os init_state: add new state OSINIT_IDLELOOP

[GUIDINGLI]

Summary

os init_state: add new state OSINIT_IDLELOOP

This is for distinguish we are in init idle or idle loop.
Use for assertion for sem_trywait & sem_wait

In the previous PR:
#5577

There is a check error:
DEBUGASSERT(OSINIT_IDLELOOP() && !sched_idletask());
should be:
DEBUGASSERT(!OSINIT_IDLELOOP() || !sched_idletask());

And the last PR regression caused by lack of testing.
This time do the fully test.

Impact

idle take sem

Testing

Test in maix-bit:kostest, VELA

[xiaoxiang781216]

@masayuki2009 could you try this one?

[masayuki2009]

@GUIDINGLI

Hmm, stm32f4discovery:kostest (DEBUG build) failed with assertion.
I've never seen this assertion before.

user_main: semaphore test
sem_test: Initializing semaphore to 0
sem_test: Starting waiter thread 1
sem_test: Set thread 1 priority to 191
waiter_func: Thread 1 Started
sem_test: Starting waiter thread 2
waiter_func: Thread 1 initial semaphore value = 0
sem_test: Set thread 2 priority to 128
waiter_func: Thread 1 waiting on semaphore
waiter_func: Thread 2 Started
sem_test: Starting poster thread 3
waiter_func: Thread 2 initial semaphore value = -1
sem_test: Set thread 3 priority to 64
waiter_func: Thread 2 waiting on semaphore
poster_func: Thread 3 started
poster_func: Thread 3 semaphore value = -2
poster_func: Thread 3 posting semaphore
waiter_func: Thread 1 awakened
poster_func: Thread 3 new semaphore value = -1
waiter_func: Thread 1 new semaphore value = -1
pup_assert: Assertion failed at file:armv7-m/arm_svcall.c line: 429 task: pt-0x8022711
arm_registerdump: R0: 20000004 R1: 00000000 R2: 40004400  R3: 00000000
arm_registerdump: R4: 20000004 R5: 00000031 R6: 00000000  FP: 00000068
arm_registerdump: R8: 00000001 SB: 2000000a SL: 00000000 R11: 00000001
arm_registerdump: IP: 08003eb7 SP: 20012358 LR: 08005353  PC: 0800ad9a
arm_registerdump: xPSR: 61000000 PRIMASK: 00000000 CONTROL: 00000000
arm_registerdump: EXC_RETURN: fffffff9
arm_dump_stack: User Stack:
arm_dump_stack: sp:     200122b0
arm_dump_stack:   base: 20011cf8
arm_dump_stack:   size: 00000800
arm_stackdump: 200122a0: 20011cf8 00000800 200122b0 08004d11 00000000 00000000 0800ad9a 00000001
arm_stackdump: 200122c0: 2001230c 1ffffffc 200003e4 00000068 00000001 08003ec7 00000001 08000c31
arm_stackdump: 200122e0: 00000000 200003e4 0000000b 0800203b 00000000 080008ad 2001230c 00000031
arm_stackdump: 20012300: 00000000 08000845 200022d0 20012358 00000000 20000004 00000031 00000000
arm_stackdump: 20012320: 00000068 00000001 2000000a 00000000 00000001 fffffff9 20000004 00000000
arm_stackdump: 20012340: 40004400 00000000 08003eb7 08005353 0800ad9a 61000000 00000001 00000000
arm_stackdump: 20012360: 20000004 0000001f 20010bab 00000001 00000011 20000010 00000068 0800b11b
arm_stackdump: 20012380: 0800b07f 20010b9c 0000001f 0000001f 00000001 200124b8 00000000 200124b8
arm_stackdump: 200123a0: 000000ed 08007073 00000001 200020f8 0000001f 00000000 00000001 08007083
arm_stackdump: 200123c0: 00000000 08000c0b 00000000 00000001 0000001f 08025e6d 00000014 20010b9c
arm_stackdump: 200123e0: 20010b7c 08025e6d 80000000 0000000a 20010b7c 00000000 00000000 08026929
arm_stackdump: 20012400: 0000000a 0a010b7c 080260ed 2001247c 0000000a 08026113 0000000a 2001247c
arm_stackdump: 20012420: 0000001f 0802640f 0802a85f 00000001 20002090 00000000 00000032 00000000
arm_stackdump: 20012440: 00000000 fffffffd 00000000 20010b84 00000000 20010b7c 0802a83f 200124b4
arm_stackdump: 20012460: 20002090 00000000 00000000 00000000 00000000 08025ef7 00000000 080260ed
arm_stackdump: 20012480: 20010b10 080260cd 0000001e 20010b7c 0802a83f 08025835 20002080 08025c1d
arm_stackdump: 200124a0: 0802a83f 200124b4 00000002 08022751 0802a83f 00000002 00000000 20002368
arm_stackdump: 200124c0: 00000002 ffffffff 00000000 08025847 00000002 08008a6d 00000080 08008a6d
arm_showtasks:    PID    PRI     STACK   COMMAND
arm_dump_task:      0      0      1000   Idle Task
arm_dump_task:      2    100      2024   ostest_main
arm_dump_task:      3    100      1992   ostest
arm_dump_task:     29    128      2048   pt-0x8022711
arm_dump_task:     30     64      2048   pt-0x8022669

