Add a new field to record the global on the basis of mm_backtrace.
When using alloc, the field is incremented by 1,
so that the memory usage can be dumped within the range
Signed-off-by: anjiahao <anjiahao@xiaomi.com>
FD (file descriptor) is widely used in system software development,
and almost all implementations of posix os (including nuttx) use FD as an index.
the value of fd needs to be allocated starting from the minimum available value of 3, and each process has a copy,
so the same fd value is very easy to reuse in the program.
In multi threaded or multi process environments without address isolation,
If the ownership, global variables, and competition relationships of fd are not properly handled,
there may be issues with fd duplication or accidental closure.
Further leading to the following issues, which are difficult to troubleshoot.
1. Security vulnerability: the fd we wrote is not the expected fd and will be accessed by hackers to obtain data
2. Program exceptions or crashes: write or read fd failures, and program logic errors
3. The structured file XML or database is damaged: the data format written to the database is not the expected format.
The implementation principle of fdsan is based on the implementation of Android
https://android.googlesource.com/platform/bionic/+/master/docs/fdsan.md
Signed-off-by: hujun5 <hujun5@xiaomi.com>
The spawn proxy thread is a special existence in NuttX, usually some developers
spend a lot of time on stack overflow of spawn proxy thread:
https://github.com/apache/nuttx/issues/9046https://github.com/apache/nuttx/pull/9081
In order to avoid similar issues, this PR will remove spawn proxy thread to simplify
the process of task/posix_spawn().
1. Postpone the related processing of spawn file actions until after task_init()
2. Delete the temporary thread of spawn proxy and related global variables
Signed-off-by: chao an <anchao@xiaomi.com>
fix the problem that stat fails to use the relative path
An error will be reported if used in the following way:
stat("//./bin", &st);
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
and remove the check of SYS_DOWN since it defines to SYS_RESTART in reboot_notifer.h:
#define SYS_DOWN 0x0001 /* Notify of system down */
#define SYS_RESTART SYS_DOWN
#define SYS_HALT 0x0002 /* Notify of system halt */
#define SYS_POWER_OFF 0x0003 /* Notify of system power off */
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
This way the mappings can be modified for any vm area, not only the
process that is running.
Why? This allows mapping pages to kernel dynamically, this functionality
will be presented later.
Refer to issue #8867 for details and rational.
Convert sigset_t to an array type so that more than 32 signals can be supported.
Why not use a uin64_t?
- Using a uin32_t is more flexible if we decide to increase the number of signals beyound 64.
- 64-bit accesses are not atomic, at least not on 32-bit ARMv7-M and similar
- Keeping the base type as uint32_t does not introduce additional overhead due to padding to achieve 64-bit alignment of uin64_t
- Some architectures still supported by NuttX do not support uin64_t
types,
Increased the number of signals to 64. This matches Linux. This will support all xsignals defined by Linux and also 32 real time signals (also like Linux).
This is is a work in progress; a draft PR that you are encouraged to comment on.
In the last change, the assert behavior test was not considered. Therefore, in this change, sync will be performed only when SYS_RESTART/SYS_DOWN situation occurs, avoiding unpredictable results caused by handling locks or semaphores in interrupts.
mount/fs_foreachmountpoint.c: In function 'mountpoint_filter':
mount/fs_foreachmountpoint.c:99:38: warning: 'sprintf' may write a terminating nul past the end of the destination [-Wformat-overflow=]
99 | sprintf(&dirpath[pathlen], "/%s", node->i_name);
| ^
In function 'mountpoint_filter',
inlined from 'mountpoint_filter' at mount/fs_foreachmountpoint.c:64:12:
mount/fs_foreachmountpoint.c:99:7: note: 'sprintf' output between 2 and 257 bytes into a destination of size 256
99 | sprintf(&dirpath[pathlen], "/%s", node->i_name);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: chao an <anchao@xiaomi.com>
NuttX kernel should not use the syscall functions, especially after
enabling CONFIG_SCHED_INSTRUMENTATION_SYSCALL, all system functions
will be traced to backend, which will impact system performance.
Signed-off-by: chao an <anchao@xiaomi.com>
Summary:
- I noticed that sometimes mqueue_test in ostest failed
(i.e. mq_open failed) with rv-virt:smp.
