When devif_loopback handles a packet (like a ping targeting at this dev), it does not call the txdone callback, breaking the tx pipeline and may left some packets unhandled, delayed until next transmit on the network interface.
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
Share one worker between multiple simulated network devices may work most of the time, but sometimes breaks the tx pipeline when sending packets on more than one interface at the same time, and leaves some packets unprocessed in network stack, delayed until next transmit on the network interface. The rx process is likely delayed in packet processing under similar situation, so keep g_avail_work and g_recv_work the same number as interfaces.
dev0 tx1 avail tx1 done
v v
work dev0 tx1 -> dev0 tx1 -> empty -> dev1 tx2 -> dev1 tx3 -> empty
^ ^ ^ ^
dev1 tx2 avail (failed to queue) tx3 avail tx2 done tx3 done
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
common/arm_backtrace_fp.c: In function 'up_backtrace':
common/arm_backtrace_fp.c:126:23: warning: assignment to 'void *' from 'uintptr_t' {aka 'unsigned int'} makes pointer from integer without a cast [-Wint-conversion]
126 | istacklimit = arm_intstack_top();
| ^
Signed-off-by: chao an <anchao@xiaomi.com>
This change adds the following:
- Rename the board configuration name from qemu-a53 to qemu-v8a.
- Add the configurations for Cortex-A57 and Cortex-A72.
Signed-off-by: Hidenori Matsubayashi <hidenori.matsubayashi@gmail.com>
Use CONFIG_ARCH_CHIP_QEMU instead of CONFIG_ARCH_CHIP_QEMU_A53. This is because these configurations depend on the hardware configuration of qemu (memory map, irq numbers, etc.).
Signed-off-by: Hidenori Matsubayashi <hidenori.matsubayashi@gmail.com>
Support more than one TAP device for debugging (future) NAT and FORWARD, left WPCAP and VPNKit unchanged (force number to 1).
Although we can support at most 31 interfaces, limit to 8 like CONFIG_TUN_NINTERFACES.
Signed-off-by: wengzhe <wengzhe@xiaomi.com>
1. rename arm_backtrace_thumb.c to arm_backtrace_sp.c
2. use EHABI stack unwinder instead of instruction unwind
Signed-off-by: chao an <anchao@xiaomi.com>
Make a separate config flag for enabling L2 cache. This is on by
default when compiling a standalone/bootloader configuration, but
can also be disabled for special cases, such as memory testing
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
This change moves the static vars to the appropriate place to follow the coding standard. It also adds comments as sections and `static` where necessary.
Signed-off-by: Hidenori Matsubayashi <hidenori.matsubayashi@gmail.com>
The default size in QEMU system is 128MB, and the size specified in chip.h is also 128MB. However, the region size for MMU was 512MB, so fixed it.
Signed-off-by: Hidenori Matsubayashi <hidenori.matsubayashi@gmail.com>
Summary:
- This commit refactors the entry point name for BUILD_KERNEL
to avoid misunderstanding the name
Impact:
- None
Testing:
- Tested with rv-virt:ksmp64
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
Summary:
- I noticed that the OS timer sometimes proceeds fast when
a task is scheduled to run on CPUO via IPI.
- Actually, qemu-rv implementation shares supervisor software
interrupt for both timer and IPI on CPU0.
- This commit fixes this issue.
Impact:
- qemu-rv only
Testing:
- Tested with qemu-6.2
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
Summary:
- This commit adds SMP support for BUILD_KERNEL
Impact:
- RISC-V: BUILD_KERNEL + SMP only
Testing:
- Tested with rv-virt:ksmp64 (will be added later)
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
Run-time check for L3 page table size, to ensure it is large enough
to map all of the kernel memory.
NOTE: The check has to be run-time, as KFLASH_SIZE/KSRAM_SIZE are really
linker relocation symbols, and thus cannot be utilized compile-time.
L3 table maps 2MB of memory, this brings an implicit requirement for
any L3 region to be aligned to 2MB. This commit adds build time sanity
checks to ensure this requirement is met.
For other SvXX architectures the boundary requirement (might be) is
different.
Summary:
- This commit introduces g_percpu_spin to avoid deadlock
in riscv_percpu.c instead of using the global spinlock.
Impact:
- None
Testing:
- Tested with rv-virt:knsh64 and rv-virt:ksmp64 (will be added later)
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
