Tom Rini [Sat, 15 Jan 2022 12:34:46 +0000 (07:34 -0500)]
Merge branch '2022-01-14-assorted-fixes'
- A number of fixes in various subsystems. This includes having the phy
uclass track power-on and init counts as this should resolve some
tricky functional problems on a number of platforms.
Eugen Hristev [Tue, 4 Jan 2022 16:20:19 +0000 (18:20 +0200)]
lib: Kconfig: fix PHANDLE_CHECK_SEQ position outside of menu
CONFIG_PHANDLE_CHECK_SEQ is outside of the menu 'Library routines'
thus it's invisible in menuconfig and cannot be selected.
Fix this by moving the 'endmenu' after the PHANDLE_CHECK_SEQ definition
Fixes: c589132a1d ("fdt: Use phandle to distinguish DT nodes with same name") Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com> Reviewed-by: Aswath Govindraju <a-govindraju@ti.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Peter Robinson [Sun, 2 Jan 2022 11:38:35 +0000 (11:38 +0000)]
env: fat: Add new lines at the end of print statements
Add some new line feeds at the end of print messages to make things
easier to read on the console. The other env options do this so
this is just an omission for FAT env.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
On boards using the RK3399 SoC, the USB OHCI and EHCI controllers share
the same PHY device instance. While these controllers are being stopped
they both attempt to power-off and deinitialize it, but trying to
power-off the deinitialized PHY device results in a hang. This usually
happens just before booting an OS, and can be explicitly triggered by
running "usb start; usb stop" in the U-Boot shell.
Implement a uclass-wide counting mechanism for PHY initialization and
power state change requests, so that we don't power-off/deinitialize a
PHY instance until all of its users want it done. The Allwinner A10 USB
PHY driver does this counting in-driver, remove those parts in favour of
this in-uclass implementation.
The sandbox PHY operations test needs some changes since the uclass will
no longer call into the drivers for actions matching its tracked state
(e.g. powering-off a powered-off PHY). Update that test, and add a new
one which simulates multiple users of a single PHY.
The major complication here is that PHY handles aren't deduplicated per
instance, so the obvious idea of putting the counts in the PHY handles
don't immediately work. It seems possible to bind a child udevice per
PHY instance to the PHY provider and deduplicate the handles in each
child's uclass-private areas, like in the CLK framework. An alternative
approach could be to use those bound child udevices themselves as the
PHY handles. Instead, to avoid the architectural changes those would
require, this patch solves things by dynamically allocating a list of
structs (one per instance) in the provider's uclass-private area.
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org> Tested-by: Peter Robinson <pbrobinson@gmail.com> - Rock960
Mark Kettenis [Tue, 21 Dec 2021 16:31:50 +0000 (17:31 +0100)]
armv8: apple: Disable PSCI reset
Apple's ARMv8 cores don't implement EL3 and therefore don't
provide a PSCI implementation. So don't attempt to use
PSCI to reset on machines using Apple SoCs.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
This happens when nvme_print_info() tries to return to the caller. It
looks like this error is caused by trying to allocate 8 KiB of memory
on the stack by the two uses of ALLOC_CACHE_ALIGN_BUFFER().
Use malloc_cache_aligned() to allocate this memory dynamically instead.
This fixes 'nvme detail' on Turris Omnia.
Note that similar change was applied to file drivers/nvme/nvme.c in past by
commit 2f83481dff9c ("nvme: use page-aligned buffer for identify command").
Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Marek Behún <marek.behun@nic.cz>
Attempt to handle cases with a downstream port of a PCIe switch where
link training never completes and the link continues switching between
speeds indefinitely with the data link layer never reaching the active
state.
It has been observed with a downstream port of the ASMedia ASM2824 Gen 3
switch wired to the upstream port of the Pericom PI7C9X2G304 Gen 2
switch, using a Delock Riser Card PCI Express x1 > 2 x PCIe x1 device,
P/N 41433, wired to a SiFive HiFive Unmatched board. In this setup the
switches are supposed to negotiate the link speed of preferably 5.0GT/s,
falling back to 2.5GT/s.
However the link continues oscillating between the two speeds, at the
rate of 34-35 times per second, with link training reported repeatedly
active ~84% of the time, e.g.:
Make use of this observation then and attempt to detect the inability to
negotiate the link speed automatically, and then handle it by hand. Use
the Data Link Layer Link Active status flag as the primary indicator of
successful link speed negotiation, but given that the flag is optional
by hardware to implement (the ASM2824 does have it though), resort to
checking for the mandatory Link Bandwidth Management Status flag showing
that the link speed or width has been changed in an attempt to correct
unreliable link operation (the ASM2824 does set it too).
