NetBSD Problem Report #7324

Received: (qmail 8170 invoked from network); 5 Apr 1999 11:48:33 -0000
Message-Id: <199904051136.MAA02078@black-star.demon.co.uk>
Date: Mon, 5 Apr 1999 12:36:40 +0100 (BST)
From: mpumford@black-star.demon.co.uk
Reply-To: mpumford@black-star.demon.co.uk
To: gnats-bugs@gnats.netbsd.org
Subject: Fails to detect IDE devices on Simtec IDE interface
X-Send-Pr-Version: 3.95

>Number:         7324
>Category:       port-arm32
>Synopsis:       Fails to detect IDE devices on Simtec IDE interface
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    port-arm32-maintainer
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Apr 05 04:50:00 +0000 1999
>Closed-Date:    Tue Apr 13 12:59:14 +0000 1999
>Last-Modified:  Tue Apr 13 13:00:13 +0000 1999
>Originator:     Mike Pumford
>Release:        All kernels supped 11/03/1999 until 01/04/1999
>Organization:
>Environment:
Acorn RiscPC 48MB RAM 2MB VRAM
Acorn SCSI Card
Simtec IDE Interface
Seagate 6.4GB UDMA drive attached to Simtec interface
Pioneer DRM-602X CDROM attached to Acorn SCSI interface.

System: NetBSD black-star.demon.co.uk 1.3K NetBSD 1.3K (BLACK-STAR) #40: Thu Mar 25 01:18:39 GMT 1999 mpumford@black-star.demon.co.uk:/usr/src/netbsd/sys/arch/arm32/compile/BLACK-STAR arm32



>Description:
The kernel fails to detect devices attached to the Simtec IDE card at boot and
as a consequence fails to find wd0 the root device.

Reverting to v1.60 of sys/dev/ic/wdc.c solves this problem and allows the boot
to continue.


>How-To-Repeat:
Boot the kernel.
>Fix:
>Release-Note:
>Audit-Trail:

From: Manuel Bouyer <bouyer@antioche.lip6.fr>
To: mpumford@black-star.demon.co.uk
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface
Date: Mon, 5 Apr 1999 16:21:14 +0200

 On Apr 5, mpumford@black-star.demon.co.uk wrote
 > 
 > >Description:
 > The kernel fails to detect devices attached to the Simtec IDE card at boot and
 > as a consequence fails to find wd0 the root device.
 > 
 > Reverting to v1.60 of sys/dev/ic/wdc.c solves this problem and allows the boot
 > to continue.
 > 

 What version of sys/dev/ic/wdc.c did you use ? Current version is
 1.66 (1.66.0.2 on the 1.4 branch), please give it a try.

 --
 Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
 --

From: Mark Brinicombe <mark@causality.com>
To: mpumford@black-star.demon.co.uk
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface
Date: Tue, 6 Apr 1999 00:06:03 -0700 (PDT)

 Hi,
 Can you try the 1.4_ALPHA kernels that are online. With these kernels I   
 see the card/cable fault error (needs fixing) but was able to probe and   
 attach the IDE disk fitted to the card.

                 Mark



From: Mike Pumford <mpumford@black-star.demon.co.uk>
To: mark@causality.com
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface 
Date: Tue, 06 Apr 1999 23:34:26 +0100

 > Hi,
 > Can you try the 1.4_ALPHA kernels that are online. With these kernels I   
 > see the card/cable fault error (needs fixing) but was able to probe and   
 > attach the IDE disk fitted to the card.
 >      
 No luck I'm afraid. I also tried the disk on the internal IDE interface. It 
 was probed correctly but the kernel panicked immediately afterwards with a 
 prefetch abort in supervisor mode.

