regmap: regmap-irq: silently ignore unsupported type settings

PATCH In review Matti Vaittinen 5 participants 9 messages Dec 27, 2018 – Jan 8, 2019

Subsystems (2)

Driver core Register map abstraction

lore

Share & tools

lore

Matti Vaittinen

author

2018-12-27 08:45 UTC #0 lore raw

PATCH 1/1 regmap: regmap-irq: silently ignore unsupported type settings

Do not return error if irq-type setting is requested for controlloer which does not support this. This is how regmap-irq has previously handled the undupported type settings and existing drivers seem to be upset if failure is now reported.

Fixes 1c2928e3e321 ("regmap: regmap-irq/gpio-max77620: add level-irq support")

Signed-off-by Matti Vaittinen <[email protected]>

---

Geert reported that 1c2928e3e321 breaks da9063-rtc on the Renesas Koelsch board:

https://lore.kernel.org/lkml/[email protected]/T/#m194616cc88d7b4c2a78f7ce07907608fdb64a092

Geert, do you know if anyone vould to test this?

 drivers/base/regmap/regmap-irq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/base/regmap/regmap-irq.c +1 -1

index 1bd1145ad8b5..d2d0014b0d23 100644
@@ -258,7 +258,7 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
 	const struct regmap_irq_type *t = &irq_data->type;
 
 	if ((t->types_supported & type) != type)
-		return -ENOTSUPP;
+		return 0;
 
 	reg = t->type_reg_offset / map->reg_stride;