If these checks indicate that link may not operate correctly, then poll
the Data Link Layer Link Active status flag along with the Link Training
flag for the duration of 200ms to see if the link has stabilised, that
is either that the Data Link Layer Link Active status flag has been set
or that Link Training has been inactive during at least the second half
of the interval.
If that has indicated failure, restrict the target speed to 2.5GT/s,
request a link retrain and check again if the link has stabilised. If
that does not work either, then restore the original speed setting and
claim defeat, otherwise we are done.
NB interestingly enough with the ASM2824 vs PI7C9X2G304 configuration
referred above asking the ASM2824 to retrain with a higher target link
speed once the 2.5GT/s speed has been negotiated makes the two devices
successfully negotiate 5.0GT/s. Lifting the 2.5GT/s speed restriction
would however prevent our workaround from working with an OS that issues
a reset and that is unaware of the problem. This is because the devices
would then try to negotiate a higher link speed from scratch and fail,
while the sticky property of the Target Link Speed setting will keep the
2.5GT/s speed restriction across a reset.
Keep the 2.5GT/s speed restriction then, conservatively, if functional
once applied.
Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk> Reviewed-by: Stefan Roese <sr@denx.de>
qianfan Zhao [Tue, 16 Nov 2021 01:35:38 +0000 (09:35 +0800)]
lib: sparse: Make CHUNK_TYPE_RAW buffer aligned
CHUNK_TYPE_RAW buffer is not aligned, and flash sparse images by
fastboot will report "Misaligned operation" if DCACHE is enabled.
Flashing Sparse Image
CACHE: Misaligned operation at range [84000028, 84001028]
CACHE: Misaligned operation at range [84001034, 84002034]
CACHE: Misaligned operation at range [8401104c, 8401304c]
Fix it
Signed-off-by: qianfan Zhao <qianfanguijin@163.com> Reviewed-by: Sean Anderson <sean.anderson@seco.com>
Pali Rohár [Wed, 12 Jan 2022 17:32:08 +0000 (18:32 +0100)]
arm: mvebu: Replace hardcoded values 0x0030/0x4030 by proper calculation
These hardcoded values were calculated from CONFIG_SPL_TEXT_BASE macro. Now
this macro is configurable via Kconfig, so calculate values 0x0030/0x4030
at compile time via CONFIG_SPL_TEXT_BASE option. Values 0x0030/0x4030
represents offset of CONFIG_SPL_TEXT_BASE from address 0x40000000.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 12 Jan 2022 17:30:09 +0000 (18:30 +0100)]
arm: mvebu: db-88f6720: Fix CONFIG_SPL_TEXT_BASE and remove wrong memory layout
Memory layout in the comment is from Armada XP platform which uses load
address 0x40004030. DB-88f6720 is Armada 375 platform which uses same load
address as Armada 38x which is 0x40000030.
Currently SPL support for Armada 375 is unfinished and does not work. There
is missing Serdes initialization and DDR3 training code. So nobody noticed
that CONFIG_SPL_* options are not correct.
Fix at least CONFIG_SPL_TEXT_BASE constant and remove incorrect comments
about memory layout. So it is not misleading.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Do not check for kwbimage configuration file when just showing information
about existing kwbimage file.
The check for kwbimage configuration file is required only when creating
kwbimage, not when showing information about image or when extracting data
from image.
With this change, it is possible to call mkimage -l and dumpimage -l also
for existing kwbimage file.
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, 12 Jan 2022 17:20:53 +0000 (18:20 +0100)]
tools: kwbimage: Extract main data image without -p arg for dumpimage
When there is no -p argument for dumpimage tool specified, extract the main
data image from kwbimage file. This makes dumpimage consistent with other
image formats.
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, 12 Jan 2022 17:20:52 +0000 (18:20 +0100)]
tools: kwbimage/kwboot: Check ext field for non-zero value
Despite the official specification, BootROM does not look at the lowest bit
of ext field but rather checks if ext field is non-zero.
Moreover original Marvell doimage tool puts into the mhdr->ext field the
number of extended headers, so basically it sets ext filed to non-zero
value if some extended header is present.
Fix U-Boot dumpimage and kwboot tools to parse correctly also kwbimage
files created by Marvell doimage tool, in the same way as the BootROM is
doing it when booting these images.
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, 12 Jan 2022 17:20:50 +0000 (18:20 +0100)]
tools: kwbimage: Dump kwbimage config file on '-p -1' option
To regenerate kwbimage from existing image, it is needed to have kwbimage
config file. Add a new option to generate kwbimage config file from
existing kwbimage when '-p 1' option is given.
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, 12 Jan 2022 17:20:49 +0000 (18:20 +0100)]
tools: kwbimage: Show binary image offset in mkimage -l, in addition to size
For debugging purposes it is good to know where the binary image would be
loaded and also it is needed to know if printed size is image size or the
size of header together with image.
