| Message ID | 20190429035515.73611-2-drinkcat@chromium.org (mailing list archive) |
|---|---|
| State | Mainlined, archived |
| Commit | 35594bc7cecf3a78504b590e350570e8f4d7779e |
| Headers | show |
| Series | pinctrl: mediatek: Fix 2 issues related to resume from wake sources. | expand |
Hi Nicolas, On Sun, Apr 28, 2019 at 8:55 PM Nicolas Boichat <drinkcat@chromium.org> wrote: > > Before suspending, mtk-eint would set the interrupt mask to the > one in wake_mask. However, some of these interrupts may not have a > corresponding interrupt handler, or the interrupt may be disabled. > > On resume, the eint irq handler would trigger nevertheless, > and irq/pm.c:irq_pm_check_wakeup would be called, which would > try to call irq_disable. However, if the interrupt is not enabled > (irqd_irq_disabled(&desc->irq_data) is true), the call does nothing, > and the interrupt is left enabled in the eint driver. > > Especially for level-sensitive interrupts, this will lead to an > interrupt storm on resume. > > If we detect that an interrupt is only in wake_mask, but not in > cur_mask, we can just mask it out immediately (as mtk_eint_resume > would do anyway at a later stage in the resume sequence, when > restoring cur_mask). > > Fixes: bf22ff45bed ("genirq: Avoid unnecessary low level irq function calls") > Signed-off-by: Nicolas Boichat <drinkcat@chromium.org> Acked-by: Sean Wang <sean.wang@kernel.org> > --- > drivers/pinctrl/mediatek/mtk-eint.c | 16 +++++++++++++++- > 1 file changed, 15 insertions(+), 1 deletion(-) > > diff --git a/drivers/pinctrl/mediatek/mtk-eint.c b/drivers/pinctrl/mediatek/mtk-eint.c > index f464f8cd274b75c..737385e86beb807 100644 > --- a/drivers/pinctrl/mediatek/mtk-eint.c > +++ b/drivers/pinctrl/mediatek/mtk-eint.c > @@ -318,7 +318,7 @@ static void mtk_eint_irq_handler(struct irq_desc *desc) > struct irq_chip *chip = irq_desc_get_chip(desc); > struct mtk_eint *eint = irq_desc_get_handler_data(desc); > unsigned int status, eint_num; > - int offset, index, virq; > + int offset, mask_offset, index, virq; > void __iomem *reg = mtk_eint_get_offset(eint, 0, eint->regs->stat); > int dual_edge, start_level, curr_level; > > @@ -328,10 +328,24 @@ static void mtk_eint_irq_handler(struct irq_desc *desc) > status = readl(reg); > while (status) { > offset = __ffs(status); > + mask_offset = eint_num >> 5; > index = eint_num + offset; > virq = irq_find_mapping(eint->domain, index); > status &= ~BIT(offset); > > + /* > + * If we get an interrupt on pin that was only required > + * for wake (but no real interrupt requested), mask the > + * interrupt (as would mtk_eint_resume do anyway later > + * in the resume sequence). > + */ > + if (eint->wake_mask[mask_offset] & BIT(offset) && > + !(eint->cur_mask[mask_offset] & BIT(offset))) { > + writel_relaxed(BIT(offset), reg - > + eint->regs->stat + > + eint->regs->mask_set); > + } > + > dual_edge = eint->dual_edge[index]; > if (dual_edge) { > /* > -- > 2.21.0.593.g511ec345e18-goog >
On Mon, Apr 29, 2019 at 5:55 AM Nicolas Boichat <drinkcat@chromium.org> wrote: > Before suspending, mtk-eint would set the interrupt mask to the > one in wake_mask. However, some of these interrupts may not have a > corresponding interrupt handler, or the interrupt may be disabled. > > On resume, the eint irq handler would trigger nevertheless, > and irq/pm.c:irq_pm_check_wakeup would be called, which would > try to call irq_disable. However, if the interrupt is not enabled > (irqd_irq_disabled(&desc->irq_data) is true), the call does nothing, > and the interrupt is left enabled in the eint driver. > > Especially for level-sensitive interrupts, this will lead to an > interrupt storm on resume. > > If we detect that an interrupt is only in wake_mask, but not in > cur_mask, we can just mask it out immediately (as mtk_eint_resume > would do anyway at a later stage in the resume sequence, when > restoring cur_mask). > > Fixes: bf22ff45bed ("genirq: Avoid unnecessary low level irq function calls") > Signed-off-by: Nicolas Boichat <drinkcat@chromium.org> Patch applied for fixes. Hm a late ACK made me miss this, sorry. Yours, Linus Walleij
diff --git a/drivers/pinctrl/mediatek/mtk-eint.c b/drivers/pinctrl/mediatek/mtk-eint.c index f464f8cd274b75c..737385e86beb807 100644 --- a/drivers/pinctrl/mediatek/mtk-eint.c +++ b/drivers/pinctrl/mediatek/mtk-eint.c @@ -318,7 +318,7 @@ static void mtk_eint_irq_handler(struct irq_desc *desc) struct irq_chip *chip = irq_desc_get_chip(desc); struct mtk_eint *eint = irq_desc_get_handler_data(desc); unsigned int status, eint_num; - int offset, index, virq; + int offset, mask_offset, index, virq; void __iomem *reg = mtk_eint_get_offset(eint, 0, eint->regs->stat); int dual_edge, start_level, curr_level; @@ -328,10 +328,24 @@ static void mtk_eint_irq_handler(struct irq_desc *desc) status = readl(reg); while (status) { offset = __ffs(status); + mask_offset = eint_num >> 5; index = eint_num + offset; virq = irq_find_mapping(eint->domain, index); status &= ~BIT(offset); + /* + * If we get an interrupt on pin that was only required + * for wake (but no real interrupt requested), mask the + * interrupt (as would mtk_eint_resume do anyway later + * in the resume sequence). + */ + if (eint->wake_mask[mask_offset] & BIT(offset) && + !(eint->cur_mask[mask_offset] & BIT(offset))) { + writel_relaxed(BIT(offset), reg - + eint->regs->stat + + eint->regs->mask_set); + } + dual_edge = eint->dual_edge[index]; if (dual_edge) { /*
Before suspending, mtk-eint would set the interrupt mask to the one in wake_mask. However, some of these interrupts may not have a corresponding interrupt handler, or the interrupt may be disabled. On resume, the eint irq handler would trigger nevertheless, and irq/pm.c:irq_pm_check_wakeup would be called, which would try to call irq_disable. However, if the interrupt is not enabled (irqd_irq_disabled(&desc->irq_data) is true), the call does nothing, and the interrupt is left enabled in the eint driver. Especially for level-sensitive interrupts, this will lead to an interrupt storm on resume. If we detect that an interrupt is only in wake_mask, but not in cur_mask, we can just mask it out immediately (as mtk_eint_resume would do anyway at a later stage in the resume sequence, when restoring cur_mask). Fixes: bf22ff45bed ("genirq: Avoid unnecessary low level irq function calls") Signed-off-by: Nicolas Boichat <drinkcat@chromium.org> --- drivers/pinctrl/mediatek/mtk-eint.c | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-)