Tero Kristo [Fri, 11 Jun 2021 08:45:22 +0000 (11:45 +0300)]
arm: mach-k3: sysfw-loader: pass boardcfg to sciserver
Copy the contents of the board config loaded from sysfw.itb into an
EXTBOOT shared memory buffer that gets passed to sciserver. This only
needs to be done if EXTBOOT area has not been populated by ROM code yet.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:20 +0000 (11:45 +0300)]
arm: mach-k3: do board config for PM only if supported
If the raw PM support is built in, we are operating in the split
firmware approach mode where PM support is not available. In this
case, skip the board config for this.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Dave Gerlach [Fri, 11 Jun 2021 08:45:18 +0000 (11:45 +0300)]
arm: mach-k3: Add platform data for j721e and j7200
Add platform clock and powerdomain data for J721e and J7200. This data
is used by the corresponding drivers to register all the required device
clocks and powerdomains.
Signed-off-by: Dave Gerlach <d-gerlach@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:17 +0000 (11:45 +0300)]
tools: k3_fit_atf: add DM binary to the FIT image
Add DM (device manager) firmware image to the fit image that is loaded by
R5 SPL. This is needed with the HSM rearch where the firmware allocation
has been changed slightly.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:16 +0000 (11:45 +0300)]
cmd: ti: pd: Add debug command for K3 power domains
Add support command for debugging K3 power domains. This is useful with
the HSM rearch setup, where power domains are directly controlled by SPL
instead of going through the TI SCI layer. The debugging support is only
available in the u-boot codebase though, so the raw register access
power domain layer must be enabled on u-boot side for this to work. By
default, u-boot side uses the TI SCI layer, and R5 SPL only uses the
direct access methods.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org> Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Tero Kristo [Fri, 11 Jun 2021 08:45:15 +0000 (11:45 +0300)]
power: domain: Introduce driver for raw TI K3 PDs
Normally, power domains are handled via TI-SCI in K3 SoCs. However,
SPL is not going to have access to sysfw resources, so it must control
them directly. Add driver for supporting this.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org> Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Tero Kristo [Fri, 11 Jun 2021 08:45:14 +0000 (11:45 +0300)]
clk: add support for TI K3 SoC clocks
Add driver to support TI K3 generation SoC clocks. This driver registers
the clocks provided via platform data, and adds support for controlling
the clocks via DT handles.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:13 +0000 (11:45 +0300)]
clk: add support for TI K3 SoC PLL
Add support for TI K3 SoC PLLs. This clock type supports
enabling/disabling/setting and querying the clock rate for the PLL. The
euclidean library routine is used to calculate divider/multiplier rates
for the PLLs.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:12 +0000 (11:45 +0300)]
clk: fix set_rate to clean up cached rates for the hierarchy
Clock rates are cached within the individual clock nodes, and right now
if one changes a clock rate somewhere in the middle of the tree, none
of its child clocks notice the change. To fix this, clear up all the
cached rates for us and our child clocks.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:11 +0000 (11:45 +0300)]
clk: fix assigned-clocks to pass with deferring provider
If a clock provider is not ready for assigning default rates/parents
during its probe, it may return -EPROBE_DEFER directly from xlate.
Handle this special case properly by skipping the entry and adjusting the
return value to pass. The defaults will be handled properly in post probe
phase then.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:09 +0000 (11:45 +0300)]
clk: add support for setting clk rate from cmdline
Add new clk subcommand "clk setfreq", for setting up a clock rate
directly from u-boot cmdline. This is handy for any debugging purposes
towards clocks.
Acked-by: Lukasz Majewski <lukma@denx.de> Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:07 +0000 (11:45 +0300)]
clk: fix clock tree dump to properly dump out every registered clock
Some clocks are not associated to a DM node, so just parsing the DM is not
enough. This is especially true for root clocks, which typically don't have
any parents. Instead, fetch every registered UCLASS_CLK instance, and dump
these out.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:06 +0000 (11:45 +0300)]
clk: fixed_rate: add API for directly registering fixed rate clocks
Current driver only supports registering fixed rate clocks from DT. Add
new API which makes it possible to register fixed rate clocks directly
from e.g. platform specific clock drivers.
Reviewed-by: Peng Fan <peng.fan@nxp.com> Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Lokesh Vutla [Fri, 11 Jun 2021 08:45:05 +0000 (11:45 +0300)]
common: fit: Update board_fit_image_post_process() to pass fit and node_offset
board_fit_image_post_process() passes only start and size of the image,
but type of the image is not passed. So pass fit and node_offset, to
derive information about image to be processed.
