Skip to contentSkip to conversation

libata fixes

PATCH In review Jeff Garzik 3 participants 5 messages Jan 27, 2009

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 libata fixes

The sata_mv 8-port fix was for a longstanding and ugly bug, for those who owned 8-port cards (4-port cards unaffected).

Please pull from 'upstream-linus' branch of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev.git upstream-linus

to receive the following updates:

 drivers/ata/Kconfig              |    6 ++--
 drivers/ata/libata-sff.c         |   12 ++++++-
 drivers/ata/pata_rb532_cf.c      |    2 -
 drivers/ata/pata_via.c           |   22 +++++++++---
 drivers/ata/sata_mv.c            |   56 ++++++++++++++----------------
 drivers/ata/sata_nv.c            |   70 +++++++++++++++++++++++--------------
 include/linux/pci_ids.h          |    4 ++
 7 files changed, 104 insertions(+), 68 deletions(-)
 delete mode 100644 arch/arm/mach-integrator/clock.h

[email protected] (1):

      [libata] pata_via: support VX855, future chips whose IDE controller use 0x0571

Mark Lord (5):
sata_mv: fix 8-port timeouts on 508x/6081 chips
sata_mv: don't read hc_irq_cause
sata_mv: remove bogus nsect restriction
sata_mv: msi masking fix (v2)
sata_mv: no longer experimental (v2)

Phil Sutter (1):
pata-rb532-cf: remove set_irq_type from finish_io

Tejun Heo (4):
libata-sff: fix incorrect EH message
libata: set NODEV_HINT for 0x7f status
sata_nv: rename nv_nf2_hardreset()
sata_nv: fix MCP5x reset

Thomas Reitmayr (1):
sata_mv: Properly initialize main irq mask

arch/arm/mach-integrator/clock.h deleted +0 -0
deleted file mode 100644
index e69de29..0000000
drivers/ata/Kconfig +3 -3
index 503a908..0bcf264 100644
@@ -112,11 +112,11 @@ config ATA_PIIX
 	  If unsure, say N.
 
 config SATA_MV
-	tristate "Marvell SATA support (HIGHLY EXPERIMENTAL)"
-	depends on EXPERIMENTAL
+	tristate "Marvell SATA support"
 	help
 	  This option enables support for the Marvell Serial ATA family.
-	  Currently supports 88SX[56]0[48][01] chips.
+	  Currently supports 88SX[56]0[48][01] PCI(-X) chips,
+	  as well as the newer [67]042 PCI-X/PCIe and SOC devices.
 
 	  If unsure, say N.
 
drivers/ata/libata-sff.c +11 -1
index 5a4aad1..0b299b0 100644
@@ -1322,7 +1322,7 @@ fsm_start:
 					 * condition.  Mark hint.
 					 */
 					ata_ehi_push_desc(ehi, "ST-ATA: "
-						"DRQ=1 with device error, "
+						"DRQ=0 without device error, "
 						"dev_stat 0x%X", status);
 					qc->err_mask |= AC_ERR_HSM |
 							AC_ERR_NODEV_HINT;
@@ -1358,6 +1358,16 @@ fsm_start:
 					qc->err_mask |= AC_ERR_HSM;
 				}
 
+				/* There are oddball controllers with
+				 * status register stuck at 0x7f and
+				 * lbal/m/h at zero which makes it
+				 * pass all other presence detection
+				 * mechanisms we have.  Set NODEV_HINT
+				 * for it.  Kernel bz#7241.
+				 */
+				if (status == 0x7f)
+					qc->err_mask |= AC_ERR_NODEV_HINT;
+
 				/* ata_pio_sectors() might change the
 				 * state to HSM_ST_LAST. so, the state
 				 * is changed after ata_pio_sectors().
drivers/ata/pata_rb532_cf.c +0 -2
index c2e6fb9..ebfcda2 100644
@@ -63,8 +63,6 @@ static inline void rb532_pata_finish_io(struct ata_port *ap)
 	   ata_sff_sync might be sufficient. */
 	ata_sff_dma_pause(ap);
 	ndelay(RB500_CF_IO_DELAY);
-
-	set_irq_type(info->irq, IRQ_TYPE_LEVEL_HIGH);
 }
 
 static void rb532_pata_exec_command(struct ata_port *ap,
drivers/ata/pata_via.c +17 -5
index 681169c..79a6c9a 100644
@@ -86,6 +86,10 @@ enum {
 	VIA_SATA_PATA	= 0x800, /* SATA/PATA combined configuration */
 };
 
+enum {
+	VIA_IDFLAG_SINGLE = (1 << 0), /* single channel controller) */
+};
+
 /*
  * VIA SouthBridge chips.
  */
@@ -97,8 +101,12 @@ static const struct via_isa_bridge {
 	u8 rev_max;
 	u16 flags;
 } via_isa_bridges[] = {
+	{ "vx855",	PCI_DEVICE_ID_VIA_VX855,    0x00, 0x2f,
+	  VIA_UDMA_133 | VIA_BAD_AST | VIA_SATA_PATA },
 	{ "vx800",	PCI_DEVICE_ID_VIA_VX800,    0x00, 0x2f, VIA_UDMA_133 |
 	VIA_BAD_AST | VIA_SATA_PATA },
+	{ "vt8261",	PCI_DEVICE_ID_VIA_8261,     0x00, 0x2f,
+	  VIA_UDMA_133 | VIA_BAD_AST },
 	{ "vt8237s",	PCI_DEVICE_ID_VIA_8237S,    0x00, 0x2f, VIA_UDMA_133 | VIA_BAD_AST },
 	{ "vt8251",	PCI_DEVICE_ID_VIA_8251,     0x00, 0x2f, VIA_UDMA_133 | VIA_BAD_AST },
 	{ "cx700",	PCI_DEVICE_ID_VIA_CX700,    0x00, 0x2f, VIA_UDMA_133 | VIA_BAD_AST | VIA_SATA_PATA },
@@ -122,6 +130,8 @@ static const struct via_isa_bridge {
 	{ "vt82c586",	PCI_DEVICE_ID_VIA_82C586_0, 0x00, 0x0f, VIA_UDMA_NONE | VIA_SET_FIFO },
 	{ "vt82c576",	PCI_DEVICE_ID_VIA_82C576,   0x00, 0x2f, VIA_UDMA_NONE | VIA_SET_FIFO | VIA_NO_UNMASK },
 	{ "vt82c576",	PCI_DEVICE_ID_VIA_82C576,   0x00, 0x2f, VIA_UDMA_NONE | VIA_SET_FIFO | VIA_NO_UNMASK | VIA_BAD_ID },
+	{ "vtxxxx",	PCI_DEVICE_ID_VIA_ANON,    0x00, 0x2f,
+	  VIA_UDMA_133 | VIA_BAD_AST },
 	{ NULL }
 };
 
@@ -460,6 +470,7 @@ static int via_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	static int printed_version;
 	u8 enable;
 	u32 timing;
+	unsigned long flags = id->driver_data;
 	int rc;
 
 	if (!printed_version++)
@@ -469,9 +480,13 @@ static int via_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	if (rc)
 		return rc;
 
+	if (flags & VIA_IDFLAG_SINGLE)
+		ppi[1] = &ata_dummy_port_info;
+
 	/* To find out how the IDE will behave and what features we
 	   actually have to look at the bridge not the IDE controller */
-	for (config = via_isa_bridges; config->id; config++)
+	for (config = via_isa_bridges; config->id != PCI_DEVICE_ID_VIA_ANON;
+	     config++)
 		if ((isa = pci_get_device(PCI_VENDOR_ID_VIA +
 			!!(config->flags & VIA_BAD_ID),
 			config->id, NULL))) {
@@ -482,10 +497,6 @@ static int via_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 			pci_dev_put(isa);
 		}
 
-	if (!config->id) {
-		printk(KERN_WARNING "via: Unknown VIA SouthBridge, disabling.\n");
-		return -ENODEV;
-	}
 	pci_dev_put(isa);
 
 	if (!(config->flags & VIA_NO_ENABLES)) {
@@ -587,6 +598,7 @@ static const struct pci_device_id via[] = {
 	{ PCI_VDEVICE(VIA, 0x1571), },
 	{ PCI_VDEVICE(VIA, 0x3164), },
 	{ PCI_VDEVICE(VIA, 0x5324), },
+	{ PCI_VDEVICE(VIA, 0xC409), VIA_IDFLAG_SINGLE },
 
 	{ },
 };
drivers/ata/sata_mv.c +26 -30

135 lines not on this page · raw message

drivers/ata/sata_nv.c +43 -27

127 lines not on this page · raw message

include/linux/pci_ids.h +4 -0
index d56ad9c..febc10e 100644
@@ -1357,6 +1357,7 @@
 #define PCI_DEVICE_ID_VIA_8783_0	0x3208
 #define PCI_DEVICE_ID_VIA_8237		0x3227
 #define PCI_DEVICE_ID_VIA_8251		0x3287
+#define PCI_DEVICE_ID_VIA_8261		0x3402
 #define PCI_DEVICE_ID_VIA_8237A		0x3337
 #define PCI_DEVICE_ID_VIA_8237S		0x3372
 #define PCI_DEVICE_ID_VIA_SATA_EIDE	0x5324
@@ -1366,10 +1367,13 @@
 #define PCI_DEVICE_ID_VIA_CX700		0x8324
 #define PCI_DEVICE_ID_VIA_CX700_IDE	0x0581
 #define PCI_DEVICE_ID_VIA_VX800		0x8353
+#define PCI_DEVICE_ID_VIA_VX855		0x8409
 #define PCI_DEVICE_ID_VIA_8371_1	0x8391
 #define PCI_DEVICE_ID_VIA_82C598_1	0x8598
 #define PCI_DEVICE_ID_VIA_838X_1	0xB188
 #define PCI_DEVICE_ID_VIA_83_87XX_1	0xB198
+#define PCI_DEVICE_ID_VIA_C409_IDE	0XC409
+#define PCI_DEVICE_ID_VIA_ANON		0xFFFF
 
 #define PCI_VENDOR_ID_SIEMENS           0x110A
 #define PCI_DEVICE_ID_SIEMENS_DSCC4     0x2102
#1 lore raw
Quoted 6 lines · Jeff Garzik

Please pull from 'upstream-linus' branch of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev.git upstream-linus

to receive the following updates:

Quoted 9 lines
 drivers/ata/Kconfig              |    6 ++--
 drivers/ata/libata-sff.c         |   12 ++++++-
 drivers/ata/pata_rb532_cf.c      |    2 -
 drivers/ata/pata_via.c           |   22 +++++++++---
 drivers/ata/sata_mv.c            |   56 ++++++++++++++----------------
 drivers/ata/sata_nv.c            |   70 +++++++++++++++++++++++--------------
 include/linux/pci_ids.h          |    4 ++
 7 files changed, 104 insertions(+), 68 deletions(-)
 delete mode 100644 arch/arm/mach-integrator/clock.h

Hmm. That delete of the empty file is "correct" (and probably came as a result of somebody doing "make distclean" which deletes empty files), but should not have been committed as part of "libata-sff: fix incorrect EH message".

Seems to be a mistake, and a small one, but still.. Be more careful.

Linus

#2 lore raw
Quoted 17 lines · Linus Torvalds
Quoted 5 lines · Jeff Garzik

Please pull from 'upstream-linus' branch of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev.git upstream-linus

to receive the following updates:

Quoted 9 lines
 drivers/ata/Kconfig              |    6 ++--
 drivers/ata/libata-sff.c         |   12 ++++++-
 drivers/ata/pata_rb532_cf.c      |    2 -
 drivers/ata/pata_via.c           |   22 +++++++++---
 drivers/ata/sata_mv.c            |   56 ++++++++++++++----------------
 drivers/ata/sata_nv.c            |   70 +++++++++++++++++++++++--------------
 include/linux/pci_ids.h          |    4 ++
 7 files changed, 104 insertions(+), 68 deletions(-)
 delete mode 100644 arch/arm/mach-integrator/clock.h
Quoted 3 lines

Hmm. That delete of the empty file is "correct" (and probably came as a result of somebody doing "make distclean" which deletes empty files), but should not have been committed as part of "libata-sff: fix incorrect EH message".

Seems to be a mistake, and a small one, but still.. Be more careful.

Er, oops. Sorry about that. Yeah, I habitually do 'make distclean' as part of my kclean and kbuild standard workflow scripts.

I also habitually _avoid_ "git commit -a", so I am surprised that I actually made this mistake. I am 100% certain that I issued neither "git rm" nor "git update-index" commands on that file.

Granted, I should have been more careful, but I am also surprised this happened at all.

Jeff

#3 lore raw
Quoted 7 lines · Jeff Garzik

I also habitually _avoid_ "git commit -a", so I am surprised that I actually made this mistake. I am 100% certain that I issued neither "git rm" nor "git update-index" commands on that file.

Granted, I should have been more careful, but I am also surprised this happened at all.

Well, it's attributed to Tejun, and he probably sent you a "git diff". The delete part of such a diff is pretty small, so you both probably missed it at the top of the diff:

	diff --git a/arch/arm/mach-integrator/clock.h b/arch/arm/mach-integrator/clock.h
	deleted file mode 100644
	index e69de29..0000000

and then you used "git am -s" or something to apply it and delete that file.

No "git commit -a" necessary - just a simple "git diff" after having done a "make distclean" and then sending the result off by email without looking any closer..

Linus

#4 lore raw

Hello, Linus, Jeff.

Quoted 17 lines · Linus Torvalds
Quoted 6 lines · Jeff Garzik

I also habitually _avoid_ "git commit -a", so I am surprised that I actually made this mistake. I am 100% certain that I issued neither "git rm" nor "git update-index" commands on that file.

Granted, I should have been more careful, but I am also surprised this happened at all.

Well, it's attributed to Tejun, and he probably sent you a "git diff". The delete part of such a diff is pretty small, so you both probably missed it at the top of the diff:

	diff --git a/arch/arm/mach-integrator/clock.h b/arch/arm/mach-integrator/clock.h
	deleted file mode 100644
	index e69de29..0000000
Quoted 6 lines

and then you used "git am -s" or something to apply it and delete that file.

No "git commit -a" necessary - just a simple "git diff" after having done a "make distclean" and then sending the result off by email without looking any closer..

Yes, that patch was generated with "git diff" and distclean was done cuz I just began to use separate output directories for build. I've restored the empty file for several times for other patches but it slipped this time. My apologies. I'll be more careful.

Thanks.

Signature · 1 line
tejun