Jagan Teki [Wed, 27 Feb 2019 18:56:57 +0000 (00:26 +0530)]
clk: sunxi: Implement EMAC, GMAC clocks, resets
- Implement EMAC, GMAC clocks via ccu_clk_gate for
all supported Allwinner SoCs.
- Implement EMAC, GMAC resets via ccu_reset for all
supported Allwinner SoCs.
Cc: Joe Hershberger <joe.hershberger@ni.com> Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Ondrej Jirman [Wed, 13 Feb 2019 17:51:05 +0000 (18:51 +0100)]
power: axp818: Fix typo in axp_set_dldo
Fix typo in axp_set_dldo() so that it correctly uses AXP818_DLDO1_CTRL
register to configure the voltage instead of setting AXP818_ELDO1_CTRL
register which is obviously incorrect.
Signed-off-by: Ondřej Jirman <megous@megous.com> Reviewed-by: Jagan Teki <jagan@openedev.com>
Tom Rini [Mon, 4 Mar 2019 18:05:53 +0000 (13:05 -0500)]
Merge git://git.denx.de/u-boot-fsl-qoriq
- Enable DHCP as boot-source in distro boot for NXP layerscape
platforms
- fix register layout for SEC on Layerscape architectures
- fixes related to DPAA2 ethernet
Jagan Teki [Wed, 27 Feb 2019 14:32:13 +0000 (20:02 +0530)]
spi: Rename sun4i_spi.c into spi-sunxi.c
Now the same SPI controller driver is reusable in all Allwinner
SoC variants, so rename the existing sun4i_spi.c into spi-sunxi.c
which eventually look like a common sunxi driver.
Also update the function, variable, structure names in driver from
sun4i into sunxi.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:12 +0000 (20:02 +0530)]
spi: sun4i: Driver cleanup
- drop unused macros.
- use base instead of base_addr, for better code readability
- move .probe and .ofdata_to_platdata functions in required
places to add platdata support in future.
- use sentinel sun4i_spi_ids.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:11 +0000 (20:02 +0530)]
spi: sun4: Add A31 spi controller support
The usual SPI transmission protocol in Allwinner A10 and A31
controllers share similar context with minimal changes in register
offsets along with few additional register bits on A31.
So, add A31 spi controller support in existing sun4i_spi with A31
specific register offsets and bits.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:10 +0000 (20:02 +0530)]
spi: sun4i: Add CLK support
Add CLK support to enable AHB and MOD SPI clocks on sun4i_spi driver.
Clock disablement could be done while releasing the bus transfer, but
the existing code doesn't disable the clocks it only taken care of clock
enablement globally in probe.
So to make a proper clock handling, the clocks should enable it in claim
and disable it in release.
This patch would also do that change, by enable and disable clock in
proper order.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:09 +0000 (20:02 +0530)]
spi: sun4i: Support fifo_depth via drvdata
Support fifo_depth via drvdata instead of macro definition, this would
eventually reduce another macro definition for new SPI controller fifo
depth support addition.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:08 +0000 (20:02 +0530)]
spi: sun4i: Access registers and bits via enum offsets
Allwinner support two different SPI controllers one for A10 and
another for A31 with minimal changes in register offsets and
respective register bits, but the logic for accessing the SPI
master via SPI slave remains nearly similar.
Add enum offsets for register set and register bits, so-that
it can access both classes of SPI controllers.
Assign same control register for global, transfer and fifo control
registers to make the same code compatible with A31 SPI controller.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Tested-by: Stefan Mavrodiev <stefan@olimex.com> # A20-SOM204
Jagan Teki [Wed, 27 Feb 2019 14:32:06 +0000 (20:02 +0530)]
clk: sunxi: Implement SPI clocks, resets
- Implement SPI AHB, MOD clocks via ccu_clk_gate for all
supported Allwinner SoCs
- Implement SPI resets via ccu_reset for all supported
Allwinner SoCs.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Jagan Teki [Wed, 27 Feb 2019 14:32:05 +0000 (20:02 +0530)]
spi: sun4i: Poll for rxfifo to be filled up
To drain rx fifo the fifo need to poll for how much data has
been filled up in rx fifo.
