Skip to contentSkip to conversation

ide: warm-plug support for IDE devices and other goodies

PATCH In review Bartlomiej Zolnierkiewicz 7 participants 39 messages Feb 8, 2008 – Mar 29, 2008

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

cover ide: warm-plug support for IDE devices and other goodies

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

- IDE warm-plug support (though it is still experimental it should work fine, unlike the older method), to warm-plug devices on a port 'idex':

  # echo -n "1" > /sys/class/ide_port/idex/delete_devices

unplug old device(s) and plug new device(s)

  # echo -n "1" > /sys/class/ide_port/idex/scan

done

- ability to add new interfaces for ide-generic host driver through /sys/class/ide_generic/add, i.e.

echo -n "0x168:0x36e:10" > /sys/class/ide_generic/add

- because of the above changes some obsoleted/broken code can be removed (thus the total effect of the patch series is -100 LOC)

diffstat:

 Documentation/ide.txt                 |   31 --
 Documentation/ide/warm-plug-howto.txt |   13 +
 block/compat_ioctl.c                  |    1 
 drivers/ide/Kconfig                   |   25 --
 drivers/ide/Makefile                  |    2 
 drivers/ide/arm/bast-ide.c            |    4 
 drivers/ide/arm/palm_bk3710.c         |    2 
 drivers/ide/arm/rapide.c              |    2 
 drivers/ide/ide-acpi.c                |    2 
 drivers/ide/ide-generic.c             |   85 +++++++
 drivers/ide/ide-pnp.c                 |    2 
 drivers/ide/ide-probe.c               |  108 ++++++++-
 drivers/ide/ide-proc.c                |    3 
 drivers/ide/ide.c                     |  386 ++++++++--------------------------
 drivers/ide/legacy/ide-cs.c           |    6 
 drivers/ide/legacy/ide_platform.c     |    2 
 drivers/ide/mips/au1xxx-ide.c         |    5 
 drivers/ide/pci/cmd640.c              |    2 
 drivers/ide/pci/delkin_cb.c           |    6 
 drivers/ide/pci/scc_pata.c            |    2 
 drivers/ide/ppc/pmac.c                |    4 
 drivers/ide/setup-pci.c               |   11 
 drivers/macintosh/mediabay.c          |   17 -
 include/asm-powerpc/mediabay.h        |    4 
 include/linux/hdreg.h                 |    4 
 include/linux/ide.h                   |   51 ----
 26 files changed, 339 insertions(+), 441 deletions(-)
#1 lore raw

PATCH 1/18 ide-generic: set hwif->chipset

This hwif->chipset fixup is already present in ide_device_add_all() but for warm-plug support we also need to reserve not currently present interfaces.

---
 drivers/ide/ide-generic.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/ide/ide-generic.c +3 -1
Index: b/drivers/ide/ide-generic.c
===================================================================
@@ -23,7 +23,9 @@ static int __init ide_generic_init(void)
 	for (i = 0; i < MAX_HWIFS; i++) {
 		ide_hwif_t *hwif = &ide_hwifs[i];
 
-		if (hwif->io_ports[IDE_DATA_OFFSET] && !hwif->present)
+		if (hwif->io_ports[IDE_DATA_OFFSET] &&
+		    (hwif->chipset == ide_unknown ||
+		     hwif->chipset == ide_forced))
 			idx[i] = i;
 		else
 			idx[i] = 0xff;
#2 lore raw

PATCH 2/18 ide: fix ide_find_port()

In reply to Bartlomiej Zolnierkiewicz

* Instead of checking for '->io_ports[IDE_DATA_OFFSET] == 0' check for '->chipset == ide_unknown' when looking for an empty ide_hwifs[] slot.

* Do ide-pnp initialization after ide-generic when IDE is built-in (ide-pnp is the only user of ide_find_port() which needs such fixup).

---
 drivers/ide/Makefile |    2 +-
 drivers/ide/ide.c    |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/ide/Makefile +1 -1
Index: b/drivers/ide/Makefile
===================================================================
@@ -36,9 +36,9 @@ ifeq ($(CONFIG_BLK_DEV_CMD640), y)
 endif
 
 obj-$(CONFIG_BLK_DEV_IDE)		+= cris/ ppc/
-obj-$(CONFIG_BLK_DEV_IDEPNP)		+= ide-pnp.o
 obj-$(CONFIG_IDE_H8300)			+= h8300/
 obj-$(CONFIG_IDE_GENERIC)		+= ide-generic.o
+obj-$(CONFIG_BLK_DEV_IDEPNP)		+= ide-pnp.o
 
 ide-cd_mod-y += ide-cd.o ide-cd_ioctl.o ide-cd_verbose.o
 
drivers/ide/ide.c +1 -1
Index: b/drivers/ide/ide.c
===================================================================
@@ -259,7 +259,7 @@ ide_hwif_t * ide_find_port(unsigned long
 
 	for (i = 0; i < MAX_HWIFS; i++) {
 		hwif = &ide_hwifs[i];
-		if (hwif->io_ports[IDE_DATA_OFFSET] == 0)
+		if (hwif->chipset == ide_unknown)
 			goto found;
 	}
 
Acked-by on v1 2/18
#3 lore raw

Re: [PATCH 02/18] ide: fix ide_find_port()

Quoted 7 lines · Bartlomiej Zolnierkiewicz

* Instead of checking for '->io_ports[IDE_DATA_OFFSET] == 0' check for '->chipset == ide_unknown' when looking for an empty ide_hwifs[] slot.

* Do ide-pnp initialization after ide-generic when IDE is built-in (ide-pnp is the only user of ide_find_port() which needs such fixup).

MBR, Sergei

#4 lore raw

PATCH 3/18 ide: use ide_find_port() instead of ide_deprecated_find_port()

In reply to Bartlomiej Zolnierkiewicz

* Use ide_find_port() instead of ide_deprecated_find_port() in bast-ide/ palm_bk3710/ide-cs/delkin_cb host drivers and in ide_register_hw().

* Remove no longer needed ide_deprecated_find_port().

---
 drivers/ide/arm/bast-ide.c    |    2 +-
 drivers/ide/arm/palm_bk3710.c |    2 +-
 drivers/ide/ide.c             |   27 +--------------------------
 drivers/ide/legacy/ide-cs.c   |    2 +-
 drivers/ide/pci/delkin_cb.c   |    2 +-
 include/linux/ide.h           |    1 -
 6 files changed, 5 insertions(+), 31 deletions(-)
drivers/ide/arm/bast-ide.c +1 -1
Index: b/drivers/ide/arm/bast-ide.c
===================================================================
@@ -46,7 +46,7 @@ bastide_register(unsigned int base, unsi
 	hw.io_ports[IDE_CONTROL_OFFSET] = aux + (6 * 0x20);
 	hw.irq = irq;
 
-	hwif = ide_deprecated_find_port(hw.io_ports[IDE_DATA_OFFSET]);
+	hwif = ide_find_port(hw.io_ports[IDE_DATA_OFFSET]);
 	if (hwif == NULL)
 		goto out;
 
drivers/ide/arm/palm_bk3710.c +1 -1
Index: b/drivers/ide/arm/palm_bk3710.c
===================================================================
@@ -378,7 +378,7 @@ static int __devinit palm_bk3710_probe(s
 	hw.irq = irq->start;
 	hw.chipset = ide_palm3710;
 
-	hwif = ide_deprecated_find_port(hw.io_ports[IDE_DATA_OFFSET]);
+	hwif = ide_find_port(hw.io_ports[IDE_DATA_OFFSET]);
 	if (hwif == NULL)
 		goto out;
 
drivers/ide/ide.c +1 -26
Index: b/drivers/ide/ide.c
===================================================================
@@ -627,31 +627,6 @@ void ide_init_port_hw(ide_hwif_t *hwif, 
 }
 EXPORT_SYMBOL_GPL(ide_init_port_hw);
 
-ide_hwif_t *ide_deprecated_find_port(unsigned long base)
-{
-	ide_hwif_t *hwif;
-	int i;
-
-	for (i = 0; i < MAX_HWIFS; i++) {
-		hwif = &ide_hwifs[i];
-		if (hwif->io_ports[IDE_DATA_OFFSET] == base)
-			goto found;
-	}
-
-	for (i = 0; i < MAX_HWIFS; i++) {
-		hwif = &ide_hwifs[i];
-		if (hwif->hold)
-			continue;
-		if (!hwif->present && hwif->mate == NULL)
-			goto found;
-	}
-
-	hwif = NULL;
-found:
-	return hwif;
-}
-EXPORT_SYMBOL_GPL(ide_deprecated_find_port);
-
 /**
  *	ide_register_hw		-	register IDE interface
  *	@hw: hardware registers
@@ -671,7 +646,7 @@ int ide_register_hw(hw_regs_t *hw, void 
 	u8 idx[4] = { 0xff, 0xff, 0xff, 0xff };
 
 	do {
-		hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
+		hwif = ide_find_port(hw->io_ports[IDE_DATA_OFFSET]);
 		index = hwif->index;
 		if (hwif)
 			goto found;
drivers/ide/legacy/ide-cs.c +1 -1
Index: b/drivers/ide/legacy/ide-cs.c
===================================================================
@@ -156,7 +156,7 @@ static int idecs_register(unsigned long 
     hw.chipset = ide_pci;
     hw.dev = &handle->dev;
 
-    hwif = ide_deprecated_find_port(hw.io_ports[IDE_DATA_OFFSET]);
+    hwif = ide_find_port(hw.io_ports[IDE_DATA_OFFSET]);
     if (hwif == NULL)
 	return -1;
 
drivers/ide/pci/delkin_cb.c +1 -1
Index: b/drivers/ide/pci/delkin_cb.c
===================================================================
@@ -78,7 +78,7 @@ delkin_cb_probe (struct pci_dev *dev, co
 	hw.irq = dev->irq;
 	hw.chipset = ide_pci;		/* this enables IRQ sharing */
 
-	hwif = ide_deprecated_find_port(hw.io_ports[IDE_DATA_OFFSET]);
+	hwif = ide_find_port(hw.io_ports[IDE_DATA_OFFSET]);
 	if (hwif == NULL)
 		goto out_disable;
 
include/linux/ide.h +0 -1
Index: b/include/linux/ide.h
===================================================================
@@ -186,7 +186,6 @@ typedef struct hw_regs_s {
 } hw_regs_t;
 
 struct hwif_s * ide_find_port(unsigned long);
-struct hwif_s *ide_deprecated_find_port(unsigned long);
 void ide_init_port_data(struct hwif_s *, unsigned int);
 void ide_init_port_hw(struct hwif_s *, hw_regs_t *);
 
#5 lore raw

PATCH 4/18 ide-acpi: add missing drive->acpidata zeroing

In reply to Bartlomiej Zolnierkiewicz

There should be no functionality changes caused by this patch.

---
 drivers/ide/ide-acpi.c |    2 ++
 1 file changed, 2 insertions(+)
drivers/ide/ide-acpi.c +2 -0
Index: b/drivers/ide/ide-acpi.c
===================================================================
@@ -710,6 +710,8 @@ void ide_acpi_port_init_devices(ide_hwif
 	for (i = 0; i < MAX_DRIVES; i++) {
 		drive = &hwif->drives[i];
 
+		memset(drive->acpidata, 0, sizeof(*drive->acpidata));
+
 		if (!drive->present)
 			continue;
 
#6 lore raw

PATCH 5/18 ide: factor out cable detection from ide_init_port()

In reply to Bartlomiej Zolnierkiewicz

* Factor out cable detection from ide_init_port() to ide_port_cable_detect().

* Move ide_port_cable_detect() call to ide_device_add_all().

---
 drivers/ide/ide-probe.c |    4 ++++
 1 file changed, 4 insertions(+)
drivers/ide/ide-probe.c +4 -0
Index: b/drivers/ide/ide-probe.c
===================================================================
@@ -1364,7 +1364,10 @@ static void ide_init_port(ide_hwif_t *hw
 	/* call chipset specific routine for each enabled port */
 	if (d->init_hwif)
 		d->init_hwif(hwif);
+}
 
+static void ide_port_cable_detect(ide_hwif_t *hwif)
+{
 	if (hwif->cable_detect && (hwif->ultra_mask & 0x78)) {
 		if (hwif->cbl != ATA_CBL_PATA40_SHORT)
 			hwif->cbl = hwif->cable_detect(hwif);
@@ -1392,6 +1395,7 @@ int ide_device_add_all(u8 *idx, const st
 		mate = (i & 1) ? NULL : hwif;
 
 		ide_init_port(hwif, i & 1, d);
+		ide_port_cable_detect(hwif);
 		ide_port_init_devices(hwif);
 	}
 
#8 lore raw

PATCH 6/18 ide: factor out code unregistering devices from ide_unregister()

In reply to Bartlomiej Zolnierkiewicz

Factor out code unregistering devices from ide_unregister() to ide_port_unregister_devices().

---
 drivers/ide/ide.c |   30 +++++++++++++++++++-----------
 1 file changed, 19 insertions(+), 11 deletions(-)
drivers/ide/ide.c +19 -11
Index: b/drivers/ide/ide.c
===================================================================
@@ -494,6 +494,23 @@ void ide_remove_port_from_hwgroup(ide_hw
 	spin_unlock_irq(&ide_lock);
 }
 
+/* Called with ide_lock held. */
+static void ide_port_unregister_devices(ide_hwif_t *hwif)
+{
+	int i;
+
+	for (i = 0; i < MAX_DRIVES; i++) {
+		ide_drive_t *drive = &hwif->drives[i];
+
+		if (drive->present) {
+			spin_unlock_irq(&ide_lock);
+			device_unregister(&drive->gendev);
+			wait_for_completion(&drive->gendev_rel_comp);
+			spin_lock_irq(&ide_lock);
+		}
+	}
+}
+
 /**
  *	ide_unregister		-	free an IDE interface
  *	@index: index of interface (will change soon to a pointer)
@@ -520,11 +537,10 @@ void ide_remove_port_from_hwgroup(ide_hw
 
 void ide_unregister(unsigned int index, int init_default, int restore)
 {
-	ide_drive_t *drive;
 	ide_hwif_t *hwif, *g;
 	static ide_hwif_t tmp_hwif; /* protected by ide_cfg_mtx */
 	ide_hwgroup_t *hwgroup;
-	int irq_count = 0, unit;
+	int irq_count = 0;
 
 	BUG_ON(index >= MAX_HWIFS);
 
@@ -535,15 +551,7 @@ void ide_unregister(unsigned int index, 
 	hwif = &ide_hwifs[index];
 	if (!hwif->present)
 		goto abort;
-	for (unit = 0; unit < MAX_DRIVES; ++unit) {
-		drive = &hwif->drives[unit];
-		if (!drive->present)
-			continue;
-		spin_unlock_irq(&ide_lock);
-		device_unregister(&drive->gendev);
-		wait_for_completion(&drive->gendev_rel_comp);
-		spin_lock_irq(&ide_lock);
-	}
+	ide_port_unregister_devices(hwif);
 	hwif->present = 0;
 
 	spin_unlock_irq(&ide_lock);
#10 lore raw

PATCH 7/18 ide: factor out devices init from ide_init_port_data()

In reply to Bartlomiej Zolnierkiewicz

* Factor out devices init from ide_init_port_data() to ide_port_init_devices_data().

While at it:

* Add explicit clearing of IDE device structure.

There should be no functionality changes caused by this patch.

---
 drivers/ide/ide.c |   23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)
drivers/ide/ide.c +17 -6
Index: b/drivers/ide/ide.c
===================================================================
@@ -109,13 +109,13 @@ ide_hwif_t ide_hwifs[MAX_HWIFS];	/* mast
 
 EXPORT_SYMBOL(ide_hwifs);
 
+static void ide_port_init_devices_data(ide_hwif_t *);
+
 /*
  * Do not even *think* about calling this!
  */
 void ide_init_port_data(ide_hwif_t *hwif, unsigned int index)
 {
-	unsigned int unit;
-
 	/* bulk initialize hwif & drive info with zeros */
 	memset(hwif, 0, sizeof(ide_hwif_t));
 
@@ -134,8 +134,20 @@ void ide_init_port_data(ide_hwif_t *hwif
 
 	default_hwif_iops(hwif);
 	default_hwif_transport(hwif);
+
+	ide_port_init_devices_data(hwif);
+}
+EXPORT_SYMBOL_GPL(ide_init_port_data);
+
+static void ide_port_init_devices_data(ide_hwif_t *hwif)
+{
+	int unit;
+
 	for (unit = 0; unit < MAX_DRIVES; ++unit) {
 		ide_drive_t *drive = &hwif->drives[unit];
+		u8 j = (hwif->index * MAX_DRIVES) + unit;
+
+		memset(drive, 0, sizeof(*drive));
 
 		drive->media			= ide_disk;
 		drive->select.all		= (unit<<4)|0xa0;
@@ -147,15 +159,14 @@ void ide_init_port_data(ide_hwif_t *hwif
 		drive->special.b.set_geometry	= 1;
 		drive->name[0]			= 'h';
 		drive->name[1]			= 'd';
-		drive->name[2]			= 'a' + (index * MAX_DRIVES) + unit;
+		drive->name[2]			= 'a' + j;
 		drive->max_failures		= IDE_DEFAULT_MAX_FAILURES;
-		drive->using_dma		= 0;
-		drive->vdma			= 0;
+
 		INIT_LIST_HEAD(&drive->list);
 		init_completion(&drive->gendev_rel_comp);
 	}
 }
-EXPORT_SYMBOL_GPL(ide_init_port_data);
+
 
 static void init_hwif_default(ide_hwif_t *hwif, unsigned int index)
 {
Acked-by on v1 7/18
#11 lore raw

Re: [PATCH 07/18] ide: factor out devices init from ide_init_port_data()

Quoted 10 lines · Bartlomiej Zolnierkiewicz

* Factor out devices init from ide_init_port_data() to ide_port_init_devices_data().

While at it:

* Add explicit clearing of IDE device structure.

There should be no functionality changes caused by this patch.

MBR, Sergei

#12 lore raw

PATCH 8/18 ide: move ide_port_setup_devices() call to ide_device_add_all()

In reply to Bartlomiej Zolnierkiewicz

Add ide_cfg_mtx lock/unlock to ide_port_setup_devices() and then move ide_port_setup_devices() call from init_irq() to ide_device_add_all().

---
 drivers/ide/ide-probe.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/ide/ide-probe.c +4 -2
Index: b/drivers/ide/ide-probe.c
===================================================================
@@ -947,6 +947,7 @@ static void ide_port_setup_devices(ide_h
 {
 	int i;
 
+	mutex_lock(&ide_cfg_mtx);
 	for (i = 0; i < MAX_DRIVES; i++) {
 		ide_drive_t *drive = &hwif->drives[i];
 
@@ -961,6 +962,7 @@ static void ide_port_setup_devices(ide_h
 
 		ide_add_drive_to_hwgroup(drive);
 	}
+	mutex_unlock(&ide_cfg_mtx);
 }
 
 /*
@@ -1086,8 +1088,6 @@ static int init_irq (ide_hwif_t *hwif)
 			hwif->sharing_irq ? "shar" : "serializ", match->name);
 	printk("\n");
 
-	ide_port_setup_devices(hwif);
-
 	mutex_unlock(&ide_cfg_mtx);
 	return 0;
 out_unlink:
@@ -1443,6 +1443,8 @@ int ide_device_add_all(u8 *idx, const st
 			continue;
 		}
 
+		ide_port_setup_devices(hwif);
+
 		ide_acpi_init(hwif);
 		ide_acpi_port_init_devices(hwif);
 	}
#14 lore raw

PATCH 9/18 ide: rework PowerMac media-bay support

In reply to Bartlomiej Zolnierkiewicz

Rework PowerMac media-bay support in such way that instead of un/registering the IDE interface we un/register IDE devices:

* Add ide_port_scan() helper for probing+registerering devices on a port.

* Rename ide_port_unregister_devices() to __ide_port_unregister_devices().

* Add ide_port_unregister_devices() helper for unregistering devices on a port.

* Add 'ide_hwif_t *cd_port' to 'struct media_bay_info', pass 'hwif' instead of hwif->index to media_bay_set_ide_infos() and use it to setup 'cd_port'.

* Use ide_port_unregister_devices() instead of ide_unregister() and ide_port_scan() instead of ide_register_hw() in media_bay_step().

* Unexport ide_register_hw() and make it static.

