| Message ID | a9b84439c28b96a9b6a4cbfca85f1e13a457d25d.1490135047.git.julia@ni.com |
|---|---|
| State | New |
| Headers | show |
On Tue, Mar 21, 2017 at 05:43:07PM -0500, Julia Cartwright wrote: >The 104-idi-48 gpio driver currently implements an irq_chip for handling >interrupts; due to how irq_chip handling is done, it's necessary for the >irq_chip methods to be invoked from hardirq context, even on a a >real-time kernel. Because the spinlock_t type becomes a "sleeping" >spinlock w/ RT kernels, it is not suitable to be used with irq_chips. > >A quick audit of the operations under the lock reveal that they do only >minimal, bounded work, and are therefore safe to do under a raw spinlock. > >Signed-off-by: Julia Cartwright <julia@ni.com> Hi Julia, This driver also uses a second spinlock_t, called ack_lock, to prevent reentrance into the idi_48_irq_handler function. Should ack_lock also be implemented as a raw_spinlock_t? Thanks, William Breathitt Gray >--- >New patch as of v2 of series. > > drivers/gpio/gpio-104-idi-48.c | 18 ++++++++++-------- > 1 file changed, 10 insertions(+), 8 deletions(-) > >diff --git a/drivers/gpio/gpio-104-idi-48.c b/drivers/gpio/gpio-104-idi-48.c >index 568375a7ebc2..337c048168d8 100644 >--- a/drivers/gpio/gpio-104-idi-48.c >+++ b/drivers/gpio/gpio-104-idi-48.c >@@ -51,7 +51,7 @@ MODULE_PARM_DESC(irq, "ACCES 104-IDI-48 interrupt line numbers"); > */ > struct idi_48_gpio { > struct gpio_chip chip; >- spinlock_t lock; >+ raw_spinlock_t lock; > spinlock_t ack_lock; > unsigned char irq_mask[6]; > unsigned base; >@@ -112,11 +112,12 @@ static void idi_48_irq_mask(struct irq_data *data) > if (!idi48gpio->irq_mask[boundary]) { > idi48gpio->cos_enb &= ~BIT(boundary); > >- spin_lock_irqsave(&idi48gpio->lock, flags); >+ raw_spin_lock_irqsave(&idi48gpio->lock, flags); > > outb(idi48gpio->cos_enb, idi48gpio->base + 7); > >- spin_unlock_irqrestore(&idi48gpio->lock, flags); >+ raw_spin_unlock_irqrestore(&idi48gpio->lock, >+ flags); > } > > return; >@@ -145,11 +146,12 @@ static void idi_48_irq_unmask(struct irq_data *data) > if (!prev_irq_mask) { > idi48gpio->cos_enb |= BIT(boundary); > >- spin_lock_irqsave(&idi48gpio->lock, flags); >+ raw_spin_lock_irqsave(&idi48gpio->lock, flags); > > outb(idi48gpio->cos_enb, idi48gpio->base + 7); > >- spin_unlock_irqrestore(&idi48gpio->lock, flags); >+ raw_spin_unlock_irqrestore(&idi48gpio->lock, >+ flags); > } > > return; >@@ -186,11 +188,11 @@ static irqreturn_t idi_48_irq_handler(int irq, void *dev_id) > > spin_lock(&idi48gpio->ack_lock); > >- spin_lock(&idi48gpio->lock); >+ raw_spin_lock(&idi48gpio->lock); > > cos_status = inb(idi48gpio->base + 7); > >- spin_unlock(&idi48gpio->lock); >+ raw_spin_unlock(&idi48gpio->lock); > > /* IRQ Status (bit 6) is active low (0 = IRQ generated by device) */ > if (cos_status & BIT(6)) { >@@ -256,7 +258,7 @@ static int idi_48_probe(struct device *dev, unsigned int id) > idi48gpio->chip.get = idi_48_gpio_get; > idi48gpio->base = base[id]; > >- spin_lock_init(&idi48gpio->lock); >+ raw_spin_lock_init(&idi48gpio->lock); > spin_lock_init(&idi48gpio->ack_lock); > > err = devm_gpiochip_add_data(dev, &idi48gpio->chip, idi48gpio); >-- >2.12.0 > -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Mar 22, 2017 at 08:44:14AM -0400, William Breathitt Gray wrote: > On Tue, Mar 21, 2017 at 05:43:07PM -0500, Julia Cartwright wrote: > >The 104-idi-48 gpio driver currently implements an irq_chip for handling > >interrupts; due to how irq_chip