* [PATCH] add spi attributes to qla1280
@ 2004-08-10 23:05 Jeremy Higdon
2004-08-11 6:31 ` Jes Sorensen
2004-08-11 7:11 ` Christoph Hellwig
0 siblings, 2 replies; 9+ messages in thread
From: Jeremy Higdon @ 2004-08-10 23:05 UTC (permalink / raw)
To: linux-scsi
This patch adds SPI attributes to the qla1280 driver.
It is similar to a patch submitted back in March by mort, except that it has
been updated to the current API.
Jes, can you confirm that this is okay?
Signed-off-by: Jeremy Higdon <jeremy@sgi.com>
Thanks,
jeremy
diff -ur -X /home/jeremy/dontdiff linux-2.6.8-rc2/drivers/scsi/Kconfig linux-2.6.8-rc2.new/drivers/scsi/Kconfig
--- linux-2.6.8-rc2/drivers/scsi/Kconfig 2004-08-10 15:57:22.000000000 -0700
+++ linux-2.6.8-rc2.new/drivers/scsi/Kconfig 2004-08-05 18:52:10.000000000 -0700
@@ -1258,6 +1258,7 @@
config SCSI_QLOGIC_1280
tristate "Qlogic QLA 1280 SCSI support"
depends on PCI && SCSI
+ select SCSI_SPI_ATTRS
help
Say Y if you have a QLogic ISP1x80/1x160 SCSI host adapter.
diff -ur -X /home/jeremy/dontdiff linux-2.6.8-rc2/drivers/scsi/qla1280.c linux-2.6.8-rc2.new/drivers/scsi/qla1280.c
--- linux-2.6.8-rc2/drivers/scsi/qla1280.c 2004-08-10 15:57:22.000000000 -0700
+++ linux-2.6.8-rc2.new/drivers/scsi/qla1280.c 2004-08-10 00:52:05.000000000 -0700
@@ -353,6 +353,8 @@
#include <scsi/scsi_device.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_tcq.h>
+#include <scsi/scsi_transport.h>
+#include <scsi/scsi_transport_spi.h>
#include "scsi.h"
#else
#include <linux/blk.h>
@@ -1447,6 +1449,7 @@
int status = 0;
struct nvram *nv;
unsigned long flags;
+ int is1x160;
ha = (struct scsi_qla_host *)device->host->hostdata;
nv = &ha->nvram;
@@ -1489,6 +1492,21 @@
status = qla1280_set_target_parameters(ha, bus, target);
qla1280_get_target_parameters(ha, device);
spin_unlock_irqrestore(HOST_LOCK, flags);
+
+ if (ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP12160 ||
+ ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP10160)
+ is1x160 = 1;
+ else
+ is1x160 = 0;
+
+ spi_period(device) = nv->bus[bus].target[target].sync_period;
+ if (is1x160)
+ spi_offset(device) = nv->bus[bus].target[target].flags.
+ flags1x160.sync_offset;
+ else
+ spi_offset(device) = nv->bus[bus].target[target].flags.
+ flags1x80.sync_offset;
+
return status;
}
@@ -4670,6 +4688,8 @@
};
#endif
+static struct scsi_transport_template *qla1280_transport_template = NULL;
+
static int __devinit
qla1280_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
@@ -4758,6 +4778,7 @@
host->max_id = MAX_TARGETS;
host->max_sectors = 1024;
host->unique_id = host->host_no;
+ host->transportt = qla1280_transport_template;
#if LINUX_VERSION_CODE < 0x020545
host->select_queue_depths = qla1280_select_queue_depth;
@@ -4890,6 +4911,12 @@
.remove = __devexit_p(qla1280_remove_one),
};
+static struct spi_function_template qla1280_transport_functions = {
+ .show_period = 1,
+ .show_offset = 1,
+};
+
+
static int __init
qla1280_init(void)
{
@@ -4899,6 +4926,10 @@
return -EINVAL;
}
+ qla1280_transport_template = spi_attach_transport(&qla1280_transport_functions);
+ if (!qla1280_transport_template)
+ return -ENODEV;
+
#ifdef MODULE
/*
* If we are called as a module, the qla1280 pointer may not be null
@@ -4923,6 +4954,7 @@
qla1280_exit(void)
{
pci_unregister_driver(&qla1280_pci_driver);
+ spi_release_transport(qla1280_transport_template);
}
module_init(qla1280_init);
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-10 23:05 [PATCH] add spi attributes to qla1280 Jeremy Higdon
@ 2004-08-11 6:31 ` Jes Sorensen
2004-08-11 7:11 ` Christoph Hellwig
1 sibling, 0 replies; 9+ messages in thread
From: Jes Sorensen @ 2004-08-11 6:31 UTC (permalink / raw)
To: Jeremy Higdon; +Cc: linux-scsi, James.Bottomley
>>>>> "Jeremy" == Jeremy Higdon <jeremy@sgi.com> writes:
Jeremy> This patch adds SPI attributes to the qla1280 driver. It is
Jeremy> similar to a patch submitted back in March by mort, except
Jeremy> that it has been updated to the current API.
