fix race condition if during reconfiguration two cache managers try

to delete old inactive entries: one of them removes a entry just locked by
other manager from the queue and the rbtree as long inactive entry,
causes the latter manager to segfault leaving cache mutex locked,
the bug has been introduced in r3727
This commit is contained in:
Igor Sysoev 2010-09-02 14:31:47 +00:00
parent 0084867ae8
commit f819e3b7ed
2 changed files with 10 additions and 1 deletions

View file

@ -43,7 +43,8 @@ typedef struct {
unsigned error:10;
unsigned exists:1;
unsigned updating:1;
/* 12 unused bits */
unsigned deleting:1;
/* 11 unused bits */
ngx_file_uniq_t uniq;
time_t expire;

View file

@ -582,6 +582,7 @@ ngx_http_file_cache_exists(ngx_http_file_cache_t *cache, ngx_http_cache_t *c)
fcn->uses = 1;
fcn->count = 1;
fcn->updating = 0;
fcn->deleting = 0;
renew:
@ -1102,6 +1103,10 @@ ngx_http_file_cache_expire(ngx_http_file_cache_t *cache)
continue;
}
if (fcn->deleting) {
continue;
}
p = ngx_hex_dump(key, (u_char *) &fcn->node.key,
sizeof(ngx_rbtree_key_t));
len = NGX_HTTP_CACHE_KEY_LEN - sizeof(ngx_rbtree_key_t);
@ -1153,6 +1158,7 @@ ngx_http_file_cache_delete(ngx_http_file_cache_t *cache, ngx_queue_t *q,
*p = '\0';
fcn->count++;
fcn->deleting = 1;
ngx_shmtx_unlock(&cache->shpool->mutex);
len = path->name.len + 1 + path->len + 2 * NGX_HTTP_CACHE_KEY_LEN;
@ -1168,6 +1174,7 @@ ngx_http_file_cache_delete(ngx_http_file_cache_t *cache, ngx_queue_t *q,
ngx_shmtx_lock(&cache->shpool->mutex);
fcn->count--;
fcn->deleting = 0;
}
if (fcn->count == 0) {
@ -1431,6 +1438,7 @@ ngx_http_file_cache_add(ngx_http_file_cache_t *cache, ngx_http_cache_t *c)
fcn->error = 0;
fcn->exists = 1;
fcn->updating = 0;
fcn->deleting = 0;
fcn->uniq = c->uniq;
fcn->valid_sec = c->valid_sec;
fcn->body_start = c->body_start;