 Mike


From: Mike Pumford <mpumford@black-star.demon.co.uk>
To: Manuel Bouyer <bouyer@antioche.lip6.fr>
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface 
Date: Wed, 07 Apr 1999 19:17:51 +0100

 > Oh well ... could you turn on debugging messages and send me the dmesg output ?
 > Set wdcdebug_atapi_mask, wdcdebug_mask and wdcdebug_wd_mask to 0x10
 > (DEBUG_PROBE). This can be done in sources, or with gdb on the binary:
 > gdb --write /netbsd
 > set wdcdebug_atapi_mask=0x10
 > set wdcdebug_mask=0x10
 > set wdcdebug_wd_mask=0x10
 > quit
 > 
 I patched both the last working kernel I have (with v1.60) and the kernel 
 which fails to probe.

 This is the standard RISCPC 1.4_ALPHA distribution kernel on ftp.netbsd.org 
 patched as described above.

 The drive is the master on the primary device:

 simide0 at podulebus0 [podule 1]:
 simide0:0: before reset, st0=0x50, st1=0x0
 simide0:0: wdcwait_reset() end, st0=0x50, st1=0x0
 simide0:0: after reset, ret_value=0x3
 simide0:0:0: after reset, sc=0x0 sn=0x27 cl=0x7a ch=0xc
 simide0:0:1: after reset, sc=0x0 sn=0x27 cl=0x7a ch=0xc
 simide0:0:1: not ready
 wdcattach: ch_drive_flags 0x4 0x0
 simide0:0: wdcwait_reset() end, st0=0x50, st1=0x0
 simide0:1: before reset, st0=0xff, st1=0xff

 This is the last working kernel I built for myself again patched as described 
 above with the same hardware configuration.

 simide0 at podulebus0 [ podule 1 ]:
 simide0:0: before reset, st0=0x50, st1=0x0
 simide0:0: after reset, ret_value=0x3
 simide0:0:0: after reset, sc=0x0 sn=0x1 cl=0x69 ch=0xd
 simide0:0:1: after reset, sc=0x0 sn=0x1 cl=0x69 ch=0xd
 simide0:0:1: not ready
 wdcattach: ch_drive_flags 0x1 0x0
 wd0 at simide0 channel 0 drive 0wdattach
 : <ST36451A>
 wd0: drive supports 16-sector pio transfers, lba addressing
 wd0: 6149MB, 13328 cyl, 15 head, 63 sec, 512 bytes/sect x 12594960 sectors
 wd0: atap_dmatiming_mimi=120, atap_dmatiming_recom=120
 wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 2
 simide0:1: before reset, st0=0xff, st1=0xff


 Mike


From: Manuel Bouyer <bouyer@antioche.lip6.fr>
To: Mike Pumford <mpumford@black-star.demon.co.uk>
Cc: gnats-bugs@gnats.netbsd.org, port-arm32@netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface
Date: Thu, 8 Apr 1999 16:22:37 +0200

 On Apr 7, Mike Pumford wrote
 > I patched both the last working kernel I have (with v1.60) and the kernel 
 > which fails to probe.
 > 
 > This is the standard RISCPC 1.4_ALPHA distribution kernel on ftp.netbsd.org 
 > patched as described above.
 > 
 > The drive is the master on the primary device:
 > 
 > simide0 at podulebus0 [podule 1]:
 > simide0:0: before reset, st0=0x50, st1=0x0
 > simide0:0: wdcwait_reset() end, st0=0x50, st1=0x0
 > simide0:0: after reset, ret_value=0x3
 > simide0:0:0: after reset, sc=0x0 sn=0x27 cl=0x7a ch=0xc
 > simide0:0:1: after reset, sc=0x0 sn=0x27 cl=0x7a ch=0xc
 > simide0:0:1: not ready
 > wdcattach: ch_drive_flags 0x4 0x0
 > simide0:0: wdcwait_reset() end, st0=0x50, st1=0x0
 > simide0:1: before reset, st0=0xff, st1=0xff

 Well, i've at last found a bug: your drive should have been attached as
 an ST506 :)

 The problem here is that after the reset we read random values in the
 registers instead of the required 'ata signature'.
 Could you try to increase the delay() values ?
 I saw this problem recently on a i386, the problem turned out to be
 that on i386, delay(1) was equivalent to delay(0) (now fixed). I bumped
 all delays to 10. Maybe it's not enouth for arm32 ?
 Can you try to bump all delay(10) to delay(100) in sys/dev/ic/wdc.c ?
 This should be safe, all delay(10) present here are not in an interrupt path
 as far as I can tell ...