Jeremy> Jes, can you confirm that this is okay?
Jeremy,
Patch looks good to me! James, feel free to push this.
Cheers,
Jes
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-10 23:05 [PATCH] add spi attributes to qla1280 Jeremy Higdon
2004-08-11 6:31 ` Jes Sorensen
@ 2004-08-11 7:11 ` Christoph Hellwig
2004-08-11 21:56 ` Jeremy Higdon
1 sibling, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2004-08-11 7:11 UTC (permalink / raw)
To: Jeremy Higdon; +Cc: linux-scsi
> + if (ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP12160 ||
> + ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP10160)
> + is1x160 = 1;
> + else
> + is1x160 = 0;
The driver in James' tree has most of my 1040-support patches merged,
sp please use the apporpinquate IS_ISP* macros like the rest of the drivers
does, and please handle the 1040 aswell in that code.
> +static struct scsi_transport_template *qla1280_transport_template = NULL;
No need to initialize to NULL here.
> static int __devinit
> qla1280_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
> {
> @@ -4758,6 +4778,7 @@
> host->max_id = MAX_TARGETS;
> host->max_sectors = 1024;
> host->unique_id = host->host_no;
> + host->transportt = qla1280_transport_template;
>
> #if LINUX_VERSION_CODE < 0x020545
> host->select_queue_depths = qla1280_select_queue_depth;
> @@ -4890,6 +4911,12 @@
> .remove = __devexit_p(qla1280_remove_one),
> };
>
> +static struct spi_function_template qla1280_transport_functions = {
> + .show_period = 1,
> + .show_offset = 1,
> +};
Any chance you could declare this just below the host template? that keeps
the code better organized.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-11 7:11 ` Christoph Hellwig
@ 2004-08-11 21:56 ` Jeremy Higdon
2004-08-11 22:05 ` Christoph Hellwig
0 siblings, 1 reply; 9+ messages in thread
From: Jeremy Higdon @ 2004-08-11 21:56 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: linux-scsi
On Wed, Aug 11, 2004 at 08:11:18AM +0100, Christoph Hellwig wrote:
> > + if (ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP12160 ||
> > + ha->pdev->device == PCI_DEVICE_ID_QLOGIC_ISP10160)
> > + is1x160 = 1;
> > + else
> > + is1x160 = 0;
>
> The driver in James' tree has most of my 1040-support patches merged,
> sp please use the apporpinquate IS_ISP* macros like the rest of the drivers
> does, and please handle the 1040 aswell in that code.
The above is consistent with the rest of the driver in 2.6.8-rc2.
So are you asking that I submit a patch to the driver in James's
tree instead of to 2.6.8-rc2?
To avoid possibility of error, can you give me the URL for James's
tree? Is it a bk tree or a kernel.org tree?
> > +static struct scsi_transport_template *qla1280_transport_template = NULL;
>
> No need to initialize to NULL here.
I thought about that. What I did was take a patch that Martin Hicks
submitted on March 10th and then make changes based on what I saw that
James had done for the 53c700 driver, which has:
drivers/scsi/53c700.c:STATIC struct scsi_transport_template *NCR_700_transport_template = NULL;
I will remove the initializer if I don't need it.
> > static int __devinit
> > qla1280_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
> > {
> > @@ -4758,6 +4778,7 @@
> > host->max_id = MAX_TARGETS;
> > host->max_sectors = 1024;
> > host->unique_id = host->host_no;
> > + host->transportt = qla1280_transport_template;
> >
> > #if LINUX_VERSION_CODE < 0x020545
> > host->select_queue_depths = qla1280_select_queue_depth;
> > @@ -4890,6 +4911,12 @@
> > .remove = __devexit_p(qla1280_remove_one),
> > };
> >
> > +static struct spi_function_template qla1280_transport_functions = {
> > + .show_period = 1,
> > + .show_offset = 1,
> > +};
>
> Any chance you could declare this just below the host template? that keeps
> the code better organized.
Sure.
jeremy
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-11 21:56 ` Jeremy Higdon
@ 2004-08-11 22:05 ` Christoph Hellwig
2004-08-11 22:16 ` Jeremy Higdon
0 siblings, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2004-08-11 22:05 UTC (permalink / raw)
To: Jeremy Higdon; +Cc: linux-scsi
On Wed, Aug 11, 2004 at 02:56:51PM -0700, Jeremy Higdon wrote:
> The above is consistent with the rest of the driver in 2.6.8-rc2.
> So are you asking that I submit a patch to the driver in James's
> tree instead of to 2.6.8-rc2?
Yes.
> To avoid possibility of error, can you give me the URL for James's
> tree? Is it a bk tree or a kernel.org tree?
bk://linux-scsi.bkbits.net/scsi-misc-2.6, it's also part of the
latest -mm tree.
> > No need to initialize to NULL here.
>
> I thought about that. What I did was take a patch that Martin Hicks
> submitted on March 10th and then make changes based on what I saw that
> James had done for the 53c700 driver, which has:
>
> drivers/scsi/53c700.c:STATIC struct scsi_transport_template *NCR_700_transport_template = NULL;
needs fixing aswell :)
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-11 22:05 ` Christoph Hellwig
@ 2004-08-11 22:16 ` Jeremy Higdon
2004-08-11 22:22 ` Christoph Hellwig
0 siblings, 1 reply; 9+ messages in thread
From: Jeremy Higdon @ 2004-08-11 22:16 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: linux-scsi
On Wed, Aug 11, 2004 at 11:05:56PM +0100, Christoph Hellwig wrote:
> On Wed, Aug 11, 2004 at 02:56:51PM -0700, Jeremy Higdon wrote:
> > The above is consistent with the rest of the driver in 2.6.8-rc2.
> > So are you asking that I submit a patch to the driver in James's
> > tree instead of to 2.6.8-rc2?
>
> Yes.
>
> > To avoid possibility of error, can you give me the URL for James's
> > tree? Is it a bk tree or a kernel.org tree?
>
> bk://linux-scsi.bkbits.net/scsi-misc-2.6, it's also part of the
> latest -mm tree.
So I'll submit a patch to 2.6.8-rc4-mm1 then, if that's okay.
Thanks,
Jeremy
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-11 22:16 ` Jeremy Higdon
@ 2004-08-11 22:22 ` Christoph Hellwig
2004-08-12 5:29 ` Jeremy Higdon
0 siblings, 1 reply; 9+ messages in thread
From: Christoph Hellwig @ 2004-08-11 22:22 UTC (permalink / raw)
To: Jeremy Higdon; +Cc: Christoph Hellwig, linux-scsi
On Wed, Aug 11, 2004 at 03:16:36PM -0700, Jeremy Higdon wrote:
> > bk://linux-scsi.bkbits.net/scsi-misc-2.6, it's also part of the
> > latest -mm tree.
>
>
> So I'll submit a patch to 2.6.8-rc4-mm1 then, if that's okay.
Sounds good. While you're at it can you also fix the compilation
error when the driver is compiled for port I/O instead of normal
mmio (e.g. SGI ViswW), you broke that with your readb_relaxed changes.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-11 22:22 ` Christoph Hellwig
@ 2004-08-12 5:29 ` Jeremy Higdon
2004-08-12 9:08 ` Christoph Hellwig
0 siblings, 1 reply; 9+ messages in thread
From: Jeremy Higdon @ 2004-08-12 5:29 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: linux-scsi
On Wed, Aug 11, 2004 at 11:22:25PM +0100, Christoph Hellwig wrote:
> On Wed, Aug 11, 2004 at 03:16:36PM -0700, Jeremy Higdon wrote:
> >
> > So I'll submit a patch to 2.6.8-rc4-mm1 then, if that's okay.
>
> Sounds good. While you're at it can you also fix the compilation
> error when the driver is compiled for port I/O instead of normal
> mmio (e.g. SGI ViswW), you broke that with your readb_relaxed changes.
Okay. Here is the new rendition, based on 2.6.8-rc4-mm1, with the
fix for port I/O. For sync period and offset, it seems that the
isp1040 is the same as the isp1280.
Thanks,
jeremy
Signed-off-by: Jeremy Higdon <jeremy@sgi.com>
diff -ur -X /home/jeremy/dontdiff linux-2.6.8-rc4/drivers/scsi/Kconfig linux-2.6.8-rc4.new/drivers/scsi/Kconfig
--- linux-2.6.8-rc4/drivers/scsi/Kconfig 2004-08-11 22:22:41.000000000 -0700
+++ linux-2.6.8-rc4.new/drivers/scsi/Kconfig 2004-08-11 17:41:57.000000000 -0700
@@ -1276,6 +1276,7 @@
config SCSI_QLOGIC_1280
tristate "Qlogic QLA 1280 SCSI support"
depends on PCI && SCSI
+ select SCSI_SPI_ATTRS
help
Say Y if you have a QLogic ISP1x80/1x160 SCSI host adapter.