- This commit fixes this issue by replacing sched_lock with
critical section.
Impact:
- file_mq_open()
Testing:
- Tested with ostest on QEMU-7.1
Signed-off-by: Masayuki Ishikawa <Masayuki.Ishikawa@jp.sony.com>
This implements the file system logic for posix shared memory:
shm_open, shm_unlink, mmap, munmap and close
For flat and protected builds the memory simply allocated from (user) heap
For kernel build the memory is allocated from page pool and mapped with MMU
This doesn't yet support protection flags or re-sizing already truncated shared
memory area.
Co-authored-by: Ville Juven <ville.juven@unikie.com>
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
| $ gcc -c "-O3" -Wall -isystem include -D__NuttX__ -D__KERNEL__ -I fs -I sched fs/mmap/fs_mmap.c -o fs_mmap.o
| include/sys/mman.h: In function ‘mmap’:
| fs/mmap/fs_mmap.c:259:13: warning: ‘mapped’ may be used uninitialized in this function [-Wmaybe-uninitialized]
| 259 | FAR void *mapped;
| | ^~~~~~
Signed-off-by: chao an <anchao@xiaomi.com>
With flat addressing the user pollfd list is given directly to the
drivers that perform the notification. This is fine when the addressing
is flat, as there is no ambiguity on who gets notified and the fds memory
is always mapped.
For kernel mode with MMU this does not work. The notification must be done
via a temporary buffer allocated from kernel memory.
Thus, create a copy of the user fds and pass the copy to the drivers.
Afterwards copy the output events back to the user.
Follow posix, return ENOENT when the named semaphore does not
exist to fix the ltp open posix sem_unlink 4-1.c test problem.
Signed-off-by: wangbowen6 <wangbowen6@xiaomi.com>
and Fix include/nuttx/mm/map.h:55:28: error: 'struct task_group_s' declared inside parameter list will not be visible outside of this definition or declaration
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
since mmap may exist in block_operations, but truncate may not,
moving mmap beforee truncate could make three struct more compatible
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
- Add mmap into file_operations and remove it from ioctl definitions.
- Add mm_map structure definitions to support future unmapping
- Modify all drivers to initialize the operations struct accordingly
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
- Add truncate into file_operations
- Move truncate to be common for mountpt_operations and file_operations
- Modify all drivers to initialize the operations struct accordingly
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
This is preparation for flushing streams from user space, like it should
be done.
- Move tg_streamlist (group, kernel space) ->
ta_streamlist (TLS, user space)
- Access stream list via tg_info in kernel
- Access stream list via TLS in user space
- Remove / rename nxsched_get_streams -> lib_getstreams
- Remove system call for nxsched_get_streams
In some cases, deleting a folder will return unexpected results. For example, ENOENT should be returned when deleting a non-existent folder, but the result will return ENOTDIR.
nx_open in timerfd_create will increase reference count,
Therefore, the reference count starts with a value of 0.
Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>
It is enought to only checking the buflen's avaiable or not.
(total size means total copied size)
Co-authored-by: Petro Karashchenko <petro.karashchenko@gmail.com>
1. epoll_ctl(EPOLL_CTL_ADD) support extend the epoll
events dynamicly;
2. enhance the epoll performance by moving some poll setup
and teardown process to the EPOLL_CTL_ADD/DEL/MOD;
Signed-off-by: wangbowen6 <wangbowen6@xiaomi.com>
In the past, header file paths were generated by the incdir command
Now they are generated by concatenating environment variables
In this way, when executing makefile, no shell command will be executed,
it will improve the speed of executing makfile
Signed-off-by: yinshengkai <yinshengkai@xiaomi.com>
1.Don't check the return value of nxsem_init or nxmutex_init
2.Fix some style issue
Signed-off-by: anjiahao <anjiahao@xiaomi.com>
Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
The default size of file cache is size of a sector, it may
not be good size for optimizing read/write speed in physical
device. So we can set the config according to speed test profile
to optimize access IO speed.
Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>
When ps cmd is invoked during the task exit process, the group information of the task may be deleted.
So need check the task flag before get the task group information.