Signature · 8 lines

2.14.3


-- 
Matti Vaittinen
ROHM Semiconductors

~~~ "I don't think so," said Rene Descartes.  Just then, he vanished ~~~

Geert Uytterhoeven

2018-12-29 11:14 UTC #1 lore raw

Hi Matti,

Quoted 8 lines · Matti Vaittinen

Do not return error if irq-type setting is requested for controlloer which does not support this. This is how regmap-irq has previously handled the undupported type settings and existing drivers seem to be upset if failure is now reported.
Fixes 1c2928e3e321 ("regmap: regmap-irq/gpio-max77620: add level-irq support")
Signed-off-by Matti Vaittinen <[email protected]>

Reported-by Geert Uytterhoeven <[email protected]>

Quoted 8 lines

---
Geert reported that 1c2928e3e321 breaks da9063-rtc on the Renesas Koelsch board:
https://lore.kernel.org/lkml/[email protected]/T/#m194616cc88d7b4c2a78f7ce07907608fdb64a092
Geert, do you know if anyone vould to test this?

Thanks, that seems to fix the issue with da9063-rtc.

I don't know how to trigger an actual interrupt, though.

Quoted 12 lines

drivers/base/regmap/regmap-irq.c +1 -2

@@ -258,7 +258,7 @@ static int regmap_irq_set_type(struct irq_data *data, unsigned int type)
        const struct regmap_irq_type *t = &irq_data->type;
 
        if ((t->types_supported & type) != type)
-               return -ENOTSUPP;
+               return 0;
 
        reg = t->type_reg_offset / map->reg_stride;
 
--

Quoted 1 line

2.14.3

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

Mark Brown

maintainer

2018-12-31 19:14 UTC #2 lore raw

Quoted 5 lines · Geert Uytterhoeven

Quoted 1 line
Geert, do you know if anyone vould to test this?
Thanks, that seems to fix the issue with da9063-rtc.
I don't know how to trigger an actual interrupt, though.

If it's a RTC does it have an alarm you can set?

Geert Uytterhoeven

2019-01-01 17:36 UTC #3 lore raw

Hi Mark,

Quoted 8 lines · Mark Brown

Quoted 5 lines · Geert Uytterhoeven
Quoted 1 line
Geert, do you know if anyone vould to test this?
Thanks, that seems to fix the issue with da9063-rtc.
I don't know how to trigger an actual interrupt, though.
If it's a RTC does it have an alarm you can set?

That's what I had expected, too, but there is no alarm file under /sys/class/rtc/.

Gr{oetje,eeting}s,

Geert

Signature · 5 lines

Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Steve Twiss

2019-01-02 15:31 UTC #4 lore raw

Hi Geert,

Quoted 9 lines · Geert Uytterhoeven

Subject: Re: [PATCH] regmap: regmap-irq: silently ignore unsupported type settings
Quoted 7 lines · Mark Brown
Quoted 5 lines · Geert Uytterhoeven
Quoted 1 line
Geert, do you know if anyone vould to test this?
Thanks, that seems to fix the issue with da9063-rtc.
I don't know how to trigger an actual interrupt, though.

Quoted 8 lines

Quoted 1 line
If it's a RTC does it have an alarm you can set?
That's what I had expected, too, but there is no alarm file under /sys/class/rtc/.
Gr{oetje,eeting}s,
Geert

To communicate with the DA9063 RTC I am use ioctl function calls

- RTC_SET_TIME
- RTC_RD_TIME
- RTC_ALM_SET
- RTC_ALM_READ
- RTC_AIE_ON
- RTC_AIE_OFF

- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/rtc.txt
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/testing/selftests/rtc

Although I don't use the test programs found in Linux, the ioctl calls I make are shown in the Linux selftests. I believe that Alexandre Belloni updated the RTC tests recently -- but I am not up to date with the latest.

git show d8da8665e8e34c14f9b20fe3f21dff29b24cbf02:tools/testing/selftests/rtc/rtctest.c

Regards,
Steve

Geert Uytterhoeven

2019-01-04 15:48 UTC #5 lore raw

()Hi Steve,

Quoted 28 lines · Steve Twiss

Quoted 9 lines · Geert Uytterhoeven
Subject: Re: [PATCH] regmap: regmap-irq: silently ignore unsupported type settings
Quoted 7 lines · Mark Brown
Quoted 5 lines · Geert Uytterhoeven
