We may just free some TCP connections before monitor stopped, e.g.
sacrificie a TCP conn in `tcp_alloc` will just call `tcp_free` and reuse
the connection. But we noticed that the TCP monitor is not released in
`tcp_free` because it is mounted on `conn->connevents` instead of
`conn->sconn.list` while `tcp_free` only release the latter.
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
Fix the problem when vfork() calls execv() (or execl()) to start a new application:
When the parent thread calls vfork() it receives and gets the pid of the vforked task,
and not the pid of the desired execv'ed application.
issue #3334
Signed-off-by: yangyalei <yangyalei@xiaomi.com>
with nonblocking sockets, POLLOUT is returned when the application
calls poll even if the send buffer is full
Signed-off-by: zhanghongyu <zhanghongyu@xiaomi.com>
When we are using newlib as libm, we can consider newlib-esp32 for esp32
hardware. An it contains some optimization for esp32 hardware.
Besides, two small issues has been fixed.
1. add the tar ball to .gitignore
2. Fix the some build warnings
These values are only provided by bream private message, should not include in
NMEA 0183 standard struct.
Signed-off-by: haopengxiang <haopengxiang@xiaomi.com>
We support search directory structure as follows:
Current directory
stm32/configs/common
stm32/common/configs
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
Normally, `SO_ERROR` for disconn events will be set by `tcp_poll_eventhandler`, but when the socket is closed before poll, we should also set the `SO_ERROR`.
On Linux, `tcp_poll` returns `EPOLLERR` event when `sk->sk_err` has value but doesn't let `poll` fail (doesn't set `errno`). https://github.com/torvalds/linux/blob/v6.5/net/ipv4/tcp.c#L594-L596
Note: `sk->sk_err` can be get by socket option `SO_ERROR` on Linux, so `POLLERR` will always be together with `SO_ERROR`.
Common libs like curl may try to read `SO_ERROR` on `POLLERR`.
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
A dead lock may happen before this patch:
rptun thread (usrsock_rpmsg_recvfrom_handler):
Gets all tx payload but waiting net_lock in psock_recvfrom
net driver thread (higher priority):
Receives a packet, holding net_lock, calling usrsock_rpmsg_send_event
But no tx buffer left, so rpmsg_send blocks, and won't release net_lock
In short:
Rptun: Hold all `tx_payload_buffer` -> want `net_lock`
Driver: Hold `net_lock` -> want `tx_payload_buffer`
This patch use net_lock to combine get_tx_payload and recvfrom together,
then:
- If it's waiting net_lock, tx payload will not be held, other threads
may get tx buffer and do their work.
- If net_lock is got, it won't be disturbed before finish receiving,
then tx payload won't have chance to be blocked before releasing.
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
If the gulp size in the stdio buffer the remaining user buffer size it will:
- Corrupt memory in dest (user memory) and
- Keep corrupting KERNEL memory via the stdio character buffer until the
whole system crashes, as the 'remaining' count underflows
This patch fixes this behavior.
Summary
include/arch/syscall.h: Assembler messages:
include/arch/syscall.h:133: Error: bad instruction `struct arm_smccc_res'
include/arch/syscall.h:134: Error: junk at end of line, first unrecognized character is `{'
include/arch/syscall.h:135: Error: bad instruction `unsigned long a0'
include/arch/syscall.h:136: Error: bad instruction `unsigned long a1'
include/arch/syscall.h:137: Error: bad instruction `unsigned long a2'
include/arch/syscall.h:138: Error: bad instruction `unsigned long a3'
include/arch/syscall.h:139: Error: bad instruction `unsigned long a4'
include/arch/syscall.h:140: Error: bad instruction `unsigned long a5'
include/arch/syscall.h:141: Error: bad instruction `unsigned long a6'
include/arch/syscall.h:142: Error: bad instruction `unsigned long a7'
include/arch/syscall.h:143: Error: junk at end of line, first unrecognized character is `}'
include/arch/syscall.h:145: Error: bad instruction `typedef struct arm_smccc_res
Signed-off-by: wangming9 <wangming9@xiaomi.com>
aio client will queue asynchronous io requests to the worker threads.
if PRIORITY_INHERITANCE is enabled, client thread's priority will be
set to worker threads. There will be multi-boost/restore of worker
threads' priority and assert the system.
No need priority multi-boot/restore to worker thread because client
thread's priority is alway the same.
Signed-off-by: fangxinyong <fangxinyong@xiaomi.com>
Note:
tcp_poll_eventhandler have logic after poll_notify, if we teardown poll
inside poll callback without clearing revents, use-after-free will
happen.
Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>
- There are occasional extra STOPs being sent due to an IP bug when using an
FPGA based I2C. Add a flag "inflight" to mask out extra STOP interrupts when
using the FPGA based implementation
- There are no MPFS_I2C_ST_STOP_SENT irq's "initally". It is just already
either success or still in progress
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
The circbuf part is implemented by the fb driver, and goldfish needs to
delete the part related to circbuf and adapt to the interface of the fb
driver.
Signed-off-by: jianglianfang <jianglianfang@xiaomi.com>
refact vsync queue to support multi fb poll and overlay poll.
Signed-off-by: pengyiqiang <pengyiqiang@xiaomi.com>
Signed-off-by: rongyichang <rongyichang@xiaomi.com>
The framebuffer of the emulator is a single buffer mode, and the pan_display method is not implemented, so as long as the fb device is created and poll notify once, the fb can be kept in a writable state.
Signed-off-by: pengyiqiang <pengyiqiang@xiaomi.com>
To solve the issue of carrying object files from previous builds,
Matias changed the archiving process to re-archive libapps.a on every compilation,
if libapps.a carries more object files, incremental compilation will waste too
many time in re-archiving, compared with the previous implement, this is a degradation
of the build system. Referring to mature engineering projects such as cmake, if there
is configuration or source file changed, the best solution should be to reconfigure
the environment.
Revert this PR to ensure the compilation speed during incremental compilation.
| commit 34b34e2d45 (tag: nuttx-20200914-172150)
| Author: Matias N <matias@protobits.dev>
| Date: Fri Sep 11 22:31:38 2020 -0300
|
| Fix: ensure archive files do not carry object files from prior builds
|
| In some cases, when NuttX configuration changes and this makes the
| object list used to build one of the .a libraries change as well,
| since the command used to build it is "ar crs" and this simply appends
| the list of object files, the library could still include object
| files from prior builds. This commit modifies the ARCHIVE macro to
| erase the .a file if it already exists.
|
| Since in some cases this behavior was actually expected (object
| files from a subdirectory were appended to a library created one
| level above) I added a ARCHIVE_ADD which works as ARCHIVE did.
|
| This change should greatly improve behavior of building after
| configuration changes.
Testing:
sim:nsh
-------------------------------
| Patched | Current
-------------------------------
|$ time make | $ time make
|real 0m1.270s | real 0m1.728s
|user 0m0.971s | user 0m1.276s
|sys 0m0.363s | sys 0m0.530s
-------------------------------
Private project (20+ 3rd library needs archive to libapps.a)
-------------------------------
| Patched | Current
-------------------------------
|$ time make | $ time make
|real 0m21.181s | real 0m39.721s
|user 0m14.638s | user 0m24.837s
|sys 0m6.919s | sys 0m14.394s
-------------------------------
Signed-off-by: chao an <anchao@xiaomi.com>
'rcS' will case files that has duplicate name been replaced.
so on the Mac platform we use 'q' instead of 'r' to achieve an
effect similar to '-P'
Signed-off-by: xuxin19 <xuxin19@xiaomi.com>
