currently esp32 protected mode requires a patched bootloader.
it's a bit cumbersome to build the bootloader for that purpose.
this commit attempts to remove the need of the patched bootloader
by applying the changes by ourselves using esp hal.
Using the ts/tick conversion functions provided in clock.h
Do this caused we want speed up the time calculation, so change:
clock_time2ticks, clock_ticks2time, clock_timespec_add,
clock_timespec_compare, clock_timespec_subtract... to MACRO
Signed-off-by: ligd <liguiding1@xiaomi.com>
Enforcing the default 48-bit VA for everyone also implies a 4 page table
translation system. However, if less than 40 bits are needed, a full
translation table level can be dropped, making the translations faster.
Thus, make this into a configurable option, instead of enforcing the same
address widht for everyone.
Attempt to service all interrupts pending in the PLIC's claim register. Ideally, this is more efficient than switching context for each interrupt received.
Provides two implementations:
- CSR_CYCLE: Cores which implement hardware performance monitoring.
- CSR_TIME: Uses the machine time registers.
Using the up_perf_xx bindings directory is more efficient than performing a nanosecond conversion on every gettime event.
There is a tiny possibility that when a process is started a trap is
taken which causes a context switch. This moves the kernel stack
unexpectedly and the task start logic no longer works.
Fix this by recording the initial context location, and use that to
trampoline into the user process with interrupts disabled. This ensures
the context stays intact AND the kernel stack is fully unwound before
the user process starts.
This patch unifies the extended context save/restore for RISC-V,
allowing the customized context save/restore to be used, for example,
the extended context in rv32m1.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
The register context is not needed, the original idea was to provide
the user stack pointer for signal handler delivery, but the user stack
can be obtained via sp_el0 so the context registers are not needed.
SP0 is not stored upon exception entry anyways, so this code is just
completely redundant and wrong.
reason:
In SMP, when a context switch occurs, restore_critical_section is executed.
To reduce the time taken for context switching, we directly pass the required
parameters to restore_critical_section instead of acquiring them repeatedly.
Signed-off-by: hujun5 <hujun5@xiaomi.com>
The vaddr field in TLBI means: Bits[55:12] of the virtual address to match.
This basically means the page offset of the virtual address, so the input
vaddr must be shifted to the page offset.
Reference TLBI VALE1IS register description from ARMv8-A reference manual.
The 12:0 bits in table descriptors are RES0 and AF is the 10th bit, so
it is not valid to set it in this case.
Fix this by moving AF to the common MMU_MT_NORMAL_FLAGS field
Make sure the user L1 page is updated to system memory when the kernel
mappings are copied.
Also, flush the I-cache when switching address environments.
The libmad and libmp3lame libraries have been ported to
apps/audioutils and will be compiled with NuttX.
Signed-off-by: shizhenghui <shizhenghui@xiaomi.com>
We can see them in ifconfig:
ap> ifconfig
wlan0 Link encap:Ethernet HWaddr 42:64:7f:b3:12:03 at UP mtu 1500
inet addr:10.0.1.2 DRaddr:10.0.1.1 Mask:255.255.255.0
inet6 DRaddr: ::
RX: Received Fragment Errors Bytes
00000b9b 00000000 00000000 21daf5
IPv4 IPv6 ARP Dropped
00000a33 00000137 00000031 00000000
TX: Queued Sent Errors Timeouts Bytes
00000ac4 00000ac4 00000000 00000000 1a2103
Total Errors: 00000000
Signed-off-by: meijian <meijian@xiaomi.com>
Since the introduction of the rp2350, the pico-sdk project has been
re-arranged and elf2uf2 has been merged in to a new 'picotool' binary
which is supplied and installed separately.
RDTSCP instruction reads the current value of the processor’s
time-stamp counter (a 64-bit MSR) into the EDX:EAX registers, and it
also reads the value of the IA32_TSC_AUX MSR (address C0000103H) into
the ECX register. However, the current RDTSCP implementation does not
provide a hint for the compiler that ECX has been changed, resulting in
register corrupted and subtle errors.
Signed-off-by: ouyangxiangzhen <ouyangxiangzhen@xiaomi.com>
Make umm & kmm heap alloc more clear, and possible support FLAT build to
enable kernel heap.