Make it unambiguous by showing that printed size is not the size of the
whole header, but only the size of executable code, and print also the
executable offset of this binary image. Load/execute address is the offset
relative to the base address (either 0x40004000 or 0x40000000).
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 12 Jan 2022 17:20:47 +0000 (18:20 +0100)]
arm: mvebu: Enable BootROM output on A38x
BootROMs on pre-A38x SoCs enabled its output on UART by default, but A38x'
BootROM has its output on UART disabled by default.
To enable BootROM output on A38x SoC, it is required to set DEBUG flag
(which only enables BootROM output and nothing more) in kwbimage. For UART
images this DEBUG flag is ignored by BootROM.
Enable kwbimage DEBUG flag for all A38x boards.
With this change BootROM prints the following (success) information on UART
before booting U-Boot kwbimage:
BootROM - 1.73
Booting from SPI flash
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, 12 Jan 2022 17:20:46 +0000 (18:20 +0100)]
tools: kwbimage: Enforce 128-bit boundary alignment only for Sheeva CPU
This alignment is required only for platforms based on Sheeva CPU core
which are A370 and AXP. Now when U-Boot build system correctly propagates
LOAD_ADDRESS there is no need to have enabled 128-bit boundary alignment on
platforms which do not need it. Previously it was required because load
address was implicitly rounded to 128-bit boundary and U-Boot build system
expected it and misused it. Now with explicit setting of LOAD_ADDRESS there
is no guessing for load address anymore.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 12 Jan 2022 17:20:45 +0000 (18:20 +0100)]
arm: mvebu: Correctly set LOAD_ADDRESS for U-Boot SPL binary in kwbimage
U-Boot SPL for mvebu platform is not compiled as position independent.
Therefore it is required to instruct BootROM to load U-Boot SPL at the
correct address. Loading of kwbimage binary code at specific address can be
now achieved by the new LOAD_ADDRESS token as part of BINARY command in
kwbimage config file.
Update mvebu Makefile to put value of $(CONFIG_SPL_TEXT_BASE) into
LOAD_ADDRESS token when generating kwbimage.cfg from kwbimage.cfg.in.
It is required to update regex for sed to find replacement tokens at any
position on a line in kwbimage config file and not only at the beginning of
the line. This is because LOAD_ADDRESS is specified at the end of line
containing the BINARY command.
It looks like all Armada boards set CONFIG_SPL_TEXT_BASE to value
0x40004030 or 0x40000030. Why this value? It is because main kwbimage
header is at address 0x40004030 or 0x40000000 and it is 32 bytes long.
After the main header there is the binary header, which consist of 1 byte
for type, 3 bytes for size, 1 byte for number of arguments, 3 reserved
bytes and then 4 bytes for each argument. After these arguments comes the
executable code.
So arguments start at address 0x40004028 or 0x40000028. Before commit e6571f38c943 ("arm: mvebu: Remove dummy BIN header arguments for SPL
binary") there were two (dummy) arguments, which resulted in load address
of 0x40004030 or 0x40000030, always. After that commit (which removed dummy
arguments), load address stayed same due to the 128-bit alignment done by
mkimage.
This patch now reflects the dependency between $(CONFIG_SPL_TEXT_BASE),
load address and dummy kwbimage arguments, and allows the user to adjust
$(CONFIG_SPL_TEXT_BASE) config option to some other value.
For unsupported values, when mkimage/kwbimage cannot set chosen load address
as specified by $(CONFIG_SPL_TEXT_BASE), the build process now fails,
instead of silently generating non-working kwbimage.
Removal of this alignment between $(CONFIG_SPL_TEXT_BASE) and LOAD_ADDRESS
can only be done by compiling U-Boot SPL as position independent. But this
currently is not possible for 32-bit ARM version of U-Boot SPL.
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, 12 Jan 2022 17:20:41 +0000 (18:20 +0100)]
tools: kwbimage: Add support for specifying LOAD_ADDRESS for BINARY command
ARM executable code included in kwbimage binary header, which is not
position independent, needs to be loaded and executed by BootROM at the
correct fixed address.
Armada BootROMs load kwbimage header (in which the executable code is also
stored) at fixed address 0x40004000 or 0x40000000 which is mapped to
L2-SRAM (L2 Cache as SRAM). Address 0x40004000 is used on Armada platforms
with Sheeva CPU core (A370 and AXP) where BootROM uses MMU with 0x4000
bytes for MMU translation table. Address 0x40000000 is used on all other
platforms.
Thus the only way to specify load and execute address of this executable
code in binary kwbimage header is by filling dummy arguments into the
binary header, using the same mechanism we already have for achieving
128-bit boundary alignment on A370 and AXP SoCs.
