Some peripheral subsystems request IRQ_TYPE_EDGE_BOTH interrupt type and report request failures on LIOINTC. To avoid such failures we support to set IRQ_TYPE_EDGE_BOTH type on LIOINTC, by setting LIOINTC_REG_INTC_EDGE to true and keep LIOINTC_REG_INTC_POL as is.
Cc: stable@vger.kernel.org Signed-off-by: Yinbo Zhu zhuyinbo@loongson.cn Signed-off-by: Huacai Chen chenhuacai@loongson.cn --- drivers/irqchip/irq-loongson-liointc.c | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/drivers/irqchip/irq-loongson-liointc.c b/drivers/irqchip/irq-loongson-liointc.c index 2b1bd4a96665..c0c8ef8d27cf 100644 --- a/drivers/irqchip/irq-loongson-liointc.c +++ b/drivers/irqchip/irq-loongson-liointc.c @@ -128,6 +128,10 @@ static int liointc_set_type(struct irq_data *data, unsigned int type) liointc_set_bit(gc, LIOINTC_REG_INTC_EDGE, mask, false); liointc_set_bit(gc, LIOINTC_REG_INTC_POL, mask, true); break; + case IRQ_TYPE_EDGE_BOTH: + liointc_set_bit(gc, LIOINTC_REG_INTC_EDGE, mask, true); + /* Requester need "both", keep LIOINTC_REG_INTC_POL as is */ + break; case IRQ_TYPE_EDGE_RISING: liointc_set_bit(gc, LIOINTC_REG_INTC_EDGE, mask, true); liointc_set_bit(gc, LIOINTC_REG_INTC_POL, mask, false);
On Wed, Apr 02 2025 at 17:25, Huacai Chen wrote:
Some peripheral subsystems request IRQ_TYPE_EDGE_BOTH interrupt type and report request failures on LIOINTC. To avoid such failures we support to set IRQ_TYPE_EDGE_BOTH type on LIOINTC, by setting LIOINTC_REG_INTC_EDGE to true and keep LIOINTC_REG_INTC_POL as is.
That's broken because it will either trigger on the rising or the falling edge depending on the setting of LIOINTC_REG_INTC_POL.
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Thanks,
tglx
Hi, Thomas,
On Thu, Apr 3, 2025 at 11:48 PM Thomas Gleixner tglx@linutronix.de wrote:
On Wed, Apr 02 2025 at 17:25, Huacai Chen wrote:
Some peripheral subsystems request IRQ_TYPE_EDGE_BOTH interrupt type and report request failures on LIOINTC. To avoid such failures we support to set IRQ_TYPE_EDGE_BOTH type on LIOINTC, by setting LIOINTC_REG_INTC_EDGE to true and keep LIOINTC_REG_INTC_POL as is.
That's broken because it will either trigger on the rising or the falling edge depending on the setting of LIOINTC_REG_INTC_POL.
Yes, this patch does exactly this.
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Yes, it won't trigger on both (not perfect), but it allows drivers that request "both" work (better than fail to request), and there are other irqchip drivers that do similar things.
For example,
drivers/irqchip/irq-mips-gic.c case IRQ_TYPE_EDGE_BOTH: pol = 0; /* Doesn't matter */ trig = GIC_TRIG_EDGE; dual = GIC_DUAL_DUAL; break;
drivers/irqchip/irq-qcom-mpm.c if (type & IRQ_TYPE_EDGE_BOTH) type = IRQ_TYPE_EDGE_RISING;
Huacai
Thanks,
tglx
On Sun, Apr 06 2025 at 17:46, Huacai Chen wrote:
On Thu, Apr 3, 2025 at 11:48 PM Thomas Gleixner tglx@linutronix.de wrote:
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Yes, it won't trigger on both (not perfect), but it allows drivers that request "both" work (better than fail to request), and there are
By some definition of 'work'. There is probably a good technical reason why those drivers expect EDGE_BOTH to work correctly and otherwise fail to load.
You completely fail to explain, why this hack actually 'works' and what the implications are for such drivers.
other irqchip drivers that do similar things.
Justifying bogosity with already existing bogosity is not a technical argument.
Thanks,
tglx
On Sun, Apr 6, 2025 at 6:18 PM Thomas Gleixner tglx@linutronix.de wrote:
On Sun, Apr 06 2025 at 17:46, Huacai Chen wrote:
On Thu, Apr 3, 2025 at 11:48 PM Thomas Gleixner tglx@linutronix.de wrote:
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Yes, it won't trigger on both (not perfect), but it allows drivers that request "both" work (better than fail to request), and there are
By some definition of 'work'. There is probably a good technical reason why those drivers expect EDGE_BOTH to work correctly and otherwise fail to load.
The real problem we encounter is the MMC driver. In drivers/mmc/core/slot-gpio.c there is devm_request_threaded_irq(host->parent, irq, NULL, ctx->cd_gpio_isr, IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING | IRQF_ONESHOT, ctx->cd_label, host);
"IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING" is an alias of "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING", and "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING" is "IRQ_TYPE_EDGE_BOTH".
Except MMC, "grep IRQ_TYPE_EDGE_BOTH drivers" can give some more examples.
Huacai
You completely fail to explain, why this hack actually 'works' and what the implications are for such drivers.
other irqchip drivers that do similar things.
