Avoid reading mhartid CSR directly, instead use the one we saved
in the global data structure before.
With this patch, BBL no longer needs to be hacked to provide the
mhartid CSR emulation for S-mode U-Boot.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
At present the hart id passed via a0 in the U-Boot entry is saved
to s0 at the beginning but does not preserve later. Save it to the
global data structure so that it can be used later.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
With this change, we can avoid a forward declaration.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
At present the trap handler returns to hardcoded M-mode/S-mode.
Change to returning to previous privilege level instead.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
sp cannot be loaded before restoring other registers.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
Currently the M-mode trap handler codes are in start.S. For future
extension, move them to a separate file mtrap.S.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
In arch_cpu_init_dm() do some basic architecture level cpu
initialization, like FPU enable, etc.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
With current csr_xxx ops, we cannot pass a macro to parameter
'csr', hence we need add another level to allow the parameter
to be a macro itself, aka indirect stringification.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This updates supports_extension() implementation to use the desc
string from the cpu driver whenever possible, which avoids the
reading of misa CSR for S-mode U-Boot.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This adds all exception codes in encoding.h.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
The standard RISC-V ISA sets aside a 12-bit encoding space for up
to 4096 CSRs. This adds all known CSR numbers as defined in the
RISC-V Privileged Architecture Version 1.10.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
With DM CPU driver, the non-DM version of print_cpuinfo() is no
longer needed.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This calls cpu_probe_all() to probe all available cpus.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
Increase the heap size for the pre-relocation stage, so that CPU
driver can be loaded.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
Add the QEMU RISC-V platform-specific Kconfig options, to include
CPU and timer drivers.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This adds an implementation of riscv_get_time() API that is using
rdtime instruction.
This is the case for S-mode U-Boot, and is useful for processors
that support rdtime in M-mode too.
Signed-off-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
This adds U-Boot syscon driver for SiFive's Core Local Interruptor
(CLINT). The CLINT block holds memory-mapped control and status
registers associated with software and timer interrupts.
This driver implements the riscv_get_time() API as required by
the generic RISC-V timer driver, as well as some other APIs that
are needed for handling IPI.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
So far we have a Kconfig option for supervisor mode. This adds an
option for the machine mode.
Signed-off-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
There is no need to expose RISCV_NDS to the Kconfig menu as it is
an ax25-specific option. Introduce a dedicated Kconfig option for
the cache ops of ax25 platform and use that to guard the cache ops.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Rick Chen <rick@andestech.com>
RISC-V privileged architecture v1.10 defines a real-time counter,
exposed as a memory-mapped machine-mode register - mtime. mtime must
run at constant frequency, and the platform must provide a mechanism
for determining the timebase of mtime. The mtime register has a
64-bit precision on all RV32, RV64, and RV128 systems.
Different platform may have different implementation of the mtime
block hence an API riscv_get_time() is required by this driver for
platform codes to hide such implementation details. For example,
on some platforms mtime is provided by the CLINT module, while on
some other platforms a simple 'rdtime' can be used to get the timer
counter.
With this timer driver the U-Boot timer functionalities like delay
works correctly now.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This adds a driver for RISC-V CPU. Note the driver will bind
a RISC-V timer driver if "timebase-frequency" property is
present in the device tree.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
To enumerate devices on the /soc/ node, create a "simple-bus"
driver to match "riscv-virtio-soc".
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
This adds a timebase_freq member to the 'struct cpu_platdata', to
hold the "timebase-frequency" value in the cpu or /cpus node.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Anup Patel <anup@brainfault.org>
RISC-V has two code models, medium low (medlow) and medium any (medany).
Medlow limits addressable memory to a single 2 GiB range between the
absolute addresses -2 GiB and +2 GiB. Medany limits addressable memory
to any single 2 GiB address range.
By default, medlow is selected for U-Boot on both 32-bit and 64-bit
systems.
The -mcmodel compiler flag is selected according to the Kconfig
configuration.
Signed-off-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
[bmeng: adjust to make medlow the default code model for U-Boot]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Android code base is growing, so since Android "Pie" the size of
system.img grew up to be about 740 MiB. Let's increase system.img to
1 GiB to accommodate for those changes and leave some margin for future
changes. We don't want to make it more than 1 GiB, because we should
keep userdata partition big enough (for user files, like media etc.),
and eMMC size on BeagleBoard-X15 is only 3.5 GiB.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Acked-by: Praneeth Bajjuri <praneeth@ti.com>
Output the device model migration warnings to stderr. This allows tools
like buildman to pick them up rather than suppressing them along with
the normal build output on stdout.
Signed-off-by: Chris Packham <judge.packham@gmail.com>
Commit a2b96ece5b breaks the enumeration
of the Image Types. New image types can be appended, but they cannot be
inserted in the list else backward compatibility is broken.
This restores the images types as before 2018.11 and move i.MX8 related
images at the end.
Signed-off-by: Robert Berger <robert.berger@ReliableEmbeddedSystems.com>
Signed-off-by: Stefano Babic <sbabic@denx.de>
CC: Tom Rini <trini@konsulko.com>
Use standard configuration logic to define EEPROM constants.
Names are based on VPD_EEPROM_ prefix because EEPROM_ is already
used by i2c_eeprom driver.
Signed-off-by: Denis Zalevskiy <denis.zalevskiy@ge.com>
Signed-off-by: Fabien Lahoudere <fabien.lahoudere@collabora.com>
The bus is statically defined, so remove redundant parameters
from read_vpd() for PPD and Bx50v3.
Signed-off-by: Denis Zalevskiy <denis.zalevskiy@ge.com>
Signed-off-by: Fabien Lahoudere <fabien.lahoudere@collabora.com>
Add the initial support for Elgin R1 board, which is based on the
RV1108 SoC and has the following features currently supported in
U-Boot:
- UART
- eMMC
- USB
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
- Introduce tools-only build for host tools
- Bugfixes to poplar, syscon and the hashtable, a tee return code
- Fix a warning on gcc-8 by reworking part of mtk_image to be not unsafe
wrt strings.
- serial_stm32 reset support
This is a sync with kernel mainline dts.
The U-Boot eMMC does not need to care about the power for Rockchip
SoCs, because if the board is using eMMC, the power will default on
(for bootrom), so the 'vmmc', 'vqmmc' is only useful for SD in U-Boot.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
vcc5v0_host and usbhub_enable share gpio4 RK_PA3,
which is a problem during probing (the second probe
will trigger a -EBUSY, when trying to get the gpio handle).
An analysis of the situation shows, that both regulators
are actually describing the same supply.
This patch removes the (currenlty not successful probing)
regulator vcc5v0_host from the DTS and adds the pinctrl-*
setting to usbhub_enable.
Signed-off-by: Christoph Muellner <christoph.muellner@theobroma-systems.com>
Reviewed-by: Phiilipp Tomsich <philipp.tomsich@theobroma-systems.com>
+ Update the function syscon_node_to_regmap() to force bound on
syscon uclass and directly use the list of device from DM.
+ Remove the static list syscon_list.
This patch avoid issue (crash) when syscon_node_to_regmap() is called
before and after reallocation (list content is invalid).
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Starting from version 8 the GCC, i.e. C compiler, starts complaining about
possible '\0' terminator loss or, as in this case, garbage copy.
In function ‘mtk_image_set_gen_header’,
inlined from ‘mtk_image_set_header’ at tools/mtk_image.c:733:3:
tools/mtk_image.c:659:2: warning: ‘strncpy’ specified bound 12 equals destination size [-Wstringop-truncation]
strncpy(hdr->boot.name, bootname, sizeof(hdr->boot.name));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In function ‘mtk_brom_parse_imagename’,
inlined from ‘mtk_image_check_params’ at tools/mtk_image.c:388:9:
tools/mtk_image.c:325:5: warning: ‘strncpy’ specified bound 32 equals destination size [-Wstringop-truncation]
strncpy(lk_name, val, sizeof(lk_name));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Replace it with snprintf() to tell compiler how much room we have in the
destination buffer for source string.
Fixes: 3b975a147c ("tools: MediaTek: add MTK boot header generation to mkimage")
Cc: Ryder Lee <ryder.lee@mediatek.com>
Cc: Weijie Gao <weijie.gao@mediatek.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
In some cases, UART is configured by early boot stage.
To be sure of the initial state of UART and to avoid
spurious chars on console, reset the serial block before
configuring it.
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Commit 4687919684 ("serial: Remove DM_FLAG_PRE_RELOC flag in various
drivers") essentially drops flag DM_FLAG_PRE_RELOC from serial_pl01x
driver for Poplar platform, because the platform falls into the
following strategy category made by the commit.