Extend kwbimage config file parser to allow to specify load address as
part of BINARY command with syntax:
If the specified load address is invalid or cannot be used, mkimage will
throw fatal error and exit. This will prevent generating kwbimage with
invalid load address for non-position independent binary code.
If no load address is specified, kwbimage will not fill any the dummy
arguments, thus it will behave the same as before this change.
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, 12 Jan 2022 17:20:39 +0000 (18:20 +0100)]
arm: mvebu: Generate kwbimage.cfg with $(call cmd, ...)
Usage of $(call cmd,...) is standard way to call other commands which
generate things.
It also has the advantage of printing build information in the form
KWBCFG arch/arm/mach-mvebu/kwbimage.cfg
if verbosity is disabled, and printing the build command otherwise.
Note that the '#' character needs to be escaped in Makefile when used as
value for make variable assignment.
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, 12 Jan 2022 17:20:37 +0000 (18:20 +0100)]
tools: kwbimage: Fix generating image with multiple DATA_DELAY commands
Register set header consists of sequence of DATA commands followed by
exactly one DATA_DELAY command. Thus if we are generating image with
multiple DATA_DELAY commands, we need to create more register set headers.
Fix calculation of image size with multiple DATA_DELAY commands and
correctly set pointer to struct register_set_hdr_v1 when initializing new
register set header.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Marek Behún [Wed, 12 Jan 2022 16:06:59 +0000 (17:06 +0100)]
ddr: marvell: a38x: fix SPLIT_OUT_MIX state decision
This is a cleaned up and fixed version of a patch
mv_ddr: a380: fix SPLIT_OUT_MIX state decision
in each pattern cycle the bus state can be changed
in order to avoide it, need to back to the same bus state on each
pattern cycle
by
Moti Boskula <motib@marvell.com>
The original patch is not in Marvell's mv-ddr-marvell repository. It was
gives to us by Marvell to fix an issues with DDR training on some
boards, but it cannot be applied as is to mv-ddr-marvell, because it is
a very dirty draft patch that would certainly break other things, mainly
DDR4 training code in mv-ddr-marvell, since it changes common functions.
I have cleaned up the patch and removed stuff that seemed unnecessary
(when removed, it still fixed things). Note that I don't understand
completely what the code does exactly, since I haven't studied the DDR
training code extensively (and I suspect that no one besides some few
people in Marvell understand the code completely).
Anyway after the cleanup the patch still fixes isssues with DDR training
on the failing boards.
There was also a problem with the original patch on some of the Allied
Telesis' x530 boards, reported by Chris Packham. I have asked Chris to
send me some logs, and managed to fix it:
- if you look at the change, you'll notice that it introduces
subtraction of cur_start_win[] and cur_end_win[] members, depending on
a bit set in the current_byte_status variable
- the original patch subtracted cur_start_win[] if either
BYTE_SPLIT_OUT_MIX or BYTE_HOMOGENEOUS_SPLIT_OUT bits were set, but
subtracted cur_end_win[] only if the first one (BYTE_SPLIT_OUT_MIX)
was set
- from Chris Packham logs I discovered that the x530 board where the
original patch introduced DDR training failure, only the
BYTE_HOMOGENEOUS_SPLIT_OUT bit was set, and on our boards where the
patch is needed only the BYTE_SPLIT_OUT_MIX is set in the
current_byte_status variable
- this led me to the hypothesis that both cur_start_win[] and
cur_end_win[] should be subtracted only if BYTE_SPLIT_OUT_MIX bit is
set, the BYTE_HOMOGENEOUS_SPLIT_OUT bit shouldn't be considered at all
- this hypothesis also gains credibility when considering the commit
title ("fix SPLIT_OUT_MIX state decision")
Hopefully this will fix things without breaking anything else.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de> Tested-by: Chris Packham <judge.packham@gmail.com>
Marek Behún [Tue, 4 Jan 2022 15:14:54 +0000 (16:14 +0100)]
board: gdsys: Drop Dirk Eibach from MAINTAINERS
I got an
<dirk.eibach@gdsys.cc>: host mxlb.ispgateway.de[80.67.18.126] said:
554 Sorry, no mailbox here by that name. (in reply to RCPT TO command)
when sending e-mail to dirk.eibach@gdsys.cc.
Drop Dirk Eibach from MAINTAINERS of board/gdsys/a38x and
board/gdsys/mpc8308. The latter would be left maintainerless, add
Mario Six <mario.six@gdsys.cc> (he is also maintainer of the former
board).
