[PATCH] aacraid: regression fix

[PATCH] aacraid: regression fix

[Mark Haverkamp]

The fixes for sparse warnings mixed in with the fixups for
the raw_srb handler resulted in a bug that showed up in the 32 bit
environments when trying to issue calls directly to the physical devices
that are part of the arrays (ioctl scsi passthrough).

Received from Mark Salyzyn at adaptec.
It applies to the scsi-misc-2.6 git tree.

Signed-off-by: Mark Haverkamp <markh@osdl.org>


diff --git a/drivers/scsi/aacraid/commctrl.c b/drivers/scsi/aacraid/commctrl.c
--- a/drivers/scsi/aacraid/commctrl.c
+++ b/drivers/scsi/aacraid/commctrl.c
@@ -104,7 +104,7 @@ static int ioctl_send_fib(struct aac_dev
 		 * Since we didn't really send a fib, zero out the state to allow 
 		 * cleanup code not to assert.
 		 */
-		kfib->header.XferState = 0;
+		kfib->header.XferState = cpu_to_le32(0);
 	} else {
 		retval = fib_send(le16_to_cpu(kfib->header.Command), fibptr,
 				le16_to_cpu(kfib->header.Size) , FsaNormal,
@@ -451,7 +451,7 @@ static int aac_send_raw_srb(struct aac_d
 	 *	Allocate and initialize a Fib then setup a BlockWrite command
 	 */
 	if (!(srbfib = fib_alloc(dev))) {
-		return -1;
+		return -ENOMEM;
 	}
 	fib_init(srbfib);
 
@@ -508,7 +508,7 @@ static int aac_send_raw_srb(struct aac_d
 	default:
 		data_dir = DMA_NONE;
 	}
-	if (le32_to_cpu(srbcmd->sg.count) > (sizeof(sg_list)/sizeof(sg_list[0]))) {
+	if (user_srbcmd->sg.count > (sizeof(sg_list)/sizeof(sg_list[0]))) {
 		dprintk((KERN_DEBUG"aacraid: too many sg entries %d\n",
 		  le32_to_cpu(srbcmd->sg.count)));
 		rcode = -EINVAL;
@@ -592,7 +592,7 @@ static int aac_send_raw_srb(struct aac_d
 		struct sgmap* psg = &srbcmd->sg;
 		byte_count = 0;
 
-		actual_fibsize = sizeof (struct aac_srb) + (((le32_to_cpu(srbcmd->sg.count) & 0xff) - 1) * sizeof (struct sgentry));
+		actual_fibsize = sizeof (struct aac_srb) + (((user_srbcmd->sg.count & 0xff) - 1) * sizeof (struct sgentry));
 		if(actual_fibsize != fibsize){ // User made a mistake - should not continue
 			dprintk((KERN_DEBUG"aacraid: Bad Size specified in Raw SRB command\n"));
 			rcode = -EINVAL;
@@ -639,7 +639,7 @@ static int aac_send_raw_srb(struct aac_d
 
 	if (status != 0){
 		dprintk((KERN_DEBUG"aacraid: Could not send raw srb fib to hba\n")); 
-		rcode = -1;
+		rcode = -ENXIO;
 		goto cleanup;
 	}
 

-- 
Mark Haverkamp <markh@osdl.org>

Re: [PATCH] aacraid: regression fix

[Christoph Hellwig]

On Wed, Jun 15, 2005 at 09:36:46AM -0700, Mark Haverkamp wrote:
> -		kfib->header.XferState = 0;
> +		kfib->header.XferState = cpu_to_le32(0);

you never need to byte-swap the 0 constant.

RE: [PATCH] aacraid: regression fix

[Salyzyn, Mark]

But then we are requested to byteswap the 1's constant ... ;-}

Sincerely -- Mark Salyzyn

-----Original Message-----
From: Christoph Hellwig [mailto:hch@infradead.org] 
Sent: Friday, June 17, 2005 7:21 AM
To: Mark Haverkamp
Cc: James Bottomley; Salyzyn, Mark; linux-scsi
Subject: Re: [PATCH] aacraid: regression fix

On Wed, Jun 15, 2005 at 09:36:46AM -0700, Mark Haverkamp wrote:
> -		kfib->header.XferState = 0;
> +		kfib->header.XferState = cpu_to_le32(0);

you never need to byte-swap the 0 constant.