Justifying bogosity with already existing bogosity is not a technical argument.
Thanks,
tglx
On Sun, Apr 06 2025 at 20:46, Huacai Chen wrote:
On Sun, Apr 6, 2025 at 6:18 PM Thomas Gleixner tglx@linutronix.de wrote:
On Sun, Apr 06 2025 at 17:46, Huacai Chen wrote:
On Thu, Apr 3, 2025 at 11:48 PM Thomas Gleixner tglx@linutronix.de wrote:
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Yes, it won't trigger on both (not perfect), but it allows drivers that request "both" work (better than fail to request), and there are
By some definition of 'work'. There is probably a good technical reason why those drivers expect EDGE_BOTH to work correctly and otherwise fail to load.
The real problem we encounter is the MMC driver. In drivers/mmc/core/slot-gpio.c there is devm_request_threaded_irq(host->parent, irq, NULL, ctx->cd_gpio_isr, IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING | IRQF_ONESHOT, ctx->cd_label, host);
"IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING" is an alias of "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING", and "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING" is "IRQ_TYPE_EDGE_BOTH".
I know that.
Except MMC, "grep IRQ_TYPE_EDGE_BOTH drivers" can give some more examples.
Sure, but you still do not explain why this works even when the driver is obviously depending on EDGE_BOTH. If it does not then the driver is bogus.
Looking at it, there is obviously a reason for this particular driver to request BOTH. Why?
This is the card change detection and it uses a GPIO. Insert raises one edge and remove the opposite one.
Which means whatever edge you chose randomly the detection will only work in one direction. Please don't tell me that this is correct by any meaning of correct. It's not.
The driver is perfectly fine, when the request fails. It then does the obvious right thing to poll the card detection pin.
So your change makes it worse as it screws up the detection mechanism.
What are you actually making "work"?
Thanks,
tglx
On Sun, Apr 6, 2025 at 10:20 PM Thomas Gleixner tglx@linutronix.de wrote:
On Sun, Apr 06 2025 at 20:46, Huacai Chen wrote:
On Sun, Apr 6, 2025 at 6:18 PM Thomas Gleixner tglx@linutronix.de wrote:
On Sun, Apr 06 2025 at 17:46, Huacai Chen wrote:
On Thu, Apr 3, 2025 at 11:48 PM Thomas Gleixner tglx@linutronix.de wrote:
But it won't trigger on both. So no, you cannot claim that this fixes anything.
Yes, it won't trigger on both (not perfect), but it allows drivers that request "both" work (better than fail to request), and there are
By some definition of 'work'. There is probably a good technical reason why those drivers expect EDGE_BOTH to work correctly and otherwise fail to load.
The real problem we encounter is the MMC driver. In drivers/mmc/core/slot-gpio.c there is devm_request_threaded_irq(host->parent, irq, NULL, ctx->cd_gpio_isr, IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING | IRQF_ONESHOT, ctx->cd_label, host);
"IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING" is an alias of "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING", and "IRQ_TYPE_EDGE_RISING | IRQ_TYPE_EDGE_FALLING" is "IRQ_TYPE_EDGE_BOTH".
I know that.
Except MMC, "grep IRQ_TYPE_EDGE_BOTH drivers" can give some more examples.
Sure, but you still do not explain why this works even when the driver is obviously depending on EDGE_BOTH. If it does not then the driver is bogus.
Looking at it, there is obviously a reason for this particular driver to request BOTH. Why?
This is the card change detection and it uses a GPIO. Insert raises one edge and remove the opposite one.
Which means whatever edge you chose randomly the detection will only work in one direction. Please don't tell me that this is correct by any meaning of correct. It's not.
From experiments, either setting to EDGE_RISING or EDGE_FALLING, card detection (inserting and removing) works. Maybe the driver request "BOTH", but it really need "ANY"? I've searched git log, but I haven't get any useful information.
Huacai
The driver is perfectly fine, when the request fails. It then does the obvious right thing to poll the card detection pin.
So your change makes it worse as it screws up the detection mechanism.
What are you actually making "work"?
Thanks,
tglx
On Tue, Apr 08 2025 at 20:37, Huacai Chen wrote:
On Sun, Apr 6, 2025 at 10:20 PM Thomas Gleixner tglx@linutronix.de wrote:
This is the card change detection and it uses a GPIO. Insert raises one edge and remove the opposite one.
Which means whatever edge you chose randomly the detection will only work in one direction. Please don't tell me that this is correct by any meaning of correct. It's not.
From experiments, either setting to EDGE_RISING or EDGE_FALLING, card detection (inserting and removing) works.
You might get lucky in that case because the switch bounces and there is no debounce mechanism in place.
Maybe the driver request "BOTH", but it really need "ANY"? I've searched git log, but I haven't get any useful information.
There is no maybe at all and speculation is not a technical argument.
It's well defined by the hardware:
Present ________________ | | Not present _____| |_______________
How can you detect that with a single edge?
If the switch bounces then the signal is:
Present __ ___________ __ | | | | | | Not present _____| |_| |_| |_______________
which makes it "work" by accident, but not by design.
There is ZERO guarantee that this will work with all other drivers which request EDGE_BOTH.
Thanks,
tglx
linux-stable-mirror@lists.linaro.org
-
Huacai Chen -
Huacai Chen -
Thomas Gleixner