gitea/SDL3_fork
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Remove almost all instances of "volatile" keyword.

Browse Source 
As Tiffany pointed out in Bugzilla, volatile is not useful for thread safety:

https://software.intel.com/en-us/blogs/2007/11/30/volatile-almost-useless-for-multi-threaded-programming/

Some of these volatiles didn't need to be, some were otherwise protected by
spinlocks or mutexes, and some got moved over to SDL_atomic_t data, etc.

Fixes Bugzilla #3220.
This commit is contained in:
Ryan C. Gordon
2016-01-03 06:50:50 -05:00
parent 21f66ea295
commit fa8c83c1c1
11 changed files with 66 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
test/torturethread.c
View File

@@ -21,7 +21,7 @@
#define NUMTHREADS 10
static char volatile time_for_threads_to_die[NUMTHREADS];
static SDL_atomic_t time_for_threads_to_die[NUMTHREADS];
/* Call this instead of exit(), so we can clean up SDL: atexit() is evil. */
static void
@@ -58,7 +58,7 @@ ThreadFunc(void *data)
}
SDL_Log("Thread '%d' waiting for signal\n", tid);
while (time_for_threads_to_die[tid] != 1) {
while (SDL_AtomicGet(&time_for_threads_to_die[tid]) != 1) {
; /* do nothing */
}
@@ -92,7 +92,7 @@ main(int argc, char *argv[])
for (i = 0; i < NUMTHREADS; i++) {
char name[64];
SDL_snprintf(name, sizeof (name), "Parent%d", i);
time_for_threads_to_die[i] = 0;
SDL_AtomicSet(&time_for_threads_to_die[i], 0);
threads[i] = SDL_CreateThread(ThreadFunc, name, (void*) (uintptr_t) i);
if (threads[i] == NULL) {
@@ -102,7 +102,7 @@ main(int argc, char *argv[])
}
for (i = 0; i < NUMTHREADS; i++) {
time_for_threads_to_die[i] = 1;
SDL_AtomicSet(&time_for_threads_to_die[i], 1);
}
for (i = 0; i < NUMTHREADS; i++) {