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com> Reviewed-by: Tom Rini <trini@konsulko.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:04 +0000 (11:45 +0300)]
remoteproc: k3-r5: remove sysfw PM calls if not supported
With the sysfw rearch, sysfw PM calls are no longer available from SPL
level. To properly support this, remove the is_on checks and the reset
assertion from the R5 remoteproc driver as these are not supported.
Attempting to access unavailable services will cause the device to hang.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:03 +0000 (11:45 +0300)]
arm: mach-k3: introduce new config option for sysfw split
On J7 family of SoCs (J721E and J7200), sysfw is being split to be run
under two cores, TIFS portion on DMSC core, and DM firmware under MCU
R5. As MCU R5 is also used to run one phase of the bootloader, we must
prevent access from here towards sysfw services. To support this, add
new config option which can be used to detect presence of RM/PM sysfw
services.
Signed-off-by: Tero Kristo <t-kristo@ti.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Tero Kristo [Fri, 11 Jun 2021 08:45:02 +0000 (11:45 +0300)]
lib: rational: copy the rational fraction lib routines from Linux
Copy the best rational approximation calculation routines from Linux.
Typical usecase for these routines is to calculate the M/N divider
values for PLLs to reach a specific clock rate.
This is based on linux kernel commit:
"lib/math/rational.c: fix possible incorrect result from rational
fractions helper"
(sha1: 323dd2c3ed0641f49e89b4e420f9eef5d3d5a881)
Signed-off-by: Tero Kristo <t-kristo@ti.com> Reviewed-by: Tom Rini <trini@konsulko.com> Signed-off-by: Tero Kristo <kristo@kernel.org>
Dario Binacchi [Wed, 2 Jun 2021 20:38:04 +0000 (22:38 +0200)]
rtc: davinci: add driver model support
Update the driver to support the device tree and the driver model.
The read / write helpers in rtc_ops allow access to scratch registers
only. The offset parameter is added to the address of the scratch0
register.
Support for non-DM has been removed as there were no users.
Dario Binacchi [Wed, 2 Jun 2021 20:38:02 +0000 (22:38 +0200)]
rtc: davinci: use unlock/lock mechanism
The RTC module contains a kicker mechanism to prevent any spurious
writes from changing the register values. To set the time, you must
first unlock the TC registers, update them and then lock.
Dario Binacchi [Wed, 2 Jun 2021 20:38:01 +0000 (22:38 +0200)]
rtc: davinci: check BUSY bit before set TC registers
To write correct data to the TC registers, the STATUS register must be
read until the BUSY bit is equal to zero. Once the BUSY flag is zero,
there is a 15 μs access period in which the TC registers can be
programmed.
The rtc_wait_not_busy() has been inspired by the Kernel.
Dario Binacchi [Wed, 2 Jun 2021 20:38:00 +0000 (22:38 +0200)]
rtc: davinci: replace 32bit access with 8bit access
Use 32-bit access only where it is needed. Most of the RTC registers
contain useful information in the 8 least significant bits, the others
are reserved.
arm: dts: k3-am64-main: Update the location of ATF in SRAM and increase its max size
Due to a limitation for USB DFU boot mode, SPL load address has to be less
than or equal to 0x70001000. So, load address of SPL and ATF have been
moved to 0x70000000 and 0x701a0000 respectively.
Also, the maximum size of ATF has been increased to 0x1c000 [1].
Therefore, update ATF's location and maximum size accordingly in the device
tree file.
configs: am64x_evm_*_defconfig: Rearrange the components in SRAM to satisfy the limitations for USB DFU boot mode
For USB DFU boot mode there is a limitation on the load address of boot
images that they have to be less than 0x70001000. Therefore, move the
SPL_TEXT_BASE address to 0x70000000.
Currently ATF is being loaded at 0x70000000, if the SPL is being loaded at
0x70000000 then ATF would overwrite SPL image when loaded. Therefore, move
the location of ATF to a latter location in SRAM, past the SPL image. Also
rearrange the EEPROM and BSS data on top of ATF.
Given below is the placement of various data sections in SRAM
AM64 EVM board has a micro USB 2.0 AB connector and the USB0_VBUS is
connected with a resistor divider in between. USB0_DRVVBUS pin is muxed
between USB0_DRVVBUS and GPIO1_79 signals.