[GUIDINGLI]

@masayuki2009
Can you share the detail cmds ?

[GUIDINGLI]

Can you revert this patch, and have a try ?

commit e0fabbfdd6a55510eae265472fd992aba1c0b588
Author: chao.an <anchao@xiaomi.com>
Date:   Tue Feb 22 15:25:56 2022 +0800

    arch/arm: replace SYS_syscall_return hardcode from syscall
    
    Signed-off-by: chao.an <anchao@xiaomi.com>

diff --git a/arch/arm/src/armv7-m/arm_svcall.c b/arch/arm/src/armv7-m/arm_svcall.c
index 6119ee6297..1330dc3a81 100644
--- a/arch/arm/src/armv7-m/arm_svcall.c
+++ b/arch/arm/src/armv7-m/arm_svcall.c
@@ -103,8 +103,9 @@ static void dispatch_syscall(void)
       " ldr r2, [sp, #16]\n"         /* Restore (orig_SP - new_SP) value */
       " add sp, sp, r2\n"            /* Restore SP */
       " mov r2, r0\n"                /* R2=Save return value in R2 */
-      " mov r0, #3\n"                /* R0=SYS_syscall_return */
-      " svc %0\n"::"i"(SYS_syscall)  /* Return from the SYSCALL */
+      " mov r0, %0\n"                /* R0=SYS_syscall_return */
+      " svc %1\n"::"i"(SYS_syscall_return),
+                   "i"(SYS_syscall)  /* Return from the SYSCALL */
     );
 }
 #endif

[masayuki2009]

Can you share the detail cmds ?

@GUIDINGLI

$ st-flash --format ihex write ./nuttx_user.hex
$ st-flash --format ihex write ./nuttx.hex
$ st-flash reset

[GUIDINGLI]

Can you share the detail cmds ?

@GUIDINGLI

$ st-flash --format ihex write ./nuttx_user.hex
$ st-flash --format ihex write ./nuttx.hex
$ st-flash reset

These cmds for real hardware board ?
Can this run on qemu ?

[masayuki2009]

@GUIDINGLI

BTW, do you have the stm32f4discovery board?

[masayuki2009]

@GUIDINGLI

These cmds for real hardware board ?

Yes.

Can this run on qemu ?

Unfortunately, no.

[GUIDINGLI]

Then, without this patch, the ostest can run pass ?
Also, need open CONFIG_DEBUG_ASSERTIONS=y

[masayuki2009]

Then, without this patch, the ostest can run pass ?
Also, need open CONFIG_DEBUG_ASSERTIONS=y

@GUIDINGLI

According to the latest upstream test logs, it passed with CONFIG_DEBUG_ASSERTIONS=y
However, I still suspect that my board has some issues.