Quoted 1 line
Geert, do you know if anyone vould to test this?
Thanks, that seems to fix the issue with da9063-rtc.
I don't know how to trigger an actual interrupt, though.
Quoted 8 lines
Quoted 1 line
If it's a RTC does it have an alarm you can set?
That's what I had expected, too, but there is no alarm file under /sys/class/rtc/.
Gr{oetje,eeting}s,
Geert
To communicate with the DA9063 RTC I am use ioctl function calls
- RTC_SET_TIME
- RTC_RD_TIME
- RTC_ALM_SET
- RTC_ALM_READ
- RTC_AIE_ON
- RTC_AIE_OFF

Quoted 8 lines

- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/rtc.txt
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/testing/selftests/rtc
Although I don't use the test programs found in Linux, the ioctl calls I make are shown in the Linux selftests. I believe that Alexandre Belloni updated the RTC tests recently -- but I am not up to date with the latest.
git show d8da8665e8e34c14f9b20fe3f21dff29b24cbf02:tools/testing/selftests/rtc/rtctest.c

root@koelsch:~# tools/testing/selftests/rtc/rtctest
[==========] Running 5 tests from 2 test cases.
[ RUN      ] rtc.date_read
rtctest.c:49:rtc.date_read:Current RTC date/time is 04/01/2019 14:44:25.
[       OK ] rtc.date_read
[ RUN      ] rtc.uie_read
[       OK ] rtc.uie_read
[ RUN      ] rtc.uie_select
rtctest.c:98:rtc.uie_select:Expected 0 (0) != rc (0)
rtc.uie_select: Test terminated by assertion
[     FAIL ] rtc.uie_select
[ RUN      ] rtc.alarm_alm_set
rtctest.c:137:rtc.alarm_alm_set:Alarm time now set to 14:47:23.
rtctest.c:148:rtc.alarm_alm_set:Expected 0 (0) != rc (0)
rtc.alarm_alm_set: Test failed at step #5
[     FAIL ] rtc.alarm_alm_set
[ RUN      ] rtc.alarm_wkalm_set
rtctest.c:198:rtc.alarm_wkalm_set:Alarm time now set to 04/01/2019 14:47:28.
rtctest.c:205:rtc.alarm_wkalm_set:Expected 0 (0) != rc (0)
rtctest.c:214:rtc.alarm_wkalm_set:Expected new (1546613934) == secs (1546613248)
rtc.alarm_wkalm_set: Test terminated by assertion
[     FAIL ] rtc.alarm_wkalm_set
[==========] 2 / 5 tests passed.
[  FAILED  ]
root@koelsch:~#

No interrupt fired, as witnessed by /proc/interrupts, and the pr_info() I had added to da9063_alarm_event().

Note that rtctest behaves the same before the regmap irq breakage, so this is not a recent regression...

Gr{oetje,eeting}s,

Geert

Signature · 5 lines

Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Steve Twiss

2019-01-07 18:52 UTC #6 lore raw

Hi Geert,

Quoted 30 lines · Geert Uytterhoeven

To: Steve Twiss <[email protected]>
Subject: Re: [PATCH] regmap: regmap-irq: silently ignore unsupported type settings
()Hi Steve,
Quoted 23 lines · Steve Twiss
Quoted 9 lines · Geert Uytterhoeven
Subject: Re: [PATCH] regmap: regmap-irq: silently ignore unsupported type settings
Quoted 5 lines · Mark Brown
Quoted 3 lines · Geert Uytterhoeven
Quoted 1 line
Geert, do you know if anyone vould to test this?
Thanks, that seems to fix the issue with da9063-rtc. I don't know how to trigger an actual interrupt, though.
If it's a RTC does it have an alarm you can set?
That's what I had expected, too, but there is no alarm file under /sys/class/rtc/.
To communicate with the DA9063 RTC I am use ioctl function calls
- RTC_SET_TIME
- RTC_RD_TIME
- RTC_ALM_SET
- RTC_ALM_READ
- RTC_AIE_ON
- RTC_AIE_OFF
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/rtc.txt
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/testing/selftests/rtc

Quoted 6 lines

Quoted 5 lines
Although I don't use the test programs found in Linux, the ioctl calls I make are shown in the Linux selftests. I believe that Alexandre Belloni updated the RTC tests recently -- but I am not up to date with the latest.
git show
d8da8665e8e34c14f9b20fe3f21dff29b24cbf02:tools/testing/selftests/rtc/rtctest.c