Add the corresponding properties and set the pinmux mode for USB subsystem
in the evm dts file.
arm: mach-k3: am642_init: Do USB fixups to facilitate host and device boot modes
U-Boot either supports USB host or device mode for a node at a time in the
device tree nodes. To support both host and dfu bootmodes, dr_mode is set
to "peripheral" by default and then fixed based on the mode selected by
the boot mode config dip switches on the board.
This needs to happen before the cdns3 generic layer binds the usb device
to a host or a device driver. Therefore, use fdtdec_setup_board()
implementation to fixup the device tree property.
usb: dfu: Migrate CONFIG_SYS_DFU_DATA_BUF_SIZE and CONFIG_SYS_DFU_MAX_FILE_SIZE to Kconfig
Currently the config options CONFIG_SYS_DFU_DATA_BUF_SIZE and
CONFIG_SYS_DFU_MAX_FILE_SIZE are being set in include/configs/<board>.h
files and also in <board_name>_defconfig files without a Kconfig option. It
is easier for users to set these configs in defconfig files than in config
header files as they are a part of the source code.
Add Kconfig symbols, and update the defconfigs by using tools/moveconfig.py
script.
Tero Kristo [Wed, 12 May 2021 08:03:04 +0000 (11:03 +0300)]
net: convert TFTP_TSIZE to proper Kconfig option
TFTP transfer size can be used to re-size the TFTP progress bar on
single line based on the server reported file size. The support for
this has been around from 2019, but it was never converted to proper
Kconfig.
While adding this new Kconfig, enable it by default for OMAP2+ and K3
devices also.
Tom Rini [Tue, 8 Jun 2021 21:18:35 +0000 (17:18 -0400)]
Merge branch '2021-06-08-display-and-logging-updates' into next
To quote Simon, first for the display changes:
At present we have two ways of showing a hex dump. Once has been in
U-Boot since the dawn of time and the other was recently added from
Linux.
They both have their own unique features.
This series makes a few changes to bring them closer together. It also
adds support for logging a buffer, which is useful since it can put it
through the same log drivers as other logging output.
Also it adds tests, so we can check the behaviour.
And then the logging changes:
At present when logging is not enabled, all log() calls become nops.
This does not seem right, since if the log level is high enough then
there should be some sort of message. So in that case, this series
updates it to print the message if the log level is above LOGL_INFO.
This mimics the behaviour for the log_...() macros like log_debug() and
log_info(), so we can drop the special case for these.
Also the current implementation does not support multiple log calls on
the same line nicely. The tags are repeated so the line is very hard to
read. This series adds that as a new feature.
Simon Glass [Sat, 8 May 2021 19:46:54 +0000 (13:46 -0600)]
log: Convert log values to printf() if not enabled
At present if logging not enabled, log_info() becomes a nop. But we want
log output at the 'info' level to be akin to printf(). Update the macro to
pass the output straight to printf() in this case.
This mimics the behaviour for the log_...() macros like log_debug() and
log_info(), so we can drop the special case for these.
Add new tests to cover this case. Signed-off-by: Simon Glass <sjg@chromium.org>
Simon Glass [Sat, 8 May 2021 19:46:53 +0000 (13:46 -0600)]
sandbox: log: Avoid build error with !CONFIG_LOG
The pr_cont_test.c test requires CONFIG_LOG since it directly accesses
fields in global_data that require it. Move the test into the CONFIG_LOG
condition to avoid build errors.
Enable CONFIG_LOG on sandbox (not sandbox_spl, etc.) so that we still run
this test. This requires resyncing of the configs.
Simon Glass [Sat, 8 May 2021 13:00:05 +0000 (07:00 -0600)]
display_options: Split print_buffer() into two functions
At present print_buffer() outputs a hex dump but it is not possible to
place this dump in a string. Refactor it into a top-level function which
does the printing and a utility function that dumps a line into a string.
This makes the code more generally useful.
Simon Glass [Sat, 8 May 2021 13:00:03 +0000 (07:00 -0600)]
hexdump: Support any rowsize
At present print_hex_dump() only supports either 16- or 32-byte lines.
With U-Boot we want to support any line length up to a maximum of 64.
Update the function to support this, with 0 defaulting to 16, as with
print_buffer().
Simon Glass [Sat, 8 May 2021 13:00:02 +0000 (07:00 -0600)]
hexdump: Add support for sandbox
The current implementation outputs an address as a pointer. Update the
code to use an address instead, respecting the 32/64 nature of the CPU.
