ide-scsi on sparc64...

ide-scsi on sparc64...

[Peter Jones]

Has anybody other than me tried this?  I've been atacking it for, well, a
day or so now, and I can't seem to make it work.  If I turn off ide-cdrom,
and turn on ide-scsi, I get an unaligned trap in idescsi_setup, at the
line listed below.

static void idescsi_setup (ide_drive_t *drive, idescsi_scsi_t *scsi, int
id)
{
        DRIVER(drive)->busy++;
        idescsi_drives[id] = drive;
        drive->driver_data = scsi;
        drive->ready_stat = 0;
        memset (scsi, 0, sizeof (idescsi_scsi_t));
        scsi->drive = drive;
        if (drive->id && (drive->id->config & 0x0060) = 0x20)
                set_bit (IDESCSI_DRQ_INTERRUPT, &scsi->flags);
        set_bit(IDESCSI_TRANSFORM, &scsi->transform);           <-- here
        clear_bit(IDESCSI_SG_TRANSFORM, &scsi->transform);
#if IDESCSI_DEBUG_LOG
        set_bit(IDESCSI_LOG_CMD, &scsi->log);
#endif /* IDESCSI_DEBUG_LOG */
        idescsi_add_settings(drive);
}

It panics claiming unaligned load/store... This is a bit past my
expertise.  Any ideas?

This is on a virgin CVS checkout.
 -- Peter

"Talking much about oneself can also be a means to conceal oneself."
		-- Nietzsche

Re: ide-scsi on sparc64...

[David Miller]

--- drivers/scsi/ide-scsi.c.~1~	Mon Feb 22 11:36:13 1999
+++ drivers/scsi/ide-scsi.c	Sun Apr 18 14:37:07 1999
@@ -67,7 +67,7 @@
 	int b_count;				/* Bytes transferred from current entry */
 	Scsi_Cmnd *scsi_cmd;			/* SCSI command */
 	void (*done)(Scsi_Cmnd *);		/* Scsi completion routine */
-	unsigned int flags;			/* Status/Action flags */
+	unsigned long flags;			/* Status/Action flags */
 	unsigned long timeout;			/* Command timeout */
 } idescsi_pc_t;
 
@@ -93,8 +93,8 @@
 	ide_drive_t *drive;
 	idescsi_pc_t *pc;			/* Current packet command */
 	unsigned int flags;			/* Status/Action flags */
-	int transform;				/* SCSI cmd translation layer */
-	int log;				/* log flags */
+	unsigned long transform;		/* SCSI cmd translation layer */
+	unsigned long log;			/* log flags */
 } idescsi_scsi_t;
 
 /*

Re: ide-scsi on sparc64...

[Peter Jones]

Oh.  Duh.  I guess that is pretty obvious.  Why not the second "unsigned
int flags" as well though?

 -- Peter

"Talking much about oneself can also be a means to conceal oneself."
		-- Nietzsche

On Sun, 18 Apr 1999, David Miller wrote:

> 
> --- drivers/scsi/ide-scsi.c.~1~	Mon Feb 22 11:36:13 1999
> +++ drivers/scsi/ide-scsi.c	Sun Apr 18 14:37:07 1999
> @@ -67,7 +67,7 @@
>  	int b_count;				/* Bytes transferred from current entry */
>  	Scsi_Cmnd *scsi_cmd;			/* SCSI command */
>  	void (*done)(Scsi_Cmnd *);		/* Scsi completion routine */
> -	unsigned int flags;			/* Status/Action flags */
> +	unsigned long flags;			/* Status/Action flags */
>  	unsigned long timeout;			/* Command timeout */
>  } idescsi_pc_t;
>  
> @@ -93,8 +93,8 @@
>  	ide_drive_t *drive;
>  	idescsi_pc_t *pc;			/* Current packet command */
>  	unsigned int flags;			/* Status/Action flags */
> -	int transform;				/* SCSI cmd translation layer */
> -	int log;				/* log flags */
> +	unsigned long transform;		/* SCSI cmd translation layer */
> +	unsigned long log;			/* log flags */
>  } idescsi_scsi_t;
>  
>  /*
> 

Re: ide-scsi on sparc64...

[David Miller]

   Date: Sun, 18 Apr 1999 18:20:47 -0400 (EDT)
   From: Peter Jones <pjones@redhat.com>

   Oh.  Duh.  I guess that is pretty obvious.  Why not the second
   "unsigned int flags" as well though?

Thanks for catching this, yes that one needs to be fixed up too.

--- ide-scsi.c.~2~	Sun Apr 18 14:37:07 1999
+++ ide-scsi.c	Sun Apr 18 15:23:37 1999
@@ -92,7 +92,7 @@
 typedef struct {
 	ide_drive_t *drive;
 	idescsi_pc_t *pc;			/* Current packet command */
-	unsigned int flags;			/* Status/Action flags */
+	unsigned long flags;			/* Status/Action flags */
 	unsigned long transform;		/* SCSI cmd translation layer */
 	unsigned long log;			/* log flags */
 } idescsi_scsi_t;