2018-12-17 14:36:04

by Benjamin GAIGNARD

[permalink] [raw]
Subject: [PATCH v3 0/3] Make STM32 interrupt controller use hwspinlock

This series allow to protect STM32 interrupt controller configuration registers
with a hwspinlock to avoid conflicting accesses between processors.

version 3:
- with bindings patch

version 2:
- rework hwspinlock locking sequence in stm32 irqchip to take care of the
cases where hwspinlock node is disabled or not yet probed

Benjamin Gaignard (3):
dt-bindings: interrupt-controller: stm32: Document hwlock properties
irqchip: stm32: protect configuration registers with hwspinlock
ARM: dts: stm32: Add hwlock for irqchip on stm32mp157

.../interrupt-controller/st,stm32-exti.txt | 4 +
arch/arm/boot/dts/stm32mp157c.dtsi | 1 +
drivers/irqchip/irq-stm32-exti.c | 116 ++++++++++++++++++---
3 files changed, 105 insertions(+), 16 deletions(-)

--
2.15.0



2018-12-17 14:38:25

by Benjamin GAIGNARD

[permalink] [raw]
Subject: [PATCH v3 1/3] dt-bindings: interrupt-controller: stm32: Document hwlock properties

Add hwlocks as optional property

Signed-off-by: Benjamin Gaignard <[email protected]>
---
.../devicetree/bindings/interrupt-controller/st,stm32-exti.txt | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.txt b/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.txt
index 6a36bf66d932..cd01b2292ec6 100644
--- a/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.txt
+++ b/Documentation/devicetree/bindings/interrupt-controller/st,stm32-exti.txt
@@ -14,6 +14,10 @@ Required properties:
(only needed for exti controller with multiple exti under
same parent interrupt: st,stm32-exti and st,stm32h7-exti)

+Optional properties:
+
+- hwlocks: reference to a phandle of a hardware spinlock provider node.
+
Example:

exti: interrupt-controller@40013c00 {
--
2.15.0


2018-12-17 18:42:42

by Benjamin GAIGNARD

[permalink] [raw]
Subject: [PATCH v3 3/3] ARM: dts: stm32: Add hwlock for irqchip on stm32mp157

Define a hwspinlock to be used by irq controller

Signed-off-by: Benjamin Gaignard <[email protected]>
---
arch/arm/boot/dts/stm32mp157c.dtsi | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/arm/boot/dts/stm32mp157c.dtsi b/arch/arm/boot/dts/stm32mp157c.dtsi
index 8bf1c17f8cef..452d252a4856 100644
--- a/arch/arm/boot/dts/stm32mp157c.dtsi
+++ b/arch/arm/boot/dts/stm32mp157c.dtsi
@@ -815,6 +815,7 @@
interrupt-controller;
#interrupt-cells = <2>;
reg = <0x5000d000 0x400>;
+ hwlocks = <&hsem 1>;
};

syscfg: syscon@50020000 {
--
2.15.0


2018-12-17 18:42:45

by Benjamin GAIGNARD

[permalink] [raw]
Subject: [PATCH v3 2/3] irqchip: stm32: protect configuration registers with hwspinlock

If a hwspinlock is defined in device tree use it to protect
configuration registers.

Do not request for hwspinlock during the exti driver init since the
hwspinlock driver is not probed yet at that stage and the exti driver
does not support deferred probe.
Instead of this, postpone the hwspinlock request at the first time the
hwspinlock is actually needed.

Use the hwspin_trylock_raw() API which is the most appropriated here
Indeed:
- hwspin_lock_() calls are under spin_lock protection (chip_data->rlock
or gc->lock).
- the _timeout() API relies on jiffies count which won't work if IRQs
are disabled which is the case here (a large part of the IRQ setup is
done atomically (see irq/manage.c))
As a consequence implement the retry/timeout lock from here. And since
all of this is done atomically, reduce the timeout delay to 1 ms.

