This patch reverts "Fix FDT relocation" because it breaks
fw_jump firmware.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

The sbi_hart_switch_mode() will be used by CPU hotplug. This means
if we have prints in sbi_hart_switch_mode() then these prints will
mix with Linux prints.

Being a runtime firmware, we should be verbose only at coldboot
time and error situations.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

commit d403c70f

 breaks Linux
booting because of fewer exceptions delegated to S-mode. This
patch fixes Linux booting by having all required exceptions
delegated to S-mode.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Avoid getting messages from multiple harts mingled into garbage text
with a spinlock serializing calls to sbi_puts() and sbi_printf().

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Use functions defined in sbi/riscv_asm.h.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Use this helper instead of hardcoding csr_read(mhartid).

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Cleanup sbi_hart_boot_nexti() code, adding messages for clarity and
rename the function to sbi_hart_switch_mode() to reflect what the
function actually does.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Booting now.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

When S mode is not supported, simply clear mideleg and medeleg.
Otherwise, set delegation from M-mode to S-mode, removing the page
fault exceptions as those would not happen in M-mode.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Make it clear that the end of the FW execution was reached without the
hand being passed to any payload.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Also disable interrupts and exceptions delegation when clearing mie and
mip.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Or rather, disable it if FW_PAYLOAD_FDT_ADDR is not defined. This
correspond to cases where the underlying FW that loaded and started
opensbi did not provide any FDT. In such case, the platform code will
need to prepare one.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Allow a platform to report its supported features in more details.
The new features defined are:
* SBI_PLATFORM_HAS_PMP
* SBI_PLATFORM_HAS_SCOUNTEREN
* SBI_PLATFORM_HAS_MCOUNTEREN

In addition, define the macro SBI_PLATFORM_DEFAULT_FEATURES as the set
of features that are generally expected to be supported by a Linux
capable platform.

Operations touching the features controlled with these falgs are not
executed if the platform does not set the corresponding feature flags.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

We should not jump to next stage if next mode (S-mode or U-mode)
is not supported by HART.

Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Anup Patel <anup.patel@wdc.com>

This patch renames "plat" to "platform" everywhere for better
readablility.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

On QEMU Virt, max supported HARTs are 8 but number of HARTs
actually depend on "-smp" command-line parameter passed to
QEMU. This creates problems in sbi_hart_wake_coldboot_harts()
because when number of HARTs are less than 8.

To tackle this, we introduce a bitmap to track HARTs waiting
for coldboot to finish. We wake only those HARTs who have
set their bit in coldboot bitmap.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

This patch renames "blob" to "firmware" everywhere for better
and intutive naming.

Signed-off-by: Anup Patel <anup@brainfault.org>

This patch removes unwanted sbi_printf() from
sbi_hart_wait_for_coldboot().

Signed-off-by: Anup Patel <anup.patel@wdc.com>

This patch adds documentation for fw_jump and fw_payload
firmware blobs.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

This patch adds more details to top-level README.md based on
current state of OpenSBI.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Various fixes for unleashed board

Signed-off-by: Atish Patra <atish.patra@wdc.com>

Signed-off-by: Atish Patra <atish.patra@wdc.com>

Signed-off-by: Atish Patra <atish.patra@wdc.com>

Signed-off-by: Atish Patra <atish.patra@wdc.com>

Signed-off-by: Atish Patra <atish.patra@wdc.com>

We can only use fw_payload blob on Kendryte K210 board
because there is no previous booting stage hence this
patch enables fw_payload for Kendryte K210 board.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Only compiles for now, completely untested.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Add comments to assembler (not trivial to follow).

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Ignore build and install directories

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

This patch extends fw_payload blob to provide FW_PAYLOAD_FDT_PATH
option using which we can embed custom FDT in .text section of
fw_payload blob. In other words, FW_PAYLOAD_FDT_PATH is an option
to forcefully override FDT passed by previous booting stage in
a1 register.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Instead of placing FDT (passed by previous booting stage) at a
location relative to next address we should have absolute
location of placing FDT so that we more freedom of placing FDT.

This will be particularly useful for platforms/boards with
very less RAM (such as Kendryte board).

Due to above motivation, we rename FW_xyz_FDT_OFFSET options
to FW_xyz_FDT_ADDR options and use FW_xyz_FDT_ADDR options as
absolute address for placing FDT.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

The PLAT_TEXT_START is only used by our firmware blobs hence
rename it to FW_TEXT_START.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

This patch remove hard-coding of payload location in
fw_payload.elf.ldS by adding compile-time option
FW_PAYLOAD_OFFSET.

With the new compile-time option, payload will be placed
at PLAT_TEXT_START + FW_PAYLOAD_OFFSET address.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Various typos fix.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>

Currently, we aggresively use atomic operation on CLINT IPI
register. This patch simplify CLINT IPI APIs by reducing use
of atomic operations. In future, we will gradually increase
use of atomic operations for CLINT IPI APIs.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

The QEMU sifive_u machine is not excatly same as HiFive Unleashed
board hence we add separate platform support for QEMU sifive_u
machine.

Signed-off-by: Anup Patel <anup.patel@wdc.com>

Emulation of CSR read for misa and mhartid was a workaround for
bootloader accessing these CSRs in S-mode. This patch removes
CSR read emulation for misa and mhartid.

Signed-off-by: Anup Patel <anup.patel@wdc.com>