GUIDINGLI commented on pull request #5504:
URL: https://github.com/apache/incubator-nuttx/pull/5504#issuecomment-1043876145
No, I don't think this crash caused by this PR.
For merge reason (my branch code base is no with
b316611ef0bd616609285c26dbc98fcf83b8aecb).
This PR is not reach my purpose.
There still have `if (up_interrupt_context)` branch at top of
`mm_takesemaphore()`.
In another word, this PR, only do the `mm_givesemaphore()` return when meet
`up_interrupt_context`.
This has no harmful for the logic.
So, could you please run more test without this PR ? Guess will also have
the problem.
```
76 bool mm_takesemaphore(FAR struct mm_heap_s *heap)
77 {
78 #if defined(CONFIG_BUILD_FLAT) || defined(__KERNEL__)
79 /* Check current environment */
80
81 if (up_interrupt_context())
// same with 105, will fix it later
82 {
83 /* Can't take semaphore in the interrupt handler */
84
85 return false;
86 }
87 else
88 #endif
89
90 /* getpid() returns the task ID of the task at the head of the
ready-to-
91 * run task list. mm_takesemaphore() may be called during context
92 * switches. There are certain situations during context switching
when
93 * the OS data structures are in flux and then can't be freed
immediately
94 * (e.g. the running thread stack).
95 *
96 * This is handled by getpid() to return the special value -ESRCH to
97 * indicate this special situation.
98 */
99
100 if (getpid() < 0)
101 {
102 return false;
103 }
104 #if defined(CONFIG_BUILD_FLAT) || defined(__KERNEL__)
105 else if (sched_idletask())
106 {
107 return false;
108 }
109 else if (up_interrupt_context())
110 {
111 #ifdef CONFIG_SMP
112 return false;
113 #else
114 int val;
115
116 /* Check the semaphore value, if held by someone, then return
false.
117 * Else, we can take it, return true.
118 */
119
120 _SEM_GETVALUE(&heap->mm_semaphore, &val);
121
122 return val > 0;
123 #endif
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