handling is done, it's necessary for the > >irq_chip methods to be invoked from hardirq context, even on a a > >real-time kernel. Because the spinlock_t type becomes a "sleeping" > >spinlock w/ RT kernels, it is not suitable to be used with irq_chips. > > > >A quick audit of the operations under the lock reveal that they do only > >minimal, bounded work, and are therefore safe to do under a raw spinlock. > > > >Signed-off-by: Julia Cartwright <julia@ni.com> > > Hi Julia, > > This driver also uses a second spinlock_t, called ack_lock, to prevent > reentrance into the idi_48_irq_handler function. Should ack_lock also be > implemented as a raw_spinlock_t? I saw this lock, and I don't even understand it's purpose. However, I think I convinced myself that it's harmless. Why? It's only ever acquired in a handler registered with request_irq(), which, on RT, is invoked in a context which can sleep. Thanks for taking a closer look! Julia -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Mar 22, 2017 at 1:44 PM, William Breathitt Gray <vilhelm.gray@gmail.com> wrote: > On Tue, Mar 21, 2017 at 05:43:07PM -0500, Julia Cartwright wrote: >>The 104-idi-48 gpio driver currently implements an irq_chip for handling >>interrupts; due to how irq_chip handling is done, it's necessary for the >>irq_chip methods to be invoked from hardirq context, even on a a >>real-time kernel. Because the spinlock_t type becomes a "sleeping" >>spinlock w/ RT kernels, it is not suitable to be used with irq_chips. >> >>A quick audit of the operations under the lock reveal that they do only >>minimal, bounded work, and are therefore safe to do under a raw spinlock. >> >>Signed-off-by: Julia Cartwright <julia@ni.com> > > Hi Julia, > > This driver also uses a second spinlock_t, called ack_lock, to prevent > reentrance into the idi_48_irq_handler function. Should ack_lock also be > implemented as a raw_spinlock_t? Hm, can I apply this one patch or not? Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 28, 2017 at 11:11:59AM +0200, Linus Walleij wrote: >On Wed, Mar 22, 2017 at 1:44 PM, William Breathitt Gray ><vilhelm.gray@gmail.com> wrote: >> On Tue, Mar 21, 2017 at 05:43:07PM -0500, Julia Cartwright wrote: >>>The 104-idi-48 gpio driver currently implements an irq_chip for handling >>>interrupts; due to how irq_chip handling is done, it's necessary for the >>>irq_chip methods to be invoked from hardirq context, even on a a >>>real-time kernel. Because the spinlock_t type becomes a "sleeping" >>>spinlock w/ RT kernels, it is not suitable to be used with irq_chips. >>> >>>A quick audit of the operations under the lock reveal that they do only >>>minimal, bounded work, and are therefore safe to do under a raw spinlock. >>> >>>Signed-off-by: Julia Cartwright <julia@ni.com> >> >> Hi Julia, >> >> This driver also uses a second spinlock_t, called ack_lock, to prevent >> reentrance into the idi_48_irq_handler function. Should ack_lock also be >> implemented as a raw_spinlock_t? > >Hm, can I apply this one patch or not? > >Linus Walleij Oops, sorry for missing this reply. Julia is correct that ack_lock does not need to be implemented as raw_spinlock_t. For reference, ack_lock is used to prevent a race condition on the device hardware itself related to how the 104-IDI-48 acknowledges IRQ (check out the commit description for it for a more in-depth explanation if you're curious). Long story short: Julia's patch is prefectly acceptable as is. Acked-by: William