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
Signed-off-by: buxiasen <buxiasen@xiaomi.com>
1. correct the armv8-r fpu flag from vfpv3-d16 to fpv5-sp-d16
2. add neon support
This commit updates the build script to specify the FPU type of
`fpv5-sp-d16` when the processor is configured with a single-precision
FPU, and `neon-fp-armv8` when the processor is configured with a
double-precision + Advanced SIMD-capable FPU.
Note that the `fp-armv8` FPU type is an alias for double-precision FPv5
with 32 double-precision registers (refer to the GCC
`gcc/config/arm/arm-cpus.in` for more details); NEON is always
specified in case of a double-precision configuration because the
Cortex-R52 can only be configured as such.
Reference:
https://developer.arm.com/Processors/Cortex-R52
Instruction set:
Arm and Thumb-2. Supports DSP instructions and a configurable Floating-Point
Unit either with single-precision or double precision and Neon.
Signed-off-by: chao an <anchao@lixiang.com>
Make this_cpu is arch independent and up_cpu_index do that.
In AMP mode, up_cpu_index() may return the index of the physical core.
Signed-off-by: fangxinyong <fangxinyong@xiaomi.com>
reason:
dynaminc create g_irqmap to reduce the use of data segments
CONFIG_ARCH_NUSER_INTERRUPTS should be one more than the number of IRQs actually used
Signed-off-by: hujun5 <hujun5@xiaomi.com>
In corner case, the pending ISR will be triggered immediately
after enable the IRQ, this PR will setting CPU affinity first
to avoid routing the unexpected IRQ to other CPUs.
Signed-off-by: chao an <anchao@lixiang.com>
Copied spi flash files from risc-v
Removed old spiflash from make.defs
KConfig modifications for common spiflash
Added new spiflash entry to make.defs
S2 Board config update
Modifications working on S2. Need fix on ROM function
Board fixes
Deleted old spiflash files
Fixes for smartfs operation
Set new HAL version
Fixed wrong change on S3
Cache flush must be done prior to the APP cpu initalization. This,
however, must be true for the case where PSRAM is not available or
not selected. To do that, this commit flushs the cache during the
device initialization.
This commit fixes a deadlock in `esp32s3-devkit:sta_softap`
defconfig: `spin_lock_irqsave` was being used to enter a critical
section that calls `nxsem_post`. In this case, it's recommended
to use `[enter|leave]_critical_section` to avoid deadlocks when a
context switch may happen, for instance.
NuttX uses PSRAM, possibly using the APP_CPU cache MMU,
way before starting the APP_CPU in up_cpu_start().
Flushing the cache when launching the APP_CPU can cause
data corruptions on PSRAM.
Eg. mm_heap structures if the PSRAM is added to a heap.
Some USB controllers can receive or send multiple data packets then
generate one interrupt. This mechanism can reduce the number of data
copies. Extend req buf to accommodate this.
Signed-off-by: yangsong8 <yangsong8@xiaomi.com>
This PR configures the BL808 MMU to cache the the User Text, Data and Heap. We enable the T-Head MMU Flags for Shareable, Bufferable and Cacheable, as explained in the previous PR: https://github.com/apache/nuttx/pull/13199
This PR fixes the Slow Memory Access for NuttX Apps on Ox64 BL808 SBC: https://github.com/apache/nuttx/issues/12696. With this fix, Ox64 NuttX CoreMark jumps from 19 to 1,104. (Close to Buildroot Linux CoreMark)
Modified Files:
`arch/risc-v/Kconfig`: Enabled `ARCH_MMU_EXT_THEAD` for BL808 SoC.
In protected build mode, the syscall returns without setting correct value for the CONTROL register, which causes the userspace threads working in privileged mode after syscall.
Signed-off-by: pangzhen1 <pangzhen1@xiaomi.com>
This PR configures the T-Head MMU to cache the the User Text, Data and Heap. We enable the MMU Flags for 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 Apps on BL808 and SG2000 SoCs: https://github.com/apache/nuttx/issues/12696. With this fix, SG2000 NuttX CoreMark jumps from 21 to 2,423. (Close to SG2000 Debian CoreMark)
We introduce a Kconfig Option: `ARCH_MMU_EXT_THEAD` ("System Type > Enable T-Head MMU extension support"). Enabling this Kconfig Option will configure the T-Head MMU to cache the User Text, Data and Heap.
This PR enables the MMU cache for only SG2000 SoC (Milk-V Duo S SBC). The next PR will apply the same settings to BL808 SoC (Pine64 Ox64 SBC).
