Rearrange some of the cpu initialization code to be runnable on each cpu
as they come up. Complete the 64bit bootstrap mechanism and call into C
code.
Makes it as far as trying to reschedule via an IPI. Need to implement
local apic based IPI mechanism.
On mac qemu, trying to use the physical timer does not seem to work, but
really the virtual timer is what you should be using most of the time,
especially when running under emulation.
Move common logic into a default routine in platform/power that other
platforms can reuse to implement the general default shutdown logic.
Add helper routines to print the cause.
Refactor the platforms that had substantial halt logic to reuse the
default implementation.
Initialize the uart by passing in the base and irq, as well as a flag
specifying if it's the debug uart and should directly put data into the
console buffer (if present).
This was the driver that triggered the whole thing, since GCC 14.1 was
starting to use more fancier addressing modes that was causing QEMU to
bomb out when using KVM.
The accessing method the compiler is emitting for the *REG32 macros on
arm32 and arm64 is occasionally generating load/stores with writeback.
Though this has worked before, it seems to be rejected with whatever
combination of qemu + linux + hardware on this Raspberry Pi 5.
Convert the register accessors to inline asm that uses basic load/store
instructions, which is really the only correct thing to do now and in
the long run. Add a TODO to move this to reg.h and start to revamp how
registers are accessed across LK, but for now keep it just here to fix
things.
Add the fundamental filesystem structure to attach a VirtualIO 9p
device. With the implementation of VirtIO 9p devices (lk/dev/virtio/9p),
we can use those APIs to connect to a shared folder as a LK filesystem.
Signed-off-by: Cody Wong <codycswong@google.com>
Three platforms had basically duplicated logic that just was pulled into
the fdtwalk library. Fix these up to call into those routines instead.
-qemu-virt-arm
-qemu-virt-riscv
-jh7110 (visionfive 2 soc)
A pretty simple mechanism, a list of extensions added to
RISCV_EXTENSION_LIST make variable is expanded to an underscore
delimited string appended to the end of -march=
Pretty simple but it should work for now.
Instead of setting a counter of the number of secondaries to start, have
platform or target code pass in a list of harts to start instead. This
allows for there to be discontinuties in the layout of the cpu harts, or
in the case of some sifive based hardware, hart 0 is otherwise offline.
- Add the v9p support for arm platform.
- Add an option `-f` to set the shared directory on the host machine to
`do-qemuarm`.
- For example, to use the v9p dir sharing, run the script as:
```
# under `lk/` run the follow command will share the current
# directory `lk/` within LK vm as VirtIO 9p device
scripts/do-qemuarm -f .
```
Signed-off-by: Cody Wong <codycswong@google.com>
Use the legacy version of the memory sizing info that just hands the
kernel 2 implicit ranges: 0 ... X and 1MB ... Y.
Not ideal, but when booting on very old machines without BIOS e820 call
implemented it's all you got.
-Reinitialize the uart bit modes
-Ignore framing/break errors on receive. Unclear why they seem to always
be set in the fifo, but seems to be safe to ignore for now
-Make sure platform_pgetc returns the right error code
-Initialize the output ports to all gpios
Clang's assembler rejects expressions containing e.g. (1u << N) in the
assembler. Instead using numeric expressions for per-privilege level
CSRs, we can prepend `m` or `s`. This also lets the compiler assign the
CSR encoding instead of having to hardcode it in the source code.
Looks like this is described in /reserved_memory/mmode_resv@<address> as
patched by opensbi on top of the existing qemu provided dtb.
Future patch should parse this instead of hard coding it.
Move the mmu_initial_mapping from platform into arch/x86. For this
architecture the default mappings are basically hard coded in arch/x86
anyway, so move ownership of this data there, closer to where it's
actually initialized.
Update the 32 and 64bit paging code to properly use the paddr_to_kvaddr
and vice versa routines.
Update to allocate page tables directly from the pmm instead of the heap
(on 32bit code).
The old method assumed that all of the tables were mapped within the
kmap area of the kernel. This basically works on 64bit machines but on a
32bit x86 its entirely likely the ACPI tables are at higher physical
addresses that can be reached, which is currently limited to 1GB.
By using the VM it means it can individually map the headers and each
individual table.
GLOBAL_CFLAGS will only work for .c files, which breaks on the first
inclusion of the headers by a .cpp file. *_COMPILEFLAGS is for all
language types (C, asm, C++)
Parse up to 16 pmm arenas from the multiboot memory data structure. Roll
the 32bit code to properly trim at 1GB as before, but using new logic.
Remove conditional checks on WITH_KERNEL_VM in x86 code, which only
really compiles with the mmu and the vm on.
The IRQ calculation for the virtio range was off by 8, which caused the
virtio code to override the interrupt registration for the RTC, which
caused it to stop firing. Clean up the #defines that define irq mappings
to fix this issue.
Based on building with --warn-undefined-variables, find a few places in
the build system where undefined variables were used incorrectly, or
never set due to unused code.
If we were booted at EL2 (e.g. when passing -machine
virt,virtualization=on), we need to use SMC instead of HVC for PSCI
calls. Change psci_call() to do this and add a flag to do-qemuarm to
allow testing this scenario.