---
 drivers/ide/ide-probe.c        |   18 ++++++++++++++++++
 drivers/ide/ide.c              |   22 ++++++++++++++++------
 drivers/ide/ppc/pmac.c         |    3 ++-
 drivers/macintosh/mediabay.c   |   17 +++++++----------
 include/asm-powerpc/mediabay.h |    4 +++-
 include/linux/ide.h            |    6 ++----
 6 files changed, 48 insertions(+), 22 deletions(-)
drivers/ide/ide-probe.c +18 -0
Index: b/drivers/ide/ide-probe.c
===================================================================
@@ -1490,3 +1490,21 @@ int ide_device_add(u8 idx[4], const stru
 	return ide_device_add_all(idx_all, d);
 }
 EXPORT_SYMBOL_GPL(ide_device_add);
+
+void ide_port_scan(ide_hwif_t *hwif)
+{
+	ide_port_cable_detect(hwif);
+	ide_port_init_devices(hwif);
+
+	if (ide_probe_port(hwif) < 0)
+		return;
+
+	hwif->present = 1;
+
+	ide_port_tune_devices(hwif);
+	ide_acpi_port_init_devices(hwif);
+	ide_port_setup_devices(hwif);
+	hwif_register_devices(hwif);
+	ide_proc_port_register_devices(hwif);
+}
+EXPORT_SYMBOL_GPL(ide_port_scan);
drivers/ide/ide.c +16 -6
Index: b/drivers/ide/ide.c
===================================================================
@@ -506,7 +506,7 @@ void ide_remove_port_from_hwgroup(ide_hw
 }
 
 /* Called with ide_lock held. */
-static void ide_port_unregister_devices(ide_hwif_t *hwif)
+static void __ide_port_unregister_devices(ide_hwif_t *hwif)
 {
 	int i;
 
@@ -522,6 +522,18 @@ static void ide_port_unregister_devices(
 	}
 }
 
+void ide_port_unregister_devices(ide_hwif_t *hwif)
+{
+	mutex_lock(&ide_cfg_mtx);
+	spin_lock_irq(&ide_lock);
+	__ide_port_unregister_devices(hwif);
+	hwif->present = 0;
+	ide_port_init_devices_data(hwif);
+	spin_unlock_irq(&ide_lock);
+	mutex_unlock(&ide_cfg_mtx);
+}
+EXPORT_SYMBOL_GPL(ide_port_unregister_devices);
+
 /**
  *	ide_unregister		-	free an IDE interface
  *	@index: index of interface (will change soon to a pointer)
@@ -562,7 +574,7 @@ void ide_unregister(unsigned int index, 
 	hwif = &ide_hwifs[index];
 	if (!hwif->present)
 		goto abort;
-	ide_port_unregister_devices(hwif);
+	__ide_port_unregister_devices(hwif);
 	hwif->present = 0;
 
 	spin_unlock_irq(&ide_lock);
@@ -657,8 +669,8 @@ EXPORT_SYMBOL_GPL(ide_init_port_hw);
  *	Returns -1 on error.
  */
 
-int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
-		    ide_hwif_t **hwifp)
+static int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
+			   ide_hwif_t **hwifp)
 {
 	int index, retry = 1;
 	ide_hwif_t *hwif;
@@ -692,8 +704,6 @@ found:
 	return hwif->present ? index : -1;
 }
 
-EXPORT_SYMBOL(ide_register_hw);
-
 /*
  *	Locks for IDE setting functionality
  */
drivers/ide/ppc/pmac.c +2 -1
Index: b/drivers/ide/ppc/pmac.c
===================================================================
@@ -1088,7 +1088,8 @@ pmac_ide_setup_device(pmac_ide_hwif_t *p
 	if (np->parent && np->parent->name
 	    && strcasecmp(np->parent->name, "media-bay") == 0) {
 #ifdef CONFIG_PMAC_MEDIABAY
-		media_bay_set_ide_infos(np->parent, pmif->regbase, pmif->irq, hwif->index);
+		media_bay_set_ide_infos(np->parent, pmif->regbase, pmif->irq,
+					hwif);
 #endif /* CONFIG_PMAC_MEDIABAY */
 		pmif->mediabay = 1;
 		if (!bidp)
drivers/macintosh/mediabay.c +7 -10
Index: b/drivers/macintosh/mediabay.c
===================================================================
@@ -79,6 +79,7 @@ struct media_bay_info {
 	int				sleeping;
 	struct semaphore		lock;
 #ifdef CONFIG_BLK_DEV_IDE_PMAC
+	ide_hwif_t			*cd_port;
 	void __iomem			*cd_base;
 	int				cd_irq;
 	int				cd_retry;
@@ -450,7 +451,7 @@ int check_media_bay_by_base(unsigned lon
 }
 
 int media_bay_set_ide_infos(struct device_node* which_bay, unsigned long base,
-			    int irq, int index)
+			    int irq, ide_hwif_t *hwif)
 {
 	int	i;
 
@@ -458,10 +459,11 @@ int media_bay_set_ide_infos(struct devic
 		struct media_bay_info* bay = &media_bays[i];
 
 		if (bay->mdev && which_bay == bay->mdev->ofdev.node) {
-			int timeout = 5000;
+			int timeout = 5000, index = hwif->index;
 			
 			down(&bay->lock);
 
+			bay->cd_port	= hwif;
  			bay->cd_base	= (void __iomem *) base;
 			bay->cd_irq	= irq;
 
@@ -553,15 +555,10 @@ static void media_bay_step(int i)
 			bay->timer = 0;
 			bay->state = mb_up;
 			if (bay->cd_index < 0) {
-				hw_regs_t hw;
-
 				printk("mediabay %d, registering IDE...\n", i);
 				pmu_suspend();
-				ide_init_hwif_ports(&hw, (unsigned long) bay->cd_base, (unsigned long) 0, NULL);
-				hw.irq = bay->cd_irq;
-				hw.chipset = ide_pmac;
-				bay->cd_index =
-					ide_register_hw(&hw, NULL, NULL);
+				ide_port_scan(bay->cd_port);
+				bay->cd_index = bay->cd_port->index;
 				pmu_resume();
 			}
 			if (bay->cd_index == -1) {
@@ -591,7 +588,7 @@ static void media_bay_step(int i)
     	        if (bay->cd_index >= 0) {
 			printk(KERN_DEBUG "Unregistering mb %d ide, index:%d\n", i,
 			       bay->cd_index);
-			ide_unregister(bay->cd_index, 1, 1);
+			ide_port_unregister_devices(bay->cd_port);
 			bay->cd_index = -1;
 		}
 	    	if (bay->cd_retry) {
include/asm-powerpc/mediabay.h +3 -1
Index: b/include/asm-powerpc/mediabay.h
===================================================================
@@ -22,10 +22,12 @@ int check_media_bay(struct device_node *
 /* Number of bays in the machine or 0 */
 extern int media_bay_count;
 
+#ifdef CONFIG_BLK_DEV_IDE_PMAC
 int check_media_bay_by_base(unsigned long base, int what);
 /* called by IDE PMAC host driver to register IDE controller for media bay */
 int media_bay_set_ide_infos(struct device_node *which_bay, unsigned long base,
-			    int irq, int index);
+			    int irq, ide_hwif_t *hwif);
+#endif
 
 #endif /* __KERNEL__ */
 #endif /* _PPC_MEDIABAY_H */
include/linux/ide.h +2 -4
Index: b/include/linux/ide.h
===================================================================
@@ -189,10 +189,6 @@ struct hwif_s * ide_find_port(unsigned l
 void ide_init_port_data(struct hwif_s *, unsigned int);
 void ide_init_port_hw(struct hwif_s *, hw_regs_t *);
 
-struct ide_drive_s;
-int ide_register_hw(hw_regs_t *, void (*)(struct ide_drive_s *),
-		    struct hwif_s **);
-
 static inline void ide_std_init_ports(hw_regs_t *hw,
 				      unsigned long io_addr,
 				      unsigned long ctl_addr)
@@ -1202,6 +1198,8 @@ void ide_undecoded_slave(ide_drive_t *);
 
 int ide_device_add_all(u8 *idx, const struct ide_port_info *);
 int ide_device_add(u8 idx[4], const struct ide_port_info *);
+void ide_port_unregister_devices(ide_hwif_t *);
+void ide_port_scan(ide_hwif_t *);
 
 static inline void *ide_get_hwifdata (ide_hwif_t * hwif)
 {
#15 lore raw

Re: [PATCH 09/18] ide: rework PowerMac media-bay support

Quoted 35 lines · Bartlomiej Zolnierkiewicz

Rework PowerMac media-bay support in such way that instead of un/registering the IDE interface we un/register IDE devices:

* Add ide_port_scan() helper for probing+registerering devices on a port.

* Rename ide_port_unregister_devices() to __ide_port_unregister_devices().

* Add ide_port_unregister_devices() helper for unregistering devices on a port.

* Add 'ide_hwif_t *cd_port' to 'struct media_bay_info', pass 'hwif' instead of hwif->index to media_bay_set_ide_infos() and use it to setup 'cd_port'.

* Use ide_port_unregister_devices() instead of ide_unregister() and ide_port_scan() instead of ide_register_hw() in media_bay_step().

* Unexport ide_register_hw() and make it static.

---
 drivers/ide/ide-probe.c        |   18 ++++++++++++++++++
 drivers/ide/ide.c              |   22 ++++++++++++++++------
 drivers/ide/ppc/pmac.c         |    3 ++-
 drivers/macintosh/mediabay.c   |   17 +++++++----------
 include/asm-powerpc/mediabay.h |    4 +++-
 include/linux/ide.h            |    6 ++----
 6 files changed, 48 insertions(+), 22 deletions(-)

Index: b/include/asm-powerpc/mediabay.h
===================================================================

include/asm-powerpc/mediabay.h +0 -0
@@ -22,10 +22,12 @@ int check_media_bay(struct device_node *
 /* Number of bays in the machine or 0 */
 extern int media_bay_count;
 

include/asm-powerpc/mediabay.h int check_media_bay(struct device_node * · line 33

Quoted 10 lines
include/asm-powerpc/mediabay.h +3 -1
+#ifdef CONFIG_BLK_DEV_IDE_PMAC
 int check_media_bay_by_base(unsigned long base, int what);
 /* called by IDE PMAC host driver to register IDE controller for media bay */
 int media_bay_set_ide_infos(struct device_node *which_bay, unsigned long base,
-			    int irq, int index);
+			    int irq, ide_hwif_t *hwif);
+#endif
 
 #endif /* __KERNEL__ */
 #endif /* _PPC_MEDIABAY_H */

This hunk breaks pmac32_defconfig for me.

include2/asm/mediabay.h:29: error: syntax error before 'ide_hwif_t' make[3]: *** [arch/powerpc/platforms/powermac/setup.o] Error 1 make[2]: *** [arch/powerpc/platforms/powermac] Error 2 make[1]: *** [arch/powerpc/platforms] Error 2
make: *** [sub-make] Error 2

cheers

Signature · 8 lines
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person
#16 lore raw

Re: [PATCH 09/18] ide: rework PowerMac media-bay support

Hi,

Quoted 36 lines · Michael Ellerman
Quoted 35 lines · Bartlomiej Zolnierkiewicz

Rework PowerMac media-bay support in such way that instead of un/registering the IDE interface we un/register IDE devices:

* Add ide_port_scan() helper for probing+registerering devices on a port.

* Rename ide_port_unregister_devices() to __ide_port_unregister_devices().

* Add ide_port_unregister_devices() helper for unregistering devices on a port.

* Add 'ide_hwif_t *cd_port' to 'struct media_bay_info', pass 'hwif' instead of hwif->index to media_bay_set_ide_infos() and use it to setup 'cd_port'.