Signed-off-by: Benjamin Gaignard <[email protected]>
Signed-off-by: Fabien Dessenne <[email protected]>
---
drivers/irqchip/irq-stm32-exti.c | 116 +++++++++++++++++++++++++++++++++------
1 file changed, 100 insertions(+), 16 deletions(-)

diff --git a/drivers/irqchip/irq-stm32-exti.c b/drivers/irqchip/irq-stm32-exti.c
index 0a2088e12d96..95933745433f 100644
--- a/drivers/irqchip/irq-stm32-exti.c
+++ b/drivers/irqchip/irq-stm32-exti.c
@@ -6,6 +6,8 @@
*/

#include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/hwspinlock.h>
#include <linux/interrupt.h>
#include <linux/io.h>
#include <linux/irq.h>
@@ -20,6 +22,9 @@

#define IRQS_PER_BANK 32

+#define HWSPNLCK_TIMEOUT 1000 /* usec */
+#define HWSPNLCK_RETRY_DELAY 100 /* usec */
+
struct stm32_exti_bank {
u32 imr_ofst;
u32 emr_ofst;
@@ -32,6 +37,12 @@ struct stm32_exti_bank {

#define UNDEF_REG ~0

+enum stm32_exti_hwspinlock {
+ HWSPINLOCK_UNKNOWN,
+ HWSPINLOCK_NONE,
+ HWSPINLOCK_READY,
+};
+
struct stm32_desc_irq {
u32 exti;
u32 irq_parent;
@@ -58,6 +69,9 @@ struct stm32_exti_host_data {
void __iomem *base;
struct stm32_exti_chip_data *chips_data;
const struct stm32_exti_drv_data *drv_data;
+ struct device_node *node;
+ enum stm32_exti_hwspinlock hwlock_state;
+ struct hwspinlock *hwlock;
};

static struct stm32_exti_host_data *stm32_host_data;
@@ -269,6 +283,64 @@ static int stm32_exti_set_type(struct irq_data *d,
return 0;
}

+static int stm32_exti_hwspin_lock(struct stm32_exti_chip_data *chip_data)
+{
+ struct stm32_exti_host_data *host_data = chip_data->host_data;
+ struct hwspinlock *hwlock;
+ int id, ret = 0, timeout = 0;
+
+ /* first time, check for hwspinlock availability */
+ if (unlikely(host_data->hwlock_state == HWSPINLOCK_UNKNOWN)) {
+ id = of_hwspin_lock_get_id(host_data->node, 0);
+ if (id >= 0) {
+ hwlock = hwspin_lock_request_specific(id);
+ if (hwlock) {
+ /* found valid hwspinlock */
+ host_data->hwlock_state = HWSPINLOCK_READY;
+ host_data->hwlock = hwlock;
+ pr_debug("%s hwspinlock = %d\n", __func__, id);
+ } else {
+ host_data->hwlock_state = HWSPINLOCK_NONE;
+ }
+ } else if (id != -EPROBE_DEFER) {
+ host_data->hwlock_state = HWSPINLOCK_NONE;
+ } else {
+ /* hwspinlock driver shall be ready at that stage */
+ ret = -EPROBE_DEFER;
+ }
+ }
+
+ if (likely(host_data->hwlock_state == HWSPINLOCK_READY)) {
+ /*
+ * Use the x_raw API since we are under spin_lock protection.
+ * Do not use the x_timeout API because we are under irq_disable
+ * mode (see __setup_irq())
+ */
+ do {
+ ret = hwspin_trylock_raw(host_data->hwlock);
+ if (!ret)
+ return 0;
+
+ udelay(HWSPNLCK_RETRY_DELAY);
+ timeout += HWSPNLCK_RETRY_DELAY;
+ } while (timeout < HWSPNLCK_TIMEOUT);
+
+ if (ret == -EBUSY)
+ ret = -ETIMEDOUT;
+ }
+
+ if (ret)
+ pr_err("%s can't get hwspinlock (%d)\n", __func__, ret);
+
+ return ret;
+}
+
+static void stm32_exti_hwspin_unlock(struct stm32_exti_chip_data *chip_data)
+{
+ if (likely(chip_data->host_data->hwlock_state == HWSPINLOCK_READY))
+ hwspin_unlock_raw(chip_data->host_data->hwlock);
+}
+
static int stm32_irq_set_type(struct irq_data *d, unsigned int type)
{
struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
@@ -279,21 +351,26 @@ static int stm32_irq_set_type(struct irq_data *d, unsigned int type)

irq_gc_lock(gc);