diff -ur -X /home/jeremy/dontdiff linux-2.6.8-rc4/drivers/scsi/qla1280.c linux-2.6.8-rc4.new/drivers/scsi/qla1280.c
--- linux-2.6.8-rc4/drivers/scsi/qla1280.c 2004-08-11 22:22:41.000000000 -0700
+++ linux-2.6.8-rc4.new/drivers/scsi/qla1280.c 2004-08-11 17:47:13.000000000 -0700
@@ -356,6 +356,8 @@
#include <scsi/scsi_device.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_tcq.h>
+#include <scsi/scsi_transport.h>
+#include <scsi/scsi_transport_spi.h>
#include "scsi.h"
#else
#include <linux/blk.h>
@@ -1490,6 +1492,15 @@
status = qla1280_set_target_parameters(ha, bus, target);
qla1280_get_target_parameters(ha, device);
spin_unlock_irqrestore(HOST_LOCK, flags);
+
+ spi_period(device) = nv->bus[bus].target[target].sync_period;
+ if (IS_ISP1x160(ha))
+ spi_offset(device) = nv->bus[bus].target[target].flags.
+ flags1x160.sync_offset;
+ else
+ spi_offset(device) = nv->bus[bus].target[target].flags.
+ flags1x80.sync_offset;
+
return status;
}
@@ -4681,6 +4692,12 @@
};
#endif
+static struct scsi_transport_template *qla1280_transport_template;
+static struct spi_function_template qla1280_transport_functions = {
+ .show_period = 1,
+ .show_offset = 1,
+};
+
static int __devinit
qla1280_probe_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
@@ -4769,6 +4786,7 @@
host->max_id = MAX_TARGETS;
host->max_sectors = 1024;
host->unique_id = host->host_no;
+ host->transportt = qla1280_transport_template;
#if LINUX_VERSION_CODE < 0x020545
host->select_queue_depths = qla1280_select_queue_depth;
@@ -4910,6 +4928,10 @@
return -EINVAL;
}
+ qla1280_transport_template = spi_attach_transport(&qla1280_transport_functions);
+ if (!qla1280_transport_template)
+ return -ENODEV;
+
#ifdef MODULE
/*
* If we are called as a module, the qla1280 pointer may not be null
@@ -4934,6 +4956,7 @@
qla1280_exit(void)
{
pci_unregister_driver(&qla1280_pci_driver);
+ spi_release_transport(qla1280_transport_template);
}
module_init(qla1280_init);
diff -ur -X /home/jeremy/dontdiff linux-2.6.8-rc4/drivers/scsi/qla1280.h linux-2.6.8-rc4.new/drivers/scsi/qla1280.h
--- linux-2.6.8-rc4/drivers/scsi/qla1280.h 2004-08-11 22:22:41.000000000 -0700
+++ linux-2.6.8-rc4.new/drivers/scsi/qla1280.h 2004-08-11 17:39:31.000000000 -0700
@@ -62,6 +62,7 @@
#define WRT_REG_WORD(addr, data) writew(data, addr)
#else /* MEMORY_MAPPED_IO */
#define RD_REG_WORD(addr) inw((unsigned long)addr)
+#define RD_REG_WORD_dmasync(addr) inw((unsigned long)addr)
#define WRT_REG_WORD(addr, data) outw(data, (unsigned long)addr)
#endif /* MEMORY_MAPPED_IO */
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] add spi attributes to qla1280
2004-08-12 5:29 ` Jeremy Higdon
@ 2004-08-12 9:08 ` Christoph Hellwig
0 siblings, 0 replies; 9+ messages in thread
From: Christoph Hellwig @ 2004-08-12 9:08 UTC (permalink / raw)
To: Jeremy Higdon; +Cc: linux-scsi
On Wed, Aug 11, 2004 at 10:29:24PM -0700, Jeremy Higdon wrote:
> On Wed, Aug 11, 2004 at 11:22:25PM +0100, Christoph Hellwig wrote:
> > On Wed, Aug 11, 2004 at 03:16:36PM -0700, Jeremy Higdon wrote:
> > >
> > > So I'll submit a patch to 2.6.8-rc4-mm1 then, if that's okay.
> >
> > Sounds good. While you're at it can you also fix the compilation
> > error when the driver is compiled for port I/O instead of normal
> > mmio (e.g. SGI ViswW), you broke that with your readb_relaxed changes.
>
> Okay. Here is the new rendition, based on 2.6.8-rc4-mm1, with the
> fix for port I/O. For sync period and offset, it seems that the
> isp1040 is the same as the isp1280.
Looks good to me.
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2004-08-12 9:08 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-08-10 23:05 [PATCH] add spi attributes to qla1280 Jeremy Higdon
2004-08-11 6:31 ` Jes Sorensen
2004-08-11 7:11 ` Christoph Hellwig
2004-08-11 21:56 ` Jeremy Higdon
2004-08-11 22:05 ` Christoph Hellwig
2004-08-11 22:16 ` Jeremy Higdon
2004-08-11 22:22 ` Christoph Hellwig
2004-08-12 5:29 ` Jeremy Higdon
2004-08-12 9:08 ` Christoph Hellwig
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.