Ensure Thread stack is 8 byte aligned #5405

c1728p9 · 2017-10-30T21:31:04Z

Ensure both the stack and stack size used in the Thread class are aligned to 8 bytes. This prevents the runtime error "Thread 0 error -11: Unknown" due to incorrect stack alignment.

c1728p9 · 2017-10-30T21:41:03Z

/morph build

mbed-ci · 2017-10-30T22:09:57Z

Build : SUCCESS

Build number : 381
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/5405/

Triggering tests

/morph test
/morph uvisor-test

mbed-ci · 2017-10-30T23:56:34Z

Test : SUCCESS

Build number : 184
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/5405/184

ghost · 2017-10-31T10:22:54Z

rtos/Thread.cpp

@@ -36,15 +45,21 @@ namespace rtos {

 void Thread::constructor(osPriority priority,
        uint32_t stack_size, unsigned char *stack_mem, const char *name) {
+
+    const uintptr_t unaligned_mem = (uintptr_t)stack_mem;


This might be better using a reinterpret_cast.

ghost · 2017-10-31T10:23:02Z

rtos/Thread.cpp

    _attr.name = name ? name : "application_unnamed_thread";
-    _attr.stack_mem = (uint32_t*)stack_mem;
+    _attr.stack_mem = (uint32_t*)aligned_mem;


and this as a static_cast

bulislaw · 2017-10-31T10:26:03Z

rtos/Thread.cpp

@@ -23,6 +23,15 @@

 #include "mbed.h"
 #include "rtos/rtos_idle.h"
+#include "mbed_assert.h"
+
+#define ALIGN_UP(pos, align) ((pos) % (align) ? (pos) +  ((align) - (pos) % (align)) : (pos))


We don't need to complicate it with conditionals, how about standard: (pos + align - 1) & ~(align - 1)

The reason I used conditionals is so this works with all numbers and not just powers of 2.

Ok fair enough.

bulislaw

Good to have!

adbridge · 2017-10-31T14:07:57Z

@c1728p9 please address the review comments

c1728p9 · 2017-10-31T15:15:31Z

Updated this pr to use reinterpret_cast and static_cast.

c1728p9 · 2017-10-31T15:17:14Z

/morph build

ghost · 2017-10-31T15:36:05Z

Hm. Sorry, my bad, the static_cast broke everything.

Ensure both the stack and stack size used in the Thread class are aligned to 8 bytes. This prevents the runtime error "Thread 0 error -11: Unknown" due to incorrect stack alignment.

c1728p9 · 2017-10-31T17:06:21Z

Changed it to use reinterpret_cast for both.

c1728p9 · 2017-10-31T17:06:23Z

/morph build

mbed-ci · 2017-10-31T18:12:04Z

Build : SUCCESS

Build number : 395
Build artifacts/logs : http://mbed-os.s3-website-eu-west-1.amazonaws.com/?prefix=builds/5405/

Triggering tests

/morph test
/morph uvisor-test

mbed-ci · 2017-10-31T20:50:28Z

Test : SUCCESS

Build number : 193
Test logs :http://mbed-os-logs.s3-website-us-west-1.amazonaws.com/?prefix=logs/5405/193

pan-

Would it be possible to have ALIGN_UP and ALIGN_DOWN as functions rather than macros ?

Those macros will yield an invalid result if the pos in input is a pointer rather than an integer type due to pointer arithmetic.

c1728p9 · 2017-11-01T15:12:21Z

@pan- these macros won't yield an invalid result if you use a pointer. You'll get a compiler error as a function or macro if pos is a pointer, as % cannot be used on pointers. Without this fix PRs on master may fail the test tests-mbedmicro-rtos-mbed-threads randomly (but deterministically) due to slight ram changes causing the stack passed to Thread being unaligned.

c1728p9 mentioned this pull request Oct 30, 2017

Add ConditionVariable to mbed rtos #3648

Merged

0xc0170 requested review from bulislaw, pan- and a user October 31, 2017 08:51

0xc0170 added the needs: review label Oct 31, 2017

ghost reviewed Oct 31, 2017

View reviewed changes

ghost approved these changes Oct 31, 2017

View reviewed changes

bulislaw reviewed Oct 31, 2017

View reviewed changes

adbridge added needs: work and removed needs: review labels Oct 31, 2017

c1728p9 force-pushed the align_stack branch from 0687fd6 to 66ed3fa Compare October 31, 2017 15:13

Ensure Thread stack is 8 byte aligned

d98a011

Ensure both the stack and stack size used in the Thread class are aligned to 8 bytes. This prevents the runtime error "Thread 0 error -11: Unknown" due to incorrect stack alignment.

c1728p9 force-pushed the align_stack branch from 66ed3fa to d98a011 Compare October 31, 2017 16:58

pan- reviewed Nov 1, 2017

View reviewed changes

0xc0170 removed the needs: work label Nov 2, 2017

0xc0170 approved these changes Nov 2, 2017

View reviewed changes

0xc0170 added the ready for merge label Nov 2, 2017

c1728p9 mentioned this pull request Nov 2, 2017

malloc test: fix out of memory problem for NUCLEO_F070RB #5338

Merged

theotherjimmy merged commit 49883c3 into ARMmbed:master Nov 2, 2017

theotherjimmy added release-version: 5.6.4 and removed ready for merge labels Nov 2, 2017

maciejbocianski mentioned this pull request Nov 3, 2017

malloc test: change stack array to two dimensional array #5430

Merged

theotherjimmy mentioned this pull request Nov 9, 2017

Disable Arm Compiler 5 and Use Arm Compiler 6 by default #5406

Merged

Ensure Thread stack is 8 byte aligned #5405

Ensure Thread stack is 8 byte aligned #5405

Uh oh!

Conversation

c1728p9 commented Oct 30, 2017

Uh oh!

c1728p9 commented Oct 30, 2017

Uh oh!

mbed-ci commented Oct 30, 2017

Triggering tests

Uh oh!

mbed-ci commented Oct 30, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bulislaw left a comment

Choose a reason for hiding this comment

Uh oh!

adbridge commented Oct 31, 2017

Uh oh!

c1728p9 commented Oct 31, 2017

Uh oh!

c1728p9 commented Oct 31, 2017

Uh oh!

ghost commented Oct 31, 2017

Uh oh!

c1728p9 commented Oct 31, 2017

Uh oh!

c1728p9 commented Oct 31, 2017

Uh oh!

mbed-ci commented Oct 31, 2017

Triggering tests

Uh oh!

mbed-ci commented Oct 31, 2017

Uh oh!

pan- left a comment

Choose a reason for hiding this comment

Uh oh!

c1728p9 commented Nov 1, 2017

Uh oh!

Uh oh!