[U-Boot] [PATCH] cmd_ide: Fix detection problem with missing devices

[Wolfgang Denk]

Dear Stefan Roese,

In message <200905141714.34154.sr at denx.de> you wrote:
>
> > U-Boot 1.3.2-rc2-g02e4b4e3 (Feb 28 2008 - 16:46:26)
>   ^^^^^^^^^^^^^
> That's very old.

Indeed. That's why I included this information as data point.

> From my understanding this worked at some point on the CPCI750 as well. 

Ah! You didn't mention this before. That's bad.

> Perhaps you could give the TOT U-Boot version a try on your board used above. 
> My guess is that this will fail as well.

Correct.

> I tried to locate a problem that was introduced at some time but couldn't 
> locate it (no, I didn't use git bisect). But looking at the current code this 

Well, I did bisect it, and found that it's an "interesting" issue, as
inbettween there obviously were some versions where it didn't work at
all; i. e.:

old:

	IDE:   Bus 0: ..OK
	  Device 0: Model: HITACHI_DK23DA-20 Firm: 00J2A0A1 Ser#: 12Y0MN
		    Type: Hard Disk
		    Capacity: 19077.1 MB = 18.6 GB (39070080 x 512)
	  Device 1: not available

then:

	IDE:   Bus 0: OK 
	  Device 0: Model: HITACHI_DK23DA-20 Firm: 00J2A0A1 Ser#: 12Y0MN
	Vendor: HITACHI_DK23DA-20 Prod.: 12Y0MN Rev: 00J2A0A1
		    Type: Hard Disk
		    Capacity: 19077.1 MB = 18.6 GB (39070080 x 512)
	  Device 1: not available

then:

	IDE:   Bus 0: OK 
	  Device 0: not available
	  Device 1: not available

	This was caused by:

	574b319512b13e10800f0045e39b993f4ca25e42 is first bad commit
	commit 574b319512b13e10800f0045e39b993f4ca25e42
	Author: Detlev Zundel <dzu at denx.de>
	Date:   Mon May 5 16:11:21 2008 +0200

	    Fix disk type output in disk/part.c
	    
	    Signed-off-by: Detlev Zundel <dzu at denx.de>

	The bug is here:

	-       if (dev_desc->type==DEV_TYPE_UNKNOWN) {
	-               puts ("not available\n");
	-               return;
	-       }
	-       if (dev_desc->if_type==IF_TYPE_SCSI)  {
	-               printf ("(%d:%d) ", dev_desc->target,dev_desc->lun);
	-       }
	-       if (dev_desc->if_type==IF_TYPE_IDE) {
	-               printf ("Model: %s Firm: %s Ser#: %s\n",
	+       switch (dev_desc->type) {
	+       case IF_TYPE_SCSI:
	...
	+       case DEV_TYPE_UNKNOWN:
	+       default:
	+               puts ("not available\n");
	+               return;

        The initial test was on "dev_desc->type", but this was
        "optimized" into the switch - not noticing that the switch
        tests for "dev_desc->if_type" which can only take "IF_*"
        values, but no "DEV_*" values.

	Later, the switch() was fixed into "switch (dev_desc->if_type)",
	but then the DEV_TYPE_UNKNOWN was converted into IF_TYPE_UNKNOWN
	which might have seemed logical at that time, but by that we
	lost all trace of the original test; see 

	commit 8ec6e332eace0ee78c71ee5f645d12b06813b86f
	Author: Tor Krill <tor at excito.com>
	Date:   Thu May 29 11:10:30 2008 +0200

	    Fix incorrect switch for IF_TYPE in part.c


and finally:

	IDE:   Bus 0: OK 
	  Device 0: Model: HITACHI_DK23DA-20 Firm: 00J2A0A1 Ser#: 12Y0MN
		    Type: Hard Disk
		    Capacity: 19077.1 MB = 18.6 GB (39070080 x 512)
	  Device 1: Model:  Firm:  Ser#: 
		    Type: # 1F #
		    Capacity: not available


Hey, that's fun, isn't it :-(


> just can't work as is for the 2nd drive. And I still think my fix is ok.

I don't think so. Could you please try the patch in the following
mail? Thanks in advance.

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Democracy is mob rule, but with income taxes.