Re: Sun QLogic SCSI HBA issues

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Thu, 10 Jan 2019, Martin Husemann wrote:

> On Wed, Jan 09, 2019 at 07:03:11PM +0900, Christian Rodriguez wrote:
> > The following values show up during detection of the SCSI HBA and drive:
> > "*0x12000", "0x22000", "0x52000", "0x62000", "0x72000", and "0x82000"*.
> > Then, after detection it starts spamming a ton of messages with values like
> > *"0x952000", "0x992000", "0x9a2000", "0xa52000", *etc. It seems to just be
> > incrementing upward.
> > 
> > Any idea what this could be about? I'm really curious.
> 
> This decodes into:
> 
> 		type	seq	cmd
> 0x012000	1	1	0
> 0x022000	1	2	0
> 0x052000	1	5	0
> 0x062000	1	6	0
> 0x072000	1	7	0
> 0x082000	1	8	0
> 
> 0x952000	1	0x95	0
> 0x992000	1	0x99	0
> 0x9a2000	1	0x9a	0
> 0xa52000	1	0xa5	0
> 
> where type = 1 means scsi initiator.
> 
> This means that in the list of open commands, the first command (all or your
> values had cmd = 0) does not match the response from the device.
> 
> I have changed the error message to give better diagnostics. Could you
> retry when the next autobuild is available?

Those values look valid (as opposed to total random garbage) so it doesn't 
look like the firmware is stepping on the tokens.  

My guess it the firmware is giving duplicate completions to the commands.  
I've seen buggy ISP firmware in the past that does this.  Although it's 
very strange that it's always command 0 that's suffering from this.

Eduardo