[PATCH 2.6.24-rc8-git6] initio module hangs on loading fix set

[PATCH 2.6.24-rc8-git6] initio module hangs on loading fix set

[Stuart Swales]

I've verified (on my Initio 9100 with a DAT drive) that the 
2.6.24-rc8-git6 initio module still hangs on loading.

These fixes (other than the printk) are needed to get the module to load 
ok (and work correctly) with my adapter & tape drive.

a) printk cosmetic fix

b) cblk->sglen needs setting for later DMA I/O routines to use

c) host->bios_addr needs setting for debug output correctness

d) semaph & semaph_lock initialisation had got lost since 2.6.22

e) since 2.6.22 the bios data address was truncated to 16 bits (needs 20 
when shifted left)

Stuart Swales

--- linux-2.6.24-rc8-git6/drivers/scsi/initio.c	2008-01-16 
04:22:48.000000000 +0000
+++ linux-2.6.24-rc8-git6-sks/drivers/scsi/initio.c	2008-01-23 
18:52:04.000000000 +0000
@@ -823,7 +823,7 @@ static void initio_append_busy_scb(struc
  {

  #if DEBUG_QUEUE
-	printk("append busy SCB %o; ", scbp);
+	printk("append busy SCB %p; ", scbp); /* SKS: not %o */
  #endif
  	if (scbp->tagmsg)
  		host->act_tags[scbp->target]++;
@@ -2609,6 +2609,7 @@ static void initio_build_scb(struct init
  		cblk->bufptr = cpu_to_le32((u32)dma_addr);
  		cmnd->SCp.dma_handle = dma_addr;

+		cblk->sglen = nseg; /* SKS: needs setting for DMA routine */

  		cblk->flags |= SCF_SG;	/* Turn on SG list flag       */
  		total_len = 0;
@@ -2869,6 +2870,7 @@ static int initio_probe_one(struct pci_d
  	host = (struct initio_host *)shost->hostdata;
  	memset(host, 0, sizeof(struct initio_host));
  	host->addr = pci_resource_start(pdev, 0);
+	host->bios_addr = bios_seg; /* SKS: required for debug output */

  	if (!request_region(host->addr, 256, "i91u")) {
  		printk(KERN_WARNING "initio: I/O port range 0x%x is busy.\n", 
host->addr);
@@ -2895,6 +2897,8 @@ static int initio_probe_one(struct pci_d

  	host->pci_dev = pdev;

+	host->semaph = 1; /* SKS */
+	spin_lock_init(&host->semaph_lock);
  	host->num_scbs = num_scb;
  	host->scb = scb;
  	host->next_pending = scb;
@@ -2911,7 +2915,7 @@ static int initio_probe_one(struct pci_d
  	host->last_avail = prev;
  	spin_lock_init(&host->avail_lock);

-	initio_init(host, phys_to_virt(bios_seg << 4));
+	initio_init(host, phys_to_virt(((u32) bios_seg << 4))); /* SKS: u16 
would lose bits when shifted left */

  	host->jsstatus0 = 0;

Re: [PATCH 2.6.24-rc8-git6] initio module hangs on loading fix set

[James Bottomley]

On Wed, 2008-01-23 at 20:00 +0000, Stuart Swales wrote:
> I've verified (on my Initio 9100 with a DAT drive) that the 
> 2.6.24-rc8-git6 initio module still hangs on loading.
> 
> These fixes (other than the printk) are needed to get the module to load 
> ok (and work correctly) with my adapter & tape drive.
> 
> a) printk cosmetic fix
> 
> b) cblk->sglen needs setting for later DMA I/O routines to use
> 
> c) host->bios_addr needs setting for debug output correctness
> 
> d) semaph & semaph_lock initialisation had got lost since 2.6.22
> 
> e) since 2.6.22 the bios data address was truncated to 16 bits (needs 20 
> when shifted left)

That's brilliant, thanks!  (Linus has been complaining about the broken
state of initio for a while, so this saves my bacon).

I made your patch actually apply (your mailer broke it in places) and
stripped out the comments, which won't make sense to anyone who doesn't
have the context of the actual patch to look at.

Thanks,

James