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
Marek Behún [Tue, 4 Jan 2022 14:57:49 +0000 (15:57 +0100)]
ddr: marvell: a38x: Fix Synchronous vs Asynchronous mode determination
Before commit 4c289425752f ("mv_ddr: a38x: add support for ddr async
mode"), Asynchornous Mode was only used when the CPU Subsystem Clock
Options[4:0] field in the SAR1 register was set to value 0x13: CPU at
2 GHz and DDR at 933 MHz.
Then commit 4c289425752f ("mv_ddr: a38x: add support for ddr async
mode") added support for Asynchornous Modes with frequencies other than
933 MHz (but at least 467 MHz), but the code it added to check for
whether Asynchornous Mode should be used is wrong: it checks whether the
frequency setting in board DDR topology map is set to value other than
MV_DDR_FREQ_SAR.
Thus boards which define a specific value, greater than 400 MHz, for DDR
frequency in their board topology (e.g. Turris Omnia defines
MV_DDR_FREQ_800), are incorrectly put into Asynchornous Mode after that
commit.
The A38x Functional Specification, section 10.12 DRAM Clocking, says:
In Synchornous mode, the DRAM and CPU clocks are edge aligned and run
in 1:2 or 1:3 CPU to DRAM frequency ratios.
Change the check for whether Asynchornous Mode should be used according
to this explanation in Functional Specification.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Tested-by: Chris Packham <judge.packham@gmail.com> Reviewed-by: Stefan Roese <sr@denx.de>
Pierre Bourdon [Sat, 25 Dec 2021 04:46:29 +0000 (05:46 +0100)]
mtd: nand: pxa3xx: use marvell, prefix for custom DT properties
The DT properties for the "enable-arbiter" and "keep-config" config
knobs were previously named inconsistently:
- The u-boot driver used "nand-enable-arbiter" and "nand-keep-config"
names, without Marvell prefixes.
- The Linux driver uses "marvell,nand-keep-config" ("enable-arbiter"
does not exist anymore in recent kernels, but it also used to be
"marvell,nand-enable-arbiter").
- The device trees almost all use "marvell," prefixed names, except for
one single instance of "nand-enable-arbiter" without vendor prefix.
This commit standardizes on the vendor prefixed version, making the
u-boot driver read from DT props "marvell,nand-enable-arbiter" and
"marvell,nand-keep-config". The one device tree using the unprefixed
version is also changed to use the new naming.
This has the side effect of making the previously no-op "marvell,"
config knobs already present in some DTs actually do something. This was
likely the original intention of the DT authors, but note that this
commit was not tested on every single impacted board.
Pali Rohár [Tue, 21 Dec 2021 11:20:19 +0000 (12:20 +0100)]
arm: mvebu: a38x: serdes: Move non-serdes PCIe code to pci_mvebu.c
As explained in commit 3bedbcc3aa18 ("arm: mvebu: a38x: serdes: Don't
overwrite read-only SAR PCIe registers") it is required to set Maximum Link
Width bits of PCIe Root Port Link Capabilities Register depending of number
of used serdes lanes. As this register is part of PCIe address space and
not serdes address space, move it into pci_mvebu.c driver.
Read number of PCIe lanes from DT property "num-lanes" which is used also
by other PCIe controller drivers in Linux kernel. If this property is
absent then it defaults to 1. This property needs to be set to 4 for every
mvebu board which use PEX_ROOT_COMPLEX_X4 or PEX_BUS_MODE_X4.
Enabling of PCIe port needs to be done afer all registers in PCIe address
space are properly configure. For this purpose use new mvebu-reset driver
(part of system-controller) and remove this code from serdes code.
Because some PCIe ports cannot be enabled individually, it is required to
first setup all PCIe ports and then enable them.
This change contains also all required "num-lanes" and "resets" DTS
properties, to make pci_mvebu.c driver work correctly.
Pali Rohár [Tue, 21 Dec 2021 11:20:18 +0000 (12:20 +0100)]
arm: mvebu: Implement simple mvebu-reset driver for enabling/disabling PCIe ports
Enabling and disabling PCIe ports is done via address space of system
controller. All 32-bit Armada SoCs use low 4 bits in SoC Control 1 Register
for enabling and disabling some or more PCIe ports. Correct mapping needs
to be set in particular DTS files.
DT API for mvebu-reset is prepared for implementing resets also for other
HW blocks, but currently only PCIe is implemented via index 0.
Currently this driver is not used as PCIe ports are automatically enabled
by SerDes code executed by U-Boot SPL. But this will change in followup
patches.
Pali Rohár [Tue, 21 Dec 2021 11:20:17 +0000 (12:20 +0100)]
pci: pci_mvebu: Wait 100ms for Link Up in mvebu_pcie_probe()
After function mvebu_pcie_probe() returns U-Boot DM expects that PCIe link
is already up. In followup patches link initialization will be moved from
SPL to proper and therefore explicitly link up delay is required.
Delay mvebu_pcie_probe() for 100ms to ensure that PCIe link is up after
function finish. In the case when no card is connected to the PCIe slot,
this will delay probe time by 100ms, which should not be problematic.
