Memory delay free kasan error fix and memory note fix

[XuNeo]

Summary

1. Hold lock when access mm struct field from sched_note_heap.
2. Fix kasan reported error when delay free feature is enabled. See comments in code.

  /* If delay free is enabled, a memory node will be freed twice.
   * The first time is to add the node to the delay list, and the second
   * time is to actually free the node. Therefore, we only colorize the
   * memory node the first time, when `delay` is set to true.
   */

3. Remove CONFIG_MM_MAP_COUNT_MAX option. User should take care of it. It causes ltp_interfaces_mmap_24_1 test to fail, remove this patch for now.

Impact

Bug fixes should have no impact on existing projects.

Testing

Local tested with qemu.

Configure

tools/configure.sh mps3-an547:nsh

Change CONFIG_MM_REGIONS = 1 to reduce the heap size, in order to make kasan faster.

Enable

CONFIG_MM_KASAN=y
CONFIG_MM_FREE_DELAYCOUNT_MAX=16
CONFIG_MM_FILL_ALLOCATIONS=y

Run

 qemu-system-arm -M mps3-an547 -m 2G -nographic -kernel nuttx -gdb tcp::1127 -serial pty

Result

• Before the fix

kasan_report: kasan detected a write access error, address at 0x1019f08,size is 1, return address: 0xf4f7                                     │start () at armv8-m/arm_vectors.c:66
kasan_show_memory: Shadow bytes around the buggy address:                                                                                     │66        asm volatile ("mov lr, #0\n\t"
kasan_show_memory:   0x1019eb0: 8c 2a f8 01 9d 89 22 ca e0 e9 90 40 d1 d3 15 0a                                                               │(gdb) c
kasan_show_memory:   0x1019ec0: 41 f2 70 95 43 f1 ea aa 89 a6 e9 fb 5a 4e fc 62                                                               │Continuing.
kasan_show_memory:   0x1019ed0: fd 86 8c c0 a5 8c dc 24 e7 4a 1a c9 e9 64 bd 52                                                               │[Inferior 1 (process 1) exited normally]
kasan_show_memory:   0x1019ee0: 70 97 68 55 71 dc a1 de 1f 68 d2 9a e5 12 78 7b                                                               │(gdb) c
kasan_show_memory:   0x1019ef0: 83 ab c2 51 32 06 2b 97 65 76 26 e2 2c 06 25 72                                                               │The program is not being run.
kasan_show_memory:   0x1019f00: aa aa aa aa 61 03 00 00[c0 97 01 01 55 55 55 55                                                               │(gdb) tar rem:1127
kasan_show_memory:   0x1019f10: 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55                                                               │`/home/neo/projects/nuttx/nuttx/nuttx' has changed; re-reading symbols.
kasan_show_memory:   0x1019f20: 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55                                                               │Remote debugging using :1127
kasan_show_memory:   0x1019f30: 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55                                                               │
kasan_show_memory:   0x1019f40: 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55                                                               │Program received signal SIGTRAP, Trace/breakpoint trap.
dump_assert_info: Current Version: NuttX  12.7.0-RC0 1e556412c0d-dirty Oct  8 2024 14:36:37 arm                                               │start () at armv8-m/arm_vectors.c:66
dump_assert_info: Assertion failed panic: at file: kasan/hook.c:187 task: memstress process: memstress 0x3e56

• After the fix
  memstress runs without kasan error.

[nuttxpr]

[Experimental Bot, please feedback here]

NuttX PR Requirements Review

This PR generally meets the NuttX requirements, but has some minor issues:

Strengths:

• Summary: Provides a clear and concise explanation of the changes made. Each point addresses a specific issue, and the code comment excerpt is helpful.
• Impact: Clearly outlines the limited impact of the changes. Removing CONFIG_MM_MAP_COUNT_MAX is acknowledged as a potential impact point, which is good.
• Testing: Demonstrates testing effort by specifying the target, configuration, and steps to reproduce the issue. The inclusion of "before" and "after" logs is very helpful.

Areas for Improvement:

• Testing - Build Host(s): While you mention "Local tested with qemu," please provide more details about your build host environment (OS, CPU architecture, compiler version).
• Testing - Logs: The provided logs are helpful, but could be more focused. Instead of large log dumps, consider highlighting the specific lines demonstrating the issue before the fix and the successful execution after the fix.
• Testing - Breadth: While testing with qemu is a good starting point, consider adding tests on real hardware if possible. This helps ensure the fix generalizes well.

Conciseness:

The PR description is concise and to the point.

Overall:

This PR demonstrates a good understanding of the NuttX requirements. Addressing the testing details will further strengthen the submission.

[XuNeo]

sim:citest reports multiple ltp tests failed in CI.
But it can pass locally.

E.g.

nsh> ltp_interfaces_mmap_24_1
Total mapped size is 1048576 bytes
Test PASSED
nsh>

[GUIDINGLI]

rebase you PR, and try again