To achieve this, the current code is using wait_for_bit logic
on control register with exchange burst mode mask, which is not
a proper way of waiting for fifo filled up.
So, add code for polling rxfifo to be filled up using fifo
status register.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
With current order of include files, the file designware_spi.c
can't see that the struct global_data has the member
board_type when CONFIG_BOARD_TYPES is defined. By not seeing this
then all the members are shifted in the struct global_data.
So when the driver is trying to read from device tree blob, it
would pass the wrong address to the function 'fdtdev_get_int'.
This will make to use the default frequency 500000.
The fix consists of changing the order of include files in
designware_spi.c to include first common.h file.
Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com> Reviewed-by: Jagan Teki <jagan@openedev.com>
Eugen Hristev [Fri, 8 Feb 2019 08:57:01 +0000 (10:57 +0200)]
configs: at91: remove SPL_GENERATE_ATMEL_PMECC_HEADER from non-nand configs
SPL_GENERATE_ATMEL_PMECC_HEADER will generate a header for the SPL for NAND
information. The initial stage 1 bootloader will use this header in case the
NAND flash doesn't support commands to retrieve sector size, etc.
However this header is bad for different boot media, like MMC or SPI.
In case SD_BOOT or SPI_BOOT is used, remove the config for this.
Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
Alexander Dahl [Wed, 6 Feb 2019 14:19:41 +0000 (15:19 +0100)]
configs: sama5d27_som1_ek: Activate misc init for uSD variant
This was already set for 'sama5d27_som1_ek_mmc_defconfig' (first SD
card, full size). Without this option set, the MAC address is not read
from the I²C EEPROM at boot, and remains unset:
U-Boot 2019.01 (Jan 01 2019 - 00:00:00 +0000)
CPU: SAMA5D27 1G bits DDR2 SDRAM
Crystal frequency: 24 MHz
CPU clock : 492 MHz
Master clock : 164 MHz
DRAM: 128 MiB
MMC: sdio-host@a0000000: 0, sdio-host@b0000000: 1
Loading Environment from FAT... OK
In: serial
Out: serial
Err: serial
Net:
Error: ethernet@f8008000 address not set.
eth-1: ethernet@f8008000
Hit any key to stop autoboot: 0
pinctrl: renesas: r8a77990: Fix MOD_SEL0 bit3 when using TX0
According to the R-Car Gen3 Hardware Manual Errata for Rev 1.50,
the MOD_SEL0 bit3 is set to 0 when TX0_A pin function is selected,
and the MOD_SEL0 bit3 is set to 1 when TX0_B pin function is
selected.
pinctrl: renesas: r8a77990: Fix rename RTSx_N_TANS_x to RTSx_N_x
This patch fixes the allocation name "RTSx_N_TANS_x" of IPSR /
MOD_SEL0/1 of r8a77990 to "RTSx_N_x". This information was
confirmed in the R-Car Gen3 Hardware Manual Rev.1.50.
pinctrl: renesas: Fix r8a779{6,65} rename sel_ndfc to sel_ndf
This patch fixes the allocation name "sel_ndfc" of MOD_SEL2[22]
of r8a7796 / r8a77965 to "sel_ndf". This information was confirmed
in the R-Car Gen3 Hardware Manual Rev.1.50.
pinctrl: renesas: Remove r8a779{5,6,65} CC5_OSCOUT of IP17
This patch removes CC5_OSCOUT assignment of IP17[3:0] of r8a7795
/ r8a7796 / r8a77965. This information was confirmed in the R-Car
Gen3 Hardware Manual Rev.1.50.
pinctrl: renesas: Fix r8a779{5,6,65} rename sel_adg_ to sel_adg
This patch fixes to the correct names, and "_" is not include after
"adg" for r8a7795/r8a7796/r8a77965. This information was confirmed
in the R-Car Gen3 Hardware Manual Rev.1.50.
pinctrl: renesas: Fix r8a779{5,6,65} assign to GP7_03/02 of GPSR7
This patch is change the bit assignment of "HDMI1_CEC" to "GP7_03",
and "HDMI0_CEC" to "GP7_02". This information was confirmed in the
R-Car Gen3 Hardware Manual Rev.1.50.