This change fixes detection and initialization of some QCA98xx cards on
the first serdes when configured in x1 mode. Default configuration of
the first serdes on A385 is x4 mode, so it looks as if some delay is
required when x4 is changed to x1 and card correctly links with A385.
Other PCIe serdes ports on A385 are x1-only, and so they don't have this
problem.
Pali Rohár [Tue, 21 Dec 2021 11:20:16 +0000 (12:20 +0100)]
pci: pci_mvebu: Split initialization of PCIe ports into 3 phases
In first phase just parse DT properties and fill struct mvebu_pcie. In
second phase setup all PCIe links (without enabling them). And in the last
third phase enable all PCIe links and create UCLASS_PCI device for each
one.
Because parsing of DT is done before UCLASS_PCI is created, we cannot use
DM for this action anymore. So remove .of_to_plat callback and replace it
by ad-hoc function for parsing DT properties and filling struct mvebu_pcie.
Pali Rohár [Tue, 21 Dec 2021 11:20:14 +0000 (12:20 +0100)]
pci: pci_mvebu: Inline mvebu_pcie_port_parse_dt() function
Function mvebu_pcie_port_parse_dt() is called only from
mvebu_pcie_of_to_plat() function. Both these function parse DT properties
required to setup mvebu pcie. So inline mvebu_pcie_port_parse_dt() function
into mvebu_pcie_of_to_plat() to have all code related to parsing DT
properties at one place.
Pali Rohár [Tue, 21 Dec 2021 11:20:13 +0000 (12:20 +0100)]
pci: pci_mvebu: Fix PCIe MEM and IO resources assignment and mbus mapping
Do not call pci_set_region() for resources which were not properly mapped.
This prevents U-Boot to access unmapped memory space.
Update MBUS_PCI_MEM_SIZE and MBUS_PCI_IO_SIZE macros to cover all PCIe MEM
and IO ranges. Previously these macros covered only address ranges for the
first PCIe port. Between MBUS_PCI_IO_BASE and MBUS_PCI_MEM_BASE there is
space for six 128 MB long address ranges. So set MBUS_PCI_MEM_SIZE to value
of 6*128 MB. Similarly set MBUS_PCI_IO_SIZE to 6*64 KB.
Function resource_size() returns zero when start address is 0 and end
address is -1. So set invalid resources to these values to indicate that
resource has no mapping.
Split global PCIe MEM and IO resources (defined by MBUS_PCI_*_* macros)
into PCIe ports in mvebu_pcie_bind() function which allocates per-port
based struct mvebu_pcie, instead of using global state variables
mvebu_pcie_membase and mvebu_pcie_iobase. This makes pci_mvebu.c driver
independent of global static variables (which store the state of
allocation) and allows to bind and unbind the driver more times.
Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Marek Behún <marek.behun@nic.cz>
Pali Rohár [Tue, 21 Dec 2021 11:20:12 +0000 (12:20 +0100)]
board: gdsys: a38x: Enable PCIe link 2 in spl_board_init()
A385 controlcenterdc board does not use PCI DM properly and touches some
PCIe devices directly in its board code.
This controlcenterdc spl_board_init() function expects that PCIe link is
already initialized. Link itself is initialized in a38x serdes code but
this will change in future and link initialization will be postponed from
U-Boot SPL to proper U-Boot.
So explicitly enable PCIe link 2 in spl_board_init() function via
SoC Control Register 1 to not break this code by future changes. This board
has PCIe link 2 just x1, so no additional initialization (except enabling
PCIe port) is needed.
Pali Rohár [Tue, 21 Dec 2021 11:20:11 +0000 (12:20 +0100)]
arm: mvebu: Convert board_pex_config() to CONFIG_SPL_BOARD_INIT
The only user of board_pex_config() weak function is A385 controlcenterdc
board. It looks like that code in its board_pex_config() function needs to
be executed after PCIe link is up. Therefore put this code into
spl_board_init() function which is called after a38x serdes initialization,
and therefore it is after the serdes hws_pex_config() function finishes
(which is the state before this change).
With this change completely remove board_pex_config() function as it is not
used anymore.
Simon Glass [Thu, 13 Jan 2022 02:26:25 +0000 (19:26 -0700)]
bloblist: Relicense to allow BSD-3-Clause
This implementation is intended to be copied to other projects and
modified, to as to foster a standard means of communcating runtime
information between firmware projects.
The GPL-2 license is too restrictive for some projects, e.g. those
intended as reference implementations rather than designed for
collaborative open-source development.