 > 
 > This is the last working kernel I built for myself again patched as described 
 > above with the same hardware configuration.
 > 
 > simide0 at podulebus0 [ podule 1 ]:
 > simide0:0: before reset, st0=0x50, st1=0x0
 > simide0:0: after reset, ret_value=0x3
 > simide0:0:0: after reset, sc=0x0 sn=0x1 cl=0x69 ch=0xd
 > simide0:0:1: after reset, sc=0x0 sn=0x1 cl=0x69 ch=0xd
 > simide0:0:1: not ready
 > wdcattach: ch_drive_flags 0x1 0x0
 > wd0 at simide0 channel 0 drive 0wdattach
 > : <ST36451A>
 > wd0: drive supports 16-sector pio transfers, lba addressing
 > wd0: 6149MB, 13328 cyl, 15 head, 63 sec, 512 bytes/sect x 12594960 sectors
 > wd0: atap_dmatiming_mimi=120, atap_dmatiming_recom=120
 > wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 2
 > simide0:1: before reset, st0=0xff, st1=0xff

 Same problem here (random values in registers) but this code didn't rely on
 ata signature.

 --
 Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
 --

From: Mike Pumford <mpumford@black-star.demon.co.uk>
To: Manuel Bouyer <bouyer@antioche.lip6.fr>
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface 
Date: Sat, 10 Apr 1999 19:23:35 +0100

 > 
 > Well, i've at last found a bug: your drive should have been attached as
 > an ST506 :)
 > 
 > The problem here is that after the reset we read random values in the
 > registers instead of the required 'ata signature'.
 > Could you try to increase the delay() values ?
 > I saw this problem recently on a i386, the problem turned out to be
 > that on i386, delay(1) was equivalent to delay(0) (now fixed). I bumped
 > all delays to 10. Maybe it's not enouth for arm32 ?
 > Can you try to bump all delay(10) to delay(100) in sys/dev/ic/wdc.c ?
 > This should be safe, all delay(10) present here are not in an interrupt path
 > as far as I can tell ...
 > 
 Changing the delay value makes no difference. Also the drive is successfully 
 detected when attached to the motherboard IDE interface of the machine.

 I do have a patch which allows my system to boot but I don't propose it as a 
 fix for the problem (but it does at least let me test out 1.4_ALPHA some more.)

 I have attached the patch as it may help others with the same problem.

 -------------------------------------------------------
 *** wdc.c       1999/04/05 18:29:25     1.1.1.19
 --- wdc.c       1999/04/10 18:17:12
 ***************
 *** 261,266 ****
 --- 261,272 ----
                 } else if (sc == 0x01 && sn == 0x01 &&
                     cl == 0x00 && ch == 0x00) {
                         chp->ch_drive[drive].drive_flags |= DRIVE_ATA;
 +               } 
 +               /* SIMIDE_HACK (assume any drive without ATA signature is
 +                * an ATA Drive).
 +                  */
 +               else {
 +                       chp->ch_drive[drive].drive_flags |= DRIVE_ATA;
                 }
         }
         /*
 -------------------------------------------------------
 Mike



From: Manuel Bouyer <bouyer@antioche.lip6.fr>
To: Mike Pumford <mpumford@black-star.demon.co.uk>
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: port-arm32/7324: Fails to detect IDE devices on Simtec IDE interface
Date: Sun, 11 Apr 1999 22:57:27 +0200

 --RYliJ5NKNqkz5buL
 Content-Type: text/plain; charset=us-ascii

 On Apr 10, Mike Pumford wrote
 > Changing the delay value makes no difference. Also the drive is successfully 
 > detected when attached to the motherboard IDE interface of the machine.