* Use ide_port_unregister_devices() instead of ide_unregister() and ide_port_scan() instead of ide_register_hw() in media_bay_step().

* Unexport ide_register_hw() and make it static.

---
 drivers/ide/ide-probe.c        |   18 ++++++++++++++++++
 drivers/ide/ide.c              |   22 ++++++++++++++++------
 drivers/ide/ppc/pmac.c         |    3 ++-
 drivers/macintosh/mediabay.c   |   17 +++++++----------
 include/asm-powerpc/mediabay.h |    4 +++-
 include/linux/ide.h            |    6 ++----
 6 files changed, 48 insertions(+), 22 deletions(-)

Index: b/include/asm-powerpc/mediabay.h
===================================================================

include/asm-powerpc/mediabay.h +0 -0
@@ -22,10 +22,12 @@ int check_media_bay(struct device_node *
 /* Number of bays in the machine or 0 */
 extern int media_bay_count;
 
Quoted 1 line

include/asm-powerpc/mediabay.h int check_media_bay(struct device_node * · line 25

Quoted 1 line
include/asm-powerpc/mediabay.h +1 -0
+#ifdef CONFIG_BLK_DEV_IDE_PMAC

Does adding:

#include <linux/ide.h>

after ifdef help?

Quoted 19 lines

include/asm-powerpc/mediabay.h int check_media_bay(struct device_node * · line 33

Quoted 9 lines
include/asm-powerpc/mediabay.h +2 -1
 int check_media_bay_by_base(unsigned long base, int what);
 /* called by IDE PMAC host driver to register IDE controller for media bay */
 int media_bay_set_ide_infos(struct device_node *which_bay, unsigned long base,
-			    int irq, int index);
+			    int irq, ide_hwif_t *hwif);
+#endif
 
 #endif /* __KERNEL__ */
 #endif /* _PPC_MEDIABAY_H */

This hunk breaks pmac32_defconfig for me.

include2/asm/mediabay.h:29: error: syntax error before 'ide_hwif_t' make[3]: *** [arch/powerpc/platforms/powermac/setup.o] Error 1 make[2]: *** [arch/powerpc/platforms/powermac] Error 2 make[1]: *** [arch/powerpc/platforms] Error 2
make: *** [sub-make] Error 2

cheers

#17 lore raw

PATCH 10/18 ide: add warm-plug support for IDE devices

In reply to Bartlomiej Zolnierkiewicz

* Add 'struct class ide_port_class' ('ide_port' class) and embedd 'struct class_device classdev' ('ide_port' class device) in ide_hwif_t.

* Register 'ide_port' class in ide_init() and unregister it in cleanup_module().

* Add class_to_ide_port() macro and ide_port_class_release().

* Register ->classdev in ide_register_port () and unregister it in ide_unregister().

* Add "delete_devices" class device attribute for unregistering IDE devices on a port and "scan" one for probing+registering IDE devices on a port.

* Add ide_sysfs_register_port() helper for registering "delete_devices" and "scan" attributes with ->classdev. Call it in ide_device_add_all().

* Document IDE warm-plug support in Documentation/ide/warm-plug-howto.txt.

---
+540 bytes (x86-32)

After seeing some 'class device' conversion patches from Greg I tried to recast this patch to just use 'device' but I couldn't figure out how to use class_device_create() with *static* objects...

 Documentation/ide/warm-plug-howto.txt |   13 ++++++
 drivers/ide/ide-probe.c               |   70 +++++++++++++++++++++++++++++++++-
 drivers/ide/ide.c                     |   22 ++++++++++
 include/linux/ide.h                   |    7 ++-
 4 files changed, 109 insertions(+), 3 deletions(-)
Documentation/ide/warm-plug-howto.txt new file +13 -0
Index: b/Documentation/ide/warm-plug-howto.txt
===================================================================
@@ -0,0 +1,13 @@
+
+IDE warm-plug HOWTO
+===================
+
+To warm-plug devices on a port 'idex':
+
+# echo -n "1" > /sys/class/ide_port/idex/delete_devices
+
+unplug old device(s) and plug new device(s)
+
+# echo -n "1" > /sys/class/ide_port/idex/scan
+
+done
drivers/ide/ide-probe.c +68 -2

96 lines not on this page · raw message

drivers/ide/ide.c +22 -0
Index: b/drivers/ide/ide.c
===================================================================
@@ -596,6 +596,7 @@ void ide_unregister(unsigned int index, 
 
 	ide_remove_port_from_hwgroup(hwif);
 
+	class_device_unregister(&hwif->classdev);
 	device_unregister(&hwif->gendev);
 	wait_for_completion(&hwif->gendev_rel_comp);
 
@@ -1613,6 +1614,16 @@ struct bus_type ide_bus_type = {
 
 EXPORT_SYMBOL_GPL(ide_bus_type);
 
+static void ide_port_class_release(struct class_device *class_dev)
+{
+	put_device(&class_to_ide_port(class_dev)->gendev);
+}
+
+struct class ide_port_class = {
+	.name		= "ide_port",
+	.release	= ide_port_class_release,
+};
+
 /*
  * This is gets invoked once during initialization, to set *everything* up
  */
@@ -1633,11 +1644,20 @@ static int __init ide_init(void)
 		return ret;
 	}
 
+	ret = class_register(&ide_port_class);
+	if (ret)
+		goto out_port_class;
+
 	init_ide_data();
 
 	proc_ide_create();
 
 	return 0;
+
+out_port_class:
+	bus_unregister(&ide_bus_type);
+
+	return ret;
 }
 
 #ifdef MODULE
@@ -1674,6 +1694,8 @@ void __exit cleanup_module (void)
 
 	proc_ide_destroy();
 
+	class_unregister(&ide_port_class);
+
 	bus_unregister(&ide_bus_type);
 }
 
include/linux/ide.h +6 -1
Index: b/include/linux/ide.h
===================================================================
@@ -581,7 +581,9 @@ typedef struct hwif_s {
 	unsigned	mmio       : 1; /* host uses MMIO */
 	unsigned	straight8  : 1;	/* Alan's straight 8 check */
 
-	struct device	gendev;
+	struct device		gendev;
+	struct class_device	classdev;
+
 	struct completion gendev_rel_comp; /* To deal with device release() */
 
 	void		*hwif_data;	/* extra hwif data */
@@ -593,6 +595,8 @@ typedef struct hwif_s {
 #endif
 } ____cacheline_internodealigned_in_smp ide_hwif_t;
 
+#define class_to_ide_port(dev)	container_of(dev, ide_hwif_t, classdev)
+
 /*
  *  internal ide interrupt handler type
  */
@@ -1275,6 +1279,7 @@ extern struct mutex ide_cfg_mtx;
 #define local_irq_set(flags)	do { local_save_flags((flags)); local_irq_enable_in_hardirq(); } while (0)
 
 extern struct bus_type ide_bus_type;
+extern struct class ide_port_class;
 
 /* check if CACHE FLUSH (EXT) command is supported (bits defined in ATA-6) */
 #define ide_id_has_flush_cache(id)	((id)->cfs_enable_2 & 0x3000)
#18 lore raw

PATCH 11/18 ide-generic: add ide_generic class and attribute for adding new interfaces

In reply to Bartlomiej Zolnierkiewicz

* Add ide_generic_sysfs_init() helper registering 'ide_generic' class (together with ide_generic_class_release() ->class_release method) and use it in ide_generic_init().

* Add "add" class attribute to 'ide_generic' class for adding new interfaces (it is intended to be a replacement for obsoleted "idex=base[,ctl[,irq]]" kernel parameters).

---
+391 bytes (x86-32)

I also tried interface based on kernel parameters but it sucked (more complex, bigger source code + resulting binary).

 drivers/ide/ide-generic.c |   78 +++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 77 insertions(+), 1 deletion(-)
drivers/ide/ide-generic.c +77 -1
Index: b/drivers/ide/ide-generic.c
===================================================================
@@ -1,17 +1,89 @@
 /*
  * generic/default IDE host driver
  *
- * Copyright (C) 2004 Bartlomiej Zolnierkiewicz
+ * Copyright (C) 2004, 2008 Bartlomiej Zolnierkiewicz
  * This code was split off from ide.c.  See it for original copyrights.
  *
  * May be copied or modified under the terms of the GNU General Public License.
  */
 
+/*
+ * For special cases new interfaces may be added using sysfs, i.e.
+ *
+ *	echo -n "0x168:0x36e:10" > /sys/class/ide_generic/add
+ *
+ * will add an interface using I/O ports 0x168-0x16f/0x36e and IRQ 10.
+ */
+
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/module.h>
 #include <linux/ide.h>
 
+#define DRV_NAME	"ide_generic"
+
+static ssize_t store_add(struct class *cls, const char *buf, size_t n)
+{
+	ide_hwif_t *hwif;
+	unsigned int base, ctl;
+	int irq;
+	hw_regs_t hw;
+	u8 idx[] = { 0xff, 0xff, 0xff, 0xff };
+
+	if (sscanf(buf, "%x:%x:%d", &base, &ctl, &irq) != 3)
+		return -EINVAL;
+
+	hwif = ide_find_port(base);
+	if (hwif == NULL)
+		return -ENOENT;
+
+	memset(&hw, 0, sizeof(hw));
+	ide_std_init_ports(&hw, base, ctl);
+	hw.irq = irq;
+	hw.chipset = ide_generic;
+
+	ide_init_port_hw(hwif, &hw);
+
+	idx[0] = hwif->index;
+
+	ide_device_add(idx, NULL);
+
+	return n;
+};
+
+static struct class_attribute ide_generic_class_attrs[] = {
+	__ATTR(add, S_IWUSR, NULL, store_add),
+	__ATTR_NULL
+};
+
+static void ide_generic_class_release(struct class *cls)
+{
+	kfree(cls);
+}
+
+static int __init ide_generic_sysfs_init(void)
+{
+	struct class *cls;
+	int rc;
+
+	cls = kzalloc(sizeof(*cls), GFP_KERNEL);
+	if (!cls)
+		return -ENOMEM;
+
+	cls->name = DRV_NAME;
+	cls->owner = THIS_MODULE;
+	cls->class_release = ide_generic_class_release;
+	cls->class_attrs = ide_generic_class_attrs;
+
+	rc = class_register(cls);
+	if (rc) {
+		kfree(cls);
+		return rc;
+	}
+
+	return 0;
+}
+
 static int __init ide_generic_init(void)
 {
 	u8 idx[MAX_HWIFS];
@@ -36,6 +108,10 @@ static int __init ide_generic_init(void)
 	if (ide_hwifs[0].io_ports[IDE_DATA_OFFSET])
 		ide_release_lock();	/* for atari only */
 
+	if (ide_generic_sysfs_init())
+		printk(KERN_ERR DRV_NAME ": failed to create ide_generic "
+					 "class\n");
+
 	return 0;
 }
 
#19 lore raw

PATCH 12/18 ide: remove needless CONFIG_BLK_DEV_HD hack from init_hwif()

In reply to Bartlomiej Zolnierkiewicz

request_irq() will fail if there is already another IRQ handler registered and IRQ flags are mismatched.