Laurentiu Tudor [Tue, 26 Feb 2019 11:18:34 +0000 (13:18 +0200)]
armv8: fsl-layerscape: avoid DT fixup warning
sec_firmware reserves JR3 for it's own usage and deletes the JR3 node
from the device tree. This causes this warning to be issued when doing
the device tree fixup:
WARNING could not find node fsl,sec-v4.0-job-ring: FDT_ERR_NOTFOUND.
Fix it by excluding the device tree fixup for the JR reserved by
sec_firmware.
Laurentiu Tudor [Tue, 26 Feb 2019 11:18:33 +0000 (13:18 +0200)]
armv8: fsl-layerscape: fix SEC QI ICID setup
The SEC QI ICID setup in the QIIC_LS register is actually an offset
that is being added to the ICID coming from the qman portal. Setting
it with a non-zero value breaks SMMU setup as the resulting ICID is
not known. On top of that, the SEC QI ICID must match the qman portal
ICIDs in order to share the isolation context.
Laurentiu Tudor [Tue, 26 Feb 2019 11:18:32 +0000 (13:18 +0200)]
fsl_sec: fix register layout on Layerscape architectures
On Layerscape architectures the SEC memory map is 1MB and the
register blocks contained in it are 64KB aligned, not 4KB as
the ccsr_sec structure currently assumes. Fix the layout of
the structure for these architectures.
Revert "ls2080ardb: remove dhcp function from env as boot source"
dhcp boot is a useful feature and works out-of-the-box for ls2088a
platforms. Moreover, no solid reason is given for disabling it.
Revert the patch to re-enable it.
Revert "ls1088a: remove dhcp function from u-boot env as boot source"
dhcp boot is a useful feature and works out-of-the-box for ls1088a
platforms. Moreover, no solid reason is given for disabling it.
Revert the patch to re-enable it.
When package types are not supported by our implementation of the HII
database protocol supported error messages are displayed.
Essentially the output is only needed for debugging. By using EFI_PRINT()
the messages are only written for in debug mode and with correct
indentation.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
efi_loader: error handling for `efidebug boot add`
In `efidebug boot add iPXE scsi 0:1 snp-arm64.efi --foo` a parameter is
missing. Hence the command should not silently return as if everything were
ok but should display the usage info.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Alexander Graf <agraf@csgraf.de>
In new_package_list() we call new_packagelist() to create a new package
list. Next we try to add the packages which fails for form packages. Due
to this error we call free_packagelist(). Now in free_packagelist()
list_del() is called for an uninitialized field hii->link. This leads to
changing random memory addresses.
To solve the problem move the initialization of hii->link to
new_packagelist().
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
In the `efidebug boot add` command we do not want an unsolicited leading
backslash added to the file name.
There is no good reason to mark a loaded file with a backslash as absolute.
Anyway when reading files the file name will be interpreted as relative to
root directory of the device.
So let's get rid of this backslash.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Peng Ma [Thu, 21 Feb 2019 02:18:27 +0000 (10:18 +0800)]
powerpc: t104xrdb: Add support of MTA9ADF1G72AZ DDR
T1040RDB has been upgraded to support new DDR ie. MTA9ADF1G72AZ-3G2, 8GB.
So adding support of new DDR part by updating board_specific_parameters
udimm0.
