When executing fork() via a system call, the parent's stack gets corrupted
by the child, as during exception return the child loads the parent's
stack pointer from the context save area.
This happens because the full parent stack (including what has been pushed
during the system call) is copied to the child. What should be copied, is
only the user stack of the parent (the kernel stack is not interesting).
Fix this by only copying the parent's user stack to the child; and make
the child return directly to userspace (not via dispatch_syscall).
Simplifies the implementation of dispatch_syscall, making it easier to
understand and maintain. Let the C-compiler do most of the work, instead
of doing everything as inline assembly.
This revises PROTECTED build for qemu-rv mainly to avoid hard-coded
addresses in linker scripts. It also added rv32 support, cleaned up
config `pnsh64` and added config `pnsh`.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
This adds NuttSBI support for rv-virt device so that to enable CI
checks for NuttSBI later. It allows using `-bios nuttx` option to
run NuttX with QEMU v6.2 w/o OpenSBI.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
This adjusts sbi_start behavior so that if SBI late initialization hook
is defined, PMP setting is assumed to be fully done in the hook, because
the default PMP operation may fail for various reasons such as lacking
free entries or the default setting doesn't work on devices like QEMU
v6.2.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
These symbols only used in esp_spi.c and too simple
to conflict with others, so mark them as static:
- cfg
- ctx
- dev_cfg
- timing_param
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
Fixes regression from https://github.com/apache/nuttx/pull/12220
Error: chip/k230_start.c:80:13: error: 'k230_copy_init_data' defined but not used [-Werror=unused-function]
80 | static void k230_copy_init_data(void)
| ^~~~~~~~~~~~~~~~~~~
Steppoint can be implemented by icount(instruction count)
from RISC-V debug extension, but it may not implemented in all RISC-V cores.
Unfortunately, the currently supported RISC-V cores do not implement it.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
This change implements a driver for the two watchdog timers on the BL808, and enables it as part of the timer config. The driver is based on the GP timer driver.
when the thread to backtrace is exiting, get_tcb and up_backtrace in
different critical section may cause try to dump invalid pointer, have
to ensure the nxsched_get_tcb and up_backtrace inside same critical
section procedure.
Signed-off-by: buxiasen <buxiasen@xiaomi.com>
This fixes the always zero time in NuttSBI case as the `error` field
contains the correct timestamp. This in turn also fixes the block of
`up_udelay()` which further leads to boot failure of `canmv230:master`
config.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
The common `up_allocate_heap` is a strong symbol, which leads to
error when building with CONFIG_DEBUG_LINK_WHOLE_ARCHIVE. Make
`up_allocate_heap` weak function to mitigate this problem.
Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
This PR configures the BL808 and SG2000 MMU (inside T-Head C906) to cache the the Kernel Text, Data and Heap. We set the MMU Flags (Shareable, Bufferable and Cacheable) as explained in this article: https://lupyuen.github.io/articles/plic3#appendix-mmu-caching-for-t-head-c906
This PR fixes the Slow Memory Access for NuttX Kernel in BL808 and SG2000: https://github.com/apache/nuttx/issues/12696
In the next PR, we will fix the Slow Memory Access for NuttX Apps, by caching the User Text and Data.
arch/risc-v/src/bl808/bl808_mm_init.c: Added MMU Flags (Shareable, Bufferable and Cacheable) for BL808 Kernel Text, Data and Heap
arch/risc-v/src/sg2000/sg2000_mm_init.c: Added MMU Flags (Shareable, Bufferable and Cacheable) for SG2000 Kernel Text, Data and Heap
This fixes stack pointer restore logic to avoid parent stack corruption
by forked child in PROTECTED build.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
This fixes `tp` value of forked child in PROTECTED build to support
vfork. Why? the optimized `dispatch_syscall` requires `tp` to hold
the task TCB since commit e6973c764c.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
This fixes the following issue when configuring canmv230:master
with Ubuntu stock toolchain and cmake 3.22:
```
CMake Error at .../arch/risc-v/src/cmake/Toolchain.cmake:218 (if):
if given arguments:
"STREQUAL" "GNU_RVG"
Unknown arguments specified
```
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
This revises canm230:pnsh in a few ways:
- adjusting linker scripts structure,
- asserting PMP setting results,
- adjusting configs for both k230d and k230 devices.
Signed-off-by: Yanfeng Liu <yfliu2008@qq.com>
X0 is always 0 but still should be transfered to the client, but it don't existed in thread context, use any other register for it, and its value will be omitted by gdb client.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
This commit implements a character driver for the general purpose ADC of the BL808, based on the hardware scanning functionality. The driver supports setting different conversion resolutions and the order of channels to be scanned via menuconfig.
Only in the non-critical region， nuttx can the respond to the irq and not hold the lock
When returning from the irq, there is no need to check whether the lock needs to be restored
test:
We can use qemu for testing.
compiling
make distclean -j20; ./tools/configure.sh -l qemu-armv8a:nsh_smp ;make -j20
running
qemu-system-aarch64 -cpu cortex-a53 -smp 4 -nographic -machine virt,virtualization=on,gic-version=3 -net none -chardev stdio,id=con,mux=on -serial chardev:con -mon chardev=con,mode=readline -kernel ./nuttx
Signed-off-by: hujun5 <hujun5@xiaomi.com>