---
 drivers/ide/ide-probe.c |    7 -------
 1 file changed, 7 deletions(-)
drivers/ide/ide-probe.c +0 -7
Index: b/drivers/ide/ide-probe.c
===================================================================
@@ -1237,13 +1237,6 @@ static int hwif_init(ide_hwif_t *hwif)
 			return 0;
 		}
 	}
-#ifdef CONFIG_BLK_DEV_HD
-	if (hwif->irq == HD_IRQ && hwif->io_ports[IDE_DATA_OFFSET] != HD_DATA) {
-		printk("%s: CANNOT SHARE IRQ WITH OLD "
-			"HARDDISK DRIVER (hd.c)\n", hwif->name);
-		return 0;
-	}
-#endif /* CONFIG_BLK_DEV_HD */
 
 	if (register_blkdev(hwif->major, hwif->name))
 		return 0;
#20 lore raw

PATCH 13/18 ide: remove CONFIG_BLK_DEV_HD_IDE config option

In reply to Bartlomiej Zolnierkiewicz

* Remove CONFIG_BLK_DEV_HD hack from init_hwif_default() ("ide0=noprobe" kernel parameter should be used instead if somebody wishes to use the old "hd" driver).

* Make CONFIG_BLK_DEV_HD_ONLY config option available also when IDE subsystem is used and update help entry.

* Remove no longer needed CONFIG_BLK_DEV_HD_IDE config option.

---
 drivers/ide/Kconfig |   25 +++++--------------------
 drivers/ide/ide.c   |    4 ----
 2 files changed, 5 insertions(+), 24 deletions(-)
drivers/ide/Kconfig +5 -20
Index: b/drivers/ide/Kconfig
===================================================================
@@ -118,24 +118,6 @@ config BLK_DEV_IDE_SATA
 
 	  If unsure, say N.
 
-config BLK_DEV_HD_IDE
-	bool "Use old disk-only driver on primary interface"
-	depends on (X86 || SH_MPC1211)
-	---help---
-	  There are two drivers for MFM/RLL/IDE disks.  Most people use just
-	  the new enhanced driver by itself.  This option however installs the
-	  old hard disk driver to control the primary IDE/disk interface in
-	  the system, leaving the new enhanced IDE driver to take care of only
-	  the 2nd/3rd/4th IDE interfaces.  Doing this will prevent you from
-	  having an IDE/ATAPI CD-ROM or tape drive connected to the primary
-	  IDE interface.  Choosing this option may be useful for older systems
-	  which have MFM/RLL/ESDI controller+drives at the primary port
-	  address (0x1f0), along with IDE drives at the secondary/3rd/4th port
-	  addresses.
-
-	  Normally, just say N here; you will then use the new driver for all
-	  4 interfaces.
-
 config BLK_DEV_IDEDISK
 	tristate "Include IDE/ATA-2 DISK support"
 	---help---
@@ -1112,7 +1094,6 @@ endif
 
 config BLK_DEV_HD_ONLY
 	bool "Old hard disk (MFM/RLL/IDE) driver"
-	depends on BLK_DEV_IDE=n
 	help
 	  There are two drivers for MFM/RLL/IDE hard disks. Most people use
 	  the newer enhanced driver, but this old one is still around for two
@@ -1124,12 +1105,16 @@ config BLK_DEV_HD_ONLY
 	  for systems with only older MFM/RLL/ESDI drives. Choosing the old
 	  driver can save 13 KB or so of kernel memory.
 
+	  If you want to use this driver together with the new one you have
+	  to use "ide0=noprobe" kernel parameter to prevent the new driver
+	  from probing the primary interface.
+
 	  If you are unsure, then just choose the Enhanced IDE/MFM/RLL driver
 	  instead of this one. For more detailed information, read the
 	  Disk-HOWTO, available from
 	  <http://www.tldp.org/docs.html#howto>.
 
 config BLK_DEV_HD
-	def_bool BLK_DEV_HD_IDE || BLK_DEV_HD_ONLY
+	def_bool BLK_DEV_HD_ONLY
 
 endif # IDE
drivers/ide/ide.c +0 -4
Index: b/drivers/ide/ide.c
===================================================================
@@ -179,10 +179,6 @@ static void init_hwif_default(ide_hwif_t
 	memcpy(hwif->io_ports, hw.io_ports, sizeof(hw.io_ports));
 
 	hwif->noprobe = !hwif->io_ports[IDE_DATA_OFFSET];
-#ifdef CONFIG_BLK_DEV_HD
-	if (hwif->io_ports[IDE_DATA_OFFSET] == HD_DATA)
-		hwif->noprobe = 1;	/* may be overridden by ide_setup() */
-#endif
 }
 
 /*
#21 lore raw

PATCH 14/18 ide: remove obsoleted "idex=base[,ctl[,irq]]" kernel parameters

In reply to Bartlomiej Zolnierkiewicz

* Remove obsoleted "idex=base[,ctl[,irq]]" kernel parameters and update Documentation/ide.txt.

* Remove no longer needed ide_forced chipset type.

---
-244 bytes (x86-32)

 Documentation/ide.txt     |   31 -------------------------------
 drivers/ide/ide-generic.c |    3 +--
 drivers/ide/ide-probe.c   |    3 +--
 drivers/ide/ide-proc.c    |    3 ---
 drivers/ide/ide.c         |   22 +++++-----------------
 drivers/ide/pci/cmd640.c  |    2 +-
 drivers/ide/setup-pci.c   |   11 -----------
 include/linux/ide.h       |    2 +-
 8 files changed, 9 insertions(+), 68 deletions(-)
Documentation/ide.txt +0 -31
Index: b/Documentation/ide.txt
===================================================================
@@ -71,29 +71,6 @@ This driver automatically probes for mos
 ones), for the drives/geometries attached to those interfaces, and for the IRQ
 lines being used by the interfaces (normally 14, 15 for ide0/ide1).
 
-For special cases, interfaces may be specified using kernel "command line"
-options.  For example,
-
-	ide3=0x168,0x36e,10	/* ioports 0x168-0x16f,0x36e, irq 10 */
-
-Normally the irq number need not be specified, as ide.c will probe for it:
-
-	ide3=0x168,0x36e	/* ioports 0x168-0x16f,0x36e */
-
-The standard port, and irq values are these:
-
-	ide0=0x1f0,0x3f6,14
-	ide1=0x170,0x376,15
-	ide2=0x1e8,0x3ee,11
-	ide3=0x168,0x36e,10
-
-Note that the first parameter reserves 8 contiguous ioports, whereas the
-second value denotes a single ioport. If in doubt, do a 'cat /proc/ioports'.
-
-In all probability the device uses these ports and IRQs if it is attached
-to the appropriate ide channel.  Pass the parameter for the correct ide
-channel to the kernel, as explained above.
-
 Any number of interfaces may share a single IRQ if necessary, at a slight
 performance penalty, whether on separate cards or a single VLB card.
 The IDE driver automatically detects and handles this.  However, this may
@@ -259,14 +236,6 @@ Summary of ide driver parameters for ker
 			  Bigger values are safer than smaller ones.
 
  "idex=noprobe"		: do not attempt to access/use this interface
- 
- "idex=base"		: probe for an interface at the addr specified,
-			  where "base" is usually 0x1f0 or 0x170
-			  and "ctl" is assumed to be "base"+0x206
-			  
- "idex=base,ctl"	: specify both base and ctl
-
- "idex=base,ctl,irq"	: specify base, ctl, and irq number
 
  "idex=serialize"	: do not overlap operations on idex. Please note
 			  that you will have to specify this option for
drivers/ide/ide-generic.c +1 -2
Index: b/drivers/ide/ide-generic.c
===================================================================
@@ -96,8 +96,7 @@ static int __init ide_generic_init(void)
 		ide_hwif_t *hwif = &ide_hwifs[i];
 
 		if (hwif->io_ports[IDE_DATA_OFFSET] &&
-		    (hwif->chipset == ide_unknown ||
-		     hwif->chipset == ide_forced))
+		    hwif->chipset == ide_unknown)
 			idx[i] = i;
 		else
 			idx[i] = 0xff;
drivers/ide/ide-probe.c +1 -2
Index: b/drivers/ide/ide-probe.c
===================================================================
@@ -1514,8 +1514,7 @@ int ide_device_add_all(u8 *idx, const st
 		hwif = &ide_hwifs[idx[i]];
 
 		if (hwif->present) {
-			if (hwif->chipset == ide_unknown ||
-			    hwif->chipset == ide_forced)
+			if (hwif->chipset == ide_unknown)
 				hwif->chipset = ide_generic;
 			hwif_register_devices(hwif);
 		}
drivers/ide/ide-proc.c +0 -3
Index: b/drivers/ide/ide-proc.c
===================================================================
@@ -46,9 +46,6 @@ static int proc_ide_read_imodel
 	int		len;
 	const char	*name;
 
-	/*
-	 * Neither ide_unknown nor ide_forced should be set at this point.
-	 */
 	switch (hwif->chipset) {
 		case ide_generic:	name = "generic";	break;
 		case ide_pci:		name = "pci";		break;
drivers/ide/ide.c +5 -17
Index: b/drivers/ide/ide.c
===================================================================
@@ -1349,8 +1349,6 @@ static int __init ide_setup(char *s)
 			"minus10", "four", "qd65xx", "ht6560b", "cmd640_vlb",
 			"dtc2278", "umc8672", "ali14xx", NULL };
 