 Ok, as some other IDE controllers seems to behave this way as well, I
 changed the probe to try to deal with this more gracefully.
 Could you try the attached patch ? I just commited it, and will ask a pullup

 --
 Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
 --

 --RYliJ5NKNqkz5buL
 Content-Type: text/plain; charset=us-ascii
 Content-Disposition: attachment; filename=wdc_diff

 Index: dev/ic/wdc.c
 ===================================================================
 RCS file: /cvsroot/src/sys/dev/ic/wdc.c,v
 retrieving revision 1.66
 diff -u -r1.66 wdc.c
 --- wdc.c	1999/04/01 21:46:29	1.66
 +++ wdc.c	1999/04/11 20:31:36
 @@ -231,9 +231,9 @@

  	/*
  	 * Test presence of drives. First test register signatures looking for
 -	 * ATAPI devices , then rescan and try an ATA command, in case it's an
 -	 * old drive.
 -	 * Fill in drive_flags accordingly
 +	 * ATAPI devices. If it's not an ATAPI and reset said there may be
 +	 * something here assume it's ATA or OLD. Ghost will be killed later in
 +	 * attach routine.
  	 */
  	for (drive = 0; drive < 2; drive++) {
  		if ((ret_value & (0x01 << drive)) == 0)
 @@ -253,63 +253,16 @@
  	    	    chp->channel, drive, sc, sn, cl, ch), DEBUG_PROBE);
  		/*
  		 * sc is supposted to be 0x1 for ATAPI but at last one drive
 -		 * set it to 0x0.
 +		 * set it to 0x0 - or maybe it's the controller.
  		 */
  		if ((sc == 0x00 || sc == 0x01) && sn == 0x01 &&
  		    cl == 0x14 && ch == 0xeb) {
  			chp->ch_drive[drive].drive_flags |= DRIVE_ATAPI;
 -		} else if (sc == 0x01 && sn == 0x01 &&
 -		    cl == 0x00 && ch == 0x00) {
 -			chp->ch_drive[drive].drive_flags |= DRIVE_ATA;
 -		}
 -	}
 -	/*
 -	 * Maybe there's an old device, try to detect it if we didn't
 -	 * find a ATA or ATAPI device.
 -	 */
 -	if ((chp->ch_drive[0].drive_flags & DRIVE) != 0 ||
 -	    (chp->ch_drive[1].drive_flags & DRIVE) != 0)
 -		return (ret_value);	
 -	for (drive = 0; drive < 2; drive++) {
 -		if ((ret_value & (0x01 << drive)) == 0)
 -			continue;
 -		bus_space_write_1(chp->cmd_iot, chp->cmd_ioh, wd_sdh,
 -		    WDSD_IBM | (drive << 4));
 -		delay(10);
 -		/*
 -		 * Test registers writability (Error register not writable,
 -		 * but cyllo is), then try an ATA command.
 -		 */
 -		bus_space_write_1(chp->cmd_iot, chp->cmd_ioh, wd_error, 0x58);
 -		bus_space_write_1(chp->cmd_iot, chp->cmd_ioh, wd_cyl_lo, 0xa5);
 -		if (bus_space_read_1(chp->cmd_iot, chp->cmd_ioh, wd_error) ==
 -		    0x58 ||
 -		    bus_space_read_1(chp->cmd_iot, chp->cmd_ioh, wd_cyl_lo) !=
 -		    0xa5) {
 -			WDCDEBUG_PRINT(("%s:%d:%d: register writability "
 -			    "failed\n",
 -			    chp->wdc ? chp->wdc->sc_dev.dv_xname : "wdcprobe",
 -			    chp->channel, drive), DEBUG_PROBE);
 -			ret_value &= ~(0x01 << drive);
 -			continue;
 -		}
 -		if (wait_for_ready(chp, 10000) != 0) {
 -			WDCDEBUG_PRINT(("%s:%d:%d: not ready\n",
 -			    chp->wdc ? chp->wdc->sc_dev.dv_xname : "wdcprobe",
 -			    chp->channel, drive), DEBUG_PROBE);
 -			ret_value &= ~(0x01 << drive);
 -			continue;
 -		}
 -		bus_space_write_1(chp->cmd_iot, chp->cmd_ioh, wd_command,
 -		    WDCC_RECAL);
 -		if (wait_for_ready(chp, 10000) == 0) {
 -			chp->ch_drive[drive].drive_flags |=
 -			    DRIVE_OLD;
  		} else {
 -			WDCDEBUG_PRINT(("%s:%d:%d: WDCC_RECAL failed\n",
 -			    chp->wdc ? chp->wdc->sc_dev.dv_xname : "wdcprobe",
 -			    chp->channel, drive), DEBUG_PROBE);
 -			ret_value &= ~(0x01 << drive);
 +			chp->ch_drive[drive].drive_flags |= DRIVE_ATA;
 +			if (chp->wdc == NULL ||
 +			    (chp->wdc->cap & WDC_CAPABILITY_PREATA) != 0)
 +				chp->ch_drive[drive].drive_flags |= DRIVE_OLD;
  		}
  	}
  	return (ret_value);	
 @@ -351,12 +304,42 @@
  		    (WDC_CAPABILITY_DATA16 | WDC_CAPABILITY_DATA32)) ==
  		    WDC_CAPABILITY_DATA32)
  			chp->ch_drive[i].drive_flags |= DRIVE_CAP32;
 +		if ((chp->ch_drive[i].drive_flags & DRIVE) == 0)
 +			continue;