Okay. So, I've got my own RTC tests which I wrote using the ioctl() commands. I've not used the rtctest from the kernel before. Also, I would need to look more closely to give you a better reply for your results.

Quoted 27 lines

root@koelsch:~# tools/testing/selftests/rtc/rtctest
[==========] Running 5 tests from 2 test cases.
[ RUN      ] rtc.date_read
rtctest.c:49:rtc.date_read:Current RTC date/time is 04/01/2019 14:44:25.
[       OK ] rtc.date_read
[ RUN      ] rtc.uie_read
[       OK ] rtc.uie_read
[ RUN      ] rtc.uie_select
rtctest.c:98:rtc.uie_select:Expected 0 (0) != rc (0)
rtc.uie_select: Test terminated by assertion
[     FAIL ] rtc.uie_select
[ RUN      ] rtc.alarm_alm_set
rtctest.c:137:rtc.alarm_alm_set:Alarm time now set to 14:47:23.
rtctest.c:148:rtc.alarm_alm_set:Expected 0 (0) != rc (0)
rtc.alarm_alm_set: Test failed at step #5
[     FAIL ] rtc.alarm_alm_set
[ RUN      ] rtc.alarm_wkalm_set
rtctest.c:198:rtc.alarm_wkalm_set:Alarm time now set to 04/01/2019 14:47:28.
rtctest.c:205:rtc.alarm_wkalm_set:Expected 0 (0) != rc (0)
rtctest.c:214:rtc.alarm_wkalm_set:Expected new (1546613934) == secs
(1546613248)
rtc.alarm_wkalm_set: Test terminated by assertion
[     FAIL ] rtc.alarm_wkalm_set
[==========] 2 / 5 tests passed.
[  FAILED  ]
root@koelsch:~#

Quoted 5 lines

No interrupt fired, as witnessed by /proc/interrupts, and the pr_info() I had added to da9063_alarm_event().
Note that rtctest behaves the same before the regmap irq breakage, so this is not a recent regression...

I've just incrementally rebased from v4.18 through to v4.20 and I see the IRQ working for the DA9063 alarm on my test system. I've not applied any extra patches and the results below are for the following vanilla kernels: v4.18, v4.19, v4.20.

The next results are only for regressions in the RTC alarm for the DA9063. I only ran one simple "TEST" to create an output for you (although I did it twice to produce two IRQs):

--- 8< ---
v4.18
-----

Quoted 1 line

cat *.res

[PASS] Set RTC alarms functional test for TEST with DA9063-TEST
[PASS] Setting test type to use RTC alarm functional test for TEST with DA9063-TEST
[PASS] Running set alarm tests for RTC { 1 }
[PASS] Running test for test_rtc_prog_set_simple_alarm_seconds()
[PASS] Setting the current date and time from the da9063-rtc as 2000-01-01 00:00:00
[PASS] Setting the alarm date and time from the da9063-rtc as 2000-01-01 00:00:05 (+5 secs into the future) [PASS] Setting the listener on da9063-rtc then waiting for elapsed timeout of 15 seconds... [PASS] The alarm was triggered on da9063-rtc within the expected time and the alarm happened at 2000-01-01 00:00:05 [PASS] Running test for test_rtc_prog_set_simple_alarm_seconds()
[PASS] Setting the current date and time from the da9063-rtc as 2000-01-01 00:00:00
[PASS] Setting the alarm date and time from the da9063-rtc as 2000-01-01 00:00:15 (+15 secs into the future) [PASS] Setting the listener on da9063-rtc then waiting for elapsed timeout of 25 seconds... [PASS] The alarm was triggered on da9063-rtc within the expected time and the alarm happened at 2000-01-01 00:00:15 [PASS] Finished running DA9063 set alarm tests for RTC { 1 } on TEST

Quoted 1 line

cat /proc/interrupts | grep da9063

249:          2          0          0          0  gpio-mxc  11 Level     da9063-irq
302:          0          0          0          0  da9063-irq   0 Level     ONKEY
303:          0          0          2          0  da9063-irq   1 Level     ALARM
310:          0          0          0          0  da9063-irq   8 Level     LDO_LIM

Quoted 1 line

uname -a

Linux test 4.18.0 #1 SMP Mon Jan 7 16:25:37 GMT 2019 armv7l GNU/Linux

v4.19
-----

Quoted 1 line

cat *.res

Quoted 1 line

cat /proc/interrupts | grep da9063

249:          2          0          0          0  gpio-mxc  11 Level     da9063-irq
302:          0          0          0          0  da9063-irq   0 Level     ONKEY
