Question regarding timeout handling in scsi_io_completion

Hi, A recent patch added a timeout into scsi_io_completion in order to avoid infinite command retries. In the case where we have a failed device and the device cannot recover from the error state: If in this case 'error' is left as 0, could this cause issues in the DM layer (mpath for example),

RE: BLKDISCARD and AIO

Dell - Internal Use - Confidential Thanks, The issue is that we run a single threaded process (we maintain our own threads) and use eventfd to driver both networking and AIO, so blocking on BLKDISCARD will block all IO done by the process. Menny -Original Message- From: Martin K. P