-		hw_regs_t hwregs;
-
 		hw = s[3] - '0';
 		hwif = &ide_hwifs[hw];
 		i = match_parm(&s[4], ide_words, vals, 3);
@@ -1451,21 +1449,11 @@ static int __init ide_setup(char *s)
 			case -1: /* "noprobe" */
 				hwif->noprobe = 1;
 				goto done;
-
-			case 1:	/* base */
-				vals[1] = vals[0] + 0x206; /* default ctl */
-			case 2: /* base,ctl */
-				vals[2] = 0;	/* default irq = probe for it */
-			case 3: /* base,ctl,irq */
-				memset(&hwregs, 0, sizeof(hwregs));
-				ide_init_hwif_ports(&hwregs, vals[0], vals[1], &hwif->irq);
-				memcpy(hwif->io_ports, hwregs.io_ports, sizeof(hwif->io_ports));
-				hwif->irq      = vals[2];
-				hwif->noprobe  = 0;
-				hwif->chipset  = ide_forced;
-				goto obsolete_option;
-
-			case 0: goto bad_option;
+			case 0:
+			case 1:
+			case 2:
+			case 3:
+				goto bad_option;
 			default:
 				printk(" -- SUPPORT NOT CONFIGURED IN THIS KERNEL\n");
 				return 1;
drivers/ide/pci/cmd640.c +1 -1
Index: b/drivers/ide/pci/cmd640.c
===================================================================
@@ -415,7 +415,7 @@ static void __init setup_device_ptrs (vo
 	cmd_hwif1 = &ide_hwifs[1]; /* default, if not found below */
 	for (i = 0; i < MAX_HWIFS; i++) {
 		ide_hwif_t *hwif = &ide_hwifs[i];
-		if (hwif->chipset == ide_unknown || hwif->chipset == ide_forced) {
+		if (hwif->chipset == ide_unknown) {
 			if (hwif->io_ports[IDE_DATA_OFFSET] == 0x1f0)
 				cmd_hwif0 = hwif;
 			else if (hwif->io_ports[IDE_DATA_OFFSET] == 0x170)
drivers/ide/setup-pci.c +0 -11
Index: b/drivers/ide/setup-pci.c
===================================================================
@@ -41,17 +41,6 @@ static ide_hwif_t *ide_match_hwif(unsign
 	ide_hwif_t *hwif;
 
 	/*
-	 * Look for a hwif with matching io_base specified using
-	 * parameters to ide_setup().
-	 */
-	for (h = 0; h < MAX_HWIFS; ++h) {
-		hwif = &ide_hwifs[h];
-		if (hwif->io_ports[IDE_DATA_OFFSET] == io_base) {
-			if (hwif->chipset == ide_forced)
-				return hwif; /* a perfect match */
-		}
-	}
-	/*
 	 * Look for a hwif with matching io_base default value.
 	 * If chipset is "ide_unknown", then claim that hwif slot.
 	 * Otherwise, some other chipset has already claimed it..  :(
include/linux/ide.h +1 -1
Index: b/include/linux/ide.h
===================================================================
@@ -169,7 +169,7 @@ enum {		ide_unknown,	ide_generic,	ide_pc
 		ide_rz1000,	ide_trm290,
 		ide_cmd646,	ide_cy82c693,	ide_4drives,
 		ide_pmac,	ide_etrax100,	ide_acorn,
-		ide_au1xxx,	ide_palm3710,	ide_forced
+		ide_au1xxx,	ide_palm3710
 };
 
 typedef u8 hwif_chipset_t;
#22 lore raw

PATCH 15/18 ide: remove broken/dangerous HDIO_[UNREGISTER,SCAN]_HWIF ioctls

In reply to Bartlomiej Zolnierkiewicz

hdparm explicitely marks HDIO_[UNREGISTER,SCAN]_HWIF ioctls as DANGEROUS and given the number of bugs we can assume that there are no real users:

* DMA has no chance of working because DMA resources are released by ide_unregister() and they are never allocated again.

* Since ide_init_hwif_ports() is used for ->io_ports[] setup the ioctls don't work for almost all hosts with "non-standard" (== non ISA-like) layout of IDE taskfile registers (there is a lot of such host drivers).

* ide_port_init_devices() is not called when probing IDE devices so:
- drive->autotune is never set and IDE host/devices are not programmed for the correct PIO/DMA transfer modes (=> possible data corruption)
- host specific I/O 32-bit and IRQ unmasking settings are not applied (=> possible data corruption)
- host specific ->port_init_devs method is not called (=> no luck with

    ht6560b, qd65xx and opti621 host drivers)

* ->rw_disk method is not preserved (=> no HPT3xxN chipsets support).

* ->serialized flag is not preserved (=> possible data corruption when using icside, aec62xx (ATP850UF chipset), cmd640, cs5530, hpt366 (HPT3xxN chipsets), rz1000, sc1200, dtc2278 and ht6560b host drivers).

* ->ack_intr method is not preserved (=> needed by ide-cris, buddha, gayle and macide host drivers).

* ->sata_scr[] and sata_misc[] is cleared by ide_unregister() and it isn't initialized again (SiI3112 support needs them).

* To issue an ioctl() there need to be at least one IDE device present in the system.

* ->cable_detect method is not preserved + it is not called when probing IDE devices so cable detection is broken (however since DMA support is also broken it doesn't really matter ;-).

* Some objects which may have already been freed in ide_unregister() are restored by ide_hwif_restore() (i.e. ->hwgroup).

* ide_register_hw() may unregister unrelated IDE ports if free ide_hwifs[] slot cannot be found.

* When IDE host drivers are modular unregistered port may be re-used by different host driver that owned it first causing subtle bugs.

Since we now have a proper warm-plug support remove these ioctls, then remove no longer needed:
- ide_register_hw() and ide_hwif_restore() functions
- 'init_default' and 'restore' arguments of ide_unregister()
- zeroeing of hwif->{dma,extra}_* fields in ide_unregister()

As an added bonus IDE core code size shrinks by ~3kB (x86-32).

---
 block/compat_ioctl.c              |    1 
 drivers/ide/arm/bast-ide.c        |    2 
 drivers/ide/arm/rapide.c          |    2 
 drivers/ide/ide-pnp.c             |    2 
 drivers/ide/ide.c                 |  198 --------------------------------------
 drivers/ide/legacy/ide-cs.c       |    4 
 drivers/ide/legacy/ide_platform.c |    2 
 drivers/ide/mips/au1xxx-ide.c     |    2 
 drivers/ide/pci/delkin_cb.c       |    4 
 drivers/ide/pci/scc_pata.c        |    2 
 include/linux/hdreg.h             |    4 
 include/linux/ide.h               |    2 
 12 files changed, 17 insertions(+), 208 deletions(-)
block/compat_ioctl.c +0 -1
Index: b/block/compat_ioctl.c
===================================================================
@@ -624,7 +624,6 @@ static int compat_blkdev_driver_ioctl(st
 	case HDIO_GET_IDENTITY:
 	case HDIO_DRIVE_TASK:
 	case HDIO_DRIVE_CMD:
-	case HDIO_SCAN_HWIF:
 	/* 0x330 is reserved -- it used to be HDIO_GETGEO_BIG */
 	case 0x330:
 	/* 0x02 -- Floppy ioctls */
drivers/ide/arm/bast-ide.c +1 -1
Index: b/drivers/ide/arm/bast-ide.c
===================================================================
@@ -53,7 +53,7 @@ bastide_register(unsigned int base, unsi
 	i = hwif->index;
 
 	if (hwif->present)
-		ide_unregister(i, 0, 0);
+		ide_unregister(i);
 	else if (!hwif->hold)
 		ide_init_port_data(hwif, i);
 
drivers/ide/arm/rapide.c +1 -1
Index: b/drivers/ide/arm/rapide.c
===================================================================
@@ -76,7 +76,7 @@ static void __devexit rapide_remove(stru
 
 	ecard_set_drvdata(ec, NULL);
 
-	ide_unregister(hwif->index, 0, 0);
+	ide_unregister(hwif->index);
 
 	ecard_release_resources(ec);
 }
drivers/ide/ide-pnp.c +1 -1
Index: b/drivers/ide/ide-pnp.c
===================================================================
@@ -62,7 +62,7 @@ static void idepnp_remove(struct pnp_dev
 	ide_hwif_t *hwif = pnp_get_drvdata(dev);
 
 	if (hwif)
-		ide_unregister(hwif->index, 0, 0);
+		ide_unregister(hwif->index);
 	else
 		printk(KERN_ERR "idepnp: Unable to remove device, please report.\n");
 }
drivers/ide/ide.c +3 -195

245 lines not on this page · raw message

drivers/ide/legacy/ide-cs.c +2 -2
Index: b/drivers/ide/legacy/ide-cs.c
===================================================================
@@ -163,7 +163,7 @@ static int idecs_register(unsigned long 
     i = hwif->index;
 
     if (hwif->present)
-	ide_unregister(i, 0, 0);
+	ide_unregister(i);
     else if (!hwif->hold)
 	ide_init_port_data(hwif, i);
 
@@ -360,7 +360,7 @@ void ide_release(struct pcmcia_device *l
     if (info->ndev) {
 	/* FIXME: if this fails we need to queue the cleanup somehow
 	   -- need to investigate the required PCMCIA magic */
-	ide_unregister(info->hd, 0, 0);
+	ide_unregister(info->hd);
     }
     info->ndev = 0;
 
drivers/ide/legacy/ide_platform.c +1 -1
Index: b/drivers/ide/legacy/ide_platform.c
===================================================================
@@ -122,7 +122,7 @@ static int __devexit plat_ide_remove(str
 {
 	ide_hwif_t *hwif = pdev->dev.driver_data;
 
-	ide_unregister(hwif->index, 0, 0);
+	ide_unregister(hwif->index);
 
 	return 0;
 }
drivers/ide/mips/au1xxx-ide.c +1 -1
Index: b/drivers/ide/mips/au1xxx-ide.c
===================================================================
@@ -677,7 +677,7 @@ static int au_ide_remove(struct device *
 	ide_hwif_t *hwif = dev_get_drvdata(dev);
 	_auide_hwif *ahwif = &auide_hwif;
 
-	ide_unregister(hwif->index, 0, 0);
+	ide_unregister(hwif->index);
 
 	iounmap((void *)ahwif->regbase);
 
drivers/ide/pci/delkin_cb.c +2 -2
Index: b/drivers/ide/pci/delkin_cb.c
===================================================================
@@ -85,7 +85,7 @@ delkin_cb_probe (struct pci_dev *dev, co
 	i = hwif->index;
 
 	if (hwif->present)
-		ide_unregister(i, 0, 0);
+		ide_unregister(i);
 	else if (!hwif->hold)
 		ide_init_port_data(hwif, i);
 
@@ -120,7 +120,7 @@ delkin_cb_remove (struct pci_dev *dev)
 	ide_hwif_t *hwif = pci_get_drvdata(dev);
 
 	if (hwif)
-		ide_unregister(hwif->index, 0, 0);
+		ide_unregister(hwif->index);
 
 	pci_disable_device(dev);
 }
drivers/ide/pci/scc_pata.c +1 -1
Index: b/drivers/ide/pci/scc_pata.c
===================================================================
@@ -736,7 +736,7 @@ static void __devexit scc_remove(struct 
 		hwif->dmatable_cpu = NULL;
 	}
 