Breathitt Gray <vilhelm.gray@gmail.com> William Breathitt Gray -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 21, 2017 at 11:43 PM, Julia Cartwright <julia@ni.com> wrote: > The 104-idi-48 gpio driver currently implements an irq_chip for handling > interrupts; due to how irq_chip handling is done, it's necessary for the > irq_chip methods to be invoked from hardirq context, even on a a > real-time kernel. Because the spinlock_t type becomes a "sleeping" > spinlock w/ RT kernels, it is not suitable to be used with irq_chips. > > A quick audit of the operations under the lock reveal that they do only > minimal, bounded work, and are therefore safe to do under a raw spinlock. > > Signed-off-by: Julia Cartwright <julia@ni.com> > --- > New patch as of v2 of series. Patch applied with William's ACK. Yours, Linus Walleij -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/gpio/gpio-104-idi-48.c b/drivers/gpio/gpio-104-idi-48.c index 568375a7ebc2..337c048168d8 100644 --- a/drivers/gpio/gpio-104-idi-48.c +++ b/drivers/gpio/gpio-104-idi-48.c @@ -51,7 +51,7 @@ MODULE_PARM_DESC(irq, "ACCES 104-IDI-48 interrupt line numbers"); */ struct idi_48_gpio { struct gpio_chip chip; - spinlock_t lock; + raw_spinlock_t lock; spinlock_t ack_lock; unsigned char irq_mask[6]; unsigned base; @@ -112,11 +112,12 @@ static void idi_48_irq_mask(struct irq_data *data) if (!idi48gpio->irq_mask[boundary]) { idi48gpio->cos_enb &= ~BIT(boundary); - spin_lock_irqsave(&idi48gpio->lock, flags); + raw_spin_lock_irqsave(&idi48gpio->lock, flags); outb(idi48gpio->cos_enb, idi48gpio->base + 7); - spin_unlock_irqrestore(&idi48gpio->lock, flags); + raw_spin_unlock_irqrestore(&idi48gpio->lock, + flags); } return; @@ -145,11 +146,12 @@ static void idi_48_irq_unmask(struct irq_data *data) if (!prev_irq_mask) { idi48gpio->cos_enb |= BIT(boundary); - spin_lock_irqsave(&idi48gpio->lock, flags); + raw_spin_lock_irqsave(&idi48gpio->lock, flags); outb(idi48gpio->cos_enb, idi48gpio->base + 7); - spin_unlock_irqrestore(&idi48gpio->lock, flags); + raw_spin_unlock_irqrestore(&idi48gpio->lock, + flags); } return; @@ -186,11 +188,11 @@ static irqreturn_t idi_48_irq_handler(int irq, void *dev_id) spin_lock(&idi48gpio->ack_lock); - spin_lock(&idi48gpio->lock); + raw_spin_lock(&idi48gpio->lock); cos_status = inb(idi48gpio->base + 7); - spin_unlock(&idi48gpio->lock); + raw_spin_unlock(&idi48gpio->lock); /* IRQ Status (bit 6) is active low (0 = IRQ generated by device) */ if (cos_status & BIT(6)) { @@ -256,7 +258,7 @@ static int idi_48_probe(struct device *dev, unsigned int id) idi48gpio->chip.get = idi_48_gpio_get; idi48gpio->base = base[id]; - spin_lock_init(&idi48gpio->lock); + raw_spin_lock_init(&idi48gpio->lock); spin_lock_init(&idi48gpio->ack_lock); err = devm_gpiochip_add_data(dev, &idi48gpio->chip, idi48gpio);
The 104-idi-48 gpio driver currently implements an irq_chip for handling interrupts; due to how irq_chip handling is done, it's necessary for the irq_chip methods to be invoked from hardirq context, even on a a real-time kernel. Because the spinlock_t type becomes a "sleeping" spinlock w/ RT kernels, it is not suitable to be used with irq_chips. A quick audit of the operations under the lock reveal that they do only minimal, bounded work, and are therefore safe to do under a raw spinlock. Signed-off-by: Julia Cartwright <julia@ni.com> --- New patch as of v2 of series. drivers/gpio/gpio-104-idi-48.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-)