tests/arch/arm/arm_thread_swap failed on frdm_k64f board. #20595
Comments
|
Need to investigate why the no-optimizations' test is failing, but I assume the regular test is still passing, @chen-png right? |
|
Adding low as this is seems to be specific to FRDM-board. Test passes on nRF platforms, and QEMU |
yes, the regular tests can pass, but if with the CONFIG_NO_OPTIMIZATIONS, it failed. |
|
FYI: |
carlescufi
added
area: ARM
|
I'm not able to reproduce this. Both the hello world commented on by @pabigot and the actual test case. @chen-png can you confirm if this problem is still occurring for you? (tagging @MaureenHelm ) |
|
yes, it still failed with latest commit 5c603d9. |
|
Yeap, still failing for me as well. |
|
Dang, I hate not being able to reproduce... 👎 |
@dleach02 are you using CONFIG_NO_OPTIMIZATIONS=y? |
|
yes I did... I'll go through it again to verify I didn't fat finger something. |
|
Okay. I see the failure on arm_thread_swap. Debugging now |
dleach02
added
priority: medium
|
The bug is that when reprogramming an NXP MPU descriptor, the hardware clears the valid bit for that memory region when the code writes the start address, end address, and attribute descriptor registers. If this memory area is accessed during this time it is invalid a memory fault occurs. In this particular bug, the code in arch/arm/core/aarch32/cortex_m/mpu/nxp_mpu.c:mpu_configure_dynamic_mpu_regions() attempts to do a temporary mapping of all of RAM before applying the new set of memory maps. The function region_init() does the actual programming of the MPU descriptor registers and when the CONFIG_NO_OPTIMIZATIONS is set, the compiler generated code for region_init() uses locations on the program stack to store the local variables used to write the MPU descriptor. If the stack memory is in the descriptor being updated, the first write to the descriptor will clear the valid bit and the next read of the stack stored variable to write the next descriptor register will result in a memory fault. At initial startup of the system, the first dynamic MPU descriptor is the only descriptor set to all of RAM. The code attempts to "reset" this descriptor to all of memory again causing the fault. A PR is being prepared which changes the temporary remapping of RAM to use the last of the dynamic descriptors to avoid this sequence. |
|
Great explanation @dleach02 and agree also with setting to Medium severity |
Race conditions exist when remapping the NXP MPU. When writing the start, end, or attribute registers of a MPU descriptor, the hardware will automatically clear the region's valid bit. If that region gets accessed before the code is able to set the valid bit, the core will fault. Issue #20595 revealled this problem with the code in region_init() when the compiler options are set to no optimizations. The code generated by the compiler put local variables on the stack and then read those stack based variables when writing the MPU descriptor registers. If that region mapped the stack a memory fault would occur. Higher compiler optimizations would store these local variables in CPU registers which avoided the memory access when programming the MPU descriptor. Because the NXP MPU uses a logic OR operation of the MPU descriptors, the fix uses the last descriptor in the MPU hardware to remap all of dynamic memory for access instead of the first of the dynamic memory descriptors as was occuring before. This allows reprogramming of the primary discriptor blocks without having a memory fault. After all the dynamic memory blocks are mapped, the unused blocks will have their valid bits cleared including this temporary one, if it wasn't alread changed during the mapping of the current set. Fixes #20595 Signed-off-by: David Leach <david.leach@nxp.com>
Describe the bug
arch.arm.swap.common.no_optimizations and arch.arm.swap.common.fp_sharing.no_optimizations test cases failed on frdm_k64f board, there is nothing output int the handler.log.
To Reproduce
Steps to reproduce the behavior:
-T tests/arch/arm/arm_thread_swap
Screenshots or console output
sanitycheck -p frdm_k64f --device-testing --device-serial /dev/ttyACM0
-T tests/arch/arm/arm_thread_swap
Renaming output directory to /home/ztest/work/zephyrproject/zephyr/sanity-out.2
JOBS: 40
Building initial testcase list...
4 test configurations selected, 0 configurations discarded due to filters.
Adding tasks to the queue...
frdm_k64f tests/arch/arm/arm_thread_swap/arch.arm.swap.common.no_optimizations FAILED: N/A
total complete: 2/ 4 50% skipped: 0, failed: 1
frdm_k64f tests/arch/arm/arm_thread_swap/arch.arm.swap.common.fp_sharing.no_optimizations FAILED: N/A
total complete: 4/ 4 100% skipped: 0, failed: 2
2 of 4 tests passed (50.00%), 2 failed, 0 skipped with 0 warnings in 121.41 seconds
In total 4 test cases were executed on 1 out of total 212 platforms (0.47%)
Hardware distribution summary:
frdm_k64f (None): 4
Environment (please complete the following information):
The text was updated successfully, but these errors were encountered: