Reenable recording_micro_allocator_test

[veblush]

A recent QEMU upgrade appears to expose an underlying unaligned memory access issue on Cortex-M3. This is causing crashes (uncaught target signal 7 (Bus error)) in two tests, TestRecordsPersistentTfLiteTensorData and TestRecordsPersistentTfLiteTensorQuantizationData, which were temporarily disabled in PR #3206.

The crash occurs when accessing the test model's Flatbuffer memory, as shown by the following callstack:

#0  0x00008e14 in flatbuffers::Vector<long long, unsigned long>::Get (this=<optimized out>, i=110405) at tensorflow/lite/micro/tools/make/downloads/flatbuffers/include/flatbuffers/vector.h:178
#1  0x00008edc in tflite::internal::InitializeTfLiteTensorFromFlatbuffer (persistent_buffer_allocator=0x407fdc80, non_persistent_buffer_allocator=0x407fdc7c, allocate_temp=false, flatbuffer_tensor=..., 
    buffers=0x15d29 <kTestConvModelData+116>, result=0x407fdbb4) at tensorflow/lite/micro/micro_allocator.cc:285
#2  0x000093c8 in tflite::MicroAllocator::PopulateTfLiteTensorFromFlatbuffer (this=this@entry=0x407fdbd4, model=model@entry=0x15cd9 <kTestConvModelData+36>, tensor=tensor@entry=0x407fdbb4, tensor_index=tensor_index@entry=0, 
    subgraph_idx=<optimized out>, subgraph_idx@entry=0, allocate_temp=<optimized out>, allocate_temp@entry=false) at tensorflow/lite/micro/micro_allocator.cc:1110
#3  0x000098ac in tflite::RecordingMicroAllocator::PopulateTfLiteTensorFromFlatbuffer (this=0x407fdbd4, model=0x15cd9 <kTestConvModelData+36>, tensor=0x407fdbb4, tensor_index=0, subgraph_index=0, allocate_temp=false)
    at tensorflow/lite/micro/recording_micro_allocator.cc:236
#4  0x0000910e in tflite::MicroAllocator::AllocatePersistentTfLiteTensor (this=0x407fdbd4, model=0x15cd9 <kTestConvModelData+36>, subgraph_allocations=0x0, tensor_index=0, subgraph_index=0) at tensorflow/lite/micro/micro_allocator.cc:814
#5  0x00008484 in main (argc=<optimized out>, argv=<optimized out>) at tensorflow/lite/micro/recording_micro_allocator_test.cc:169

My hypothesis was that the pre-generated test model file has an alignment defect but it turned out that just adding alignas(16) to the model variable was enough to fix the tests.

BUG=TestFix

[ddavis-2015]

@veblush All the required corrections are in PR #3197

[veblush]

@veblush All the required corrections are in PR #3197

Great! I'll try capturing your changes in this PR if your PR is not directly related to fixing this bug.

[veblush]

@ddavis-2015 I'd like to merge this separately from #3197 to ensure the two changes are atomic. I've ported over the relevant fixes from your PR to address the issue here. Thanks for your work on that!

[veblush]

@ddavis-2015 I'd like to merge this separately from #3197 to ensure the two changes are atomic. I've ported over the relevant fixes from your PR to address the issue here. Thanks for your work on that!

As described above, it doesn't seem to need to regenerate a test model. So I removed that part from the PR.

[ddavis-2015]

@ddavis-2015 I'd like to merge this separately from #3197 to ensure the two changes are atomic. I've ported over the relevant fixes from your PR to address the issue here. Thanks for your work on that!

As described above, it doesn't seem to need to regenerate a test model. So I removed that part from the PR.

@veblush
Oh! Duh. Of course that is the problem. Thanks for finding this. I'll change my PR accordingly.