Modified Files:
`arch/risc-v/Kconfig`: Added Kconfig Option `ARCH_MMU_EXT_THEAD` that will configure the T-Head MMU. Enabled `ARCH_MMU_EXT_THEAD` for SG2000 SoC.
`arch/risc-v/src/common/riscv_mmu.h`: Set the T-Head MMU Flags (Shareable, Bufferable and Cacheable) for User Text, Data and Heap, if `ARCH_MMU_EXT_THEAD` is enabled
`arch/risc-v/src/common/riscv_addrenv.c`: Extended the MMU Flags from 32 bits to 64 bits, to accommodate the T-Head MMU Flags
`arch/risc-v/src/common/riscv_exception.c`: Extended the MMU Flags from 32 bits to 64 bit, to accommodate the T-Head MMU Flags. This code is enabled only for MMU Paging (`CONFIG_PAGING`).
Currently concerns only arm.
tools/ci/testlist/msys2.dat:
At the moment I only added the board nucleo-l152re:nsh
.github/workflows/build.yml:
Enabled cmake for msys2
since not all Armv7-m platform support double-precision floating
operations, so we add an extra option that help those specific platform
could define the single-precision floating operations.
for example, qemu-mps platform(armv7-m) support double-precision
floating operations, so we do not need to open
CONFIG_ARM_FPU_SINGLE_HARDWARE build options;
but for fc7300 platform(armv7-m) only support single-precision floating
operations, so if we pass "-fhard" build option to "ghs" compiler, then
the following instruction will be generated:
vcvt.f64.s32 d9, s18
which will fail to execute on fc7300 platform.
with "-fsingle" build option, the "ghs" compiler will generate hardware
instruction for single-precision floating operation, and generate
software instruction for double-precision floating operation.
Signed-off-by: guoshichao <guoshichao@xiaomi.com>
To support greenhill compilation, LDFLAGS removes the code that adds -T through addprefix to ARCHSCRIPT(greenhill requires a space between -T and ldscript files, but addprefix didn't do it). However,if we modified like this, when ARCHSCRIPT has more than one ldscript, the -T is only added to the first ldscript, but not to the following ldscripts, which results in a warning when linking.
To solve this problem, we can just add a space after $(SCRIPT_OPT) when using addprefix, instead of deleting addprefix.
Signed-off-by: pangzhen1 <pangzhen1@xiaomi.com>
Wait for CMDDONE in esp32s3_waitresponse even an error response is received.
Otherwise, the CMDDONE event of this command will disrupt later command.
Signed-off-by: Yinzhe Wu <Yinzhe.Wu@sony.com>
Reviewed-by: Yuezhang Mo <Yuezhang.Mo@sony.com>
Reviewed-by: Jacky Cao <Jacky.Cao@sony.com>
Tested-by: Yinzhe Wu <Yinzhe.Wu@sony.com>
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 released
we also need keep restore_critical_section in svc call
test:
Configuring NuttX and compile:
$ ./tools/configure.sh -l qemu-armv8a:nsh_smp
$ make
Running with qemu
$ 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>
This is the initial version for kernel mode build on the arm64 platform.
It works much in the same way as the risc-v implementation so any
highlights can be read from there.
Features that have been tested working:
- Creating address environments
- Loading init (nsh) from elf file
- Booting to nsh
- Starting other processes from nsh
- ostest runs to completion
Features that are not tested / do not work:
- SHM / shared memory support
- Kernel memory mapping (MM_KMAP)
- fork/vfork
An example qemu target is provided as a separate patch:
tools/configure.sh qemu-armv8a:knsh
Why? Because this allows optimizing the user system call path in such
a way that the parameter registers don't have to be read from the saved
integer register context when the system call is executed.
Some hardware devices use discontinuous framebuffers, which require SIM support for simulating discontinuous framebuffers.
Signed-off-by: jianglianfang <jianglianfang@xiaomi.com>
There is a false positive in asan of gcc-13, we need to mask it to ensure that other parts work properly
sanitizer_common/sanitizer_common_interceptors.inc:
// FIXME: under ASan the call below may write to freed memory and corrupt
// its metadata. See
// https://github.com/google/sanitizers/issues/321.
