linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v3] pci: Wait for up to 1000ms after FLR reset
@ 2016-02-22 20:05 Alex Williamson
  2016-03-10 20:50 ` Bjorn Helgaas
  0 siblings, 1 reply; 2+ messages in thread
From: Alex Williamson @ 2016-02-22 20:05 UTC (permalink / raw)
  To: linux-pci; +Cc: allen.m.kay, linux-kernel, kvm

Some devices take longer than the spec indicates to return from FLR
reset, a notable case of this is Intel integrated graphics (IGD),
which can often take an additional 300ms powering down an attached
LCD panel as part of the FLR.  Allow devices up to 1000ms, testing
every 100ms whether the second dword of config space is read as -1.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
---
v3: Read 2nd dword since SR-IOV VFs don't implement the 1st dword
v2: Change loop to incorporate initial 100ms delay, replace delay in
    callee functions, add comment.

 drivers/pci/pci.c |   27 +++++++++++++++++++++++++--
 1 file changed, 25 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 602eb42..365ac76 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -3414,6 +3414,29 @@ int pci_wait_for_pending_transaction(struct pci_dev *dev)
 }
 EXPORT_SYMBOL(pci_wait_for_pending_transaction);
 
+/*
+ * We should only need to wait 100ms after FLR, but some devices take longer.
+ * Wait for up to 1000ms for config space to return something other than -1.
+ * Intel IGD requires this when an LCD panel is attached.  We read the 2nd
+ * dword because VFs don't implement the 1st dword.
+ */
+static void pci_flr_wait(struct pci_dev *dev)
+{
+	int i = 0;
+	u32 id;
+
+	do {
+		msleep(100);
+		pci_read_config_dword(dev, PCI_COMMAND, &id);
+	} while (i++ < 10 && id == ~0);
+
+	if (id == ~0)
+		dev_warn(&dev->dev, "Failed to return from FLR\n");
+	else if (i > 1)
+		dev_info(&dev->dev, "Required additional "
+			 "%dms to return from FLR\n", (i - 1) * 100);
+}
+
 static int pcie_flr(struct pci_dev *dev, int probe)
 {
 	u32 cap;
@@ -3429,7 +3452,7 @@ static int pcie_flr(struct pci_dev *dev, int probe)
 		dev_err(&dev->dev, "timed out waiting for pending transaction; performing function level reset anyway\n");
 
 	pcie_capability_set_word(dev, PCI_EXP_DEVCTL, PCI_EXP_DEVCTL_BCR_FLR);
-	msleep(100);
+	pci_flr_wait(dev);
 	return 0;
 }
 
@@ -3459,7 +3482,7 @@ static int pci_af_flr(struct pci_dev *dev, int probe)
 		dev_err(&dev->dev, "timed out waiting for pending transaction; performing AF function level reset anyway\n");
 
 	pci_write_config_byte(dev, pos + PCI_AF_CTRL, PCI_AF_CTRL_FLR);
-	msleep(100);
+	pci_flr_wait(dev);
 	return 0;
 }
 

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH v3] pci: Wait for up to 1000ms after FLR reset
  2016-02-22 20:05 [PATCH v3] pci: Wait for up to 1000ms after FLR reset Alex Williamson
@ 2016-03-10 20:50 ` Bjorn Helgaas
  0 siblings, 0 replies; 2+ messages in thread
From: Bjorn Helgaas @ 2016-03-10 20:50 UTC (permalink / raw)
  To: Alex Williamson; +Cc: linux-pci, allen.m.kay, linux-kernel, kvm

On Mon, Feb 22, 2016 at 01:05:48PM -0700, Alex Williamson wrote:
> Some devices take longer than the spec indicates to return from FLR
> reset, a notable case of this is Intel integrated graphics (IGD),
> which can often take an additional 300ms powering down an attached
> LCD panel as part of the FLR.  Allow devices up to 1000ms, testing
> every 100ms whether the second dword of config space is read as -1.
> 
> Signed-off-by: Alex Williamson <alex.williamson@redhat.com>

Applied to pci/virtualization for v4.6, thanks, Alex!

> ---
> v3: Read 2nd dword since SR-IOV VFs don't implement the 1st dword
> v2: Change loop to incorporate initial 100ms delay, replace delay in
>     callee functions, add comment.
> 
>  drivers/pci/pci.c |   27 +++++++++++++++++++++++++--
>  1 file changed, 25 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 602eb42..365ac76 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -3414,6 +3414,29 @@ int pci_wait_for_pending_transaction(struct pci_dev *dev)
>  }
>  EXPORT_SYMBOL(pci_wait_for_pending_transaction);
>  
> +/*
> + * We should only need to wait 100ms after FLR, but some devices take longer.
> + * Wait for up to 1000ms for config space to return something other than -1.
> + * Intel IGD requires this when an LCD panel is attached.  We read the 2nd
> + * dword because VFs don't implement the 1st dword.
> + */
> +static void pci_flr_wait(struct pci_dev *dev)
> +{
> +	int i = 0;
> +	u32 id;
> +
> +	do {
> +		msleep(100);
> +		pci_read_config_dword(dev, PCI_COMMAND, &id);
> +	} while (i++ < 10 && id == ~0);
> +
> +	if (id == ~0)
> +		dev_warn(&dev->dev, "Failed to return from FLR\n");
> +	else if (i > 1)
> +		dev_info(&dev->dev, "Required additional "
> +			 "%dms to return from FLR\n", (i - 1) * 100);
> +}
> +
>  static int pcie_flr(struct pci_dev *dev, int probe)
>  {
>  	u32 cap;
> @@ -3429,7 +3452,7 @@ static int pcie_flr(struct pci_dev *dev, int probe)
>  		dev_err(&dev->dev, "timed out waiting for pending transaction; performing function level reset anyway\n");
>  
>  	pcie_capability_set_word(dev, PCI_EXP_DEVCTL, PCI_EXP_DEVCTL_BCR_FLR);
> -	msleep(100);
> +	pci_flr_wait(dev);
>  	return 0;
>  }
>  
> @@ -3459,7 +3482,7 @@ static int pci_af_flr(struct pci_dev *dev, int probe)
>  		dev_err(&dev->dev, "timed out waiting for pending transaction; performing AF function level reset anyway\n");
>  
>  	pci_write_config_byte(dev, pos + PCI_AF_CTRL, PCI_AF_CTRL_FLR);
> -	msleep(100);
> +	pci_flr_wait(dev);
>  	return 0;
>  }
>  
> 

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2016-03-10 20:50 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-02-22 20:05 [PATCH v3] pci: Wait for up to 1000ms after FLR reset Alex Williamson
2016-03-10 20:50 ` Bjorn Helgaas

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).