FAILED: patch "[PATCH] PCI: dwc: Move interrupt acking into the proper callback" failed to apply to 4.14-stable tree

PATCH [email protected] 3 participants 6 messages Jan 15, 2019 – Jan 19, 2019

Subsystems (4)

Stable branch PCI PCI native host bridge and endpoint drivers PCI driver for synopsys DesignWare

lore

Share & tools

lore

[email protected]

author maintainer

2019-01-15 08:24 UTC #0 lore raw

The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 3f7bb2ec20ce07c02b2002349d256c91a463fcc5 Mon Sep 17 00:00:00 2001
From: Marc Zyngier <[email protected]>
Date: Tue, 13 Nov 2018 22:57:34 +0000
Subject: [PATCH] PCI: dwc: Move interrupt acking into the proper callback

The write to the status register is really an ACK for the HW, and should be treated as such by the driver. Let's move it to the irq_ack() callback, which will prevent people from moving it around in order to paper over other bugs.

Fixes 8c934095fa2f ("PCI: dwc: Clear MSI interrupt status after it is handled,

not before")

Fixes 7c5925afbc58 ("PCI: dwc: Move MSI IRQs allocation to IRQ domains

hierarchical API")

Link https://lore.kernel.org/linux-pci/[email protected]/ open here

Reported-by Trent Piepho <[email protected]>

Tested-by Niklas Cassel <[email protected]>

Tested-by Gustavo Pimentel <[email protected]>

Tested-by Stanimir Varbanov <[email protected]>

Signed-off-by Marc Zyngier <[email protected]>

[[email protected]: updated commit log]

Signed-off-by Lorenzo Pieralisi <[email protected]>

Cc [email protected]

drivers/pci/controller/dwc/pcie-designware-host.c +7 -6

index 610a5e018d27..0fa9e8fdce66 100644
@@ -99,9 +99,6 @@ irqreturn_t dw_handle_msi_irq(struct pcie_port *pp)
 					       (i * MAX_MSI_IRQS_PER_CTRL) +
 					       pos);
 			generic_handle_irq(irq);
-			dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS +
-						(i * MSI_REG_CTRL_BLOCK_SIZE),
-					    4, 1 << pos);
 			pos++;
 		}
 	}

@@ -200,14 +197,18 @@ static void dw_pci_bottom_unmask(struct irq_data *data)
 
 static void dw_pci_bottom_ack(struct irq_data *d)
 {
-	struct msi_desc *msi = irq_data_get_msi_desc(d);
-	struct pcie_port *pp;
+	struct pcie_port *pp  = irq_data_get_irq_chip_data(d);
+	unsigned int res, bit, ctrl;
 	unsigned long flags;
 
-	pp = msi_desc_to_pci_sysdata(msi);
+	ctrl = d->hwirq / MAX_MSI_IRQS_PER_CTRL;
+	res = ctrl * MSI_REG_CTRL_BLOCK_SIZE;
+	bit = d->hwirq % MAX_MSI_IRQS_PER_CTRL;
 
 	raw_spin_lock_irqsave(&pp->lock, flags);
 
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + res, 4, 1 << bit);
+
 	if (pp->ops->msi_irq_ack)
 		pp->ops->msi_irq_ack(d->hwirq, pp);

Niklas Cassel

2019-01-16 23:43 UTC #1 lore raw

Quoted 5 lines · On Tue, Jan 15, 2019 at 09:24:37AM +0100, [email protected] wrote:

The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

Hello Marc,

You said that you were going to backport this to 4.14, right?

IIRC, you said something about creating a simple dw_pci_bottom_ack() (since 4.14 lacks a ack() function).