Simon Glass [Thu, 13 Jan 2022 02:26:22 +0000 (19:26 -0700)]
bloblist: Refactor Kconfig to support alloc or fixed
At present we do support allocating the bloblist but the Kconfig is a bit
strange, since we still have to specify an address in that case. Partly
this is because it is a pain to have CONFIG options that disappears when
its dependency is enabled. It means that we must have #ifdefs in the code,
either in the C code or header file.
Make use of IF_ENABLED_INT() and its friend to solve that problem, so we
can separate out the location of bloblist into a choice. Put the address
and size into variables so we can log the result.
Add the options for SPL as well, so we can use CONFIG_IS_ENABLED().
Simon Glass [Thu, 13 Jan 2022 02:26:21 +0000 (19:26 -0700)]
bloblist: Use 'phase' consistently for bloblists
We typically refer to the different U-Boot builds that a board runs
through as phases. This avoids confusion with the word 'stage' which is
used with bootstage, for example. Fix up some bloblist Kconfig help
which uses the wrong term.
Simon Glass [Thu, 13 Jan 2022 02:26:19 +0000 (19:26 -0700)]
bloblist: Use explicit numbering for the tags
At present if someone adds a tag in the middle of the list it works well
enough within a U-Boot build. But if these tags are used in another
project, or with an older version of SPL, the numbers make become
inconsistent.
Use explicit tag numbers that never change, to resolve this problem.
Allocate areas for existing U-Boot tags and set up an area for use by
projects and vendors, as well as for private use. Keep tags above
0x10000 unallocated for now.
Update bloblist_tag_name() and the tests to work with this new setup.
Simon Glass [Thu, 13 Jan 2022 13:47:55 +0000 (06:47 -0700)]
stddef: Avoid warning with clang with offsetof()
Some bright sparks have decided that a cast on a constant cannot be a
constant, so offsetof() produces this warning on clang-10:
include/intel_gnvs.h:113:1: error: static_assert expression is not an
integral constant expression
check_member(acpi_global_nvs, unused2, GNVS_CHROMEOS_ACPI_OFFSET);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/linux/kernel.h:284:2: note: expanded from macro 'check_member'
offsetof(struct structure, member) == (offset), \
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/linux/stddef.h:20:32: note: expanded from macro 'offsetof'
^
include/intel_gnvs.h:113:1: note: cast that performs the conversions of
a reinterpret_cast is ot allowed in a constant expression
include/linux/stddef.h:20:33: note: expanded from macro 'offsetof'
Fix it by using the compiler built-in version, if available. This syncs
the function to the same implementation as Linux v5.16 in this header
file.
Andre Przywara [Tue, 11 Jan 2022 15:34:50 +0000 (15:34 +0000)]
genboardscfg: limit to 240 jobs
When genboardscfg.py is run on machines with 255 or more cores, the
process will consume more than 1024 file descriptors, which is a common
standard ulimit for user processes. As a consequence it will fail with a
lenghty Python trace, with the almost hidden message:
OSError: [Errno 24] Too many open files
It's somewhat questionable whether that level of parallelity is actually
useful for genboardscfg, so we limit the *default* number of jobs to the
safe number of 240, to avoid the problem.
If a user persists, she can still force a higher number via the -j
parameter - hopefully having raised the ulimit accordingly beforehand.
Signed-off-by: Andre Przywara <andre.przywara@arm.com> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
U-Boot define loff_t as long long. But the header
/usr/include/linux/types.h may not define it.
This has lead to a build error on Alpine Linux.
So let's use long long instead of loff_t for
the size parameter of function os_get_filesize().
Reported-by: Milan P. Stanić <mps@arvanta.net> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com> Tested-by: Milan P. Stanić <mps@arvanta.net> Reviewed-by: Simon Glass <sjg@chromium.org>
Brian Norris [Fri, 7 Jan 2022 23:15:55 +0000 (15:15 -0800)]
patman: Support absolute and ~user-relative alias files
Python doesn't naturally support tilde (~) as a user-home marker in
paths, but git-config does. So we need to resolve it before continuing.
We also shouldn't blindly join the top-level tree with the aliasesfile
path, because it might be an absolute path.
This resolves warnings like the following:
Warning: Cannot find alias file '/path/to/source/tree/~/.git-email'
Seen when git-config is like:
$ git config sendemail.aliasesfile
~/.git-email
Signed-off-by: Brian Norris <briannorris@chromium.org> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Otavio Salvador <otavio@ossystems.com.br>
Michal Simek [Fri, 7 Jan 2022 15:38:09 +0000 (16:38 +0100)]
dm: core: Switch order of pinctrl and power domain calls
The commit 3ad307784847 ("dm: core: device: enable power domain in probe")
introduced enabling power domain when device is probed.
By checking this sequence in Linux kernel was found that power domain is
handled first followed by pinctrl setting.
This patch is switching this order to follow Linux kernel that power
domains are handled first follow by pinctrl setting.