-	ide_unregister(hwif->index, 0, 0);
+	ide_unregister(hwif->index);
 
 	hwif->chipset = ide_unknown;
 	iounmap((void*)ports->dma);
include/linux/hdreg.h +3 -1
Index: b/include/linux/hdreg.h
===================================================================
@@ -422,9 +422,11 @@ struct hd_geometry {
 #define HDIO_SET_NOWERR		0x0325	/* change ignore-write-error flag */
 #define HDIO_SET_DMA		0x0326	/* change use-dma flag */
 #define HDIO_SET_PIO_MODE	0x0327	/* reconfig interface to new speed */
+#ifndef __KERNEL__
 #define HDIO_SCAN_HWIF		0x0328	/* register and (re)scan interface */
-#define HDIO_SET_NICE		0x0329	/* set nice flags */
 #define HDIO_UNREGISTER_HWIF	0x032a  /* unregister interface */
+#endif
+#define HDIO_SET_NICE		0x0329	/* set nice flags */
 #define HDIO_SET_WCACHE		0x032b	/* change write cache enable-disable */
 #define HDIO_SET_ACOUSTIC	0x032c	/* change acoustic behavior */
 #define HDIO_SET_BUSSTATE	0x032d	/* set the bus state of the hwif */
include/linux/ide.h +1 -1
Index: b/include/linux/ide.h
===================================================================
@@ -1193,7 +1193,7 @@ static inline void ide_acpi_set_state(id
 void ide_remove_port_from_hwgroup(ide_hwif_t *);
 extern int ide_hwif_request_regions(ide_hwif_t *hwif);
 extern void ide_hwif_release_regions(ide_hwif_t* hwif);
-void ide_unregister(unsigned int, int, int);
+void ide_unregister(unsigned int);
 
 void ide_register_region(struct gendisk *);
 void ide_unregister_region(struct gendisk *);
Acked-by on v1 15/18
#23 lore raw

Re: [PATCH 15/18] ide: remove broken/dangerous HDIO_[UNREGISTER,SCAN]_HWIF ioctls

Quoted 52 lines · Bartlomiej Zolnierkiewicz

hdparm explicitely marks HDIO_[UNREGISTER,SCAN]_HWIF ioctls as DANGEROUS and given the number of bugs we can assume that there are no real users:

* DMA has no chance of working because DMA resources are released by ide_unregister() and they are never allocated again.

* Since ide_init_hwif_ports() is used for ->io_ports[] setup the ioctls don't work for almost all hosts with "non-standard" (== non ISA-like) layout of IDE taskfile registers (there is a lot of such host drivers).

* ide_port_init_devices() is not called when probing IDE devices so:
- drive->autotune is never set and IDE host/devices are not programmed for the correct PIO/DMA transfer modes (=> possible data corruption)
- host specific I/O 32-bit and IRQ unmasking settings are not applied (=> possible data corruption)
- host specific ->port_init_devs method is not called (=> no luck with

    ht6560b, qd65xx and opti621 host drivers)

* ->rw_disk method is not preserved (=> no HPT3xxN chipsets support).

* ->serialized flag is not preserved (=> possible data corruption when using icside, aec62xx (ATP850UF chipset), cmd640, cs5530, hpt366 (HPT3xxN chipsets), rz1000, sc1200, dtc2278 and ht6560b host drivers).

* ->ack_intr method is not preserved (=> needed by ide-cris, buddha, gayle and macide host drivers).

* ->sata_scr[] and sata_misc[] is cleared by ide_unregister() and it isn't initialized again (SiI3112 support needs them).

* To issue an ioctl() there need to be at least one IDE device present in the system.

* ->cable_detect method is not preserved + it is not called when probing IDE devices so cable detection is broken (however since DMA support is also broken it doesn't really matter ;-).

* Some objects which may have already been freed in ide_unregister() are restored by ide_hwif_restore() (i.e. ->hwgroup).

* ide_register_hw() may unregister unrelated IDE ports if free ide_hwifs[] slot cannot be found.

* When IDE host drivers are modular unregistered port may be re-used by different host driver that owned it first causing subtle bugs.

Since we now have a proper warm-plug support remove these ioctls, then remove no longer needed:
- ide_register_hw() and ide_hwif_restore() functions
- 'init_default' and 'restore' arguments of ide_unregister()
- zeroeing of hwif->{dma,extra}_* fields in ide_unregister()

Quoted 3 lines

As an added bonus IDE core code size shrinks by ~3kB (x86-32).

Quoted 4 lines

Index: b/drivers/ide/ide-pnp.c
===================================================================

drivers/ide/ide-pnp.c +0 -0

[...]

Quoted 17 lines
drivers/ide/ide-pnp.c +0 -13
@@ -655,52 +530,6 @@ void ide_init_port_hw(ide_hwif_t *hwif, 
 }
 EXPORT_SYMBOL_GPL(ide_init_port_hw);
 
-/**
- *	ide_register_hw		-	register IDE interface
- *	@hw: hardware registers
- *	@quirkproc: quirkproc function
- *	@hwifp: pointer to returned hwif
- *
- *	Register an IDE interface, specifying exactly the registers etc.
- *
- *	Returns -1 on error.
- */
-
-static int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
-			   ide_hwif_t **hwifp)

drivers/ide/ide-pnp.c void ide_init_port_hw(ide_hwif_t *hwif, · line 680

Quoted 10 lines
drivers/ide/ide-pnp.c +0 -10
-{
-	int index, retry = 1;
-	ide_hwif_t *hwif;
-	u8 idx[4] = { 0xff, 0xff, 0xff, 0xff };
-
-	do {
-		hwif = ide_find_port(hw->io_ports[IDE_DATA_OFFSET]);
-		index = hwif->index;
-		if (hwif)
-			goto found;

Hm, I remember there was a patch that fixed the above bug where hwif is
dereferenced before being checked for NULL, I wonder how come it was lost?

WBR, Sergei

#24 lore raw

Re: [PATCH 15/18] ide: remove broken/dangerous HDIO_[UNREGISTER,SCAN]_HWIF ioctls

Quoted 4 lines · Sergei Shtylyov
Quoted 2 lines · Bartlomiej Zolnierkiewicz

hdparm explicitely marks HDIO_[UNREGISTER,SCAN]_HWIF ioctls as DANGEROUS and given the number of bugs we can assume that there are no real users:

..

There is the odd user of these, actually.

But the most recent to email me (a few weeks ago), reported that the SCAN function was no longer working on his kernel.

I'll remove the -R and -U flags completely from hdparm-8.7.

Cheers

#25 lore raw

Re: [PATCH 15/18] ide: remove broken/dangerous HDIO_[UNREGISTER,SCAN]_HWIF ioctls

Quoted 5 lines · Mark Lord
Quoted 4 lines · Sergei Shtylyov
Quoted 2 lines · Bartlomiej Zolnierkiewicz

hdparm explicitely marks HDIO_[UNREGISTER,SCAN]_HWIF ioctls as DANGEROUS and given the number of bugs we can assume that there are no real users:

Quoted 8 lines

..

There is the odd user of these, actually.

But the most recent to email me (a few weeks ago), reported that the SCAN function was no longer working on his kernel.

I'll remove the -R and -U flags completely from hdparm-8.7.

This will be quite helpful, thanks!

Bart

#26 lore raw

Re: [PATCH 15/18] ide: remove broken/dangerous HDIO_[UNREGISTER,SCAN]_HWIF ioctls

In reply to Sergei Shtylyov

On Thursday 27 March 2008, Sergei Shtylyov wrote:

[...]

Quoted 3 lines
Quoted 1 line

Thanks for reviewing it.

Quoted 27 lines
Quoted 4 lines

Index: b/drivers/ide/ide-pnp.c
===================================================================

drivers/ide/ide-pnp.c +0 -0

[...]

Quoted 22 lines
drivers/ide/ide-pnp.c +0 -18
@@ -655,52 +530,6 @@ void ide_init_port_hw(ide_hwif_t *hwif, 
 }
 EXPORT_SYMBOL_GPL(ide_init_port_hw);
 
-/**
- *	ide_register_hw		-	register IDE interface
- *	@hw: hardware registers
- *	@quirkproc: quirkproc function
- *	@hwifp: pointer to returned hwif
- *
- *	Register an IDE interface, specifying exactly the registers etc.
- *
- *	Returns -1 on error.
- */
-
-static int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
-			   ide_hwif_t **hwifp)
-{
-	int index, retry = 1;
-	ide_hwif_t *hwif;
-	u8 idx[4] = { 0xff, 0xff, 0xff, 0xff };
-
Quoted 8 lines

drivers/ide/ide-pnp.c void ide_init_port_hw(ide_hwif_t *hwif, · line 680

Quoted 5 lines
drivers/ide/ide-pnp.c +0 -5
-	do {
-		hwif = ide_find_port(hw->io_ports[IDE_DATA_OFFSET]);
-		index = hwif->index;
-		if (hwif)
-			goto found;

Hm, I remember there was a patch that fixed the above bug where hwif is
dereferenced before being checked for NULL, I wonder how come it was lost?

It has been already merged into Linus' tree (commit 0c6025d44448bd688dfd351a09bc620aafa4d1ff).

Thanks,
Bart

#27 lore raw

PATCH 16/18 ide: remove ->hold field from ide_hwif_t

In reply to Bartlomiej Zolnierkiewicz

->hold is write-only now, remove it.

---
 drivers/ide/mips/au1xxx-ide.c |    3 ---
 drivers/ide/ppc/pmac.c        |    1 -
 include/linux/ide.h           |    1 -
 3 files changed, 5 deletions(-)
drivers/ide/mips/au1xxx-ide.c +0 -3
Index: b/drivers/ide/mips/au1xxx-ide.c
===================================================================
@@ -617,9 +617,6 @@ static int au_ide_probe(struct device *d
 
 	hwif->dev = dev;
 
-	/* hold should be on in all cases */
-	hwif->hold                      = 1;
-
 	hwif->mmio  = 1;
 
 	/* If the user has selected DDMA assisted copies,
drivers/ide/ppc/pmac.c +0 -1
Index: b/drivers/ide/ppc/pmac.c
===================================================================
@@ -1120,7 +1120,6 @@ pmac_ide_setup_device(pmac_ide_hwif_t *p
 	hwif->hwif_data = pmif;
 	ide_init_port_hw(hwif, hw);
 	hwif->noprobe = pmif->mediabay;
-	hwif->hold = pmif->mediabay;
 	hwif->cbl = pmif->cable_80 ? ATA_CBL_PATA80 : ATA_CBL_PATA40;
 	hwif->set_pio_mode = pmac_ide_set_pio_mode;
 	if (pmif->kind == controller_un_ata6
include/linux/ide.h +0 -1
Index: b/include/linux/ide.h
===================================================================
@@ -573,7 +573,6 @@ typedef struct hwif_s {
 
 	unsigned	noprobe    : 1;	/* don't probe for this interface */
 	unsigned	present    : 1;	/* this interface exists */
-	unsigned	hold       : 1; /* this interface is always present */
 	unsigned	serialized : 1;	/* serialized all channel operation */
 	unsigned	sharing_irq: 1;	/* 1 = sharing irq with another hwif */
 	unsigned	reset      : 1;	/* reset after probe */
#28 lore raw

PATCH 17/18 ide: remove init_hwif_default()

In reply to Bartlomiej Zolnierkiewicz

init_hwif_default() is only used by init_ide_data() now, inline it there.