  		/* Issue a IDENTIFY command, to try to detect slave ghost */
 -		if (ata_get_params(&chp->ch_drive[i], AT_POLL, &params) !=
 +		if (ata_get_params(&chp->ch_drive[i], AT_POLL, &params) ==
  		    CMD_OK) {
 +			/* If IDENTIFY succeded, this is not an OLD ctrl */
 +			chp->ch_drive[0].drive_flags &= ~DRIVE_OLD;
 +			chp->ch_drive[1].drive_flags &= ~DRIVE_OLD;
 +		} else {
  			chp->ch_drive[i].drive_flags &=
  			    ~(DRIVE_ATA | DRIVE_ATAPI);
 +			WDCDEBUG_PRINT(("%s:%d:%d: IDENTIFY failed\n",
 +			    chp->wdc->sc_dev.dv_xname,
 +			    chp->channel, i), DEBUG_PROBE);
 +			if ((chp->ch_drive[i].drive_flags & DRIVE_OLD) == 0)
 +				continue;
 +			/* Pre-ATA drive ? */
 +			bus_space_write_1(chp->cmd_iot, chp->cmd_ioh, wd_sdh,
 +			    WDSD_IBM | (i << 4));
 +			delay(100);
 +			if (wait_for_ready(chp, 10000) != 0) {
 +				WDCDEBUG_PRINT(("%s:%d:%d: not ready\n",
 +				    chp->wdc->sc_dev.dv_xname,
 +				    chp->channel, i), DEBUG_PROBE);
 +				chp->ch_drive[i].drive_flags &= ~DRIVE_OLD;
 +				continue;
 +			}
 +			bus_space_write_1(chp->cmd_iot, chp->cmd_ioh,
 +			    wd_command, WDCC_RECAL);
 +			if (wait_for_ready(chp, 10000) != 0) {
 +				WDCDEBUG_PRINT(("%s:%d:%d: WDCC_RECAL failed\n",
 +				    chp->wdc->sc_dev.dv_xname,
 +				    chp->channel, i), DEBUG_PROBE);
 +				chp->ch_drive[i].drive_flags &= ~DRIVE_OLD;
 +			}
  		}
  	}
  	ctrl_flags = chp->wdc->sc_dev.dv_cfdata->cf_flags;
 @@ -366,6 +349,11 @@
  	    chp->ch_drive[0].drive_flags, chp->ch_drive[1].drive_flags),
  	    DEBUG_PROBE);