Signed-off-by: Peng Ma <peng.ma@nxp.com> Reviewed-by: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Moving to the driver model requires CONFIG_DM to be enabled. Currently
several boards like kmeter1_defconfig produce a build error when CONFIG_DM
is enabled:
In file included from include/common.h:35,
from ./arch/powerpc/include/asm/fsl_lbc.h:10,
from include/mpc83xx.h:10,
from ./arch/powerpc/include/asm/ppc.h:27,
from ./arch/powerpc/include/asm/u-boot.h:18,
from include/dm/of.h:10,
from include/dm/ofnode.h:12,
from include/dm/device.h:13,
from include/linux/mtd/mtd.h:26,
from drivers/mtd/mtdconcat.c:25:
include/image.h: In function ‘image_check_target_arch’:
include/image.h:846:3: error: #error "please define IH_ARCH_DEFAULT in
your arch asm/u-boot.h"
# error "please define IH_ARCH_DEFAULT in your arch asm/u-boot.h"
^~~~~
include/image.h:848:31: error: ‘IH_ARCH_DEFAULT’ undeclared (first use in
this function); did you mean ‘IH_ARCH_COUNT’?
return image_check_arch(hdr, IH_ARCH_DEFAULT);
The error can be avoided by moving the definition of IH_ARCH_DEFAULT before
#include <asm/ppc.h>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
To find out how big the early malloc heap must be in SPL, add a debug
print statement that dumps its usage before switching to relocated heap
in spl_relocate_stack_gd() via CONFIG_SPL_STACK_R_MALLOC_SIMPLE_LEN.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Stephen Warren [Tue, 26 Feb 2019 19:20:26 +0000 (12:20 -0700)]
kbuild: fix DTB .cmd source variable
*.dts are processed using a custom command, then the C pre-processor is
run on them, then they are compiled using dtc. Thus, the dependency
files generated by both cpp and dtc reference a temporary file name
rather than the actual source file. While this information isn't used
for any purpose by the build system, and hence this causes no functional
issue, it does cause the dependency files to contain invalid and
confusing data, which is unhelpful while debugging build problems. Fix
this using sed.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Reviewed-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Stephen Warren [Tue, 26 Feb 2019 19:20:25 +0000 (12:20 -0700)]
kbuild: make arch-dtbs target PHONY
Without this, the arch-dtbs target only gets evaluated when building
U-Boot the first time, not when re-building (incrementally building)
U-Boot. Thus incremental builds ignore changes to DTB files.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Reviewed-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Patrick Delaunay [Tue, 26 Feb 2019 12:09:00 +0000 (13:09 +0100)]
doc: binding: rename directory ram to memory-controller
Alignment with kernel directory name as it have already bindings for
DDR controllers in the directory:
Documentation/devicetree/bindings/memory-controller
PS: the drivers using RAM u-class should be associated with
this binding directory
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Marek Vasut [Tue, 19 Feb 2019 00:43:51 +0000 (01:43 +0100)]
ARM: cache: Fix incorrect bitwise operation
The loop implemented in the code is supposed to check whether the
PL310 operation register has any bit from the mask set. Currently,
the code checks whether the PL310 operation register has any bit
set AND whether the mask is non-zero, which is incorrect. Fix the
conditional.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Dalon Westergreen <dwesterg@gmail.com> Cc: Dinh Nguyen <dinguyen@kernel.org> Cc: Tom Rini <trini@konsulko.com> Fixes: 93bc21930a1b ("armv7: add PL310 support to u-boot") Reviewed-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com> Reviewed-by: Dinh Nguyen <dinguyen@kernel.org>
David Rivshin [Mon, 18 Feb 2019 23:04:29 +0000 (18:04 -0500)]
spi: omap3: fix set_wordlen() reading from incorrect address for CHCONF
_omap3_spi_set_wordlen() indexed the regs->channel[] array with the
old wordlen (instead of the chipselect number) when reading the current
CHCONF register value. This meant it read from the wrong memory location,
modified that value, and then wrote it back to the correct CHCONF
register. The end result is that most slave configuration settings would
be lost, such as clock divisor, clock/chipselect polarities, etc.