I'm now preparing lm3s6965-ek:kostest with QEMU, so please wait.

[masayuki2009]

I'm now preparing lm3s6965-ek:kostest with QEMU, so please wait.

@GUIDINGLI
See #5595
I confirmed that this PR works with the above lm3s6965-ek:qemu-kostest.

So let me merge this PR.

[masayuki2009]

@GUIDINGLI

I found a new issue with sabre-6quad:netknsh which is not merged to the upstream.
Actually, if I run the getprime command, DEBUGASSERT happens when it exits.

The call stack is

(gdb) where                                                                                                                                                                                                                                                                                
#0  up_assert (filename=0x10856468 "semaphore/sem_wait.c", lineno=80) at common/arm_assert.c:478                                                                                                                                                                                           
#1  0x1080d38c in _assert (filename=0x10856468 "semaphore/sem_wait.c", linenum=80) at assert/lib_assert.c:36                                                                                                                                                                               
#2  0x10804354 in nxsem_wait (sem=0x1085ab34 <g_putc_lock>) at semaphore/sem_wait.c:80                                                                                                                                                                                                     
#3  0x10800e30 in up_putc (ch=91) at chip/imx_serial.c:1135                                                                                                                                                                                                                                
#4  0x10849344 in syslog_default_putc (channel=0x1085ac20 <g_default_channel>, ch=91) at syslog/syslog_channel.c:177                                                                                                                                                                       
#5  0x10848f70 in syslog_putc (ch=91) at syslog/syslog_putc.c:93                                                                                                                                                                                                                           
#6  0x10848e18 in syslogstream_putc (this=0x1086b9f4, ch=91) at syslog/syslog_stream.c:157                                                                                                                                                                                                 
#7  0x1080f780 in vsprintf_internal (stream=0x1086b9f4, arglist=0x0, numargs=0, fmt=0x1085a101 "%5jd.%06ld] ", ap=...) at stdio/lib_libvsprintf.c:236                                                                                                                                      
#8  0x10810da8 in lib_vsprintf (stream=0x1086b9f4, fmt=0x1085a100 "[%5jd.%06ld] ", ap=...) at stdio/lib_libvsprintf.c:1373                                                                                                                                                                 
#9  0x1084a590 in lib_sprintf (obj=0x1086b9f4, fmt=0x1085a100 "[%5jd.%06ld] ") at stdio/lib_libsprintf.c:45                                                                                                                                                                                
#10 0x10848db8 in nx_vsyslog (priority=6, fmt=0x108577ac "%s: tcb=%p\n", ap=0x1086ba24) at syslog/vsyslog.c:136                                                                                                                                                                            
#11 0x10813500 in vsyslog (priority=6, fmt=0x108577ac "%s: tcb=%p\n", ap=...) at syslog/lib_syslog.c:68                                                                                                                                                                                    
#12 0x10813538 in syslog (priority=6, fmt=0x108577ac "%s: tcb=%p\n") at syslog/lib_syslog.c:100                                                                                                                                                                                            
#13 0x10819b40 in up_addrenv_kstackfree (tcb=0x10865c40) at armv7-a/arm_addrenv_kstack.c:167                                                                                                                                                                                               
#14 0x108073e4 in nxsched_release_tcb (tcb=0x10865c40, ttype=1 '\001') at sched/sched_releasetcb.c:159                                                                                                                                                                                     
#15 0x10846cfc in nxtask_terminate (pid=6, nonblocking=1 '\001') at task/task_terminate.c:184                                                                                                                                                                                              
#16 0x108465fc in nxtask_exit () at task/task_exit.c:168                                                                                                                                                                                                                                   
#17 0x10816cd8 in up_exit (status=0) at common/arm_exit.c:125                                                                                                                                                                                                                              
#18 0x10805ffc in _exit (status=0) at task/exit.c:56                                                                                                                                                                                                                                       
#19 0x10842384 in nx_pthread_exit (exit_value=0x0) at pthread/pthread_exit.c:112                                                                                                                                                                                                           
#20 0x1080bec0 in STUB_nx_pthread_exit (nbr=107, parm1=0) at stubs/STUB_nx_pthread_exit.c:11                                                                                                                                                                                               
#21 0x10801f4c in dispatch_syscall () at armv7-a/arm_syscall.c:86        

