconfig for NanoPi R5C and R5S

Update and sync Kconfig options for NanoPi R5C and NanoPi R5S with other
RK3568 boards.

SPL_FIT_SIGNATURE is enabled to add a checksum validation of the FIT
payload, also add LEGACY_IMAGE_FORMAT to keep boot scripts working.

OF_SPL_REMOVE_PROPS, SPL_DM_SEQ_ALIAS and SPL_PINCTRL change ensure
pinctrl for eMMC, SD-card and UART2 is applied in SPL.

MMC_HS200_SUPPORT and SPL counterpart is enabled to speed up eMMC load
times from on-board eMMC 5.1 modules.

Drop remaining unused or unsupported options to sync with other RK3568
boards.

Also sync device tree from linux v6.4 and drop u-boot,spl-boot-order and
use the default from rk356x-u-boot.dtsi.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agophy: rockchip: naneng-combphy: Use signal from comb PHY on RK3588
Jonas Karlman [Wed, 2 Aug 2023 19:41:22 +0000 (19:41 +0000)]
phy: rockchip: naneng-combphy: Use signal from comb PHY on RK3588

Route signal from comb PHY instead of PCIe3 PHY to PCIe1l0 and PCIe1l1.

Fixes use of pcie2x1l0 on ROCK 5B.

Code imported from mainline linux driver.

Fixes: c5b4a012bca8 ("phy: rockchip: naneng-combphy: Support rk3588")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agophy: rockchip: snps-pcie3: Add support for RK3588
Jonas Karlman [Wed, 2 Aug 2023 19:04:32 +0000 (19:04 +0000)]
phy: rockchip: snps-pcie3: Add support for RK3588

Add support for the RK3588 variant to the driver.

Code imported almost 1:1 from mainline linux driver.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agophy: rockchip: snps-pcie3: Add bifurcation support for RK3568
Jonas Karlman [Wed, 2 Aug 2023 19:28:33 +0000 (19:28 +0000)]
phy: rockchip: snps-pcie3: Add bifurcation support for RK3568

Configure aggregation or bifurcation mode on RK3568 based on the value
of data-lanes property.

Code imported almost 1:1 from mainline linux driver.

Fixes: 6ec62b6ca698 ("phy: rockchip: Add Rockchip Synopsys PCIe 3.0 PHY")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agophy: rockchip: snps-pcie3: Refactor to use a phy_init ops
Jonas Karlman [Wed, 2 Aug 2023 19:04:30 +0000 (19:04 +0000)]
phy: rockchip: snps-pcie3: Refactor to use a phy_init ops

Add a phy_init ops in preparation for upcoming support of a RK3588
variant in the driver.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agophy: rockchip: snps-pcie3: Refactor to use clk_bulk API
Jonas Karlman [Wed, 2 Aug 2023 19:04:29 +0000 (19:04 +0000)]
phy: rockchip: snps-pcie3: Refactor to use clk_bulk API

Change to use clk_bulk API and syscon_regmap_lookup_by_phandle to
simplify in preparation for upcoming support of a RK3588 variant.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agopci: pcie_dw_rockchip: Configure number of lanes and link width speed
Jonas Karlman [Wed, 2 Aug 2023 19:25:51 +0000 (19:25 +0000)]
pci: pcie_dw_rockchip: Configure number of lanes and link width speed

Set number of lanes and link width speed control register based on the
num-lanes property.

Code imported almost 1:1 from dw_pcie_setup in mainline linux.

Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
14 months agoMerge branch '2023-10-06-spl-prepare-for-universal-payload'
Tom Rini [Fri, 6 Oct 2023 21:23:47 +0000 (17:23 -0400)]
Merge branch '2023-10-06-spl-prepare-for-universal-payload'

To quote the author:
This series tidies up SPL a little and adds some core ofnode functions
needed to support Universal Payload. It also includes a few minor
fix-ups for sandbox.

For SPL the changes include CONFIG naming, removing various #ifdefs and
tidying up the FIT code.

One notable piece of the ofnode improvements is support for flattening a
livetree. This should be useful in future as we move FDT fixups to use
the ofnode API.

14 months agopci: serial: Support reading PCI-register size with base
Simon Glass [Tue, 26 Sep 2023 14:14:58 +0000 (08:14 -0600)]
pci: serial: Support reading PCI-register size with base

The PCI helpers read only the base address for a PCI region. In some cases
the size is needed as well, e.g. to pass along to a driver which needs to
know the size of its register area.