+ err = stm32_exti_hwspin_lock(chip_data);
+ if (err)
+ goto unlock;
+
rtsr = irq_reg_readl(gc, stm32_bank->rtsr_ofst);
ftsr = irq_reg_readl(gc, stm32_bank->ftsr_ofst);

err = stm32_exti_set_type(d, type, &rtsr, &ftsr);
- if (err) {
- irq_gc_unlock(gc);
- return err;
- }
+ if (err)
+ goto unspinlock;

irq_reg_writel(gc, rtsr, stm32_bank->rtsr_ofst);
irq_reg_writel(gc, ftsr, stm32_bank->ftsr_ofst);

+unspinlock:
+ stm32_exti_hwspin_unlock(chip_data);
+unlock:
irq_gc_unlock(gc);

- return 0;
+ return err;
}

static void stm32_chip_suspend(struct stm32_exti_chip_data *chip_data,
@@ -460,20 +537,27 @@ static int stm32_exti_h_set_type(struct irq_data *d, unsigned int type)
int err;

raw_spin_lock(&chip_data->rlock);
+
+ err = stm32_exti_hwspin_lock(chip_data);
+ if (err)
+ goto unlock;
+
rtsr = readl_relaxed(base + stm32_bank->rtsr_ofst);
ftsr = readl_relaxed(base + stm32_bank->ftsr_ofst);

err = stm32_exti_set_type(d, type, &rtsr, &ftsr);
- if (err) {
- raw_spin_unlock(&chip_data->rlock);
- return err;
- }
+ if (err)
+ goto unspinlock;

writel_relaxed(rtsr, base + stm32_bank->rtsr_ofst);
writel_relaxed(ftsr, base + stm32_bank->ftsr_ofst);
+
+unspinlock:
+ stm32_exti_hwspin_unlock(chip_data);
+unlock:
raw_spin_unlock(&chip_data->rlock);

- return 0;
+ return err;
}

static int stm32_exti_h_set_wake(struct irq_data *d, unsigned int on)
@@ -599,6 +683,8 @@ stm32_exti_host_data *stm32_exti_host_init(const struct stm32_exti_drv_data *dd,
return NULL;

host_data->drv_data = dd;
+ host_data->node = node;
+ host_data->hwlock_state = HWSPINLOCK_UNKNOWN;
host_data->chips_data = kcalloc(dd->bank_nr,
sizeof(struct stm32_exti_chip_data),
GFP_KERNEL);
@@ -625,8 +711,7 @@ stm32_exti_host_data *stm32_exti_host_init(const struct stm32_exti_drv_data *dd,

static struct
stm32_exti_chip_data *stm32_exti_chip_init(struct stm32_exti_host_data *h_data,
- u32 bank_idx,
- struct device_node *node)
+ u32 bank_idx)
{
const struct stm32_exti_bank *stm32_bank;
struct stm32_exti_chip_data *chip_data;
@@ -656,8 +741,7 @@ stm32_exti_chip_data *stm32_exti_chip_init(struct stm32_exti_host_data *h_data,
if (stm32_bank->fpr_ofst != UNDEF_REG)
writel_relaxed(~0UL, base + stm32_bank->fpr_ofst);

- pr_info("%s: bank%d, External IRQs available:%#x\n",
- node->full_name, bank_idx, irqs_mask);
+ pr_info("%pOF: bank%d\n", h_data->node, bank_idx);

return chip_data;
}
@@ -697,7 +781,7 @@ static int __init stm32_exti_init(const struct stm32_exti_drv_data *drv_data,
struct stm32_exti_chip_data *chip_data;

stm32_bank = drv_data->exti_banks[i];
- chip_data = stm32_exti_chip_init(host_data, i, node);
+ chip_data = stm32_exti_chip_init(host_data, i);

gc = irq_get_domain_generic_chip(domain, i * IRQS_PER_BANK);

