Support for separate firmware and kernel payload is added
by updating BBL to read optional preloaded kernel address
attributes from device-tree using a similar mechanism to
that used to pass init ramdisk addresses to linux kernel.
chosen {
riscv,kernel-start = <0x00000000 0x80200000>;
riscv,kernel-end = <0x00000000 0x80590634>;
};
These attributes are added by QEMU and read by BBL when combining
-bios <firmware-image> and -kernel <kernel-image> options. e.g.
$ qemu-system-riscv64 -machine virt -bios bbl -kernel vmlinux
With this change, bbl can be compiled without --with-payload
and the dummy payload alignment is altered to make the memory
footprint of the firmware-only bbl smaller. The dummy payload
message is updated to indicate the alternative load method.
This load method could also be supported by a first stage boot
loader that reads seperate firmware and kernel from SPI flash.
The main advantage of this new mechanism is that it eases kernel
development by avoiding the riscv-pk packaging step after kernel
builds, makes building per repository artefacts for CI simpler,
and mimics bootloaders on other platforms that can load a kernel
image file directly. Ultimately BBL should use an SPI driver to
load the kernel image however this mechanism supports use cases
such such as QEMU's -bios, -kernel and -initrd options following
examples from other platforms that pass kernel entry to firmware
via device-tree.
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: Alistair Francis <Alistair.Francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
* If BBL emulates the FPU, the trap handler will load emulated FCSR from
x0's save slot into tp. The emulated FCSR should be initialized, else
the field of rounding mode will contain garbage codes. This will
lead to raising SIGABRT for a user mode program which tries to print a
floating point variable. In glibc, __printf_fp_l() (defined in
riscv-glibc/stdio-common/printf_fp.c) will call round_away() (defined
in riscv-glibc/include/rounding-mode.h). With a garbage rounding mode
in emulated FCSR, round_away() may call abort().
The E51 core on the U54-MC lacks supervisor mode, thus the plic_s_ie and plic_s_thresh are NULL when running on this core. This adds checks for this case.
QEMU currently provides the console via HTIF and the SBI. That's a bit
messy because BBL polls for serial input, which means that typing too
quickly loses characters. While QEMU has a standard 16550 device model,
there's no way to have two consoles share the console in QEMU (as they'd
step all over each other) so that means we can't have both the HTIF
console and the 16550 console.
With this patch, QEMU can be changed to use a 16650 instead of the HTIF
for serial output. Linux will use the SBI for early printk support
(which is fine, polling for output is stable) and then swap over as soon
as it detects the UART. When Linux swaps it prints out the whole
history, but there's probably a way to get around that. There's a few
lines that are output to both, but it appears the Linux driver is close
enough to ours that nothing catastrophic happens -- there's not much to
the device, so hopefully that pans out on real hardware too.
Once Linux swaps over to natively using the driver we get reliable
console input. If you don't have the in-kernel driver then Linux never
swaps over and keeps using the SBI console just like before.
Some platforms can't boot Linux on all the harts. This commit allows
platforms to define the set of harts that should be prevented from
booting past BBL. This is essentially just a new mechanism for defining
the DISABLED_HART_MASK.
The clint was wiping out information discovered by the plic.
Initialize hart stacks as they are discovered.
Then fill in clint+plic info
Then wake the harts.
Andrew Waterman
Michael Clark
Zihao Yu
Jim Straus
Wesley W. Terpstra
Zong Li
Palmer Dabbelt
Andrew Waterman