Add some initial tests copied from print_test_display_buffer(), just the
ones that can pass with the current implementation.
Note that for this case print_hex_dump() and print_bufffer() produce the
same result. For now the tests are duplicated sine we have separate
functions.
Simon Glass [Sat, 8 May 2021 13:00:01 +0000 (07:00 -0600)]
hexdump: Move API to header file
Move the comments to the header file so people can find the function info
without digging in the implementation. Fix up the code style and add an
enum for the first arg.
Simon Glass [Sat, 8 May 2021 12:59:59 +0000 (06:59 -0600)]
test: Add a test for print_buffer()
Add a test for this function, to cover the various features. Expand the
expect_str length to take acount of the ~300-bytes lines generated in one
case.
Simon Glass [Sat, 8 May 2021 12:59:58 +0000 (06:59 -0600)]
test: Convert print tests to use ut framework
This test predates the test framework in U-Boot. It uses #define DEBUG and
assert() to check the result. Update it to use the framework so it can
report failure constitent with other tests.
Simon Glass [Sat, 8 May 2021 12:59:57 +0000 (06:59 -0600)]
test: Detect when expect_str is too small
If a line of more than 256 bytes is generated, the test will fail but the
reason is not clear. Add a check for this condition and print a helpful
message.
Simon Glass [Sat, 8 May 2021 12:59:56 +0000 (06:59 -0600)]
console: Report an error when output buffer is exhausted
If the console output buffer is exhausted, characters are silently dropped
from the end. Detect this condition and report an error when reading back
the characters.
Tom Rini [Mon, 7 Jun 2021 23:12:04 +0000 (19:12 -0400)]
Merge branch '2021-06-07-arm-cache-cp15-improvements' into next
To quote the series author, Patrick Delaunay:
On STM32MP15x platform we can use OP-TEE, loaded in DDR in a region
protected by a firewall. This region is reserved in the device with
the "no-map" property as defined in the binding file
doc/device-tree-bindings/reserved-memory/reserved-memory.txt.
Sometime the platform boot failed in U-Boot on a Cortex A7 access to
this region (depending of the binary and the issue can change with compiler
version or with code alignment), then the firewall raise an error,
for example:
After investigation, the forbidden access is a speculative request performed
by the Cortex A7 because all the DDR is mapped as MEMORY with CACHEABLE
property.
The issue is solved only when the region reserved by OP-TEE is no more
mapped in U-Boot as it is already done in Linux kernel.
Tested on DK2 board with OP-TEE 3.12 / TF-A 2.4:
With hard-coded address for OP-TEE reserved memory,
the error doesn't occur.
void dram_bank_mmu_setup(int bank)
{
....
for (i = start >> MMU_SECTION_SHIFT;
i < (start >> MMU_SECTION_SHIFT) + (size >> MMU_SECTION_SHIFT);
i++) {
option = DCACHE_DEFAULT_OPTION;
if (i >= 0xde0)
option = INVALID_ENTRY;
set_section_dcache(i, option);
}
}
Just by modifying the test on 0xde0 to 0xdf0, the OP-TEE memory protected
by firewall is mapped cacheable and the error occurs.
I think that it can be a general issue for ARM architecture: the "no-map" tag
of reserved memory in device should be respected by U-Boot if firewall
is configured before U-Boot execution.
But I don't propose a generic solution in
arm/lib/cache-cp15.c:dram_bank_mmu_setup()
because the device tree parsing done in lmb_init_and_reserve() takes a
long time when it is executed without data cache.
stm32mp: don't map the reserved region with no-map property
No more map the reserved region with "no-map" property by marking
the corresponding TLB entries with invalid entry (=0) to avoid
speculative access.
The device tree parsing done in lmb_init_and_reserve() takes a
long time when it is executed without data cache, so it is called in
enable_caches() before to disable it.
This patch fixes an issue where predictive read access on secure DDR
OP-TEE reserved area are caught by firewall.
image-fdt: save no-map parameter of reserve-memory
Save the 'no-map' information present in 'reserved-memory' node to allow
correct handling when the MMU is configured in board to avoid
speculative access.
This binding is defined in
doc/device-tree-bindings/reserved-memory/reserved-memory.txt
Additional properties:
...
no-map (optional) - empty property
- Indicates the operating system must not create a virtual mapping
of the region as part of its standard mapping of system memory,
nor permit speculative access to it under any circumstances other
than under the control of the device driver using the region.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Add "flags" in lmb_property to save the "no-map" property of
reserved region and a new function lmb_reserve_flags() to check
this flag.
