If cancellation points are enabled, then the following logic is activated in sem_wait(). This causes ECANCELED to be returned every time that sem_wait is called.
int sem_wait(FAR sem_t *sem)
{
...
/* sem_wait() is a cancellation point */
if (enter_cancellation_point())
{
#ifdef CONFIG_CANCELLATION_POINTS
/* If there is a pending cancellation, then do not perform
* the wait. Exit now with ECANCELED.
*/
errcode = ECANCELED;
goto errout_with_cancelpt;
#endif
}
...
Normally this works fine. sem_wait() is the OS API called by the application and will cancel the thread just before it returns to the application. Since it is cancellation point, it should never be called from within the OS.
There there is is one perverse cases where sem_wait() may be nested within another cancellation point. If open() is called, it will attempt to lock a VFS data structure and will eventually call nxmutex_lock(). nxmutex_lock() waits on a semaphore:
int nxmutex_lock(FAR mutex_t *mutex)
{
...
for (; ; )
{
/* Take the semaphore (perhaps waiting) */
ret = _SEM_WAIT(&mutex->sem);
if (ret >= 0)
{
mutex->holder = _SCHED_GETTID();
break;
}
ret = _SEM_ERRVAL(ret);
if (ret != -EINTR && ret != -ECANCELED)
{
break;
}
}
...
}
In the FLAT build, _SEM_WAIT expands to sem_wait(). That causes the error in the logic: It should always expand to nxsem_wait(). That is because sem_wait() is cancellation point and should never be called from with the OS or the C library internally.
The failure occurs because the cancellation point logic in sem_wait() returns -ECANCELED (via _SEM_ERRVAL) because sem_wait() is nested; it needs to return the -ECANCELED error to the outermost cancellation point which is open() in this case. Returning -ECANCELED then causes an infinite loop to occur in nxmutex_lock().
The correct behavior in this case is to call nxsem_wait() instead of sem_wait(). nxsem_wait() is identical to sem_wait() except that it is not a cancelation point. It will return -ECANCELED if the thread is canceled, but only once. So no infinite loop results.
In addition, an nxsem_wait() system call was added to support the call from nxmutex_lock().
This resolves Issue #9695
Signal must be masked when it is delivered to a signal handler per:
https://pubs.opengroup.org/onlinepubs/007904875/functions/sigaction.html:
When a signal is caught by a signal-catching function installed by sigaction(), a new signal mask is calculated and installed for the duration of the signal-catching function (or until a call to either sigprocmask() or sigsuspend() is made). This mask is formed by taking the union of the current signal mask and the value of the sa_mask for the signal being delivered [XSI] [Option Start] unless SA_NODEFER or SA_RESETHAND is set, [Option End] and then including the signal being delivered. If and when the user's signal handler returns normally, the original signal mask is restored.
Any action queued for that signal while the signal is masked should be deferred. It should go into the group pending signal list and should not be processed until until the signal is unmasked (which should occur when the signal handler returns).
Assert in nxsem_post if:
- Priority inheritance is enabled on a semaphore
- A thread that does not hold the semaphore attempts to post it
This will detect an error condition described in https://cwiki.apache.org/confluence/display/NUTTX/Signaling+Semaphores+and+Priority+Inheritance
None. The debug instrumentation is only enabled if CONFIG_DEBUG_ASSERTIONS is enabled.
Use sim:ostest. Verify that no assertions occur.
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.
There are number problems with the implementation of realtime signals in NuttX as discussed in Issue #8869. A first step to correcting any of these is to correct the definitions of SIGRTMIN, SIGRTMAX, and RTSIG_SIX.
SIGRTMIN is the first real-time signals. Real-time signal numbers must not overlap the standard signal numbers. Before this fix, it was set equal to the first standard signal. Real-time signals differ from standard signals in that (1) they have no default actions, and (2) real time signal actions are prioritized whereas standard signal actions are processed FIFO.
SIGRTMAX is the last real-time signal.