Fixes: 77b8d04854f4 ("spi: omap3: Convert to driver model") Signed-off-by: David Rivshin <drivshin@allworx.com>
Anup Patel [Mon, 25 Feb 2019 08:15:33 +0000 (08:15 +0000)]
riscv: Enable CONFIG_SYS_BOOT_RAMDISK_HIGH for using initrd
This patch enables CONFIG_SYS_BOOT_RAMDISK_HIGH for RISC-V
because bootm will update initrd location in DTB only if
CONFIG_SYS_BOOT_RAMDISK_HIGH is enabled. If we don't enable
this option then bootm assumes DTB already has initrd details
which is not the case most of the time.
Atish Patra [Mon, 25 Feb 2019 08:15:27 +0000 (08:15 +0000)]
doc: Add a readme guide for SiFive FU540
The readme guide describes the procedure to build, flash and boot Linux
using U-Boot on HiFive Unleashed. It also explains the current state of
U-boot support and future action items.
Anup Patel [Mon, 25 Feb 2019 08:15:19 +0000 (08:15 +0000)]
riscv: Add SiFive FU540 board support
This patch adds SiFive FU540 board support. For now, only
SiFive serial, SiFive PRCI, and Cadance MACB drivers are
only enabled. The SiFive FU540 defconfig by default builds
U-Boot for S-Mode because U-Boot on SiFive FU540 will run
in S-Mode as payload of BBL or OpenSBI.
Signed-off-by: Atish Patra <atish.patra@wdc.com> Signed-off-by: Anup Patel <anup.patel@wdc.com> Reviewed-by: Alexander Graf <agraf@suse.de> Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Atish Patra [Mon, 25 Feb 2019 08:15:14 +0000 (08:15 +0000)]
cpu: Bind timer driver for boot hart
Currently, timer driver is bound only for hart0.
There is no mandatory requirement that hart0 should always
come up. In fact, HiFive Unleashed SoC hart0 doesn't boot
in S-mode because it only has M-mode.
The timer driver should be bound for boot hart.
Signed-off-by: Atish Patra <atish.patra@wdc.com> Signed-off-by: Anup Patel <anup.patel@wdc.com> Reviewed-by: Alexander Graf <agraf@suse.de> Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Anup Patel [Mon, 25 Feb 2019 08:14:49 +0000 (08:14 +0000)]
clk: Add SiFive FU540 PRCI clock driver
Add driver code for the SiFive FU540 PRCI IP block. This IP block
handles reset and clock control for the SiFive FU540 device and
implements SoC-level clock tree controls and dividers.
Anup Patel [Mon, 25 Feb 2019 08:14:30 +0000 (08:14 +0000)]
riscv: generic: Ensure that U-Boot runs within 4GB for 64bit systems
On 64bit systems, the DRAM top can be easily beyond 4GB and U-Boot
DMA mapping APIs will generate DMA addresses beyond 4GB. This
breaks DMA programming in 32bit DMA capable devices (such as
Cadence MACB ethernet). For example, If DRAM is more then 2GB
on QEMU sifive_u machine then Cadence MACB ethernet stops working
for U-Boot because it is a 32bit DMA capable device.
To handle 32bit DMA capable devices on 64bit systems, we provide
custom implementation of board_get_usable_ram_top() which ensures
that usable ram top is not more then 4GB. This in-turn ensures
that U-Boot always runs within 4GB hence DMA addresses generated
by DMA mapping APIs will be within 4GB too.
Signed-off-by: Atish Patra <atish.patra@wdc.com> Signed-off-by: Anup Patel <anup.patel@wdc.com> Reviewed-by: Alexander Graf <agraf@suse.de> Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Anup Patel [Mon, 25 Feb 2019 08:14:24 +0000 (08:14 +0000)]
riscv: Add place-holder asm/arch/clk.h for driver compilation
Some of the drivers (such as Cadence MACB ethernet driver) expect
asm/arch/clk.h to be provided by arch support so we add place-holder
asm/arch-generic/clk.h for RISC-V generic CPU.