 +	/* If no drives, abort here */
 +	if ((chp->ch_drive[0].drive_flags & DRIVE) == 0 &&
 +	    (chp->ch_drive[1].drive_flags & DRIVE) == 0)
 +		return;
 +
  	/*
  	 * Attach an ATAPI bus, if needed.
  	 */
 @@ -391,7 +379,8 @@
  	}

  	for (i = 0; i < 2; i++) {
 -		if ((chp->ch_drive[i].drive_flags & DRIVE_ATA) == 0) {
 +		if ((chp->ch_drive[i].drive_flags &
 +		    (DRIVE_ATA | DRIVE_OLD)) == 0) {
  			continue;
  		}
  		memset(&aa_link, 0, sizeof(struct ata_atapi_attach));
 Index: dev/ic/wdcvar.h
 ===================================================================
 RCS file: /cvsroot/src/sys/dev/ic/wdcvar.h,v
 retrieving revision 1.16
 diff -u -r1.16 wdcvar.h
 --- wdcvar.h	1999/04/01 21:46:30	1.16
 +++ wdcvar.h	1999/04/11 20:31:36
 @@ -90,6 +90,7 @@
  #define	WDC_CAPABILITY_ATA_NOSTREAM 0x0040 /* Don't use stream funcs on ATA */
  #define	WDC_CAPABILITY_ATAPI_NOSTREAM 0x0080 /* Don't use stream f on ATAPI */
  #define WDC_CAPABILITY_NO_EXTRA_RESETS 0x0100 /* only reset once */
 +#define WDC_CAPABILITY_PREATA 0x0200 /* ctrl can be a pre-ata one */
  	u_int8_t      PIO_cap; /* highest PIO mode supported */
  	u_int8_t      DMA_cap; /* highest DMA mode supported */
  	u_int8_t      UDMA_cap; /* highest UDMA mode supported */
 Index: dev/isa/wdc_isa.c
 ===================================================================
 RCS file: /cvsroot/src/sys/dev/isa/wdc_isa.c,v
 retrieving revision 1.13
 diff -u -r1.13 wdc_isa.c
 --- wdc_isa.c	1999/02/22 03:24:33	1.13
 +++ wdc_isa.c	1999/04/11 20:31:36
 @@ -150,7 +150,8 @@
  		sc->sc_wdcdev.dma_finish = wdc_isa_dma_finish;
  		wdc_isa_dma_setup(sc);
  	}
 -	sc->sc_wdcdev.cap |= WDC_CAPABILITY_DATA16 | WDC_CAPABILITY_DATA32;
 +	sc->sc_wdcdev.cap |= WDC_CAPABILITY_DATA16 | WDC_CAPABILITY_DATA32 |
 +	    WDC_CAPABILITY_PREATA;
  	sc->sc_wdcdev.PIO_cap = 0;
  	sc->wdc_chanptr = &sc->wdc_channel;
  	sc->sc_wdcdev.channels = &sc->wdc_chanptr;

 --RYliJ5NKNqkz5buL--
State-Changed-From-To: open->feedback 
State-Changed-By: bouyer 
State-Changed-When: Mon Apr 12 02:22:23 PDT 1999 
State-Changed-Why:  
I just commited a fix to the 1.4 branch (which should be identical to      
the patch I sent you yesterday). Could you compile a new kernel and tell me how 
it goes ? 
State-Changed-From-To: feedback->closed 
State-Changed-By: bouyer 
State-Changed-When: Tue Apr 13 05:59:14 PDT 1999 
State-Changed-Why:  
Mike said the patch I sent him solved the problem. Since then this patch has 
been pulled up to the branch 
>Unformatted:

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.