Update the functions to allow the size to be returned. For serial, record
the information and provided it with the serial_info() call.

A limitation still exists in that the size is not available when OF_LIVE
is enabled, so take account of that in the tests.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agodm: core: Tweak device_is_on_pci_bus() for code size
Simon Glass [Tue, 26 Sep 2023 14:14:57 +0000 (08:14 -0600)]
dm: core: Tweak device_is_on_pci_bus() for code size

This function cannot return true if PCI is not enabled, since no PCI
devices will have been bound. Add a check for this to reduce code size
where it is used.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agoserial: Drop ns16550 serial_getinfo() in SPL
Simon Glass [Tue, 26 Sep 2023 14:14:56 +0000 (08:14 -0600)]
serial: Drop ns16550 serial_getinfo() in SPL

This is typically not needed in SPL/TPL and increases the code size.
Drop it.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agospl: Add C-based runtime detection of SPL
Simon Glass [Tue, 26 Sep 2023 14:14:55 +0000 (08:14 -0600)]
spl: Add C-based runtime detection of SPL

The spl_phase() function indicates whether U-Boot is in SPL and before
or after relocation. But sometimes it is useful to check for SPL with
zero code-size impact. Since spl_phase() checks the global_data flags,
it does add a few bytes.

Add a new spl_in_proper() function to check if U-Boot proper is
running, regardless of the relocation status.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agocommand: Include a required header in command.h
Simon Glass [Wed, 27 Sep 2023 14:22:37 +0000 (08:22 -0600)]
command: Include a required header in command.h

This uses ARRAY_SIZE() but does not include the header file which declares
it. Fix this, so that command.h can be included without common.h

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
14 months agobloblist: Add missing name
Simon Glass [Tue, 26 Sep 2023 14:14:52 +0000 (08:14 -0600)]
bloblist: Add missing name

Add a missing bloblist name.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agobloblist: Support initing from multiple places
Simon Glass [Tue, 26 Sep 2023 14:14:51 +0000 (08:14 -0600)]
bloblist: Support initing from multiple places

Typically the bloblist is set up after the devicetree is present. This
makes sense because bloblist may use malloc() to allocate the space it
needs.

However sometimes the devicetree itself may be present in the bloblist.
In that case it is at a known location in memory so we can init the
bloblist very early, before devicetree.

Add a flag to indicate whether the bloblist has been inited. Add a
function to init it only if needed. Use that in the init sequence.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agosandbox: Move the bloblist down a little in memory
Simon Glass [Tue, 26 Sep 2023 14:14:50 +0000 (08:14 -0600)]
sandbox: Move the bloblist down a little in memory

Move this down by 4KB so that it is large enough to hold the devicetree.

Also fix up the devicetree address in the documetation while we are here.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agosandbox: Only read the state if we have a state file
Simon Glass [Tue, 26 Sep 2023 14:14:49 +0000 (08:14 -0600)]
sandbox: Only read the state if we have a state file

We should not read this unless requested. Make it conditional on the
option being provided.

Add some debugging to show the state being written.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agosandbox: Init the EC properly even if no state file is available
Simon Glass [Tue, 26 Sep 2023 14:14:48 +0000 (08:14 -0600)]
sandbox: Init the EC properly even if no state file is available

This currently relies on sandbox attempting to read a state file. At
present it always does, even when there is no state file, in which case it
fails, but still inits the EC.

That is a bug, so update this driver to set the current image always, even
if no state is read.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agosandbox: Move reading the RAM buffer into a better place
Simon Glass [Tue, 26 Sep 2023 14:14:47 +0000 (08:14 -0600)]
sandbox: Move reading the RAM buffer into a better place

This should not happen in the argument-parsing function. Move it to the
main program.

Add some debugging for reading/writing.

Signed-off-by: Simon Glass <sjg@chromium.org>
14 months agodm: core: Add tests for oftree_path()
Simon Glass [Tue, 26 Sep 2023 14:14:46 +0000 (08:14 -0600)]
dm: core: Add tests for oftree_path()

Add a few simple tests for getting the root node, since this is handled
as a special case in the implementation.

Signed-off-by: Simon Glass <sjg@chromium.org>