RTSIG_MAX must be set equal to maximum number of realtime signals reserved for application use
The change corrects the definitons but has not impact at all becasuse none of there definitions are currently used in the OS. But they will be when prioritized real time signal handling is implemented.
arch/arm/src/efm32/efm32_start.c:
/* For the case of the separate user-/kernel-space build, perform whatever
* platform specific initialization of the user memory is required.
* Normally this just means initializing the user space .data and .bss
* segments.
*/
#ifdef CONFIG_NUTTX_KERNEL
efm32_userspace();
showprogress('E');
#endif
But there is no CONFIG_NUTTX_KERNEL configuration setting. Comparing this to other architectures it is clear this should be
#ifdef CONFIG_BUILD_PROTECTED
Add a comment briefly explaining why one timer tick is added to the watchdog delay. This has confused a lot of people and has resulted in numerous awkward discussions and ill-informed attempts to modify the code to remove that addition. Perhaps a rather lengthy comment will reduce that confusion and put an end to this thrashing.
Because of this, it will take some time to detangle the licensing under net/. Many new features, original features were added to the NuttX network. Clearly, any references to Adam Dunkels in the files that implement these new features that have no counterpart in uIP 1.0 are errors.
This PR removes the references and converts the license headers to Apache 2.0 where possible. The affected files include only (1) the implementation of IPv6 (including neighbor support under ICMPv6) and (2) Raw sockets. Neither of these features are present in uIP 1.0 and the licenses can be freely updated.
I not two problems in handling of the return error values in PR #3858:
1. In KERNEL mode, the error return value of _SEM_WAIT() will be a negated errno value; in all other modes, it will be -1 (ERROR) with the errno variable set. This must be handled in the test of the returned value: Don't compare with -1; rather check if < 0
2. Also, conversion of the returned value to a negated errno value must be handled differently. This is handled by replacing -get_errno() with the macro _ERRVAL(ret)
This effects only error handling (it fixes it) and no other impacts are expected.
OS functions in syscall.csv are ordered alphabetically. However, two recently added functions are not in the correct location. This PR simply corrects that ordering.
The ordering of one entry was also corrected in libs/libc/libc.csv. Same issue.
This change is only cosmetic.
Verified only by CI
NuttX provides the UDP_BINDTODEVICE socket option. This is a UDP protocol-specific implementation of the semi-standard Linux SO_BINDTODEVICE socket option: "SO_BINDTODEVICE forces packets on the socket to only egress the bound interface, regardless of what the IP routing table would normally choose. Similarly only packets which ingress the bound interface will be received on the socket, packets from other interfaces will not be delivered to the socket." https://codingrelic.geekhold.com/2009/10/code-snippet-sobindtodevice.html
If CONFIG_NET_UDP_BINDTODEVICE is selected and a UDP socket is bound to the device, then unrecognized packets UDP packets must not be dropped, but must be forwarded along to the bound socket unconditionally.
It the typical case, this should have no impact. It does effect the applications that use DHCP and do select the UDP_BINDTODEVICE socket option.
This PR replace existing improper logic in the code and also the improper attempts to fix problems from PR #3601 and PR #3598. Those changes are improper because they expose DHCP appliction dependencies in the OS, breaking modularity and independence of the OS and application.
Tested with stm32f4discovery:netnsh with CONFIG_NET_UDP_BINDTODEVICE. A proper DHCP test setup is needed, however.
Fix comment in libs/libc/tls/tls_getinfo.c: The TLS data must lie at the beginning of the allocated stack memory for both push-up and push-down stacks.
That field is never used and, in most cases, is never initialized correctly.
This should have no impact to anything with the possible exception of free-running drivers.
Verified using CI builds only.
Per the Linux man page, "By default, getopt() permutes the contents of argv as it scans, so that eventually all the nonoptions are at the end." This behavior, however, is not implemented in the NuttX getopt() logic.
The Linux man page requires that the getopt_long() and getopt_long_only() functions accept arguments to options in a form like:
--option=argument
This PR adds that missing functionality.