The default allocation use flags = LMB_NONE.
The adjacent reserved memory region are merged only when they have
the same flags value.
This patch is partially based on flags support done in Linux kernel
mm/memblock .c (previously lmb.c); it is why LMB_NOMAP = 0x4, it is
aligned with MEMBLOCK_NOMAP value.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Patrick Delaunay [Fri, 21 May 2021 10:24:59 +0000 (12:24 +0200)]
net: luton: remove address translation after ofnode_read_resource
Removed call of ofnode_translate_address() after ofnode_read_resource
in luton_switch.c:luton_probe(); it is unnecessary since
the commit feb7ac457c20 ("dm: core: Add address translation in
fdt_get_resource").
Fixes: feb7ac457c20 ("dm: core: Add address translation in fdt_get_resource") Reviewed-by: Ramon Fried <rfried.dev@gmail.com> Reported-by: Horatiu Vultur <horatiu.vultur@microchip.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
pwm: cros_ec: Rename "priv_auto_alloc_size" to "priv_auto"
With commit 41575d8e4c33 ("dm: treewide: Rename auto_alloc_size members
to be shorter") "priv_auto_alloc_size" was renamed to "priv_auto". This
driver was sent to the mailing list before that change, merged after it,
and still has the old form. Apply the rename here as well.
Fixes: 1b9ee2882e6b ("pwm: Add a driver for Chrome OS EC PWM") Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
os_find_text_base() assumes that first line of /proc/self/maps holds
information about the text. Hence we must call the function before calling
os_malloc() which calls mmap(0x10000000,).
Failure to do so has led to incorrect values for pc_reloc when an
exception was reported
=> exception undefined
Illegal instruction
pc = 0x5628d82e9d3c, pc_reloc = 0x5628c82e9d3c
as well as incorrect output of the bdinfo command
=> bdinfo
relocaddr = 0x0000000007858000
reloc off = 0x0000000010000000
Fixes: b308d9fd18fa ("sandbox: Avoid using malloc() for system state") Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Bin Meng [Fri, 30 Apr 2021 13:16:58 +0000 (21:16 +0800)]
of: addr: Translate 'dma-ranges' for parent nodes missing 'dma-ranges'
'dma-ranges' frequently exists without parent nodes having 'dma-ranges'.
While this is an error for 'ranges', this is fine because DMA capable
devices always have a translatable DMA address. Also, with no
'dma-ranges' at all, the assumption is that DMA addresses are 1:1 with
no restrictions unless perhaps the device itself has implicit
restrictions.
This keeps in sync with Linux kernel commit:
81db12ee15cb: of/address: Translate 'dma-ranges' for parent nodes missing 'dma-ranges'
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Fri, 28 May 2021 08:00:49 +0000 (10:00 +0200)]
arm: mvebu: turris_omnia: support invoking rescue boot from console
Make it possible to invoke rescue boot from U-Boot console, without
having to press the factory reset button. This is needed when accessing
the device remotely, for example.
Achieve this by putting rescue command into `bootcmd_rescue` default
environment variable and setting some distroboot environment variables
to their default values when the factory button is pressed.
Rescue boot from console can be invoked by running
run bootcmd_rescue
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 26 May 2021 15:59:40 +0000 (17:59 +0200)]
arm: a37xx: pci: Fix configuring PCIe resources
The `ranges` DT property of the PCIe node is currently ignored by
Aardvark driver - all entries are used as transparent PCIe MEM, despite
some of them being defined for IO in DT.
This is because the driver does not setup PCIe outbound windows and thus
a default configuration is used.
This can cause an external abort on CPU when a device driver tries to
access non-MEM space.
Setup the PCIe windows according to the `ranges` property for all
non-MEM resources (currently only IO) and also non-transparent MEM
resources.
Because Linux expects that bootloader does not setup Aardvark PCIe
windows, disable them before booting Linux.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 26 May 2021 15:59:39 +0000 (17:59 +0200)]
arm: a37xx: pci: Increase PCIe MEM size from 16 MiB to 127 MiB
For some configurations with more PCIe cards and PCIe bridges, 16 MiB of
PCIe MEM space may not be enough. Since TF-A already allocates a 128 MiB
CPU window for PCIe, and since IO port space is only 64 KiB in total,
use all the remaining space (64 + 32 + 16 + 8 + 4 + 2 + 1 = 127 MiB) for
PCIe MEM.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 26 May 2021 15:59:38 +0000 (17:59 +0200)]
arm: a37xx: pci: Fix a3700_fdt_fix_pcie_regions() function
Current version of this function uses a lot of incorrect assumptions about
the `ranges` DT property:
* parent(#address-cells) == 2
* #size-cells == 2
* number of entries == 2
* address size of first entry == 0x1000000
* second child address entry == base + 0x1000000
Trying to increase PCIe MEM space to more than 16 MiB leads to an overlap
with PCIe IO space, and trying to define additional MEM space (as a third
entry in the `ranges` DT property) causes U-Boot to crash when booting the
kernel.