303:          0          0          0          2  da9063-irq   1 Level     ALARM
310:          0          0          0          0  da9063-irq   8 Level     LDO_LIM

Quoted 1 line

uname -a

Linux test 4.19.0 #1 SMP Mon Jan 7 17:34:43 GMT 2019 armv7l GNU/Linux

v4.20
-----

Quoted 1 line

cat *.res

Quoted 1 line

cat /proc/interrupts | grep da9063

249:          2          0          0          0  gpio-mxc  11 Level     da9063-irq
302:          0          0          0          0  da9063-irq   0 Level     ONKEY
303:          0          0          2          0  da9063-irq   1 Level     ALARM
310:          0          0          0          0  da9063-irq   8 Level     LDO_LIM

Quoted 1 line

uname -a

Linux test 4.20.0 #1 SMP Mon Jan 7 17:46:59 GMT 2019 armv7l GNU/Linux --- 8< ---

So, with these results, I don't think this is a regression going back very far.

But when I get to vanilla v5.0-rc1, there are lots of IRQ failures, so I can't test because nothing gets loaded. All of the IRQs in each of the DA9063 MFD cells are failing, not just for the RTC. I get the following from the console logs for v5.0-rc1 ...

da9063 1-0058: Device detected (chip-ID: 0x61, var-ID: 0x60)
genirq: Setting trigger mode 8 for irq 310 failed (regmap_irq_set_type+0x0/0x164) da9063-regulators da9063-regulators: Failed to request LDO_LIM IRQ.
da9063-regulators: probe of da9063-regulators failed with error -524 [...]
da9063-onkey da9063-onkey: DMA mask not set
genirq: Setting trigger mode 8 for irq 302 failed (regmap_irq_set_type+0x0/0x164) da9063-onkey da9063-onkey: Failed to request IRQ 302: -524
da9063-onkey: probe of da9063-onkey failed with error -524 da9063-rtc da9063-rtc: DMA mask not set
da9063-rtc da9063-rtc: registered as rtc0
genirq: Setting trigger mode 8 for irq 303 failed (regmap_irq_set_type+0x0/0x164) da9063-rtc da9063-rtc: Failed to request ALARM IRQ 303: -524
da9063-rtc: probe of da9063-rtc failed with error -524
[...]
da9063-watchdog da9063-watchdog: DMA mask not set

I'll test the proposed fix tomorrow.

Regards,
Steve

Steve Twiss

Acked-by Tested-by

2019-01-08 10:22 UTC #7 lore raw

In reply to Geert Uytterhoeven

Hi Geert,

Quoted 34 lines · On 29 December 2018 11:14 Geert Uytterhoeven wrote:

To: Matti Vaittinen <[email protected]>
Cc: [email protected]; Mark Brown <[email protected]>; Greg KH <[email protected]>; Rafael J. Wysocki <[email protected]>; Linux Kernel Mailing List <[email protected]>; Support Opensource <[email protected]>
Subject: Re: [PATCH] regmap: regmap-irq: silently ignore unsupported type settings
Hi Matti,
Quoted 8 lines · Matti Vaittinen
Do not return error if irq-type setting is requested for controlloer which does not support this. This is how regmap-irq has previously handled the undupported type settings and existing drivers seem to be upset if failure is now reported.
Fixes 1c2928e3e321 ("regmap: regmap-irq/gpio-max77620: add level-irq support")
Signed-off-by Matti Vaittinen <[email protected]>
Reported-by Geert Uytterhoeven <[email protected]>
Quoted 6 lines
---
Geert reported that 1c2928e3e321 breaks da9063-rtc on the Renesas Koelsch board:
https://lore.kernel.org/lkml/[email protected]/T/#
m194616cc88d7b4c2a78f7ce07907608fdb64a092
Quoted 2 lines
Geert, do you know if anyone vould to test this?

Quoted 19 lines

Thanks, that seems to fix the issue with da9063-rtc.
I don't know how to trigger an actual interrupt, though.
drivers/base/regmap/regmap-irq.c static int regmap_irq_set_type(struct irq_data *data,
Quoted 3 lines
drivers/base/regmap/regmap-irq.c +0 -0
@@ -258,7 +258,7 @@ static int regmap_irq_set_type(struct irq_data *data,
unsigned int type)
Quoted 10 lines
drivers/base/regmap/regmap-irq.c +1 -2
        const struct regmap_irq_type *t = &irq_data->type;
 
        if ((t->types_supported & type) != type)
-               return -ENOTSUPP;