This change effects only getopt_long() and getopt_long_only()
Tested on a simulator NSH configuration with a modified version of the getopt() test in apps/testing/ostest.
If an unrecognized long option is encountered, we must skip over that argv[] entry or getopt_long() will seriously misbehave.
Affects getopt_long() and getopt_long_only()
Problem found and fix verified with an updated version of the OS test.
Found a place in getopt_common() where the option string is used before it is checked if it is NULL. This can happen because the short option string is optional for getopt_long() and getopt_long_only()
If optstring is NULL, that would be an ERROR for getopt(), but not for the getopt_long() versions.
Should effect only the getopt() APIs
Tested on the simulator using apps/testing/ostest.
1. Null pointer dereference:
- for (ndx = 0; longopts[ndx].name[0] != '\0'; ndx++)
+ for (ndx = 0; longopts[ndx].name != NULL; ndx++)
2. Handle single character long options. An option like -x could be either a short option or a long option (under getopt_long_only()). This case was not being handled correctly.
3. Add missing support for optional arguments to short options (indicated with two "::"
This effects all members of the getopt() family of APIs.
Tested on the simulator using extensions to apps/testing/ostest.
Add implementations of getopt_long() and getopt_long_only()
There could be impacts to getop() since that implementation is shared withe new getopts_long() and getops_long_only() implementation.
Tested using a modified version of sim:nsh
getopt() in the FLAT build environment is not thread safe. This is because global variables that are process-specific in Unix are truly global in the FLAT build. Moving the getopt() variables into TLS resolves this issue.
No side-effects are expected other than to getopt()
Tested with sim:nsh
-CONFIG_USEC_PER_TICK=10000
+CONFIG_USEC_PER_TICK=1000
A system timer with a 10 MS period is not sufficient to run the dual thread sporadic scheduler test since the timings in that test are also around 10 MS. Apparently there is a race condition when both sporadic thread's budgets complete on the same clock time. This change does not eliminate the race, but reduces its effect greatly.
Fix missing semicolon at the end of a DEBUGASSERT statement:
sched/sched_sporadic.c: In function 'sporadic_budget_expire':
sched/sched_sporadic.c:512:15: error: expected ';' before 'period'
512 | period = (sporadic->repl_period >> 1) - unrealized;
| ^~~~~~
sched/sched_sporadic.c: In function 'nxsched_resume_sporadic':
sched/sched_sporadic.c:1078:19: error: expected ';' before 'period'
1078 | period = (sporadic->repl_period >> 1) - unrealized;
| ^~~~~~
Fix use of uninitialized variable in DEBUGASSERT statement:
sched/sched_sporadic.c:466:27: warning: 'sporadic' may be used uninitialized in this function [-Wmaybe-uninitialized]
466 | sporadic->nrepls > 0);
Also fixes some typos.
There should be no unexpected side-effects of this changed.
Tested with the stm32f4discovery:sporadic configuration (see PR #3097
This PR is the companion to #apache/incubator_nuttx_apps/620 . See that PR for further information.
No impact is anticipated
Tested using the (new) stm32f4discovery:sporadic configuration.
Urgent data preceded "normal" data in the TCP payload. If the urgent data is larger than the size of the TCP payload, this indicates that the entire payload is urgent data and that urgent data continues in the next packet.
This case was handled correctly for the case where urgent data was present but was not being handled correctly in the case where the urgent data was NOT present.
Endian-ness issues are mind boggling sometimes. I have been confused by the logic in inet_ntop() a few times so I thought I would add some comments so that I (and others) will understand the endian-ness issues in the future. No change to logic, only comments changed.