## Flattened Device Tree blob at 04f00000
Booting using the fdt blob at 0x4f00000
Loading Device Tree to 000000001fb01000, end 000000001fb08f12 ... OK
ERROR: board-specific fdt fixup failed: <unknown error>
- must RESET the board to recover.
Fix a3700_fdt_fix_pcie_regions() to properly parse and update all addresses
in the `ranges` property according to
https://elinux.org/Device_Tree_Usage#PCI_Address_Translation
Now it is possible to increase PCIe MEM space from 16 MiB to maximal value
of 127 MiB.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Fixes: cb2ddb291ee6 ("arm64: mvebu: a37xx: add device-tree fixer for PCIe regions") Reviewed-by: Stefan Roese <sr@denx.de>
Change DT compatible string for A3700 PCIe from 'marvell,armada-37xx-pcie'
to 'marvell,armada-3700-pcie' to make U-Boot A3700 PCIe DT node compatible
with Linux' DT node.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 26 May 2021 15:59:35 +0000 (17:59 +0200)]
arm: a37xx: pci: Disable bus mastering when unloading driver
Disable Root Bridge I/O space, memory space and bus mastering in Aardvark's
remove method, which is called before booting Linux kernel.
This ensures that PCIe device which was initialized and used by U-Boot
cannot do new DMA transfers until Linux initializes PCI subsystem and loads
appropriate drivers for the device.
During initialization of PCI subsystem Linux in fact disables this bus
mastering on Root Bridge (and later enables it when driver is loaded and
configured), but there is a possibility of a small window after U-Boot
boots Linux when bus mastering is enabled, which is not correct.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 26 May 2021 15:59:34 +0000 (17:59 +0200)]
arm: a37xx: pci: Don't put link into LTSSM Recovery state during probe
During our debugging of the Aardvark driver in Linux we have discovered
that the PCIE_CORE_LINK_CTRL_STAT_REG register in fact controls standard
PCIe Link Control Register for PCIe Root Bridge. This led us to discover
that the name of the PCIE_CORE_LINK_TRAINING macro and the corresponding
comment by this macro's usage is misleading; this bit in fact controls
Retrain Link, which, according to PCIe base spec is defined as:
A write of 1b to this bit initiates Link retraining by directing the
Physical Layer LTSSM to the Recovery state. If the LTSSM is already in
Recovery or Configuration, re-entering Recovery is permitted but not
required.
Entering Recovery state is normally done from LTSSM L0, L0s and L1 states.
But since the pci-aardvark.c driver enables Link Training just a few lines
above, the controller is not in L0 ready state yet. So setting aardvark bit
PCIE_CORE_LINK_TRAINING does not actually enter Recovery state at this
place.
Moreover, trying to enter LTSSM Recovery state without other configuration
is causing issues for some cards (e.g. Atheros AR9xxx and QCA9xxx). Since
Recovery state is not entered, these issues are not triggered.
Remove code which tries to enter LTSSM Recovery state completely.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Sean Anderson [Thu, 27 May 2021 16:02:34 +0000 (12:02 -0400)]
fastboot: Fix overflow when calculating chunk size
If a chunk was larger than 4GiB, then chunk_data_sz would overflow and
blkcnt would not be calculated correctly. Upgrade it to a u64 and cast
its multiplicands as well. Also fix bytes_written while we're at it.
Signed-off-by: Sean Anderson <sean.anderson@seco.com> Reviewed-by: Heiko Schocher <hs@denx.de>
This contains the fix to bring back the SD card as MMC0. In the long run
we are looking into a more robust solution, but for now we need to fix
this, as this breaks the user experience left, right, and centre.
Also add the one MAINTAINERS path addition from Samuel.
Signed-off-by: Green Wan <green.wan@sifive.com> Reviewed-by: Sean Anderson <seanga2@gmail.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Rick Chen <rick@andestech.com>
projects / u-boot.git / log