I am sharing a set of globals between different threads, so I allocate the memory as needed depending on how much is being placed into it. Kinda like a buffer. Whenever I try to free it from the same thread, but different function, glib goes crazy and segfaults.
I'm not a c programmer by far. I am trying to clear up memory leaks in the code that I hacked together.
char *keybuffer[], *valuesbuffer[], *returnbuffer[], *bufferaction;
int memcache_lock=1,*bufferelements, bufferowner=0;
void memcache_clear_buffers(){
free(&bufferaction); //I get the same reaction when I have the & and without the &
// free(*keybuffer); //I thought maybe I need to use the pointer. Then
// free(*valuesbuffer); // I thought maybe I needed to reference the address in memory.
// free(*returnbuffer); // I appreciate any help.
// free(*bufferelements);
}
void memcache_allocate_buffers(int size){
*keybuffer = (char *)malloc(size * sizeof(char *));
*valuesbuffer = (char *)malloc(size * sizeof(char *));
*returnbuffer = (char *)malloc(size * sizeof(char *));
bufferelements = malloc(sizeof(int));
bufferaction = (char *)malloc(sizeof(char*));
}
Here is the output from valgrind.
3500== Invalid free() / delete / delete[]
==3500== at 0x4027C02: free (vg_replace_malloc.c:366)
==3500== by 0x4191D4C: memcache_clear_buffers (dm_memcache.c:254)
==3500== by 0x41921ED: memcache_set (dm_memcache.c:328)
==3500== by 0x4192281: memcache_sid (dm_memcache.c:151)
==3500== by 0x418316D: db_user_exists (dm_db.c:3208)
==3500== by 0x403F7F2: auth_user_exists (authsql.c:49)
==3500== by 0x419ADA6: auth_user_exists (authmodule.c:130)
==3500== by 0x4040D4E: auth_validate (authsql.c:333)
==3500== by 0x419B106: auth_validate (authmodule.c:163)
==3500== by 0x74656CFF: ???
==3500== Address 0x41b55c0 is 0 bytes inside data symbol "bufferaction"
How is it an invalid free?