This commit corrects the following compilation error:
/usr/include/cygwin/socket.h:27:8: error: redefinition of 'struct sockaddr'
27 | struct sockaddr {
| ^~~~~~~~
In file included from /usr/include/w32api/winsock2.h:57,
from sim/up_wpcap.c:48:
/usr/include/w32api/psdk_inc/_ip_types.h:70:8: note: originally defined here
70 | struct sockaddr {
| ^~~~~~~~
In file included from /usr/include/sys/socket.h:13,
from /usr/include/cygwin/in.h:21,
from /usr/include/netinet/in.h:12,
from sim/up_wpcap.c:57:
/usr/include/cygwin/socket.h:39:8: error: redefinition of 'struct sockaddr_storage'
39 | struct sockaddr_storage {
| ^~~~~~~~~~~~~~~~
In file included from sim/up_wpcap.c:48:
/usr/include/w32api/winsock2.h:269:10: note: originally defined here
269 | struct sockaddr_storage {
| ^~~~~~~~~~~~~
The compilation was broken by a couple of recent blind, unverified changes to up_wpcap.c. Most were introduced with commit: 8ce0ff5ce4 with this change:
diff --git a/arch/sim/src/sim/up_wpcap.c b/arch/sim/src/sim/up_wpcap.c
index ef7b4b3a0c..a15421e80c 100644
--- a/arch/sim/src/sim/up_wpcap.c
+++ b/arch/sim/src/sim/up_wpcap.c
@@ -55,6 +55,8 @@
#include <netinet/in.h>
+#include "up_internal.h"
+
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
up_internal.h includes:
47 # include <sys/types.h>
48 # include <stdbool.h>
49 # include <netinet/in.h>
And netinet/in.h includes:
46 #include <sys/types.h>
47 #include <sys/socket.h>
48 #include <stdint.h>
Which is where the collision error is introduced since up_wpcap.c includes winsock2.h already. There were additional problems introduced to the file by other changes:
- A malformed syslog() call was added
- Some issues with netdriver_setmacaddr()
Move sched/task/task/task_gettid.c to libs/libc/unistd/lib_gettid.c. gettid() is a dumb wrapper around getpid(). It is wasteful of resources to support TWO systme calls, one for getpid() and one for gettid(). Instead, move gettid() in the C library where it calls the single sysgtem call, getpid(). Much cleaner.
Commit 34b34e2d45 uses the full path to libapps.a and introduced the use of the Make notdir command. That command breaks the Cygwin build because when a native Windows toolchain is used, the full path to libapps.a is a Windows-sytle path and the Make notdir command (like most other GNU Make commands) fails if it is passed a Windows-style path. This commit replaces the Make notdir command with the Bash basename command which can handle Windows paths.
PR #1450 broke the Cygwin build. Refer to Issue #1672.
The use of of logic like:
EXTRA_LIBPATHS += -L "${dir ${shell $(CC) $(ARCHCPUFLAGS) --print-file-name=libgcc.a}}"
fails when the Toolchain $(CC) is a native Windows toolchain. That is because the returned path is a Windows-style patch which cannot be handled by the make 'dir' command. Commit 4910d43ab0 reorganized a lot of definitions and replaced the correct code with the use of the limit make 'dir' command. The original code used the Bash dirname command which does not suffer from this limitation; it can handle both POSIX and Windows paths.
This was verified using the stm32f4discover:nsh toolchain with the Windows native ARM Embedded toolchain. That toolchain returns:
arm-none-eabi-gcc --print-file-name=libgcc.a
c:/program files (x86)/gnu tools arm embedded/9 2019-q4-major/bin/../lib/gcc/arm-none-eabi/9.2.1/libgcc.a
The pic32mx_usbpullup() and pic32mz_usbpullup() callbacks were referred to with various incorrect names in comments: pic32mx_pullup(), pic32mz_pullup(), stm32_pullup().
fs/procfs/fs_procfs.c contained an unused reference to an STM32 procfs structure. This is wrong in two ways: (1) There should be not STM32 references outside of arch/arm and boards/arm and (2) the declare STM32 structure is not used anyway.