And

(gdb) up                                                                                                                                                                                                                                                                                   
#1  0x1080d38c in _assert (filename=0x10856468 "semaphore/sem_wait.c", linenum=80) at assert/lib_assert.c:36                                                                                                                                                                               
36        up_assert(filename, linenum);                                                                                                                                                                                                                                                    
(gdb) up                                                                                                                                                                                                                                                                                   
#2  0x10804354 in nxsem_wait (sem=0x1085ab3c <g_putc_lock>) at semaphore/sem_wait.c:80                                                                                                                                                                                                     
80        DEBUGASSERT(!OSINIT_IDLELOOP() || !sched_idletask());                                                                                                                                                                                                                            
(gdb) p *rtcb                                                                                                                                                                                                                                                                              
$1 = {flink = 0x0, blink = 0x0, group = 0x10862b10, pid = 0, sched_priority = 0 '\000', init_priority = 0 '\000', start = 0x1080336c <nx_start>, entry = {pthread = 0x1080336c <nx_start>, main = 0x1080336c <nx_start>}, task_state = 2 '\002', flags = 6, lockcount = 1, errcode = 0, timeslice = 0, waitdog = {next = 0x0, func = 0x0, lag = 0, arg = 0}, adj_stack_size = 2024, stack_alloc_ptr = 0x1086115c, stack_base_ptr = 0x10861174, waitsem = 0x0, sigprocmask = 0, sigwaitmask = 0, sigpendactionq = {head = 0x0, tail = 0x0}, sigpostedq = {head = 0x0, tail = 0x0}, sigunbinfo = {si_signo = 0 '\000', si_code = 0 '\000', si_errno = 0 '\000', si_value = {sival_int = 0, sival_ptr = 0x0}}, msgwaitq = 0x0, mhead = 0x0, xcp = {sigdeliver = 0x0, saved_pc = 0, saved_cpsr = 0, sigreturn = 0, regs = {0, 277200652, 277230636, 0, 0, 9682944, 0, 0, 0, 0, 0, 277223732, 277223352, 277223732, 276839228, 276923088, 2147483987, 0 <repeats 33 times>}, nsyscalls = 0 '\000', syscall = {{cpsr = 0, sysreturn = 0}, {cpsr = 0, sysreturn = 0}}, ustkptr = 0x0, kstack = 0x0, kstkptr = 0x0}, name = "Idle Task", '\000' <repeats 22 times>}      

It seems that syslog flushes the internal buffers from the Idle task.

[masayuki2009]

It seems that syslog flushes the internal buffers from the Idle task.

@GUIDINGLI @xiaoxiang781216

Actual code is inside up_addrenv_kstackfre.
So do you mean that this PR does not permit such a debug log?
If so, this would be a big restriction for debugging.

int up_addrenv_kstackfree(FAR struct tcb_s *tcb)                                                                                                                                                                                                                                           
{                                                                                                                                                                                                                                                                                          
=>binfo("tcb=%p\n", tcb);                                                                                                                                                                                                                                                                  
  DEBUGASSERT(tcb);   

[masayuki2009]

So do you mean that this PR does not permit such a debug log?
If so, this would be a big restriction for debugging.

@GUIDINGLI @xiaoxiang781216

I found that up_putc() in imx_serial.c uses sem_wait().
I fixed the issue in #5633

[xiaoxiang781216]

It's good to know that this patch could catch the potential issue.