The issue was found on Xilinx Kria SOM where firmware is blocking setting
up pin configuration/muxes without enabling power domain for the specific
IP first.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Tom Rini [Thu, 13 Jan 2022 12:34:27 +0000 (07:34 -0500)]
Merge branch '2022-01-12-assorted-updates'
- Fix binman fake blob support to write outside source directory
- Azure now has stages in the pipeline
- Update to latest focal tag for containers in CI.
- Finish dropping LynxOS
- Add migration message for timer code
Mark Kettenis [Wed, 12 Jan 2022 18:55:17 +0000 (19:55 +0100)]
arm: apple: Use watchdog timer for system reset
Rely on the new watchdog timer driver and the sysreset uclass to
reset the system. This gets rid of hard-coded addresses and
should work on systems based on the new M1 Pro and M1 Max SoCs
as well.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Stefan Roese <sr@denx.de>
Tested-on: Apple M1 Macbook Tested-by: Simon Glass <sjg@chromium.org>
Mark Kettenis [Wed, 12 Jan 2022 18:55:16 +0000 (19:55 +0100)]
arm: dts: apple: Add watchdog timer node
Add a node for the watchdog timer based on the proposed Linux
device tree bindings.
Remove the old reboot node which was a watchdog timert node in
disguise using a preliminary device tree binding.
Signed-off-by: Sven Peter <sven@svenpeter.dev> Signed-off-by: Mark Kettenis <kettenis@openbsd.org> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Stefan Roese <sr@denx.de>
Tested-on: Apple M1 Macbook Tested-by: Simon Glass <sjg@chromium.org>
Mark Kettenis [Wed, 12 Jan 2022 18:55:15 +0000 (19:55 +0100)]
watchdog: Add a driver for the Apple watchdog
This driver supports the watchdog timer found on Apple's M1 SoC.
On systems that use these SoC, the watchdog timer is the primary
way to reboot the system.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org> Reviewed-by: Sven Peter <sven@svenpeter.dev> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Stefan Roese <sr@denx.de>
Tested-on: Apple M1 Macbook Tested-by: Simon Glass <sjg@chromium.org>
Thomas Huth [Sat, 13 Nov 2021 17:13:50 +0000 (18:13 +0100)]
LynxOS is no longer supported
LynxOS needed the do_bootm_lynxkdi() function that got removed in 7e713067ee ("Remove LYNX KDI remainders") - and that function needed
a lynxkdi_boot() function, where the last implementation had been
removed in 98f705c9ce ("powerpc: remove 4xx support") already. Looks
like this OS is definitely not supported anymore, so remove it from
the corresponding lists.
Signed-off-by: Thomas Huth <thuth@redhat.com> Reviewed-by: Stefan Roese <sr@denx.de>
Simon Glass [Wed, 12 Jan 2022 20:10:36 +0000 (13:10 -0700)]
binman: Write fake blobs to the output directory
At present binman writes fake blobs to the current directory. This is not
very helpful, since the files serve no useful purpose once binman has
finished. They clutter up the source directory and affect future runs,
since the files in the current directory are often used in preference to
those in the board directory.
To avoid these problems, write them to the output directory instead.
Move the file-creation code to the Entry base class, so it can be used by
any entry type that needs it. This is required since some entry types,
such as Entry_blob_ext_list, are not subclasses of Entry_blob.
Tom Rini [Wed, 12 Jan 2022 00:14:28 +0000 (19:14 -0500)]
ci: azure: Update to use stages
Follow what we do in GitLab CI where we break the jobs up in to stages
such that if earlier and often quicker sanity tests fail we don't run
everything else.
Pali Rohár [Fri, 26 Nov 2021 10:42:52 +0000 (11:42 +0100)]
pci: sh7751: Fix access to config space via PCI_CONF1_ADDRESS() macro
sh7751 platform uses standard format of Config Address for PCI
Configuration Mechanism #1.
Commit 72c2f4acd76f ("pci: sh7751: Convert to DM and DT probing") which did
conversion of PCI sh7751 driver to DM, broke access to config space as that
commit somehow swapped device and function bits in config address.
Fix all these issues by using new U-Boot macro PCI_CONF1_ADDRESS() which
calculates Config Address correctly.
Also remove nonsense function sh7751_pci_addr_valid() which was introduced
in commit 72c2f4acd76f ("pci: sh7751: Convert to DM and DT probing")
probably due to workarounded issues with mixing/swapping device and
function bits of config address which probably resulted in non-working
access to some devices. With correct composing of config address there
should not be such issue anymore.
Signed-off-by: Pali Rohár <pali@kernel.org> Fixes: 72c2f4acd76f ("pci: sh7751: Convert to DM and DT probing") Cc: Marek Vasut <marek.vasut+renesas@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
projects / u-boot.git / log