Fixes:
##[error]net/tun.c:837:13: error: conflicting types for 'tun_poll_expiry'
837 | static void tun_poll_expiry(FAR void *arg)
| ^~~~~~~~~~~~~~~
net/tun.c:196:13: note: previous declaration of 'tun_poll_expiry' was here
196 | static void tun_poll_expiry(wdparm_t arg);
| ^~~~~~~~~~~~~~~
And
##[error]net/tun.c:881:57: error: passing argument 4 of 'wd_start' makes integer from pointer without a cast [-Werror=int-conversion]
881 | wd_start(&priv->txpoll, TUN_WDDELAY, tun_poll_expiry, priv);
| ^~~~
| |
| struct tun_device_s *
In file included from /github/workspace/sources/nuttx/include/nuttx/sched.h:41,
from /github/workspace/sources/nuttx/include/sched.h:34,
from /github/workspace/sources/nuttx/include/nuttx/arch.h:81,
from net/tun.c:60:
/github/workspace/sources/nuttx/include/nuttx/wdog.h:134:42: note: expected 'wdparm_t' {aka 'long unsigned int'} but argument is of type 'struct tun_device_s *'
134 | wdentry_t wdentry, wdparm_t arg);
| ~~~~~~~~~^~~
And
chip/cxd56_rtc.c: In function 'up_rtc_initialize':
##[error]chip/cxd56_rtc.c:358:3: error: too many arguments to function 'cxd56_rtc_initialize'
358 | cxd56_rtc_initialize(1, NULL);
| ^~~~~~~~~~~~~~~~~~~~
chip/cxd56_rtc.c:253:13: note: declared here
253 | static void cxd56_rtc_initialize(wdparm_t arg)
| ^~~~~~~~~~~~~~~~~~~~
This will resolve numerous problems with the way that hard links, in particular "." and ".." are handled. Instead of trying to fudge the stat flags, the correct implementation is to follow the hard link to the final link target node. That is what must determine the attributes of the directory entry.
This PR corrects an error in the ROMFS file system. The error occurred after following a hard link (depending on how the ROMFS image is organized). The error occurred because some of the information buffered before following the links was stale and, hence, out of sync after following the hard link. This would cause random errors when paths containing hardlinks were used with ROMFS.
This PR resolves Issue #1543. Please compare the following output with the output in Issue #1543 to see how the problem was resolved:
NuttShell (NSH) NuttX-9.1.0
nsh> mount
/etc type romfs
/proc type procfs
/tmp type vfat
nsh> ls -Rl /etc
/etc:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 20 group
dr-xr-xr-x 0 init.d/
-r-xr-xr-x 35 passwd
/etc/init.d:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 71 rcS
nsh> ls -l /etc/init.d
/etc/init.d:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 71 rcS
nsh> ls -l /etc/init.d/.
/etc/init.d/.:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 71 rcS
nsh> ls -l /etc/init.d/..
/etc/init.d/..:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 20 group
dr-xr-xr-x 0 init.d/
-r-xr-xr-x 35 passwd
nsh> ls -l /etc/init.d/../.
/etc/init.d/../.:
dr-xr-xr-x 0 .
dr-xr-xr-x 0 ..
-r-xr-xr-x 20 group
dr-xr-xr-x 0 init.d/
-r-xr-xr-x 35 passwd
nsh>
drivers/net/ftmac100.c, libs/libc/stdlib/lib_aligned_alloc.c
A continuation of PRs #1507, #1510, and #1512. See Issue #1481 for additional information.
malloc() and free() should never be used within the OS. This will work in the FLAT build because there is only a single heap, but will cause problems in PROTECTED and KERNEL build modes where there are separate heaps for user and kernel memory.
Typically kmm_malloc(), kmm_zalloc(), and kmm_free() should be called within the kernel in those build modes to use the kernel heap.
