Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932591Ab3JOPzA (ORCPT ); Tue, 15 Oct 2013 11:55:00 -0400 Received: from iolanthe.rowland.org ([192.131.102.54]:58620 "HELO iolanthe.rowland.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1759343Ab3JOPy4 (ORCPT ); Tue, 15 Oct 2013 11:54:56 -0400 Date: Tue, 15 Oct 2013 11:54:55 -0400 (EDT) From: Alan Stern X-X-Sender: stern@iolanthe.rowland.org To: Ming Lei cc: Vishal Annapurve , Linux Kernel Mailing List , linux-usb Subject: Re: [PATCH] usb-storage: scsiglue: Changing the command result In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3019 Lines: 72 On Tue, 15 Oct 2013, Ming Lei wrote: > On Thu, Sep 26, 2013 at 7:35 PM, Vishal Annapurve wrote: > > Hi, > > > > There was a recent commit in mainline for the scsi devices which do not > > respond properly to medium access command: > > > > commit 18a4d0a22ed6c54b67af7718c305cd010f09ddf8 > > > > [SCSI] Handle disk devices which can not process medium access commands > > We have experienced several devices which fail in a fashion we do not > > currently handle gracefully in SCSI. After a failure these devices will > > respond to the SCSI primary command set (INQUIRY, TEST UNIT READY, etc.) > > but any command accessing the storage medium will time out. > > > > I came across a USB drive which showed similar problem and what I see is > > usb storage is still not able to cope with such devices properly. > > > > The control flow downwards is like: > > scsi_times_out --> Setting cmd->result as DID_TIME_OUT > > scsi_error_handler > > scsi_unjam_host > > scsi_eh_abort_cmds > > command_abort (sets US_FLIDX_TIMED_OUT for us->dflags > > calls stop_transport, > > and waits for) usb_stor_control_thread > > (which is waiting for > > transport > > call to return inside > > > > usb_stor_invoke_transport) > > both > > usb_stor_control_thread and > > > > usb_stor_invoke_transport > > check for us->dflags > > timed_out bit and > > set the result as > > DID_ABORT > > and signal completion > > for command_abort > > to complete > > ...... > > sd_eh_action > > checks for cmd->result and finds out that it's DID_ABORT rather than > > DID_TIME_OUT. > > > > This patch updates the command result to be TIME_OUT explicitly before > > returning from command_abort in scsiglue.c. > > > > I would like to know if this patch can work out for such USB Storage > > devices? What would be the better way to do the same? > > Looks your diagnose is correct, and patch should be doable, the > only side effect is that previous returned DID_ABORT in srb->result > is replaced with DID_TIME_OUT now. > > Another way is to implement .eh_timed_out callback to return different > error for the two failure, but looks not a big deal. Instead of overriding the result code, a better way to do this would simply be to change the places where srb->result is currently set to DID_ABORT << 16. Just make it DID_TIME_OUT << 16 instead. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/