Kind regards,
Niklas

Quoted 68 lines

thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 3f7bb2ec20ce07c02b2002349d256c91a463fcc5 Mon Sep 17 00:00:00 2001
From: Marc Zyngier <[email protected]>
Date: Tue, 13 Nov 2018 22:57:34 +0000
Subject: [PATCH] PCI: dwc: Move interrupt acking into the proper callback
The write to the status register is really an ACK for the HW, and should be treated as such by the driver. Let's move it to the irq_ack() callback, which will prevent people from moving it around in order to paper over other bugs.
Fixes 8c934095fa2f ("PCI: dwc: Clear MSI interrupt status after it is handled,
not before")
Fixes 7c5925afbc58 ("PCI: dwc: Move MSI IRQs allocation to IRQ domains
hierarchical API")
Link https://lore.kernel.org/linux-pci/[email protected]/ open here
Reported-by Trent Piepho <[email protected]>
Tested-by Niklas Cassel <[email protected]>
Tested-by Gustavo Pimentel <[email protected]>
Tested-by Stanimir Varbanov <[email protected]>
Signed-off-by Marc Zyngier <[email protected]>
[[email protected]: updated commit log]
Signed-off-by Lorenzo Pieralisi <[email protected]>
Cc [email protected]
drivers/pci/controller/dwc/pcie-designware-host.c +7 -6
index 610a5e018d27..0fa9e8fdce66 100644
@@ -99,9 +99,6 @@ irqreturn_t dw_handle_msi_irq(struct pcie_port *pp)
 					       (i * MAX_MSI_IRQS_PER_CTRL) +
 					       pos);
 			generic_handle_irq(irq);
-			dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS +
-						(i * MSI_REG_CTRL_BLOCK_SIZE),
-					    4, 1 << pos);
 			pos++;
 		}
 	}
@@ -200,14 +197,18 @@ static void dw_pci_bottom_unmask(struct irq_data *data)
 
 static void dw_pci_bottom_ack(struct irq_data *d)
 {
-	struct msi_desc *msi = irq_data_get_msi_desc(d);
-	struct pcie_port *pp;
+	struct pcie_port *pp  = irq_data_get_irq_chip_data(d);
+	unsigned int res, bit, ctrl;
 	unsigned long flags;
 
-	pp = msi_desc_to_pci_sysdata(msi);
+	ctrl = d->hwirq / MAX_MSI_IRQS_PER_CTRL;
+	res = ctrl * MSI_REG_CTRL_BLOCK_SIZE;
+	bit = d->hwirq % MAX_MSI_IRQS_PER_CTRL;
 
 	raw_spin_lock_irqsave(&pp->lock, flags);
 
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + res, 4, 1 << bit);
+
 	if (pp->ops->msi_irq_ack)
 		pp->ops->msi_irq_ack(d->hwirq, pp);
 

Marc Zyngier

2019-01-17 09:58 UTC #2 lore raw

Quoted 10 lines · Niklas Cassel

Quoted 5 lines · On Tue, Jan 15, 2019 at 09:24:37AM +0100, [email protected] wrote:
The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.
Hello Marc,
You said that you were going to backport this to 4.14, right?

Me, or anyone else. Preferably someone who, like you, has the HW at hand (I don't).

Quoted 2 lines

IIRC, you said something about creating a simple dw_pci_bottom_ack() (since 4.14 lacks a ack() function).

Indeed. Something like this:

drivers/pci/dwc/pcie-designware-host.c +11 -2

index bc3e2d8d0cce..f4f3eeee10af 100644
@@ -45,8 +45,19 @@ static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size,
 	return dw_pcie_write(pci->dbi_base + where, size, val);
 }
 
+static void dwc_irq_ack(struct irq_data *d)
+{
+	struct msi_desc *msi = irq_data_get_msi_desc(d);
+	struct pcie_port *pp = msi_desc_to_pci_sysdata(msi);
+	int pos = d->hwirq % 32;
+	int i = d->hwirq / 32;
+
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12, 4, BIT(pos));
+}
+
 static struct irq_chip dw_msi_irq_chip = {
 	.name = "PCI-MSI",
+	.irq_ack = dwc_irq_ack,
 	.irq_enable = pci_msi_unmask_irq,
 	.irq_disable = pci_msi_mask_irq,
 	.irq_mask = pci_msi_mask_irq,

@@ -72,8 +83,6 @@ irqreturn_t dw_handle_msi_irq(struct pcie_port *pp)
 					    pos)) != 32) {
 			irq = irq_find_mapping(pp->irq_domain, i * 32 + pos);
 			generic_handle_irq(irq);
-			dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12,
-					    4, 1 << pos);
 			pos++;
 		}
 	}

All I can say is that it compiles.

Thanks,

M.

Signature · 1 line

Jazz is not dead. It just smells funny...

