signal handler stack must be properly aligned, otherwise vector instructions doesn't work in signal handler
Signed-off-by: p-szafonimateusz <p-szafonimateusz@xiaomi.com>
move initial RSP for AP cores below regs area.
otherwise IDLE thread for AP cores can be corrupted
XCP region now match regs allocation in up_initial_state()
Signed-off-by: p-szafonimateusz <p-szafonimateusz@xiaomi.com>
So the driver can get the driver pointer by vdev/hdev->priv,
and later vdev/hdev->priv can be used to store other data such as the
virtio/vhost drivers' private data.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
vhost-rng is a simple virtio device implementation, it receives
the buffer receviced from the virtio-rng driver and fill the random
number to this buffer and return to virtio-rng.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
virtio is a framework to implement the virtio drivers
vhost (Virtual Host) is a framework to implement the virtio devices
With the virtio and vhost framework, we can use the virtio drivers
and vhost drivers to implement the cross-core communication.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
When building a branch like `releases/12.7`, the CI Workflow incorrectly checks out the `master` branch of `nuttx-apps`, instead of `releases/12.7`. This PR fixes a typo in `apps_ref`, to check out the correct branch.
This PR proposes to enhance the CI Workflow, to skip the unnecessary NuttX Builds. Currently, NuttX Devs wait for the CI Builds to complete across All Architectures (Arm32, Arm64, RISC-V, Xtensa), even though they modified a Single Architecture. With this PR, the CI Workflow will build only the Modified Architecture.
The solution uses the Arch Labels for PRs. We target only the Simple PRs: One Arch Label + One Size Label (e.g. "Arch: risc-v, Size: XS")
- If "Arch: arm / arm64" is the only non-size label, then we build `other`, `arm-01`, `arm-02`, ...
- If "Arch: risc-v" is the only non-size label, then build `risc-v-01`, `risc-v-02`
- If "Arch: xtensa" is the only non-size label, then build `xtensa-01`, `xtensa-02`
- The above rules apply when the PR is Created or Modified
- When the PR is Merged: All targets shall be built
The code is explained here: https://github.com/apache/nuttx/issues/13775
use reserved[2] in struct resource_table as the command, avoid the
resource table format do not follow the standard.
Signed-off-by: Yongrong Wang <wangyongrong@xiaomi.com>
Since FPU is now always saved into the current process stack location
upon exception entry, there is no need to keep fpu_regs (or saved_fpu_regs)
in the TCB.
1. Add cmake support for rptun_serure;
2. Move rptun_secure.c before rptun_ivshmem.c in Make.defs to follow
the order in Kconfig.
Signed-off-by: Bowen Wang <wangbowen6@xiaomi.com>
The request length may not exceed CONFIG_CDCACM_BULKIN_ lenghts, otherwise buffer overflow will occur
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
Correct some of the cache operations:
- EP0 request length was handled incorrectly
- Received data cache invalidate was exceeding the received buffer
- writedtd is also called with no data (EP0 ACK/NACK). Don't touch cache in that case.
Fix trip wire handling to conform with the IMX93 reference manual
Also add DEBUGASSERTS for future to check the validity of pointers and sizes
Co-authored-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
read content of undefinedinsn address, and compare it with what it is in elf to check if there is a ram bit flip
Signed-off-by: liaoao <liaoao@xiaomi.com>
add_compile_options(--param asan-globals=1) is recognized as
--param-lasan-globals=1, which causes compilation exception:
Signed-off-by: W-M-R <Mike_0528@163.com>
wd_timer handler will change current task, so need process
scheduler before wd timer handler, Ensure that the current
task is not updated before the RR process
Signed-off-by: zhangyuan29 <zhangyuan29@xiaomi.com>
As the handling of sp_el0 was moved from the context switch routine
to exception entry/exit, we must set sp_el0 explicitly when the user
process is first started.
Example:
When executing "df -h" on Core A to view mount information, this
process will traverse inode nodes, thereby holding the inode_lock.
Since the inode type of the mount point may be rpmsgfs, it will fetch statfs
information from another Core B.
Meanwhile, rcS on Core B needs to obtain file information from Core A,
which will be achieved by fetching stat information through rpmsgfs.
When this message arrives at Core A, a deadlock can occur between Core A's
rptun ap and nsh task.
However, both of these places involve read operations only, thus a reader-writer lock
can be utilized to prevent such a deadlock.
Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>
Matteo Golin