---
 drivers/ide/ide.c |   25 +++++++++----------------
 1 file changed, 9 insertions(+), 16 deletions(-)
drivers/ide/ide.c +9 -16
Index: b/drivers/ide/ide.c
===================================================================
@@ -168,19 +168,6 @@ static void ide_port_init_devices_data(i
 }
 
 
-static void init_hwif_default(ide_hwif_t *hwif, unsigned int index)
-{
-	hw_regs_t hw;
-
-	memset(&hw, 0, sizeof(hw_regs_t));
-
-	ide_init_hwif_ports(&hw, ide_default_io_base(index), 0, &hwif->irq);
-
-	memcpy(hwif->io_ports, hw.io_ports, sizeof(hw.io_ports));
-
-	hwif->noprobe = !hwif->io_ports[IDE_DATA_OFFSET];
-}
-
 /*
  * init_ide_data() sets reasonable default values into all fields
  * of all instances of the hwifs and drives, but only on the first call.
@@ -201,9 +188,9 @@ static void init_hwif_default(ide_hwif_t
 #define MAGIC_COOKIE 0x12345678
 static void __init init_ide_data (void)
 {
-	ide_hwif_t *hwif;
 	unsigned int index;
 	static unsigned long magic_cookie = MAGIC_COOKIE;
+	hw_regs_t hw;
 
 	if (magic_cookie != MAGIC_COOKIE)
 		return;		/* already initialized */
@@ -211,9 +198,15 @@ static void __init init_ide_data (void)
 
 	/* Initialise all interface structures */
 	for (index = 0; index < MAX_HWIFS; ++index) {
-		hwif = &ide_hwifs[index];
+		ide_hwif_t *hwif = &ide_hwifs[index];
+
 		ide_init_port_data(hwif, index);
-		init_hwif_default(hwif, index);
+
+		memset(&hw, 0, sizeof(hw));
+		ide_init_hwif_ports(&hw, ide_default_io_base(index), 0,
+				    &hwif->irq);
+		memcpy(hwif->io_ports, hw.io_ports, sizeof(hw.io_ports));
+		hwif->noprobe = !hwif->io_ports[IDE_DATA_OFFSET];
 #if !defined(CONFIG_PPC32) || !defined(CONFIG_PCI)
 		hwif->irq =
 			ide_init_default_irq(hwif->io_ports[IDE_DATA_OFFSET]);
#29 lore raw

PATCH 18/18 ide: remove ide_init_hwif_ports()

In reply to Bartlomiej Zolnierkiewicz

ide_init_hwif_ports() is only used by init_ide_data() now, inline it there.

---
 drivers/ide/ide.c   |   11 +++++++++--
 include/linux/ide.h |   32 --------------------------------
 2 files changed, 9 insertions(+), 34 deletions(-)
drivers/ide/ide.c +9 -2
Index: b/drivers/ide/ide.c
===================================================================
@@ -199,13 +199,20 @@ static void __init init_ide_data (void)
 	/* Initialise all interface structures */
 	for (index = 0; index < MAX_HWIFS; ++index) {
 		ide_hwif_t *hwif = &ide_hwifs[index];
+		unsigned long io_addr = ide_default_io_base(index);
+		unsigned long ctl_addr = ide_default_io_ctl(io_addr);
 
 		ide_init_port_data(hwif, index);
 
+#ifdef CONFIG_IDE_ARCH_OBSOLETE_INIT
 		memset(&hw, 0, sizeof(hw));
-		ide_init_hwif_ports(&hw, ide_default_io_base(index), 0,
-				    &hwif->irq);
+		ide_std_init_ports(&hw, io_addr, ctl_addr);
+# ifdef CONFIG_PPC32
+		if (ppc_ide_md.ide_init_hwif)
+			ppc_ide_md.ide_init_hwif(&hw, io_addr, 0, &hwif->irq);
+# endif
 		memcpy(hwif->io_ports, hw.io_ports, sizeof(hw.io_ports));
+#endif
 		hwif->noprobe = !hwif->io_ports[IDE_DATA_OFFSET];
 #if !defined(CONFIG_PPC32) || !defined(CONFIG_PCI)
 		hwif->irq =
include/linux/ide.h +0 -32
Index: b/include/linux/ide.h
===================================================================
@@ -215,38 +215,6 @@ static inline void ide_std_init_ports(hw
 # define ide_init_default_irq(base)	(0)
 #endif
 
-#ifdef CONFIG_IDE_ARCH_OBSOLETE_INIT
-static inline void ide_init_hwif_ports(hw_regs_t *hw,
-				       unsigned long io_addr,
-				       unsigned long ctl_addr,
-				       int *irq)
-{
-	if (!ctl_addr)
-		ide_std_init_ports(hw, io_addr, ide_default_io_ctl(io_addr));
-	else
-		ide_std_init_ports(hw, io_addr, ctl_addr);
-
-	if (irq)
-		*irq = 0;
-
-	hw->io_ports[IDE_IRQ_OFFSET] = 0;
-
-#ifdef CONFIG_PPC32
-	if (ppc_ide_md.ide_init_hwif)
-		ppc_ide_md.ide_init_hwif(hw, io_addr, ctl_addr, irq);
-#endif
-}
-#else
-static inline void ide_init_hwif_ports(hw_regs_t *hw,
-				       unsigned long io_addr,
-				       unsigned long ctl_addr,
-				       int *irq)
-{
-	if (io_addr || ctl_addr)
-		printk(KERN_WARNING "%s: must not be called\n", __FUNCTION__);
-}
-#endif /* CONFIG_IDE_ARCH_OBSOLETE_INIT */
-
 /* Currently only m68k, apus and m8xx need it */
 #ifndef IDE_ARCH_ACK_INTR
 # define ide_ack_intr(hwif) (1)
#30 lore raw
In reply to Bartlomiej Zolnierkiewicz
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

I'll have a look & test next week hopefully.

Ben.

#31 lore raw
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

Well, I have serious problem on a Pegasos which appeared some time between 2.6.24 and 2.6.25-rc1: at boot I get an infinite string of

hdb: empty DMA table?

I'm trying to bisect it right now.

Gabriel

Quoted 10 lines

I'll have a look & test next week hopefully.

Ben.

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://ozlabs.org/mailman/listinfo/linuxppc-dev

#32 lore raw
Quoted 11 lines · Gabriel Paubert
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

Quoted 6 lines

Well, I have serious problem on a Pegasos which appeared some time between 2.6.24 and 2.6.25-rc1: at boot I get an infinite string of

hdb: empty DMA table?

I'm trying to bisect it right now.

Argh, the first bisect point ended up with timeouts on hdb...

Flagged as bad, to try to see when the problems started, but I suspect that there are several.

Gabriel

#33 lore raw
Quoted 15 lines · Gabriel Paubert
Quoted 14 lines · Gabriel Paubert
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

Well, I have serious problem on a Pegasos which appeared some time between 2.6.24 and 2.6.25-rc1: at boot I get an infinite string of

Quoted 8 lines
Quoted 3 lines

hdb: empty DMA table?

I'm trying to bisect it right now.

Argh, the first bisect point ended up with timeouts on hdb...

Flagged as bad, to try to see when the problems started, but I suspect that there are several.

Well, it's worse than this. After 7 or 8 bisections (all flagged bad) the symptoms changed: the system boots up to the point where it does not recognize LVM volumes. So there are at least 3 bugs:
1) the timeout on interrupts (only seen on hdb)
2) the empty DMA table messages (seen on hda and hdb)
3) the inability to see logical volumes ("pvs" does not find them, they are back when rebooting into 2.6.24). This is the apparently the earliest introduced bug (no problems with the disks with that version).

I am away from that machine until next Tuesday. IIRC the last git bisect was something like 160 revisions left.

Gabriel

Quoted 6 lines

Gabriel
_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://ozlabs.org/mailman/listinfo/linuxppc-dev

#34 lore raw
In reply to Gabriel Paubert
Quoted 11 lines · Gabriel Paubert
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

Quoted 6 lines

Well, I have serious problem on a Pegasos which appeared some time between 2.6.24 and 2.6.25-rc1: at boot I get an infinite string of

hdb: empty DMA table?

I'm trying to bisect it right now.

Pegasos doesn't use the pmac ide which is what we were discussing. It uses a VIA IDE which uses a normal PRD table. So something else must have broken...

Ben.

#35 lore raw
In reply to Benjamin Herrenschmidt
Quoted 10 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

I'll have a look & test next week hopefully.

Also, the above would have the advantage of not relying on drivers/ide infrastructure, and thus working with libata (once somebody has ported pmac ide to libata).

Ben.

#36 lore raw
Quoted 8 lines · Benjamin Herrenschmidt
Quoted 6 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Quoted 8 lines
Quoted 4 lines

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

I'll have a look & test next week hopefully.

Also, the above would have the advantage of not relying on drivers/ide infrastructure, and thus working with libata (once somebody has ported pmac ide to libata).

Neither of these things exist at the moment so lets stick to the existing code which is scheduled for 2.6.26 and which finally allows removal of ppc specific ide hacks from arch/{powerpc,ppc} (500 LOC gone, patches to be posted this week), not to mention other nice changes in the future...

Thanks,
Bart

#37 lore raw
In reply to Benjamin Herrenschmidt
Quoted 8 lines · On Wed, 13 Feb 2008 08:04:07 +1100 Benjamin Herrenschmidt…
Quoted 6 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Quoted 8 lines
Quoted 4 lines

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

I'll have a look & test next week hopefully.

Also, the above would have the advantage of not relying on drivers/ide infrastructure, and thus working with libata (once somebody has ported pmac ide to libata).

Unfortunately you need a degree in dentistry to open a Macintosh up and fix it otherwise we would have support by now.

Alan

#38 lore raw
Quoted 14 lines · Alan Cox
Quoted 11 lines · On Wed, 13 Feb 2008 08:04:07 +1100 Benjamin Herrenschmidt…
Quoted 9 lines · Benjamin Herrenschmidt
Quoted 4 lines · Bartlomiej Zolnierkiewicz

- couple of fixes and preparatory patches

- rework of PowerMac media-bay support ([un]register IDE devices instead of [un]registering IDE interface) [ it is the main reason for spamming PPC ML ]

Interesting... I was thinking about doing a full remove of the device at a higher level instead but I suppose what you propose is easier.

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

I'll have a look & test next week hopefully.

Also, the above would have the advantage of not relying on drivers/ide infrastructure, and thus working with libata (once somebody has ported pmac ide to libata).

Unfortunately you need a degree in dentistry to open a Macintosh up and fix it otherwise we would have support by now.

Heh :-)

Recent powermacs are trivial to open ! But yeah, I do need to produce a driver for libata one of these days. On my todo list ...

Cheers,
Ben.