+               return 0;
 
        reg = t->type_reg_offset / map->reg_stride;
 
--
2.14.3

I've run this patch on the DA9063 set-up I have here. And this fixes the problem.

Acked-by Steve Twiss <[email protected]>

Tested-by Steve Twiss <[email protected]>

Regards,
Steve

My previous report using the unmodified v5.0-rc1 Linux kernel is here:
https://lore.kernel.org/lkml/6ED8E3B22081A4459DAC7699F3695FB7022179F018@SW-EX-MBX02.diasemi.com

This v5.0-rc1 Linux kernel failed to allow IRQs for the MFD cells in the DA9063. But, altering the v5.0-rc1 kernel with the patch given above (to return 0 instead of -ENOTSUPP in regmap-irq.c), fixes the problem.

A cut-down console output for this result is given below.

--- 8< ---
[...]
da9063 1-0058: Device detected (chip-ID: 0x61, var-ID: 0x60)
[...]
da9063-onkey da9063-onkey: DMA mask not set
input: da9063-onkey as /devices/soc0/soc/2100000.aips-bus/21a4000.i2c/i2c-1/1-0058/da9063-onkey/input/input2 da9063-rtc da9063-rtc: DMA mask not set
da9063-rtc da9063-rtc: registered as rtc0
[...]
da9063-watchdog da9063-watchdog: DMA mask not set
[...]
da9063-rtc da9063-rtc: setting system clock to 2000-01-01T00:00:00 UTC (946684800)
[...]
--- 8< ---

and the results of setting an RTC alarm and waiting for the interrupt outputs the following:

--- 8< ---

Quoted 1 line

cat *.res

[PASS] Set RTC alarms functional test for TEST with DA9063-TEST
[PASS] Setting test type to use BB RTC alarm functional test for TEST with DA9063-TEST
[PASS] Running set alarm tests for RTC { 1 }
[PASS] Running test for test_rtc_prog_set_simple_alarm_seconds()
[PASS] Setting the current date and time from the da9063-rtc as 2000-01-01 00:00:00
[PASS] Setting the alarm date and time from the da9063-rtc as 2000-01-01 00:00:05 (+5 secs into the future) [PASS] Setting the listener on da9063-rtc then waiting for elapsed timeout of 15 seconds... [PASS] The alarm was triggered on da9063-rtc within the expected time and the alarm happened at 2000-01-01 00:00:05 [PASS] Running test for test_rtc_prog_set_simple_alarm_seconds()
[PASS] Setting the current date and time from the da9063-rtc as 2000-01-01 00:00:00
[PASS] Setting the alarm date and time from the da9063-rtc as 2000-01-01 00:00:15 (+15 secs into the future) [PASS] Setting the listener on da9063-rtc then waiting for elapsed timeout of 25 seconds... [PASS] The alarm was triggered on da9063-rtc within the expected time and the alarm happened at 2000-01-01 00:00:15 [PASS] Finished running DA9063 set alarm tests for RTC { 1 } on TEST

Quoted 1 line

cat /proc/interrupts | grep da9063

249:          2          0          0          0  gpio-mxc  11 Level     da9063-irq
302:          0          0          0          0  da9063-irq   0 Level     ONKEY
303:          0          0          0          2  da9063-irq   1 Level     ALARM
310:          0          0          0          0  da9063-irq   8 Level     LDO_LIM

Quoted 1 line

uname -a

Linux test 5.0.0-rc1 #1 SMP Tue Jan 8 09:39:30 GMT 2019 armv7l GNU/Linux --- 8< ---

Charles Keepax

2019-01-03 17:27 UTC #8 lore raw

In reply to Matti Vaittinen

Quoted 8 lines · Matti Vaittinen

Do not return error if irq-type setting is requested for controlloer which does not support this. This is how regmap-irq has previously handled the undupported type settings and existing drivers seem to be upset if failure is now reported.
Fixes 1c2928e3e321 ("regmap: regmap-irq/gpio-max77620: add level-irq support")
Signed-off-by Matti Vaittinen <[email protected]>

Quoted 9 lines

---
Geert reported that 1c2928e3e321 breaks da9063-rtc on the Renesas Koelsch board:
https://lore.kernel.org/lkml/[email protected]/T/#m194616cc88d7b4c2a78f7ce07907608fdb64a092
Geert, do you know if anyone vould to test this?

Bit late since its already been applied but this patch fixes the issues I had.

Thanks,
Charles