Memory is never free. Possible memory leak:
./boards/arm/cxd56xx/common/src/cxd56_crashdump.c: pdump = malloc(sizeof(fullcontext_t));
Memory allocated with malloc(), but freed with kmm_free():
./drivers/usbhost/usbhost_composite.c: cfgbuffer = (FAR uint8_t *)malloc(CUSTOM_CONFIG_BUFSIZE);
Memory is never freed in these cases. It is allocated in the driver initialization logic, but there is no corresponding uninitialization logic; memory is not freed on error conditions:
./arch/arm/src/lc823450/lc823450_i2s.c: priv = (struct lc823450_i2s_s *)zalloc(sizeof(struct lc823450_i2s_s));
./arch/arm/src/sam34/sam_spi.c: spics = (struct sam_spics_s *)zalloc(sizeof(struct sam_spics_s));
./arch/arm/src/sama5/sam_spi.c: spics = (struct sam_spics_s *)zalloc(sizeof(struct sam_spics_s));
./arch/arm/src/samv7/sam_spi.c: spics = (struct sam_spics_s *)zalloc(sizeof(struct sam_spics_s));
Memory is allocated with zalloc() but freed on error conditions with kmm_free():
./arch/arm/src/sama5/sam_ssc.c: priv = (struct sam_ssc_s *)zalloc(sizeof(struct sam_ssc_s));
./arch/arm/src/samv7/sam_ssc.c: priv = (struct sam_ssc_s *)zalloc(sizeof(struct sam_ssc_s));
./arch/arm/src/stm32/stm32_i2s.c: priv = (struct stm32_i2s_s *)zalloc(sizeof(struct stm32_i2s_s));
Memory is never freed:
./drivers/spi/spi_bitbang.c: priv = (FAR struct spi_bitbang_s *)zalloc(sizeof(struct spi_bitbang_s));
arch/arm/src/cxd56xx/cxd56_gnss.c, arch/arm/src/xmc4/xmc4_spi.c,
crypto/blake2s.c, drivers/lcd/pcf8574_lcd_backpack.c, drivers/lcd/st7032.c
User space memory should not be used within the OS and, when it is absolutely necessary to use user-space memory, it should be allocated using kumm_malloc().
drivers/net/telnet.c
drivers/wireless/bluetooth/bt_uart_bcm4343x.c
drivers/wireless/ieee802154/mrf24j40/mrf24j40.c
Kernel memory was allocated using kmm_malloc() or kmm_zalloc() but freed with with the user-space allocator free(). In the FLAT build, this is bad style, but not harmful because there is only a single, heap and malloc() and kmm_malloc() map to the same function.
But that is not true in the case of the PROTECTED or KERNEL builds. In those cases, there are separate heaps. kmm_malloc() will allocate from the kernel heap. free() will attempt to free the kernel memory from the user heap and will cause an assertion (or other obscure failure if assertions are disabled).
Prohibit use of pthread_cleanup API's by kernel threads. The pthread pthread_cleanup functions MUST run in user mode, making them unusable for kernel threads.
See Issue #1263
This backs out a part of PR 1179 which has a very serious error: If C buffered I/O is available, then printf() and vprintf MUST use it. Otherwise, the ordering of the I/O will be screwed up. They must not use direct file descriptor I/O UNLESS C buffered I/O is disabled.
arch/z80/src/Makefile: Correct inclusion of non-existent file. This was not a problem before because there was '-' before the include. Problem revealed with '-' removed.
arch/z80/src/ez80/Toolchain.defs: Apparently there are not too many '"' in path definition.
tools/incdir.c: No space between -usrinc: or -sysinc: and the list of paths.
Move global variable declaration out of include/nuttx/video/video.h and into the file where it is initialized. With some toolchains/environments, declaring globals in header files results in multiply defined symobl errors at link time. This corrects that build problem.
In the FLAT build if CONFIG_LIB_SYSCALL=y, then the function task_spawn() will be duplicated.: One version in libs/libc/spawn and one version in sched/task.
The version of task_spawn in lib/libc/spawn exists only if CONFIG_LIB_SYSCALL is selected. In that case, the one in sched/task/task_spawn.c should be static, at least in the FLAT build.