==572161==ERROR: AddressSanitizer: dynamic-stack-buffer-overflow on address 0xed3d3f00 at pc 0xef46af64 bp 0xed3d3de8 sp 0xed3d39bc
WRITE of size 128 at 0xed3d3f00 thread T0
#0 0xef46af63 in __interceptor_pthread_sigmask ../../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:4419
#1 0x5486aa7d in up_irq_save sim/posix/sim_hostirq.c:97
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
It can only read the contents of the first buffer, so fblen should be changed to ensure that it can read the second buffer as well.
Signed-off-by: jianglianfang <jianglianfang@xiaomi.com>
armv7-m/arm_mpu.c: In function 'mpu_dump_region':
armv7-m/arm_mpu.c:621:13: warning: format '%X' expects argument of type 'unsigned int', but argument 4 has type 'long unsigned int' [-Wformat=]
621 | _info("MPU-%d, alignedbase=0%08X l2size=%"PRIu32" SRD=%X"
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
armv7-m/arm_mpu.c:621:13: warning: format '%X' expects argument of type 'unsigned int', but argument 6 has type 'long unsigned int' [-Wformat=]
621 | _info("MPU-%d, alignedbase=0%08X l2size=%"PRIu32" SRD=%X"
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
armv7-m/arm_mpu.c:621:13: warning: format '%X' expects argument of type 'unsigned int', but argument 7 has type 'long unsigned int' [-Wformat=]
621 | _info("MPU-%d, alignedbase=0%08X l2size=%"PRIu32" SRD=%X"
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
armv7-m/arm_mpu.c:621:13: warning: format '%u' expects argument of type 'unsigned int', but argument 8 has type 'long unsigned int' [-Wformat=]
621 | _info("MPU-%d, alignedbase=0%08X l2size=%"PRIu32" SRD=%X"
Signed-off-by: fangxinyong <fangxinyong@xiaomi.com>
Due to the xtensa mpu feature, the size of the Region depends on the Base of the next Region. e.g.
Region[1] = 0x20000000
Region[0] = 0x30000000
Then Region[1] length = Region[0] - Region[1]
So this approach is not suitable to implement the behavior of cleaning
up the Region and such a configuration will result in affecting the very beginning (the higher) Region
Therefore, to address this feature, in this change we return the Region
value and implement the ability to modify the target Region's attributes
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
Modified the input parameters of mpu_initialize to require the caller to provide the number of entries in the table
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
Summary:
1.Modify the conditions for entering different include header files
2.Added pre-definition for _Atomic _Bool when it is missing
3.Added nuttx for stdatomic implementation. When toolchain does not support atomic, use lib/stdatomic to implement it
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
Summary:
1. Add nuttx/atomic.h in include
2. Use nuttx/atomic.h instead of stdatomic.h in the nuttx directory
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
Update internal reference to get the most updated Espressif's
libraries. Those libraries are based on branch `release/v5.1` of
the ESP-IDF and include `v5.1.4` version of it.
sim/posix/sim_alsa.c:728:24: runtime error: member
access within null pointer of type 'const struct sim_codec_ops_s'
Signed-off-by: jinxiuxu <jinxiuxu@xiaomi.com>
In a multicore task scenario, there may be a situation where the task runs on different cores at different time slices (when the task is not bound to a particular core).
When the task calls cache clean/flush(range > cache size), depending on the optimization, clean_all, flush_all are called. however, at this point, there may be dirty data or incomplete data profiles in the cache on the kernel that is running the task, which may result in dirty data being flushed into memory or make the application think that the flushed data should be successfully flushed into memory, leading to unknown consequences.
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
In a multicore task scenario, there may be a situation where the task runs on different cores at different time slices (when the task is not bound to a particular core).
When the task calls cache clean/flush(range > cache size), depending on the optimization, clean_all, flush_all are called. however, at this point, there may be dirty data or incomplete data profiles in the cache on the kernel that is running the task, which may result in dirty data being flushed into memory or make the application think that the flushed data should be successfully flushed into memory, leading to unknown consequences.
Signed-off-by: chenrun1 <chenrun1@xiaomi.com>
Execute data and instruction sync barriers after writing MPU register,
to ensure MPU setting take effects that the new changes are seen.
testing in lm3s6965-ek:qemu-protected
Signed-off-by: fangxinyong <fangxinyong@xiaomi.com>
when the vsync comes, fb drivers should call fb_vsync_pollnotify to notify POLLPRI, so that users can catch the synchronization time of vsync and do something.
Signed-off-by: jianglianfang <jianglianfang@xiaomi.com>
