Hi Dan,
On 6/15/2016 8:33 PM, Dan Carpenter wrote:
Hello John Hsu,
The patch b50455fab459: "ASoC: nau8825: cross talk suppression measurement function" from Jun 7, 2016, leads to the following static checker warning:
sound/soc/codecs/nau8825.c:265 nau8825_sema_acquire() warn: 'sem:&nau8825->xtalk_sem' is sometimes locked here and sometimes unlocked.
sound/soc/codecs/nau8825.c 240 /** 241 * nau8825_sema_acquire - acquire the semaphore of nau88l25 242 * @nau8825: component to register the codec private data with 243 * @timeout: how long in jiffies to wait before failure or zero to wait 244 * until release 245 * 246 * Attempts to acquire the semaphore with number of jiffies. If no more 247 * tasks are allowed to acquire the semaphore, calling this function will 248 * put the task to sleep. If the semaphore is not released within the 249 * specified number of jiffies, this function returns. 250 * Acquires the semaphore without jiffies. If no more tasks are allowed 251 * to acquire the semaphore, calling this function will put the task to 252 * sleep until the semaphore is released. 253 * It returns if the semaphore was acquired.
Sometimes it returns without the lock held.
254 */ 255 static void nau8825_sema_acquire(struct nau8825 *nau8825, long timeout) 256 { 257 int ret; 258 259 if (timeout) 260 ret = down_timeout(&nau8825->xtalk_sem, timeout); 261 else 262 ret = down_interruptible(&nau8825->xtalk_sem); 263 264 if (ret < 0) 265 dev_warn(nau8825->dev, "Acquire semaphone fail\n");
This function makes me sad. Warn and corrupt is my least favourite anti-pattern.
Is it better to separate this function to two case, one locked with timeout and the other not? Do you have any suggestion about it? Thank you.