The version of task_spawn.c in libs/libc/spawn simply marshals the parameters into a structure and calls nx_task_spawn(). If CONFIG_LIB_SYSCALL is defined then nx_task_spawn() will un-marshal the data can call the real task spawn. This nonsense is only necessary because task_spawn has 8 parameters and the maximum number of parameters in a system call is only 6.
Without syscalls: Application should call directly in task_spawn() in sched/task/task_spawn.c and, hence, it must not be static
With syscalls: Application should call the marshalling task_spawn() in libs/libc/spawn/lib_task_spawn.c -> That will call the autogenerated nx_task_spawn() proxy -> And generate a system call -> The system call will the unmarshalling nx_task_spawn() in sched/task/task_spawn.c -> Which will, finally, call the real task_spawn().
The side-effect of making task_spawn() static is that it then cannot be used within the OS. But as far as I can tell, nothing in the OS itself currently uses task_spawn() so I think it is safe to make it conditionally static. But that only protects from duplicate symbols in the useless case mentioned above.
Reported by 권석근 <kwonsksj@gmail.com>:
I found a bug at "pty.c" during ssh server implementation.
When I turn on CONFIG_SERIAL_TERMIOS and OPOST|ONLCR on pty device
for nsh console's stdin/stdout (ssh shell service), I've got system crash.
Bugs at line 687 of pty.c, pty_write()
ntotal++;
when converting '\n' to '\r\n', pty_write() will return more than requested
(+1, for example) length. and this will break caller lib_fflush(), line 150
of lib_libfflush.c.
When she get (libfflush()) bytes_nwritten which is greater than nbuffer,
nbuffer goes to negative at line 150 and eventually destroys
*stream->fs_bufpos at line 163 of lib_libflush.c
Removing ntotal++; line 687 of pty.c will fix this bug.
BTW, nsh using ptm/pty as a ssh shell service works great with libssh +
mbedtls.
Only Make.defs files tht followed the same pattern as the ARM Make.defs were modified. This excludes some of the sim and renesas Make.defs files and all of the z80 Make.defs files.
Before any other Make logic can be used, we must immediately build the tools/incdir binary. It will be used as soon as Make.defs is included and will generate errors otherwise.
Remove the other locations in the tools/incdir binary was being build from tools/Makefile.unix and tools/Makefile.win
This is a change suggested by Xiao Xiang in an email thread. Some make variables with depend on forking and shell and running a script to get the value of the variable. Using := we can force the calculation to occur only once. This leads to a small but consistent improvement in build performance.
This change really applies to ALL Make.defs files but is applied only to one here so that it can be thoroughly verified and possbily leveraged to other Make.defs files in the future.
1. If config.c is compiled on any platform other than Cygwin, then the variable wintools is not used.
2. Add more debug output so we can see what is going on in the PR checks.
Block and MTD drivers may be opened and managed as though they were character drivers. But this is really sleight of hand; there is a hidden character driver proxy that mediates the interface to the block and MTD drivers in this case.
fstat(), however, did not account for this. It would report the characteristics of the proxy character driver, not of the underlying block or MTD driver.
This change corrects that. fstat now checks if the character driver is such a proxy and, if so, reports the characteristics of the underlying block or MTD driver, not the proxy character driver.
sched_releasetcb() will normally free the stack allocated for a task. However, a task with a custom, user-managed stack may be created using nxtask_init() followed by nxtask_activer. If such a custom stack is used then it must not be free in this many or a crash will most likely result.
This chagne addes a flag call TCB_FLAG_CUSTOM_STACK that may be passed in the the pre-allocted TCB to nxtask_init(). This flag is not used internally anywhere in the OS except that if set, it will prevent sched_releasetcb() from freeing that custom stack.
Add trivial function nxtask_uninit(). This function will undo all operations on a TCB performed by task_init() and release the TCB by calling kmm_free(). This is intended primarily to support error recovery operations after a successful call to task_init() such was when a subsequent call to task_activate fails.
