Skip to contentSkip to conversation

staging driver s626 clashes with philips SAA7146 media/dvb based cards

DISCUSSION Herton Ronaldo Krzesinski 5 participants 17 messages Jun 16, 2009 – Jun 18, 2009 1 message not in the archive Staging

lore
Share & tools
lore
#0 lore raw

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

--
[]'s
Herton

#1 lore raw
Quoted 7 lines · Herton Ronaldo Krzesinski

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

If you can propose the proper sub ids, or the needed checks, please send a patch.

thanks,

greg k-h

#2 lore raw

Em Terça-feira 16 Junho 2009, às 17:51:21, Greg KH escreveu:

Quoted 12 lines · Herton Ronaldo Krzesinski
Quoted 7 lines

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 6 lines
Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

If you can propose the proper sub ids, or the needed checks, please send a patch.

Can't propose proper sub ids here etc., as I don't know about/don't have s626 device, s626 author is CC'ed here to check this. But I could send a patch to disable just the build of s626 if acceptable/desired for the moment.

Quoted 4 lines

thanks,

greg k-h

--
[]'s
Herton

#3 lore raw
Quoted 15 lines · Herton Ronaldo Krzesinski

Em Terça-feira 16 Junho 2009, às 17:51:21, Greg KH escreveu:

Quoted 14 lines · Herton Ronaldo Krzesinski
Quoted 7 lines

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

Quoted 6 lines
Quoted 2 lines

If you can propose the proper sub ids, or the needed checks, please send a patch.

Can't propose proper sub ids here etc., as I don't know about/don't have s626 device, s626 author is CC'ed here to check this. But I could send a patch to disable just the build of s626 if acceptable/desired for the moment.

The Windows driver (<http://www.sensoray.com/downloads/sdk626.zip>) has this in the models section of the INF file:

%sx26.DeviceDesc%=sxdrv.Device,PCI\VEN_1131&DEV_7146&SUBSYS_02726000

And it looks like the correct device because this the strings section contains:

sx26.DeviceDesc= "Sensoray Model 626 Analog/Digital I/O"

Interpreting the above information gives us:

PCI Vendor ID = 0x1131
PCI Device ID = 0x7146
PCI Subvendor ID = 0x6000
PCI Subdevice ID = 0x0272 (626)

The Linux SDK for this board
(<http://www.sensoray.com/downloads/s626-1.0.1.tar.gz> has the same info in the s626core.h file.

Signature · 2 lines
-=( Ian Abbott @ MEV Ltd.    E-mail: <[email protected]>        )=-
-=( Tel: +44 (0)161 477 1898   FAX: +44 (0)161 718 3587         )=-
#4 lore raw

Em Quarta-feira 17 Junho 2009, às 09:26:00, Ian Abbott escreveu:

Quoted 35 lines
Quoted 15 lines · Herton Ronaldo Krzesinski

Em Terça-feira 16 Junho 2009, às 17:51:21, Greg KH escreveu:

Quoted 14 lines · Herton Ronaldo Krzesinski
Quoted 7 lines

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

Quoted 6 lines
Quoted 2 lines

If you can propose the proper sub ids, or the needed checks, please send a patch.

Can't propose proper sub ids here etc., as I don't know about/don't have s626 device, s626 author is CC'ed here to check this. But I could send a patch to disable just the build of s626 if acceptable/desired for the moment.

The Windows driver (<http://www.sensoray.com/downloads/sdk626.zip>) has this in the models section of the INF file:

%sx26.DeviceDesc%=sxdrv.Device,PCI\VEN_1131&DEV_7146&SUBSYS_02726000

And it looks like the correct device because this the strings section contains:

sx26.DeviceDesc= "Sensoray Model 626 Analog/Digital I/O"

Interpreting the above information gives us:

Quoted 8 lines

PCI Vendor ID = 0x1131
PCI Device ID = 0x7146
PCI Subvendor ID = 0x6000
PCI Subdevice ID = 0x0272 (626)

The Linux SDK for this board
(<http://www.sensoray.com/downloads/s626-1.0.1.tar.gz> has the same info in the s626core.h file.

Ok thanks. So lets limit s626 by its subvendor:subdevice id, patch follows:

Quoted 1 line

From 6f4d2430959a378ab754c5dbd3903fdcf33abe36 Mon Sep 17 00:00:00 2001

From: Herton Ronaldo Krzesinski <[email protected]>
Date: Wed, 17 Jun 2009 13:31:15 -0300
Subject: [PATCH] Staging: comedi: use subvendor:subdevice ids for its pci id device table

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)
drivers/staging/comedi/drivers/s626.c +6 -2
index 30dec9d..b4b7713 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * Note: always specify subvendor:subdevice ids in table below, to avoid
+ * clash with Philips SAA7146 media/dvb based cards which have same
+ * vendor:device ids as S626
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
Signature · 5 lines
1.6.3.2

--
[]'s
Herton
#5 lore raw
Quoted 88 lines · Herton Ronaldo Krzesinski

Em Quarta-feira 17 Junho 2009, às 09:26:00, Ian Abbott escreveu:

Quoted 33 lines
Quoted 12 lines · Herton Ronaldo Krzesinski

Em Terça-feira 16 Junho 2009, às 17:51:21, Greg KH escreveu:

Quoted 11 lines · Herton Ronaldo Krzesinski
Quoted 7 lines

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

If you can propose the proper sub ids, or the needed checks, please send a patch.

Can't propose proper sub ids here etc., as I don't know about/don't have s626 device, s626 author is CC'ed here to check this. But I could send a patch to disable just the build of s626 if acceptable/desired for the moment.

The Windows driver (<http://www.sensoray.com/downloads/sdk626.zip>) has this in the models section of the INF file:

%sx26.DeviceDesc%=sxdrv.Device,PCI\VEN_1131&DEV_7146&SUBSYS_02726000

And it looks like the correct device because this the strings section contains:

sx26.DeviceDesc= "Sensoray Model 626 Analog/Digital I/O"

Interpreting the above information gives us:

Quoted 8 lines

PCI Vendor ID = 0x1131
PCI Device ID = 0x7146
PCI Subvendor ID = 0x6000
PCI Subdevice ID = 0x0272 (626)

The Linux SDK for this board
(<http://www.sensoray.com/downloads/s626-1.0.1.tar.gz> has the same info in the s626core.h file.

Ok thanks. So lets limit s626 by its subvendor:subdevice id, patch follows:

From 6f4d2430959a378ab754c5dbd3903fdcf33abe36 Mon Sep 17 00:00:00 2001
From: Herton Ronaldo Krzesinski <[email protected]>
Date: Wed, 17 Jun 2009 13:31:15 -0300
Subject: [PATCH] Staging: comedi: use subvendor:subdevice ids for its pci id device table

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)
drivers/staging/comedi/drivers/s626.c +6 -3
index 30dec9d..b4b7713 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * Note: always specify subvendor:subdevice ids in table below, to avoid
+ * clash with Philips SAA7146 media/dvb based cards which have same
+ * vendor:device ids as S626
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
--
Quoted 1 line

1.6.3.2

There are also a couple of calls to pci_get_device() that need changing to pci_get_subsys() in the s626_attach() function (so it also might be worth #define'ing the subsys numbers to avoid repeating these magic numbers).

--
-=( Ian Abbott @ MEV Ltd. E-mail: <[email protected]> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-

#6 lore raw

Em Qua 17 Jun 2009, às 15:21:48, Ian Abbott escreveu:

Quoted 88 lines
Quoted 87 lines · Herton Ronaldo Krzesinski

Em Quarta-feira 17 Junho 2009, às 09:26:00, Ian Abbott escreveu:

Quoted 33 lines
Quoted 12 lines · Herton Ronaldo Krzesinski

Em Terça-feira 16 Junho 2009, às 17:51:21, Greg KH escreveu:

Quoted 11 lines · Herton Ronaldo Krzesinski
Quoted 7 lines

Hi,

The s626 (comedi) driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

So a PCI device that does different things has the same device ids? ick, stupid vendors...

Quoted 8 lines
Quoted 5 lines
Quoted 3 lines

Looks like s626 shouldn't claim all 1131:7146 devices, either by specifying specific subdevice/subvendor ids specific to s626 devices or doing additional checks in its probe/attach function.

If you can propose the proper sub ids, or the needed checks, please send a patch.

Can't propose proper sub ids here etc., as I don't know about/don't have s626 device, s626 author is CC'ed here to check this. But I could send a patch to disable just the build of s626 if acceptable/desired for the moment.

The Windows driver (<http://www.sensoray.com/downloads/sdk626.zip>) has this in the models section of the INF file:

%sx26.DeviceDesc%=sxdrv.Device,PCI\VEN_1131&DEV_7146&SUBSYS_02726000

And it looks like the correct device because this the strings section contains:

sx26.DeviceDesc= "Sensoray Model 626 Analog/Digital I/O"

Interpreting the above information gives us:

Quoted 8 lines

PCI Vendor ID = 0x1131
PCI Device ID = 0x7146
PCI Subvendor ID = 0x6000
PCI Subdevice ID = 0x0272 (626)

The Linux SDK for this board
(<http://www.sensoray.com/downloads/s626-1.0.1.tar.gz> has the same info in the s626core.h file.

Ok thanks. So lets limit s626 by its subvendor:subdevice id, patch follows:

From 6f4d2430959a378ab754c5dbd3903fdcf33abe36 Mon Sep 17 00:00:00 2001
From: Herton Ronaldo Krzesinski <[email protected]>
Date: Wed, 17 Jun 2009 13:31:15 -0300
Subject: [PATCH] Staging: comedi: use subvendor:subdevice ids for its pci id device table

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)
drivers/staging/comedi/drivers/s626.c +6 -2
index 30dec9d..b4b7713 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * Note: always specify subvendor:subdevice ids in table below, to avoid
+ * clash with Philips SAA7146 media/dvb based cards which have same
+ * vendor:device ids as S626
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
Quoted 7 lines
Quoted 2 lines

--
1.6.3.2

There are also a couple of calls to pci_get_device() that need changing to pci_get_subsys() in the s626_attach() function (so it also might be worth #define'ing the subsys numbers to avoid repeating these magic numbers).

Indeed, using pci_get_device() would still cause problems if s626 is manually modprobed. Check new patch below, it now uses pci_get_subsys and ensures we use subvendor or subdevice. I didn't removed the define, may be there could be more s626 cards available in future with same id? If not, it could be cleaned up and may be logic at s626_attach() simplified.

Quoted 1 line

From 6c83b8665da770c2d60fe692819107e3d0bc329d Mon Sep 17 00:00:00 2001

From: Herton Ronaldo Krzesinski <[email protected]>
Date: Wed, 17 Jun 2009 19:56:33 -0300
Subject: [PATCH v2] Staging: comedi (s626): always use subvendor:subdevice ids in pci id table

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, additionaly ensuring that subvendor or subdevice id is set in pci id table entries.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |   35 ++++++++++++++++++++------------
 1 files changed, 22 insertions(+), 13 deletions(-)
drivers/staging/comedi/drivers/s626.c +22 -13
index 30dec9d..3ec4407 100644
@@ -110,9 +110,9 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/* sub pci id must be specified, see s626_attach for more details */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,24 +498,33 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
 	resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
+	/*
+	 * Require also one of sub pci ids to be defined (see check below),
+	 * otherwise there will be a clash with Philips SAA7146 media/dvb
+	 * based cards (they have same vendor:device == 0x1131:0x7146 pair
+	 * as main S626 cards)
+	 */
+	for (ids = s626_pci_table;
+	     (ids->vendor && (ids->subvendor || ids->subdevice)) && !pdev;
+	     ids++) {
+		pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+				      ids->subdevice, NULL);
+		if (!pdev)
+			continue;
+
 		if (it->options[0] || it->options[1]) {
+			/* matches requested bus/slot */
 			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
-				/* matches requested bus/slot */
+			    PCI_SLOT(pdev->devfn) == it->options[1])
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
+			pci_dev_put(pdev);
+			pdev = NULL;
 		}
 	}
 	devpriv->pdev = pdev;
Signature · 1 line
1.6.3.2
Quoted 7 lines

--
-=( Ian Abbott @ MEV Ltd. E-mail: <[email protected]> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-

--
[]'s
Herton

#7 lore raw
Quoted 61 lines · Herton Ronaldo Krzesinski

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, additionaly ensuring that subvendor or subdevice id is set in pci id table entries.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |   35

++++++++++++++++++++------------ 1 files changed, 22 insertions(+), 13 deletions(-)

(quoted diff) +0 -0

b/drivers/staging/comedi/drivers/s626.c index 30dec9d..3ec4407 100644

drivers/staging/comedi/drivers/s626.c +20 -9
@@ -110,9 +110,9 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/* sub pci id must be specified, see s626_attach for more details */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,24 +498,33 @@ static int s626_attach(struct comedi_device *dev,
 struct comedi_devconfig *it) resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
+	/*
+	 * Require also one of sub pci ids to be defined (see check below),
+	 * otherwise there will be a clash with Philips SAA7146 media/dvb
+	 * based cards (they have same vendor:device == 0x1131:0x7146 pair
+	 * as main S626 cards)
+	 */
+	for (ids = s626_pci_table;
+	     (ids->vendor && (ids->subvendor || ids->subdevice)) && !pdev;
+	     ids++) {
+		pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+				      ids->subdevice, NULL);
+		if (!pdev)
+			continue;
+
 		if (it->options[0] || it->options[1]) {
+			/* matches requested bus/slot */
 			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
-				/* matches requested bus/slot */
+			    PCI_SLOT(pdev->devfn) == it->options[1])

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, · line 530

Quoted 10 lines
drivers/staging/comedi/drivers/s626.c +2 -4
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
+			pci_dev_put(pdev);
+			pdev = NULL;
 		}
 	}
 	devpriv->pdev = pdev;

This patch looks buggy. It's changing the logic beyond just checking for subvendor/subdevice ids.

#8 lore raw

Em Qua 17 Jun 2009, às 20:35:13, Frank Mori Hess escreveu:

Quoted 68 lines · Herton Ronaldo Krzesinski
Quoted 67 lines

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, additionaly ensuring that subvendor or subdevice id is set in pci id table entries.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |   35

++++++++++++++++++++------------ 1 files changed, 22 insertions(+), 13 deletions(-)

(quoted diff) +0 -0

b/drivers/staging/comedi/drivers/s626.c index 30dec9d..3ec4407 100644

drivers/staging/comedi/drivers/s626.c +21 -13
@@ -110,9 +110,9 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/* sub pci id must be specified, see s626_attach for more details */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,24 +498,33 @@ static int s626_attach(struct comedi_device *dev,
 struct comedi_devconfig *it) resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
+	/*
+	 * Require also one of sub pci ids to be defined (see check below),
+	 * otherwise there will be a clash with Philips SAA7146 media/dvb
+	 * based cards (they have same vendor:device == 0x1131:0x7146 pair
+	 * as main S626 cards)
+	 */
+	for (ids = s626_pci_table;
+	     (ids->vendor && (ids->subvendor || ids->subdevice)) && !pdev;
+	     ids++) {
+		pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+				      ids->subdevice, NULL);
+		if (!pdev)
+			continue;
+
 		if (it->options[0] || it->options[1]) {
+			/* matches requested bus/slot */
 			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
-				/* matches requested bus/slot */
+			    PCI_SLOT(pdev->devfn) == it->options[1])
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
+			pci_dev_put(pdev);
Quoted 9 lines

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it) · line 530

Quoted 4 lines
drivers/staging/comedi/drivers/s626.c +1 -0
+			pdev = NULL;
 		}
 	}
 	devpriv->pdev = pdev;

This patch looks buggy. It's changing the logic beyond just checking for subvendor/subdevice ids.

That's the intention here, so that it avoids someone adding a new pci id without specifying either subvendor or subdevice id for 0x1131:0x7146 boards, but yes there will be a problem if boards with vendor:id not equal to 0x1131:0x7146 appear in future, as you will be obliged to add subvendor:subdevice id even if not needed.

If not wanted and it gone too far, I can revert to use the same logic as pci_match_id, or just simplify this in case it's unlikely more s626 boards appear.

The current situation is ugly, comedi subsystem could have a better way to deal with hotplug and probe of devices, without you having to reimplement what pci subsystem functions already does.

--
[]'s
Herton

#9 lore raw

Em Qua 17 Jun 2009, às 21:05:56, Herton Ronaldo Krzesinski escreveu:

Quoted 85 lines

Em Qua 17 Jun 2009, às 20:35:13, Frank Mori Hess escreveu:

Quoted 68 lines · Herton Ronaldo Krzesinski
Quoted 67 lines

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, additionaly ensuring that subvendor or subdevice id is set in pci id table entries.

Reference: http://lkml.org/lkml/2009/6/16/552

---
 drivers/staging/comedi/drivers/s626.c |   35

++++++++++++++++++++------------ 1 files changed, 22 insertions(+), 13 deletions(-)

(quoted diff) +0 -0

b/drivers/staging/comedi/drivers/s626.c index 30dec9d..3ec4407 100644

drivers/staging/comedi/drivers/s626.c +21 -13
@@ -110,9 +110,9 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/* sub pci id must be specified, see s626_attach for more details */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,24 +498,33 @@ static int s626_attach(struct comedi_device *dev,
 struct comedi_devconfig *it) resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
+	/*
+	 * Require also one of sub pci ids to be defined (see check below),
+	 * otherwise there will be a clash with Philips SAA7146 media/dvb
+	 * based cards (they have same vendor:device == 0x1131:0x7146 pair
+	 * as main S626 cards)
+	 */
+	for (ids = s626_pci_table;
+	     (ids->vendor && (ids->subvendor || ids->subdevice)) && !pdev;
+	     ids++) {
+		pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+				      ids->subdevice, NULL);
+		if (!pdev)
+			continue;
+
 		if (it->options[0] || it->options[1]) {
+			/* matches requested bus/slot */
 			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
-				/* matches requested bus/slot */
+			    PCI_SLOT(pdev->devfn) == it->options[1])
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
+			pci_dev_put(pdev);
Quoted 9 lines

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it) · line 530

Quoted 4 lines
drivers/staging/comedi/drivers/s626.c +1 -0
+			pdev = NULL;
 		}
 	}
 	devpriv->pdev = pdev;

This patch looks buggy. It's changing the logic beyond just checking for subvendor/subdevice ids.

That's the intention here, so that it avoids someone adding a new pci id without specifying either subvendor or subdevice id for 0x1131:0x7146 boards, but yes there will be a problem if boards with vendor:id not equal to 0x1131:0x7146 appear in future, as you will be obliged to add subvendor:subdevice id even if not needed.

Quoted 7 lines

If not wanted and it gone too far, I can revert to use the same logic as pci_match_id, or just simplify this in case it's unlikely more s626 boards appear.

The current situation is ugly, comedi subsystem could have a better way to deal with hotplug and probe of devices, without you having to reimplement what pci subsystem functions already does.

Erm... forget that, you mean the "/* no bus/slot specified */" I removed. That previous logic to me didn't made much sense, why you must always request a particular bus/slot?

--
[]'s
Herton

A message that is not in the archive (sent off-list or not yet received): [email protected]

#11 lore raw

Em Qua 17 Jun 2009, às 21:21:51, Frank Mori Hess escreveu:

Quoted 11 lines · you
Quoted 9 lines
Quoted 3 lines

This patch looks buggy. It's changing the logic beyond just checking for subvendor/subdevice ids.

That's the intention here, so that it avoids someone adding a new pci id without specifying either subvendor or subdevice id for 0x1131:0x7146 boards, but yes there will be a problem if boards with vendor:id not equal to 0x1131:0x7146 appear in future, as you will be obliged to add subvendor:subdevice id even if not needed.

Quoted 4 lines

Your patch breaks configuration of the board unless the bus and slot are explicitly specified. Just make a minimal patch that replaces pci_get_device with pci_get_subsys and fixes the problem that was reported.

Hmm that's not what the patch does, it doesn't break configuration, keeps the same logic as before (I was wrong in my last email replying to myself), check it, if it->options[0] and it->options[1] isn't specified, the pdev is valid so the for loop exits (see !pdev check).

Quoted 16 lines
Quoted 8 lines

If not wanted and it gone too far, I can revert to use the same logic as pci_match_id, or just simplify this in case it's unlikely more s626 boards appear.

The current situation is ugly, comedi subsystem could have a better way to deal with hotplug and probe of devices, without you having to reimplement what pci subsystem functions already does.

Agreed, it's currently in a limbo between trying to support auto probing of devices and supporting comedi's old way of configuring hardware. But I don't anticipate you are going to refactor the comedi core and all the drivers, so just make a minimal patch that doesn't change any logic it doesn't need to.

#12 lore raw
Quoted 10 lines · Herton Ronaldo Krzesinski
Quoted 5 lines

Your patch breaks configuration of the board unless the bus and slot are explicitly specified. Just make a minimal patch that replaces pci_get_device with pci_get_subsys and fixes the problem that was reported.

Hmm that's not what the patch does, it doesn't break configuration, keeps the same logic as before (I was wrong in my last email replying to myself), check it, if it->options[0] and it->options[1] isn't specified, the pdev is valid so the for loop exits (see !pdev check).

Your right. However, it also turns a loop over pci devices into a loop over pci ids, which appears to break the case of multiple s626 boards where the bus/slot of the second s626 board is specified. If you're not willing to provide a minimal patch that just fixes the reported problem, just say so. It would have been less effort for me to do it myself than analyze what your changes are breaking.

#13 lore raw

Em Quinta-feira 18 Junho 2009, às 04:28:04, Frank Mori Hess escreveu:

Quoted 12 lines · Herton Ronaldo Krzesinski
Quoted 10 lines
Quoted 5 lines

Your patch breaks configuration of the board unless the bus and slot are explicitly specified. Just make a minimal patch that replaces pci_get_device with pci_get_subsys and fixes the problem that was reported.

Hmm that's not what the patch does, it doesn't break configuration, keeps the same logic as before (I was wrong in my last email replying to myself), check it, if it->options[0] and it->options[1] isn't specified, the pdev is valid so the for loop exits (see !pdev check).

Quoted 6 lines

Your right. However, it also turns a loop over pci devices into a loop over pci ids, which appears to break the case of multiple s626 boards where the bus/slot of the second s626 board is specified. If you're not willing to provide a minimal patch that just fixes the reported problem, just say so. It would have been less effort for me to do it myself than analyze what your changes are breaking.

That's part of reviewing process, I just wanted to enhance it in case more pci ids are added in the future along with the switch to pci_get_subsys, I don't know why you act like that, you don't want the code to be enhanced? Other comedi driver loop over ids, for example gsc_hpdi

And indeed what you say it's true, there is a bug in the patch now that you checked it out properly, it has a problem with this multiple s626 case, fixed that now. Pointing out the problem instead just saying it's broken helps :)

I also removed the obligation to add sub ids and re-factored the patch, that was too much and not good, just a simpler loop over pci id array is used now, and a comment was added in pci id list telling people that they should add subvendor:subdevice ids for boards with vendor:device == 0x1131:0x7146, in case new boards with this id appear.

Let me know if there is any problem remaining with this version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

drivers/staging/comedi/drivers/s626.c +20 -15
index 30dec9d..4210590 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
 	resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
-		if (it->options[0] || it->options[1]) {
-			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
+	for (i = 0; i < ARRAY_SIZE(s626_pci_table) && !pdev; i++) {
+		ids = &s626_pci_table[i];
+		do {
+			pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+					      ids->subdevice, pdev);
+
+			if ((it->options[0] || it->options[1]) && pdev) {
 				/* matches requested bus/slot */
+				if (pdev->bus->number == it->options[0] &&
+				    PCI_SLOT(pdev->devfn) == it->options[1])
+					break;
+			} else
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
-		}
+		} while (1);
 	}
 	devpriv->pdev = pdev;
 
Signature · 2 lines
[]'s
Herton
#14 lore raw
Quoted 71 lines · Herton Ronaldo Krzesinski

Let me know if there is any problem remaining with this version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

drivers/staging/comedi/drivers/s626.c +18 -10
index 30dec9d..4210590 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
        resource_size_t resourceStart;
        dma_addr_t appdma;
        struct comedi_subdevice *s;
-       struct pci_dev *pdev;
+       const struct pci_device_id *ids;
+       struct pci_dev *pdev = NULL;
 
        if (alloc_private(dev, sizeof(struct s626_private)) < 0)
                return -ENOMEM;
 
-       for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-                       NULL); pdev != NULL;
-               pdev = pci_get_device(PCI_VENDOR_ID_S626,
-                       PCI_DEVICE_ID_S626, pdev)) {
-               if (it->options[0] || it->options[1]) {
-                       if (pdev->bus->number == it->options[0] &&
-                               PCI_SLOT(pdev->devfn) == it->options[1]) {
+       for (i = 0; i < ARRAY_SIZE(s626_pci_table) && !pdev; i++) {
+               ids = &s626_pci_table[i];
+               do {
+                       pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+                                             ids->subdevice, pdev);
+
+                       if ((it->options[0] || it->options[1]) && pdev) {
                                /* matches requested bus/slot */
+                               if (pdev->bus->number == it->options[0] &&
+                                   PCI_SLOT(pdev->devfn) == it->options[1])
+                                       break;

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it) · line 526

Quoted 10 lines
drivers/staging/comedi/drivers/s626.c +2 -5
+                       } else
                                break;
-                       }
-               } else {
-                       /* no bus/slot specified */
-                       break;
-               }
+               } while (1);
        }
        devpriv->pdev = pdev;

The outer for loop iterates once too often - it doesn't need to iterate over the sentinel at the end of the id table as that shouldn't match any PCI device.

Signature · 2 lines
-=( Ian Abbott @ MEV Ltd.    E-mail: <[email protected]>        )=-
-=( Tel: +44 (0)161 477 1898   FAX: +44 (0)161 718 3587         )=-
#15 lore raw

Em Quinta-feira 18 Junho 2009, às 14:32:31, Ian Abbott escreveu:

Quoted 78 lines · Herton Ronaldo Krzesinski
Quoted 77 lines

Let me know if there is any problem remaining with this version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

drivers/staging/comedi/drivers/s626.c +19 -14
index 30dec9d..4210590 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
        resource_size_t resourceStart;
        dma_addr_t appdma;
        struct comedi_subdevice *s;
-       struct pci_dev *pdev;
+       const struct pci_device_id *ids;
+       struct pci_dev *pdev = NULL;
 
        if (alloc_private(dev, sizeof(struct s626_private)) < 0)
                return -ENOMEM;
 
-       for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-                       NULL); pdev != NULL;
-               pdev = pci_get_device(PCI_VENDOR_ID_S626,
-                       PCI_DEVICE_ID_S626, pdev)) {
-               if (it->options[0] || it->options[1]) {
-                       if (pdev->bus->number == it->options[0] &&
-                               PCI_SLOT(pdev->devfn) == it->options[1]) {
+       for (i = 0; i < ARRAY_SIZE(s626_pci_table) && !pdev; i++) {
+               ids = &s626_pci_table[i];
+               do {
+                       pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+                                             ids->subdevice, pdev);
+
+                       if ((it->options[0] || it->options[1]) && pdev) {
                                /* matches requested bus/slot */
+                               if (pdev->bus->number == it->options[0] &&
+                                   PCI_SLOT(pdev->devfn) == it->options[1])
+                                       break;
+                       } else
                                break;
-                       }
-               } else {
-                       /* no bus/slot specified */
-                       break;
Quoted 8 lines

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it) · line 526

Quoted 4 lines
drivers/staging/comedi/drivers/s626.c +1 -1
-               }
+               } while (1);
        }
        devpriv->pdev = pdev;

The outer for loop iterates once too often - it doesn't need to iterate over the sentinel at the end of the id table as that shouldn't match any PCI device.

Ouch, yep didn't saw that, new version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

drivers/staging/comedi/drivers/s626.c +20 -15
index 30dec9d..7bf2a79 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-		0},
+	{PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
 	{0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
 	resource_size_t resourceStart;
 	dma_addr_t appdma;
 	struct comedi_subdevice *s;
-	struct pci_dev *pdev;
+	const struct pci_device_id *ids;
+	struct pci_dev *pdev = NULL;
 
 	if (alloc_private(dev, sizeof(struct s626_private)) < 0)
 		return -ENOMEM;
 
-	for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-			NULL); pdev != NULL;
-		pdev = pci_get_device(PCI_VENDOR_ID_S626,
-			PCI_DEVICE_ID_S626, pdev)) {
-		if (it->options[0] || it->options[1]) {
-			if (pdev->bus->number == it->options[0] &&
-				PCI_SLOT(pdev->devfn) == it->options[1]) {
+	for (i = 0; i < (ARRAY_SIZE(s626_pci_table) - 1) && !pdev; i++) {
+		ids = &s626_pci_table[i];
+		do {
+			pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+					      ids->subdevice, pdev);
+
+			if ((it->options[0] || it->options[1]) && pdev) {
 				/* matches requested bus/slot */
+				if (pdev->bus->number == it->options[0] &&
+				    PCI_SLOT(pdev->devfn) == it->options[1])
+					break;
+			} else
 				break;
-			}
-		} else {
-			/* no bus/slot specified */
-			break;
-		}
+		} while (1);
 	}
 	devpriv->pdev = pdev;
 
Signature · 2 lines
[]'s
Herton
#16 lore raw
Quoted 101 lines · Herton Ronaldo Krzesinski

Em Quinta-feira 18 Junho 2009, às 14:32:31, Ian Abbott escreveu:

Quoted 77 lines · Herton Ronaldo Krzesinski
Quoted 76 lines

Let me know if there is any problem remaining with this version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

drivers/staging/comedi/drivers/s626.c +19 -13
index 30dec9d..4210590 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
        resource_size_t resourceStart;
        dma_addr_t appdma;
        struct comedi_subdevice *s;
-       struct pci_dev *pdev;
+       const struct pci_device_id *ids;
+       struct pci_dev *pdev = NULL;
 
        if (alloc_private(dev, sizeof(struct s626_private)) < 0)
                return -ENOMEM;
 
-       for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-                       NULL); pdev != NULL;
-               pdev = pci_get_device(PCI_VENDOR_ID_S626,
-                       PCI_DEVICE_ID_S626, pdev)) {
-               if (it->options[0] || it->options[1]) {
-                       if (pdev->bus->number == it->options[0] &&
-                               PCI_SLOT(pdev->devfn) == it->options[1]) {
+       for (i = 0; i < ARRAY_SIZE(s626_pci_table) && !pdev; i++) {
+               ids = &s626_pci_table[i];
+               do {
+                       pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+                                             ids->subdevice, pdev);
+
+                       if ((it->options[0] || it->options[1]) && pdev) {
                                /* matches requested bus/slot */
+                               if (pdev->bus->number == it->options[0] &&
+                                   PCI_SLOT(pdev->devfn) == it->options[1])
+                                       break;
+                       } else
                                break;
-                       }
-               } else {
-                       /* no bus/slot specified */
Quoted 8 lines

drivers/staging/comedi/drivers/s626.c static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it) · line 526

Quoted 5 lines
drivers/staging/comedi/drivers/s626.c +1 -2
-                       break;
-               }
+               } while (1);
        }
        devpriv->pdev = pdev;

The outer for loop iterates once too often - it doesn't need to iterate over the sentinel at the end of the id table as that shouldn't match any PCI device.

Ouch, yep didn't saw that, new version:

Staging: comedi: s626: use subvendor:subdevice ids for SAA7146 board

The current s626 comedi driver in staging conflicts with philips SAA7146 media/dvb based cards, because it claims the same vendor:device pci id for all subdevice/subvendor ids. What happens is that for people that have a philips SAA7146 media/dvb based card, s626 if available gets loaded by udev and makes system freeze (https://qa.mandriva.com/show_bug.cgi?id=51445).

The s626 driver shouldn't claim all 1131:7146 devices. Fix this by specifying specific known subvendor:subdevice ids in its pci id table list.

Quoted 7 lines

Also s626_attach is modified to use now pci_get_subsys instead of pci_get_device as reported by Ian Abbott, and now we loop over pci id table entries in case more ids are added in the future.

Reference: http://lkml.org/lkml/2009/6/16/552

This looks fine now, though I think it's unlikely this driver will support more than one subdevice ID.

Quoted 66 lines
drivers/staging/comedi/drivers/s626.c +20 -16
index 30dec9d..7bf2a79 100644
@@ -110,9 +110,13 @@ static const struct s626_board s626_boards[] = {
 #define PCI_VENDOR_ID_S626 0x1131
 #define PCI_DEVICE_ID_S626 0x7146
 
+/*
+ * For devices with vendor:device id == 0x1131:0x7146 you must specify
+ * also subvendor:subdevice ids, because otherwise it will conflict with
+ * Philips SAA7146 media/dvb based cards.
+ */
 static DEFINE_PCI_DEVICE_TABLE(s626_pci_table) = {
-       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, PCI_ANY_ID, PCI_ANY_ID, 0, 0,
-               0},
+       {PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626, 0x6000, 0x0272, 0, 0, 0},
        {0}
 };
 
@@ -498,25 +502,26 @@ static int s626_attach(struct comedi_device *dev, struct comedi_devconfig *it)
        resource_size_t resourceStart;
        dma_addr_t appdma;
        struct comedi_subdevice *s;
-       struct pci_dev *pdev;
+       const struct pci_device_id *ids;
+       struct pci_dev *pdev = NULL;
 
        if (alloc_private(dev, sizeof(struct s626_private)) < 0)
                return -ENOMEM;
 
-       for (pdev = pci_get_device(PCI_VENDOR_ID_S626, PCI_DEVICE_ID_S626,
-                       NULL); pdev != NULL;
-               pdev = pci_get_device(PCI_VENDOR_ID_S626,
-                       PCI_DEVICE_ID_S626, pdev)) {
-               if (it->options[0] || it->options[1]) {
-                       if (pdev->bus->number == it->options[0] &&
-                               PCI_SLOT(pdev->devfn) == it->options[1]) {
+       for (i = 0; i < (ARRAY_SIZE(s626_pci_table) - 1) && !pdev; i++) {
+               ids = &s626_pci_table[i];
+               do {
+                       pdev = pci_get_subsys(ids->vendor, ids->device, ids->subvendor,
+                                             ids->subdevice, pdev);
+
+                       if ((it->options[0] || it->options[1]) && pdev) {
                                /* matches requested bus/slot */
+                               if (pdev->bus->number == it->options[0] &&
+                                   PCI_SLOT(pdev->devfn) == it->options[1])
+                                       break;
+                       } else
                                break;
-                       }
-               } else {
-                       /* no bus/slot specified */
-                       break;
-               }
+               } while (1);
        }
        devpriv->pdev = pdev;
 
 
 
--

[]'s
Herton

Signature · 2 lines
-=( Ian Abbott @ MEV Ltd.    E-mail: <[email protected]>        )=-
-=( Tel: +44 (0)161 477 1898   FAX: +44 (0)161 718 3587         )=-
#17 lore raw
In reply to Herton Ronaldo Krzesinski

Hi Herton,

Quoted 29 lines · Herton Ronaldo Krzesinski

Em Quinta-feira 18 Junho 2009, às 04:28:04, Frank Mori Hess escreveu:

Quoted 12 lines · Herton Ronaldo Krzesinski
Quoted 10 lines
Quoted 5 lines

Your patch breaks configuration of the board unless the bus and slot are explicitly specified.  Just make a minimal patch that replaces pci_get_device with pci_get_subsys and fixes the problem that was reported.

Hmm that's not what the patch does, it doesn't break configuration, keeps the same logic as before (I was wrong in my last email replying to myself), check it, if it->options[0] and it->options[1] isn't specified, the pdev is valid so the for loop exits (see !pdev check).

Quoted 6 lines

Your right.  However, it also turns a loop over pci devices into a loop over pci ids, which appears to break the case of multiple s626 boards where the bus/slot of the second s626 board is specified.  If you're not willing to provide a minimal patch that just fixes the reported problem, just say so.  It would have been less effort for me to do it myself than analyze what your changes are breaking.

That's part of reviewing process, I just wanted to enhance it in case more pci ids are added in the future along with the switch to pci_get_subsys, I don't know why you act like that, you don't want the code to be enhanced? Other comedi driver loop over ids, for example gsc_hpdi

And indeed what you say it's true, there is a bug in the patch now that you checked it out properly, it has a problem with this multiple s626 case, fixed that now. Pointing out the problem instead just saying it's broken helps :)

Quoted 5 lines

I also removed the obligation to add sub ids and re-factored the patch, that was too much and not good, just a simpler loop over pci id array is used now, and a comment was added in pci id list telling people that they should add subvendor:subdevice ids for boards with vendor:device == 0x1131:0x7146, in case new boards with this id appear.

Since there is valid subsystem information: why do you have to loop over the ID's
manually in the driver ?

Can't the ID's be set into the PCI ID table and a normal pci_probe be used instead ?
Not that i am familiar with the comedi stuff though ...

We use the probe method directly (rather than pci_get_ ) for most of the multimedia drivers. As an example this is what i do:
http://jusst.de/hg/saa716x/file/59dd985d4473/linux/drivers/media/dvb/saa716x/saa716x_budget.c

Maybe it helps you in some way.

Regards,
Manu