Surround DM_FLAG_PRE_RELOC flag with OF_CONTROL check, for
drivers that support both statically declared devices and
configuration from device tree
Before the commit lands, Poplar platform works by statically declaring
pl011 serial device via U_BOOT_DEVICE() with DM_FLAG_PRE_RELOC flag set
in the driver. But since Poplar also supports device configuration from
device tree, the commit practically drops the flag for Poplar, and hence
breaks the platform from booting.
This patch changes platform code and device tree to initiate pl011
serial device from device tree rather than static declaration, so that
above strategy about DM_FLAG_PRE_RELOC applies to Poplar, and therefore
the reported boot failure gets fixed.
Reported-by: Igor Opaniuk <igor.opaniuk@linaro.org>
Fixes: 4687919684 ("serial: Remove DM_FLAG_PRE_RELOC flag in various drivers")
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Igor Opaniuk <igor.opaniuk@linaro.org>
Tested-by: Igor Opaniuk <igor.opaniuk@linaro.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The motivation for this is to allow distributions to distribute all
possible tools in a generic way, avoiding the need of specific tools
building for each machine.
Especially on OpenEmbedded / Yocto Project ecosystem, it is very
common each BSP to end providing their specific tools when they need
to generate images for some SoC (e.g MX23 / MX28 in meta-freescale
case).
Using this, we can package the tools doing:
$: make tools-only_defconfig
$: make tools-only
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
[trini: Add MAINTAINERS entry for myself, add to .travis.yml, make
U-Boot itself buildable to not trip up other frameworks]
Signed-off-by: Tom Rini <trini@konsulko.com>
The contents of the test subdirectories only make sense when we have
CONFIG_UNIT_TEST set. We will otherwise attempt to build code on for
example sandbox that needs CONFIG_UNIT_TEST otherwise and rather than
complicate the Makefiles simply leave them out when we can.
Signed-off-by: Tom Rini <trini@konsulko.com>
The commit below incorrectly fixed hexport_r();
> size = totlen + 1;
One extra byte is necessary to NULL-terminate a whole buffer, "resp."
Fixes: f1b20acb4a ("hashtable: Fix length calculation in hexport_r")
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
If OP-TEE core is compiled with support of REE FS and RPMB
at the same time (CFG_RPMB_FS ?= y; CFG_RPMB_FS ?= y), and persistent
storage API is used with TEE_STORAGE_PRIVATE storage id, it will
lead to TA panic.
E/TC:? 0 TA panicked with code 0xffff0009
.....
E/TC:? 0 Call stack:
E/TC:? 0 0x000000004002f2f8 TEE_OpenPersistentObject at
lib/libutee/tee_api_objects.c:422
In this particular case TEE_ERROR_STORAGE_NOT_AVAILABLE is more suitable
than TEE_ERROR_NOT_IMPLEMENTED, as it provides to a TA a possibility
to handle this error code [1].
>From GPD TEE Internal Core specification [2]:
TEE_ERROR_STORAGE_NOT_AVAILABLE - if the persistent object is stored in a
storage area which is currently inaccessible. It may be associated with
the device but unplugged, busy, or inaccessible for some other reason.
[1]: 94db01ef44/lib/libutee/tee_api_objects.c (L419)
[2]: https://globalplatform.org/wp-content/uploads/2018/06/GPD_TEE_Internal_Core_API_Specification_v1.1.2.50_PublicReview.pdf
Signed-off-by: Igor Opaniuk <igor.opaniuk@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>
When the device is in peripheral mode there is no
struct usb_bus_priv allocated pointer, as the uclass driver
("usb_dev_generic") doesn't call per_device_auto_alloc_size.
This results in writing to the internal SDRAM at
priv->desc_before_addr = true;
Signed-off-by: Stefan Mavrodiev <stefan@olimex.com>
For USB3.0, the max packetsize for GET_DESCRIPTOR should be
sent as exponent value for 2. This means for 512, max packet
size should be filled with 9(2^9=512). Also, fill the USB
version field with 3.0 if speed is negotiated to Superspeed.
This fixes the issue of DFU gadget download failure with
superspeed. Without this patch, the max packet size is
overflowed to zero as the bMaxPacketsize is of u8 and hence
host is not able to detect this device.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
