On 10/16/23 15:38, Christoph Hellwig wrote:
On Mon, Oct 16, 2023 at 11:24:25AM +0200, Hannes Reinecke wrote:+ req = scsi_alloc_request(sdev->request_queue, REQ_OP_DRV_IN, + BLK_MQ_REQ_NOWAIT); + if (!req) { + SNIC_HOST_ERR(snic->shost, + "Devrst: TMF busy\n"); + goto dev_rst_end;So if we fail this allocation, which can easily happen, we just fail the reset and escalate. If that's fine we probably want a big fat comment about that here.
Yes, that is precisely the intention. If scsi_alloc_request() fails it means the we're on a quite busy system, where _all_ requests are in error (remember, this is SCSI EH where all commands are completed except for the failed ones). And in that case a host reset is probably the better option. But okay, I'll be adding a warning / comment here. Cheers, Hannes -- Dr. Hannes Reinecke Kernel Storage Architect hare@xxxxxxx +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman
