[GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[James Bottomley]

The patch is here

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

However, as there's only a single patch in there, I attach it in case
you'd prefer simply to apply it as a patch rather than merge a single
patch tree.

James

---

commit 15d100224c123f0f993ef88e95fd5d46bb0bd085
Author: James Smart <James.Smart@Emulex.Com>
Date:   Thu Aug 31 12:27:57 2006 -0400

    [SCSI] lpfc: don't free mempool if mailbox is busy
    
    Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

diff --git a/drivers/scsi/lpfc/lpfc_init.c b/drivers/scsi/lpfc/lpfc_init.c
index f6948ff..c31fe41 100644
--- a/drivers/scsi/lpfc/lpfc_init.c
+++ b/drivers/scsi/lpfc/lpfc_init.c
@@ -387,7 +387,8 @@ lpfc_config_port_post(struct lpfc_hba * 
 
 	lpfc_init_link(phba, pmb, phba->cfg_topology, phba->cfg_link_speed);
 	pmb->mbox_cmpl = lpfc_sli_def_mbox_cmpl;
-	if (lpfc_sli_issue_mbox(phba, pmb, MBX_NOWAIT) != MBX_SUCCESS) {
+	rc = lpfc_sli_issue_mbox(phba, pmb, MBX_NOWAIT);
+	if (rc != MBX_SUCCESS) {
 		lpfc_printf_log(phba,
 				KERN_ERR,
 				LOG_INIT,
@@ -404,7 +405,8 @@ lpfc_config_port_post(struct lpfc_hba * 
 		readl(phba->HAregaddr); /* flush */
 
 		phba->hba_state = LPFC_HBA_ERROR;
-		mempool_free(pmb, phba->mbox_mem_pool);
+		if (rc != MBX_BUSY)
+			mempool_free(pmb, phba->mbox_mem_pool);
 		return -EIO;
 	}
 	/* MBOX buffer will be freed in mbox compl */

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[Andrew Morton]

On Thu, 14 Sep 2006 16:12:58 -0500
James Bottomley <James.Bottomley@SteelEye.com> wrote:

> The patch is here
> 
> master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git
> 
> However, as there's only a single patch in there, I attach it in case
> you'd prefer simply to apply it as a patch rather than merge a single
> patch tree.

<digs around>

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc6/2.6.18-rc6-mm2/broken-out/fix-panic-when-reinserting-adaptec-pcmcia-scsi-card.patch
might be handy too.  Your ack is my command.

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[James Bottomley]

On Thu, 2006-09-14 at 14:20 -0700, Andrew Morton wrote:
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc6/2.6.18-rc6-mm2/broken-out/fix-panic-when-reinserting-adaptec-pcmcia-scsi-card.patch
> might be handy too.  Your ack is my command.

Well ... no, not really, on the grounds that the patch is wrong.

The correct fix is to eliminate the aha152x_host array by converting the
driver to the correct driver model ... I just haven't had time to look
at doing that yet.

James

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[Andrew Morton]

On Thu, 14 Sep 2006 16:37:39 -0500
James Bottomley <James.Bottomley@SteelEye.com> wrote:

> On Thu, 2006-09-14 at 14:20 -0700, Andrew Morton wrote:
> > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.18-rc6/2.6.18-rc6-mm2/broken-out/fix-panic-when-reinserting-adaptec-pcmcia-scsi-card.patch
> > might be handy too.  Your ack is my command.
> 
> Well ... no, not really, on the grounds that the patch is wrong.

So is oopsing ;)

> The correct fix is to eliminate the aha152x_host array by converting the
> driver to the correct driver model ... I just haven't had time to look
> at doing that yet.

<lachrymose>If we'd been told that on July 8 and/or August 14, this might
be fixed by now</>

Perhaps Alex might like to take a look into doing this (please)?

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[James Bottomley]

On Thu, 2006-09-14 at 14:47 -0700, Andrew Morton wrote:
> <lachrymose>If we'd been told that on July 8 and/or August 14, this might
> be fixed by now</>

OK, so mea culpa, I'll make amends

> Perhaps Alex might like to take a look into doing this (please)?

Just test out this patch, please ...

James

diff --git a/drivers/scsi/aha152x.c b/drivers/scsi/aha152x.c
index f974869..7fd56b8 100644
--- a/drivers/scsi/aha152x.c
+++ b/drivers/scsi/aha152x.c
@@ -253,6 +253,7 @@ #include <linux/kernel.h>
 #include <linux/isapnp.h>
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
+#include <linux/list.h>
 #include <asm/semaphore.h>
 #include <scsi/scsicam.h>
 
@@ -262,6 +263,8 @@ #include <scsi/scsi_host.h>
 #include <scsi/scsi_transport_spi.h>
 #include "aha152x.h"
 
+static LIST_HEAD(aha152x_host_list);
+
 
 /* DEFINES */
 
@@ -423,8 +426,6 @@ #endif /* ISAPNP */
 
 #endif /* !PCMCIA */
 
-static int registered_count=0;
-static struct Scsi_Host *aha152x_host[2];
 static struct scsi_host_template aha152x_driver_template;
 
 /*
@@ -541,6 +542,7 @@ #endif
 #ifdef __ISAPNP__
 	struct pnp_dev *pnpdev;
 #endif
+	struct list_head host_list;
 };
 
 
@@ -755,20 +757,9 @@ static inline Scsi_Cmnd *remove_SC(Scsi_
 	return ptr;
 }
 
-static inline struct Scsi_Host *lookup_irq(int irqno)
-{
-	int i;
-
-	for(i=0; i<ARRAY_SIZE(aha152x_host); i++)
-		if(aha152x_host[i] && aha152x_host[i]->irq==irqno)
-			return aha152x_host[i];
-
-	return NULL;
-}
-
 static irqreturn_t swintr(int irqno, void *dev_id, struct pt_regs *regs)
 {
-	struct Scsi_Host *shpnt = lookup_irq(irqno);
+	struct Scsi_Host *shpnt = (struct Scsi_Host *)dev_id;
 
 	if (!shpnt) {
         	printk(KERN_ERR "aha152x: catched software interrupt %d for unknown controller.\n", irqno);
@@ -791,10 +782,11 @@ struct Scsi_Host *aha152x_probe_one(stru
 		return NULL;
 	}
 
-	/* need to have host registered before triggering any interrupt */
-	aha152x_host[registered_count] = shpnt;
-
 	memset(HOSTDATA(shpnt), 0, sizeof *HOSTDATA(shpnt));
+	INIT_LIST_HEAD(&HOSTDATA(shpnt)->host_list);
+
+	/* need to have host registered before triggering any interrupt */
+	list_add_tail(&HOSTDATA(shpnt)->host_list, &aha152x_host_list);
 
 	shpnt->io_port   = setup->io_port;
 	shpnt->n_io_port = IO_RANGE;
@@ -907,12 +899,10 @@ #endif
 
 	scsi_scan_host(shpnt);
 
-	registered_count++;
-
 	return shpnt;
 
 out_host_put:
-	aha152x_host[registered_count]=NULL;
+	list_del(&HOSTDATA(shpnt)->host_list);
 	scsi_host_put(shpnt);
 
 	return NULL;
@@ -937,6 +927,7 @@ #ifdef __ISAPNP__
 #endif
 
 	scsi_remove_host(shpnt);
+	list_del(&HOSTDATA(shpnt)->host_list);
 	scsi_host_put(shpnt);
 }
 
@@ -1459,9 +1450,12 @@ static struct work_struct aha152x_tq;
  */
 static void run(void)
 {
-	int i;
-	for (i = 0; i<ARRAY_SIZE(aha152x_host); i++) {
-		is_complete(aha152x_host[i]);
+	struct aha152x_hostdata *hd;
+
+	list_for_each_entry(hd, &aha152x_host_list, host_list) {
+		struct Scsi_Host *shost = container_of((void *)hd, struct Scsi_Host, hostdata);
+		
+		is_complete(shost);
 	}
 }
 
@@ -1471,7 +1465,7 @@ static void run(void)
  */
 static irqreturn_t intr(int irqno, void *dev_id, struct pt_regs *regs)
 {
-	struct Scsi_Host *shpnt = lookup_irq(irqno);
+	struct Scsi_Host *shpnt = (struct Scsi_Host *)dev_id;
 	unsigned long flags;
 	unsigned char rev, dmacntrl0;
 
@@ -3953,16 +3947,17 @@ #if defined(__ISAPNP__)
 #endif
 	}
 
-	return registered_count>0;
+	return 1;
 }
 
 static void __exit aha152x_exit(void)
 {
-	int i;
+	struct aha152x_hostdata *hd;
 
-	for(i=0; i<ARRAY_SIZE(setup); i++) {
-		aha152x_release(aha152x_host[i]);
-		aha152x_host[i]=NULL;
+	list_for_each_entry(hd, &aha152x_host_list, host_list) {
+		struct Scsi_Host *shost = container_of((void *)hd, struct Scsi_Host, hostdata);
+		
+		aha152x_release(shost);
 	}
 }
 

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[Douglas Gilbert]

James,
Is the door already closed for patches to lk 2.6.19 ?

Doug Gilbert

Re: [GIT PATCH] (hopefully) final SCSI fixes for 2.6.19

[James Bottomley]

On Thu, 2006-09-14 at 17:51 -0400, Douglas Gilbert wrote:
> Is the door already closed for patches to lk 2.6.19 ?

It ain't over until the fat penguin sings.

But I know of no major outstanding issues which would justify an
emergency fix at this stage.

James