[BUG] RTCB can't be null since g_readytorun is in bss section

[PetervdPerk-NXP]

Description / Steps to reproduce the issue

https://github.com/apache/nuttx/blob/master/sched/sched/sched_lock.c#L80-L85
Mentions that rtcb may be NULL only during early boot-up phases however RTCB ends up in bss.

20002450 g     O .bss   00000008 g_readytorun

Therefore rtcb is a random value instead of NULL.

Not sure what would be better remove the sched_lock check or initialize g_readytorun as 0.

On which OS does this issue occur?

[OS: Linux]

What is the version of your OS?

Ubuntu

NuttX Version

Master

Issue Architecture

[Arch: arm]

Issue Area

[Area: OS Components]

Verification

• I have verified before submitting the report.

[acassis]

@hujun260 @anchao please take a look!

[acassis]

@PetervdPerk-NXP .bss area is suppose to be zeroed early in the boot process, so it will not be a random value as you said.

[acassis]

This is from stm32_start.c:

  stm32_clockconfig();
  arm_fpuconfig();
  stm32_lowsetup();
  stm32_gpioinit();
  showprogress('A');

  /* Clear .bss.  We'll do this inline (vs. calling memset) just to be
   * certain that there are no issues with the state of global variables.
   */

  for (dest = (uint32_t *)_START_BSS; dest < (uint32_t *)_END_BSS; )
    {
      *dest++ = 0;
    }

  showprogress('B');

[anchao]

It looks similar to the problem @PeterBee97 solved before #14889 .
Before BSS is initialized, the boot-up code should avoid using any kernel API.

This is a common problem, we need to adjust the startup code to ensure BSS initialization earlier. @GUIDINGLI @hujun260 could you arrange resources to invest in this? I am not working on nuttx projects recently. CC @xiaoxiang781216

[xiaoxiang781216]

The problem is clear that .bss need to be initialized before os, even before any c code, except the code which zero memory.