Niklas Cassel

2019-01-18 16:00 UTC #3 lore raw

Quoted 27 lines · Marc Zyngier

Quoted 10 lines · Niklas Cassel
Quoted 5 lines · On Tue, Jan 15, 2019 at 09:24:37AM +0100, [email protected] wrote:
The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.
Hello Marc,
You said that you were going to backport this to 4.14, right?
Me, or anyone else. Preferably someone who, like you, has the HW at hand (I don't).
Quoted 2 lines
IIRC, you said something about creating a simple dw_pci_bottom_ack() (since 4.14 lacks a ack() function).
Indeed. Something like this:
drivers/pci/dwc/pcie-designware-host.c +0 -0
index bc3e2d8d0cce..f4f3eeee10af 100644
@@ -45,8 +45,19 @@ static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size,
 	return dw_pcie_write(pci->dbi_base + where, size, val);
 }
 

drivers/pci/dwc/pcie-designware-host.c static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size, · line 57

Quoted 10 lines

drivers/pci/dwc/pcie-designware-host.c +10 -0

+static void dwc_irq_ack(struct irq_data *d)
+{
+	struct msi_desc *msi = irq_data_get_msi_desc(d);
+	struct pcie_port *pp = msi_desc_to_pci_sysdata(msi);
+	int pos = d->hwirq % 32;
+	int i = d->hwirq / 32;
+
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12, 4, BIT(pos));
+}
+

Thanks Marc.

This does not seem to work.
It appears that the ack function is never called.
I assume that this is because of:

irq_set_chip_and_handler(irq, &dw_msi_irq_chip, handle_simple_irq);

is it safe to simply change this to handle_edge_irq? (Which seems to be the case after this after this driver's MSI handling was heavily refactored in 4.17.)

I know that you were against reverting 8c934095fa2f ("PCI: dwc: Clear MSI interrupt status after it is handled, not before") on mainline, since that wouldn't hinder people from moving around stuff in the future, but perhaps reverting this commit on 4.14 is safer than changing the irq handler type?

Kind regards,
Niklas

Quoted 23 lines

drivers/pci/dwc/pcie-designware-host.c +1 -2

 static struct irq_chip dw_msi_irq_chip = {
 	.name = "PCI-MSI",
+	.irq_ack = dwc_irq_ack,
 	.irq_enable = pci_msi_unmask_irq,
 	.irq_disable = pci_msi_mask_irq,
 	.irq_mask = pci_msi_mask_irq,

@@ -72,8 +83,6 @@ irqreturn_t dw_handle_msi_irq(struct pcie_port *pp)
 					    pos)) != 32) {
 			irq = irq_find_mapping(pp->irq_domain, i * 32 + pos);
 			generic_handle_irq(irq);
-			dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12,
-					    4, 1 << pos);
 			pos++;
 		}
 	}

All I can say is that it compiles.

Thanks,

M.

Signature · 1 line

Jazz is not dead. It just smells funny...

Marc Zyngier

2019-01-18 16:22 UTC #4 lore raw

Quoted 41 lines · Niklas Cassel

Quoted 27 lines · Marc Zyngier
Quoted 10 lines · Niklas Cassel
Quoted 5 lines · On Tue, Jan 15, 2019 at 09:24:37AM +0100, [email protected] wrote:
The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.
Hello Marc,
You said that you were going to backport this to 4.14, right?
Me, or anyone else. Preferably someone who, like you, has the HW at hand (I don't).
Quoted 2 lines
IIRC, you said something about creating a simple dw_pci_bottom_ack() (since 4.14 lacks a ack() function).
Indeed. Something like this:
drivers/pci/dwc/pcie-designware-host.c +0 -0
index bc3e2d8d0cce..f4f3eeee10af 100644
@@ -45,8 +45,19 @@ static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size,
 	return dw_pcie_write(pci->dbi_base + where, size, val);
 }
 
drivers/pci/dwc/pcie-designware-host.c static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size, · line 57
Quoted 10 lines
drivers/pci/dwc/pcie-designware-host.c +10 -0
+static void dwc_irq_ack(struct irq_data *d)
+{
+	struct msi_desc *msi = irq_data_get_msi_desc(d);
+	struct pcie_port *pp = msi_desc_to_pci_sysdata(msi);
+	int pos = d->hwirq % 32;
+	int i = d->hwirq / 32;
+
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12, 4, BIT(pos));
+}
+
Thanks Marc.