@@ -760,7 +844,7 @@ __init stm32_exti_hierarchy_init(const struct stm32_exti_drv_data *drv_data,
return -ENOMEM;

for (i = 0; i < drv_data->bank_nr; i++)
- stm32_exti_chip_init(host_data, i, node);
+ stm32_exti_chip_init(host_data, i);

domain = irq_domain_add_hierarchy(parent_domain, 0,
drv_data->bank_nr * IRQS_PER_BANK,
--
2.15.0


2018-12-18 15:08:51

by Rob Herring (Arm)

[permalink] [raw]
Subject: Re: [PATCH v3 1/3] dt-bindings: interrupt-controller: stm32: Document hwlock properties

On Mon, 17 Dec 2018 15:22:13 +0100, Benjamin Gaignard wrote:
> Add hwlocks as optional property
>
> Signed-off-by: Benjamin Gaignard <[email protected]>
> ---
> .../devicetree/bindings/interrupt-controller/st,stm32-exti.txt | 4 ++++
> 1 file changed, 4 insertions(+)
>

Reviewed-by: Rob Herring <[email protected]>

2018-12-18 15:40:49

by Marc Zyngier

[permalink] [raw]
Subject: Re: [PATCH v3 0/3] Make STM32 interrupt controller use hwspinlock

On 17/12/2018 14:22, Benjamin Gaignard wrote:
> This series allow to protect STM32 interrupt controller configuration registers
> with a hwspinlock to avoid conflicting accesses between processors.
>
> version 3:
> - with bindings patch
>
> version 2:
> - rework hwspinlock locking sequence in stm32 irqchip to take care of the
> cases where hwspinlock node is disabled or not yet probed
>
> Benjamin Gaignard (3):
> dt-bindings: interrupt-controller: stm32: Document hwlock properties
> irqchip: stm32: protect configuration registers with hwspinlock
> ARM: dts: stm32: Add hwlock for irqchip on stm32mp157
>
> .../interrupt-controller/st,stm32-exti.txt | 4 +
> arch/arm/boot/dts/stm32mp157c.dtsi | 1 +
> drivers/irqchip/irq-stm32-exti.c | 116 ++++++++++++++++++---
> 3 files changed, 105 insertions(+), 16 deletions(-)
>

I've taken the first two patches. Please route the DTS patch to the
appropriate tree.

Thanks,

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

2018-12-19 17:30:10

by Alexandre Torgue

[permalink] [raw]
Subject: Re: [PATCH v3 0/3] Make STM32 interrupt controller use hwspinlock

Hi Marc,

On 12/18/18 4:39 PM, Marc Zyngier wrote:
> On 17/12/2018 14:22, Benjamin Gaignard wrote:
>> This series allow to protect STM32 interrupt controller configuration registers
>> with a hwspinlock to avoid conflicting accesses between processors.
>>
>> version 3:
>> - with bindings patch
>>
>> version 2:
>> - rework hwspinlock locking sequence in stm32 irqchip to take care of the
>> cases where hwspinlock node is disabled or not yet probed
>>
>> Benjamin Gaignard (3):
>> dt-bindings: interrupt-controller: stm32: Document hwlock properties
>> irqchip: stm32: protect configuration registers with hwspinlock
>> ARM: dts: stm32: Add hwlock for irqchip on stm32mp157
>>
>> .../interrupt-controller/st,stm32-exti.txt | 4 +
>> arch/arm/boot/dts/stm32mp157c.dtsi | 1 +
>> drivers/irqchip/irq-stm32-exti.c | 116 ++++++++++++++++++---
>> 3 files changed, 105 insertions(+), 16 deletions(-)
>>
>
> I've taken the first two patches. Please route the DTS patch to the
> appropriate tree.
>

I'll take DTS patch in stm32-next branch.

Regards
Alex

> Thanks,
>
> M.
>