That error recovery is trivial but not obvious. This helper function should eliminate confusion about what to do to recover after calling nxtask_init()
-Move task_init() and task_activate() prototypes from include/sched.h to include/nuttx/sched.h. These are internal OS functions and should not be exposed to the user.
-Remove references to task_init() and task_activate() from the User Manual.
-Rename task_init() to nxtask_init() since since it is an OS internal function
-Rename task_activate() to nxtask_activate since it is an OS internal function
The build gets broken when both the NFS and DEBUG_FS_ERROR configurations enabled. This tiny change fixes the compilation error.
Run file through nxstyle and correct issues.
NOTE: This line:
/* This logic just supports "//" sequences in the path name */
generates the false alarm error:
fs/nfs/nfs_util.c:84:39: error: C++ style comment
Functions within the OS must never set the errno value. fs_fdopen() was setting the errno value. Now, after some parameter changes, it reports errors via a negated errno integer return value as do most all other internal OS functions.
Documentation/NuttXCCodingStandard.html:
- Fix a typo: "be aligned so that the comment begins in the same comment on each line." Of course that should have been column.
- Update a coding standard requirement. With regard to Comments to the Right of Statements, the document said: "If such comments are used, they should ... fit on one line." This is not true any more. In common practice (include the checks in nxstyle), comments to the right of statements may be continued in the following line. The standard way of continuing such comments was also documented.
In this case the coding standard actually contradicted iteself since first it says the that the comments to the right must fit on one line, but later illustrates (without description) how to continue a long right hand comment on the following line.
Documents/style.css: It appears that a variable width font was being used for <pre> and <code>, that was really screwing up the presentation, especially in the coding standared where there are many examples illustrating correct vertical alignment. The font type "Bitstream Vera Sans Mono" was replaced with "courier" for <pre> and <code>
Remove support for the Codesourcery, Atollic, DevKitArm, Raisonance, and CodeRed toolchains. Not only are these tools old and no longer used but they are all equivalent to standard ARM EABI toolchains. Retaining specific support has no effect (they are still supported, but now just as generic EABI toolchains).
1. Internal scheduler functions should begin with nxsched_, not sched_
2. Follow the consistent naming patter of https://cwiki.apache.org/confluence/display/NUTTX/Naming+of+OS+Internal+Functions
# clock_systimer -> clock_systime_tick
# clock_systimespec -> clock_systime_timespec
sched_oneshot_extclk -> nxsched_oneshot_extclk
sched_period_extclk -> nxsched_period_extclk
# nxsem_setprotocol -> nxsem_set_protocol
# nxsem_getprotocol -> nxsem_get_protocol
# nxsem_getvalue -> nxsem_get_value
nxsem_initholders -> nxsem_initialize_holders
nxsem_addholder -> nxsem_add_holder
nxsem_addholder_tcb -> nxsem_add_holder_tcb
nxsem_boostpriority -> nxsem_boost_priority
nxsem_releaseholder -> nxsem_release_holder
nxsem_restorebaseprio -> nxsem_restore_baseprio
Some planned name changed were skipped for now because they effect too many files (and would require many hours of coding style fixups).
All complaints fixed except for those that were not possible to fix:
- Used of Mixed case identifier in ESP32 files. These are references to Expressif ROM functions which are outside of the scope of NuttX.
1. Add missing conditional logic in include/sys/syscall_lookup.h
2. CONFIG_NPTHREAD_KEYS removed from code but was still in sched/Kconfig
3. Refresh all configurations affected by PR 1007
4. syscall/syscall_funclookup.c needs to include nuttx/tls.h
1. Move pthread-specific data files from sched/pthread/ to libs/libc/pthread.
2. Remove pthread-specific data functions from syscalls.
3. Implement tls_alloc() and tls_free() with system calls.
4. Reimplement pthread_key_create() and pthread_key_free() using tls_alloc() and tls_free().
5. Reimplement pthread_set_specific() and pthread_get_specicif() using tls_set_value() and tls_get_value()