Quoted 5 lines

This does not seem to work.
It appears that the ack function is never called.
I assume that this is because of:
irq_set_chip_and_handler(irq, &dw_msi_irq_chip, handle_simple_irq);

Ah, indeed, I forgot about that particular nugget.

Quoted 4 lines

is it safe to simply change this to handle_edge_irq? (Which seems to be the case after this after this driver's MSI handling was heavily refactored in 4.17.)

Yes, this should be just a matter of using the right interrupt flow. Please let me know if that works for you.

Quoted 6 lines

I know that you were against reverting 8c934095fa2f ("PCI: dwc: Clear MSI interrupt status after it is handled, not before") on mainline, since that wouldn't hinder people from moving around stuff in the future, but perhaps reverting this commit on 4.14 is safer than changing the irq handler type?

I still don't think that it is a good idea. Using the correct flow is just as important in order to properly honors the interrupt masking which the "simple" flow completely ignores.

Thanks,

M.

Signature · 1 line

Jazz is not dead. It just smells funny...

Niklas Cassel

2019-01-19 11:52 UTC #5 lore raw

Quoted 51 lines · Marc Zyngier

Quoted 41 lines · Niklas Cassel
Quoted 27 lines · Marc Zyngier
Quoted 10 lines · Niklas Cassel
Quoted 5 lines · On Tue, Jan 15, 2019 at 09:24:37AM +0100, [email protected] wrote:
The patch below does not apply to the 4.14-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.
Hello Marc,
You said that you were going to backport this to 4.14, right?
Me, or anyone else. Preferably someone who, like you, has the HW at hand (I don't).
Quoted 2 lines
IIRC, you said something about creating a simple dw_pci_bottom_ack() (since 4.14 lacks a ack() function).
Indeed. Something like this:
drivers/pci/dwc/pcie-designware-host.c +0 -0
index bc3e2d8d0cce..f4f3eeee10af 100644
@@ -45,8 +45,19 @@ static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size,
 	return dw_pcie_write(pci->dbi_base + where, size, val);
 }
 
drivers/pci/dwc/pcie-designware-host.c static int dw_pcie_wr_own_conf(struct pcie_port *pp, int where, int size, · line 57
Quoted 10 lines
drivers/pci/dwc/pcie-designware-host.c +10 -0
+static void dwc_irq_ack(struct irq_data *d)
+{
+	struct msi_desc *msi = irq_data_get_msi_desc(d);
+	struct pcie_port *pp = msi_desc_to_pci_sysdata(msi);
+	int pos = d->hwirq % 32;
+	int i = d->hwirq / 32;
+
+	dw_pcie_wr_own_conf(pp, PCIE_MSI_INTR0_STATUS + i * 12, 4, BIT(pos));
+}
+
Thanks Marc.
Quoted 5 lines
This does not seem to work.
It appears that the ack function is never called.
I assume that this is because of:
irq_set_chip_and_handler(irq, &dw_msi_irq_chip, handle_simple_irq);
Ah, indeed, I forgot about that particular nugget.
Quoted 1 line

Quoted 6 lines

Quoted 3 lines
is it safe to simply change this to handle_edge_irq? (Which seems to be the case after this after this driver's MSI handling was heavily refactored in 4.17.)
Yes, this should be just a matter of using the right interrupt flow. Please let me know if that works for you.

Hello Marc,

I ran a simple stress test during the night with your patch plus:

- irq_set_chip_and_handler(irq, &dw_msi_irq_chip, handle_simple_irq);
+ irq_set_chip_and_handler(irq, &dw_msi_irq_chip, handle_edge_irq);

without any issues. (Feel free to add my Tested-by).

So this looks good to me :) Thanks!

Kind regards,
Niklas

Quoted 17 lines

Quoted 7 lines
I know that you were against reverting 8c934095fa2f ("PCI: dwc: Clear MSI interrupt status after it is handled, not before") on mainline, since that wouldn't hinder people from moving around stuff in the future, but perhaps reverting this commit on 4.14 is safer than changing the irq handler type?
I still don't think that it is a good idea. Using the correct flow is just as important in order to properly honors the interrupt masking which the "simple" flow completely ignores.
Thanks,
	M.
Signature · 1 line
Jazz is not dead. It just smells funny...