Skip to contentSkip to conversation

pci: Use a bus-global mutex to protect VPD operations

PATCH New Mark D Rustad 3 participants 3 messages May 16, 2015 – May 18, 2015 PCI

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 pci: Use a bus-global mutex to protect VPD operations

Some devices have a problem with concurrent VPD access to different functions of the same physical device, so move the protecting mutex from the pci_vpd structure to the pci_bus structure. There are a number of reports on support sites for a variety of devices from various vendors getting the "vpd r/w failed" message. This is likely to at least fix some of them. Thanks to Shannon Nelson for helping to come up with this approach.

---
 drivers/pci/access.c |   10 ++++------
 drivers/pci/probe.c  |    1 +
 include/linux/pci.h  |    1 +
 3 files changed, 6 insertions(+), 6 deletions(-)
drivers/pci/access.c +4 -6
index d9b64a175990..6a1c8d6f95f1 100644
@@ -281,7 +281,6 @@ PCI_USER_WRITE_CONFIG(dword, u32)
 
 struct pci_vpd_pci22 {
 	struct pci_vpd base;
-	struct mutex lock;
 	u16	flag;
 	bool	busy;
 	u8	cap;
@@ -340,7 +339,7 @@ static ssize_t pci_vpd_pci22_read(struct pci_dev *dev, loff_t pos, size_t count,
 	if (pos < 0 || pos > vpd->base.len || end > vpd->base.len)
 		return -EINVAL;
 
-	if (mutex_lock_killable(&vpd->lock))
+	if (mutex_lock_killable(&dev->bus->vpd_mutex))
 		return -EINTR;
 
 	ret = pci_vpd_pci22_wait(dev);
@@ -376,7 +375,7 @@ static ssize_t pci_vpd_pci22_read(struct pci_dev *dev, loff_t pos, size_t count,
 		}
 	}
 out:
-	mutex_unlock(&vpd->lock);
+	mutex_unlock(&dev->bus->vpd_mutex);
 	return ret ? ret : count;
 }
 
@@ -392,7 +391,7 @@ static ssize_t pci_vpd_pci22_write(struct pci_dev *dev, loff_t pos, size_t count
 	if (pos < 0 || (pos & 3) || (count & 3) || end > vpd->base.len)
 		return -EINVAL;
 
-	if (mutex_lock_killable(&vpd->lock))
+	if (mutex_lock_killable(&dev->bus->vpd_mutex))
 		return -EINTR;
 
 	ret = pci_vpd_pci22_wait(dev);
@@ -424,7 +423,7 @@ static ssize_t pci_vpd_pci22_write(struct pci_dev *dev, loff_t pos, size_t count
 		pos += sizeof(u32);
 	}
 out:
-	mutex_unlock(&vpd->lock);
+	mutex_unlock(&dev->bus->vpd_mutex);
 	return ret ? ret : count;
 }
 
@@ -453,7 +452,6 @@ int pci_vpd_pci22_init(struct pci_dev *dev)
 
 	vpd->base.len = PCI_VPD_PCI22_SIZE;
 	vpd->base.ops = &pci_vpd_pci22_ops;
-	mutex_init(&vpd->lock);
 	vpd->cap = cap;
 	vpd->busy = false;
 	dev->vpd = &vpd->base;
drivers/pci/probe.c +1 -0
index 6675a7a1b9fc..40c2a5a751d0 100644
@@ -494,6 +494,7 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus *parent)
 	INIT_LIST_HEAD(&b->devices);
 	INIT_LIST_HEAD(&b->slots);
 	INIT_LIST_HEAD(&b->resources);
+	mutex_init(&b->vpd_mutex);
 	b->max_bus_speed = PCI_SPEED_UNKNOWN;
 	b->cur_bus_speed = PCI_SPEED_UNKNOWN;
 #ifdef CONFIG_PCI_DOMAINS_GENERIC
include/linux/pci.h +1 -0
index 353db8dc4c6e..f8a51d172255 100644
@@ -454,6 +454,7 @@ struct pci_bus {
 	struct msi_controller *msi;	/* MSI controller */
 	void		*sysdata;	/* hook for sys-specific extension */
 	struct proc_dir_entry *procdir;	/* directory entry in /proc/bus/pci */
+	struct mutex	vpd_mutex;	/* bus-wide VPD access mutex */
 
 	unsigned char	number;		/* bus number */
 	unsigned char	primary;	/* number of primary bridge */
#1 lore raw

PATCH 1/1 pci: Use a bus-global mutex to protect VPD operations

Quoted 16 lines

-----Original Message-----
From: Intel-wired-lan [mailto:intel-wired-lan-bounces at lists.osuosl.org] On Behalf Of Mark D Rustad
Sent: Friday, May 15, 2015 5:14 PM
To: intel-wired-lan at lists.osuosl.org
Subject: [Intel-wired-lan] [PATCH] pci: Use a bus-global mutex to protect VPD operations

Some devices have a problem with concurrent VPD access to different functions of the same physical device, so move the protecting mutex from the pci_vpd structure to the pci_bus structure. There are a number of reports on support sites for a variety of devices from various vendors getting the "vpd r/w failed" message. This is likely to at least fix some of them. Thanks to Shannon Nelson for helping to come up with this approach.

Quoted 1 line
Quoted 97 lines
---
 drivers/pci/access.c |   10 ++++------
 drivers/pci/probe.c  |    1 +
 include/linux/pci.h  |    1 +
 3 files changed, 6 insertions(+), 6 deletions(-)
drivers/pci/access.c +1 -2
index d9b64a175990..6a1c8d6f95f1 100644
@@ -281,7 +281,6 @@ PCI_USER_WRITE_CONFIG(dword, u32)
 
 struct pci_vpd_pci22 {
 	struct pci_vpd base;
-	struct mutex lock;
 	u16	flag;
 	bool	busy;
 	u8	cap;
@@ -340,7 +339,7 @@ static ssize_t pci_vpd_pci22_read(struct pci_dev
 *dev, loff_t pos, size_t count,
 	if (pos < 0 || pos > vpd->base.len || end > vpd->base.len)
 		return -EINVAL;
 
-	if (mutex_lock_killable(&vpd->lock))
+	if (mutex_lock_killable(&dev->bus->vpd_mutex))
 		return -EINTR;
 
drivers/pci/access.c +2 -2
 	ret = pci_vpd_pci22_wait(dev);
@@ -376,7 +375,7 @@ static ssize_t pci_vpd_pci22_read(struct pci_dev
 *dev, loff_t pos, size_t count,
 		}
 	}
 out:
-	mutex_unlock(&vpd->lock);
+	mutex_unlock(&dev->bus->vpd_mutex);
 	return ret ? ret : count;
 }
@@ -392,7 +391,7 @@ static ssize_t pci_vpd_pci22_write(struct pci_dev
 *dev, loff_t pos, size_t count
 	if (pos < 0 || (pos & 3) || (count & 3) || end > vpd->base.len)
 		return -EINVAL;
 
-	if (mutex_lock_killable(&vpd->lock))
+	if (mutex_lock_killable(&dev->bus->vpd_mutex))
 		return -EINTR;
 
drivers/pci/access.c +1 -2
 	ret = pci_vpd_pci22_wait(dev);
@@ -424,7 +423,7 @@ static ssize_t pci_vpd_pci22_write(struct pci_dev
 *dev, loff_t pos, size_t count
 		pos += sizeof(u32);
 	}
 out:
-	mutex_unlock(&vpd->lock);
+	mutex_unlock(&dev->bus->vpd_mutex);
 	return ret ? ret : count;
 }
@@ -453,7 +452,6 @@ int pci_vpd_pci22_init(struct pci_dev *dev)
 
 	vpd->base.len = PCI_VPD_PCI22_SIZE;
 	vpd->base.ops = &pci_vpd_pci22_ops;
-	mutex_init(&vpd->lock);
 	vpd->cap = cap;
 	vpd->busy = false;
 	dev->vpd = &vpd->base;
drivers/pci/probe.c +1 -0
index 6675a7a1b9fc..40c2a5a751d0 100644
@@ -494,6 +494,7 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus
 *parent)
 	INIT_LIST_HEAD(&b->devices);
 	INIT_LIST_HEAD(&b->slots);
 	INIT_LIST_HEAD(&b->resources);
+	mutex_init(&b->vpd_mutex);
 	b->max_bus_speed = PCI_SPEED_UNKNOWN;
 	b->cur_bus_speed = PCI_SPEED_UNKNOWN;

#ifdef CONFIG_PCI_DOMAINS_GENERIC

include/linux/pci.h +1 -0
index 353db8dc4c6e..f8a51d172255 100644
@@ -454,6 +454,7 @@ struct pci_bus {
 	struct msi_controller *msi;	/* MSI controller */
 	void		*sysdata;	/* hook for sys-specific extension */
 	struct proc_dir_entry *procdir;	/* directory entry in /proc/bus/pci
 */
+	struct mutex	vpd_mutex;	/* bus-wide VPD access mutex */
 
 	unsigned char	number;		/* bus number */

unsigned char primary; /* number of primary bridge */

_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan at lists.osuosl.org
http://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#2 lore raw

PATCH 1/1 pci: Use a bus-global mutex to protect VPD operations

Quoted 17 lines · Shannon
Quoted 2 lines

-----Original Message-----
From: Intel-wired-lan

[mailto:intel-wired-lan-bounces at lists.osuosl.org]

Quoted 14 lines

On Behalf Of Mark D Rustad
Sent: Friday, May 15, 2015 5:14 PM
To: intel-wired-lan at lists.osuosl.org
Subject: [Intel-wired-lan] [PATCH] pci: Use a bus-global mutex to protect VPD operations

Some devices have a problem with concurrent VPD access to different functions of the same physical device, so move the protecting mutex from the pci_vpd structure to the pci_bus structure. There are a number of reports on support sites for a variety of devices from various vendors getting the "vpd r/w failed" message. This is likely to at least fix some of them. Thanks to Shannon Nelson for helping to come up with this approach.

Quoted 11 lines
Quoted 1 line
Quoted 5 lines
---
 drivers/pci/access.c |   10 ++++------
 drivers/pci/probe.c  |    1 +
 include/linux/pci.h  |    1 +
 3 files changed, 6 insertions(+), 6 deletions(-)

Since this patch really needs to be sent to linux-pci & netdev, I have let Mark know to re-send this to Bjorn and the linux-pci mailing list.

Dropping this patch so that Mark can resend to the more appropriate
mailing lists.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part
URL: <http://lists.osuosl.org/pipermail/intel-wired-lan/attachments/20150518/8ee31a78/attachment-0001.asc>