Anup Patel [Mon, 25 Feb 2019 08:14:10 +0000 (08:14 +0000)]
riscv: Rename cpu/qemu to cpu/generic
The QEMU CPU support under arch/riscv is pretty much generic
and works fine for SiFive Unleashed as well. In fact, there
will be quite a few RISC-V SOCs for which QEMU CPU support
will work fine.
This patch renames cpu/qemu to cpu/generic to indicate the
above fact. If there are SOC specific errata workarounds
required in cpu/generic then those can be done at runtime
in cpu/generic based on CPU vendor specific DT compatible
string.
Signed-off-by: Anup Patel <anup.patel@wdc.com> Reviewed-by: Alexander Graf <agraf@suse.de> Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tom Rini [Tue, 26 Feb 2019 13:45:08 +0000 (08:45 -0500)]
Merge tag 'efi-2019-04-rc3' of https://github.com/xypron2/u-boot
Pull request for the UEFI sub-system for v2019.04-rc3
A new option -e is added to the env command which allows to display and
change UEFI variables in a user friendly way.
A new command efidebug is introduced to edit the UEFI boot sequence and to
display different aspects of the state of the UEFI sub-system: memory map,
loaded images, handles, drivers and devices.
Marek Vasut [Tue, 19 Feb 2019 18:32:28 +0000 (19:32 +0100)]
mmc: renesas: Unconditionally set DTCNTL TAPNUM to 8
According to latest specification rev.0026 and after confirmation with
HW engineer, the DTCNTL register TAPNUM field must be set to 8 even on
H3 ES2.0 SoC. Make it so.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Marek Vasut [Tue, 19 Feb 2019 18:20:14 +0000 (19:20 +0100)]
mmc: tmio: Clear BUSWIDTH bit when WMODE bit is set
According to latest specification rev.0026, when HOST_MODE bit 0
(WMODE) is not set, HOST_MODE bit 8 (BUSWIDTH) is ignored. Clear
HOST_MODE bit 8 in such case and align the code with Linux and
avoid possible unforeseen issues.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com> Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Marek Vasut [Tue, 19 Feb 2019 00:07:21 +0000 (01:07 +0100)]
ARM: socfpga: Clear PL310 early in SPL
On SoCFPGA Gen5 systems, it can rarely happen that a reboot from Linux
will result in stale data in PL310 L2 cache controller. Even if the L2
cache controller is disabled via the CTRL register CTRL_EN bit, those
data can interfere with operation of devices using DMA, like e.g. the
DWMMC controller. This can in turn cause e.g. SPL to fail reading data
from SD/MMC.
The obvious solution here would be to fully reset the L2 cache controller
via the reset manager MPUMODRST L2 bit, however this causes bus hang even
if executed entirely from L1 I-cache to avoid generating any bus traffic
through the L2 cache controller.
This patch thus configures and enables the L2 cache controller very early
in the SPL boot process, clears the L2 cache and disables the L2 cache
controller again.
The reason for doing it in SPL is because we need to avoid accessing any
of the potentially stale data in the L2 cache, and we are certain any of
the stale data will be below the OCRAM address range. To further reduce
bus traffic during the L2 cache invalidation, we enable L1 I-cache and
run the invalidation code entirely out of the L1 I-cache.
Marek Vasut [Wed, 13 Feb 2019 20:50:25 +0000 (21:50 +0100)]
ARM: cache: Fix incorrect bitwise operation
The loop implemented in the code is supposed to check whether the
PL310 operation register has any bit from the mask set. Currently,
the code checks whether the PL310 operation register has any bit
set AND whether the mask is non-zero, which is incorrect. Fix the
conditional.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Dalon Westergreen <dwesterg@gmail.com> Cc: Dinh Nguyen <dinguyen@kernel.org> Cc: Tom Rini <trini@konsulko.com> Fixes: 93bc21930a1b ("armv7: add PL310 support to u-boot")
projects / u-boot.git / log