Skip to contentSkip to conversation

ti-sysc: probe of 4a318000.target-module failed with error -16

DISCUSSION Andreas Kemnade 2 participants 5 messages Oct 29, 2023 – Dec 1, 2023

lore
Share & tools
lore
#0 lore raw

Hi,

I am seeing the error message from $subject on omap4 devices regularly, on omap4430-panda and on the epson bt200 (omap4460).
While not having the timer seems not critical in itself. I am wondering whether something nasty is behind it which might cause more severe problems.
Is this really seen everywhere?

Regards,
Andreas

#1 lore raw
Quoted 6 lines · Andreas Kemnade

I am seeing the error message from $subject on omap4 devices regularly, on omap4430-panda and on the epson bt200 (omap4460).
While not having the timer seems not critical in itself. I am wondering whether something nasty is behind it which might cause more severe problems.
Is this really seen everywhere?

That should be for gptimer1 that is used as a clocksource by the drivers/clocksource/timer-ti-dm-systimer.c driver, and ti-sysc will ignore it.

Maybe we should not show the error for timers, or change it to dev_info if EBUSY and timer?

Regards,

Tony

#2 lore raw

Am Tue, 31 Oct 2023 09:07:08 +0200
schrieb Tony Lindgren <[email protected]>:

Quoted 8 lines · Andreas Kemnade
Quoted 6 lines

I am seeing the error message from $subject on omap4 devices regularly, on omap4430-panda and on the epson bt200 (omap4460). While not having the timer seems not critical in itself. I am wondering whether something nasty is behind it which might cause more severe problems.
Is this really seen everywhere?

Quoted 4 lines

That should be for gptimer1 that is used as a clocksource by the drivers/clocksource/timer-ti-dm-systimer.c driver, and ti-sysc will ignore it.

hmm, it is about this I think:

       timer1_target: target-module@8000 {     /* 0x4a318000, ap 9 1c.0
       */ compatible = "ti,sysc-omap2-timer", "ti,sysc";

but I see no ti,no-reset-on-init or ti,no-idle as checked by sysc_check_active_timer(). It is a bit strange. Well, we have some alwon below.
It sysc-omap2-timer here right instead of sysc-omap4-timer?

Quoted 3 lines

Maybe we should not show the error for timers, or change it to dev_info if EBUSY and timer?

Well, I am not sure yet whether I understand that -ENXIO vs. -EBUSY business there fully.
I want to really have a checkmark behind that issue in my head...

Regards,
Andreas

#3 lore raw
Quoted 17 lines · Andreas Kemnade

Am Tue, 31 Oct 2023 09:07:08 +0200
schrieb Tony Lindgren <[email protected]>:

Quoted 8 lines · Andreas Kemnade
Quoted 6 lines

I am seeing the error message from $subject on omap4 devices regularly, on omap4430-panda and on the epson bt200 (omap4460). While not having the timer seems not critical in itself. I am wondering whether something nasty is behind it which might cause more severe problems.
Is this really seen everywhere?

Quoted 4 lines

That should be for gptimer1 that is used as a clocksource by the drivers/clocksource/timer-ti-dm-systimer.c driver, and ti-sysc will ignore it.

hmm, it is about this I think:

Quoted 7 lines
       timer1_target: target-module@8000 {     /* 0x4a318000, ap 9 1c.0
       */ compatible = "ti,sysc-omap2-timer", "ti,sysc";

but I see no ti,no-reset-on-init or ti,no-idle as checked by sysc_check_active_timer(). It is a bit strange. Well, we have some alwon below.
It sysc-omap2-timer here right instead of sysc-omap4-timer?

For timers in the wkup domain yes they are "ti,sysc-omap2-timer", that's typically timer1 and possibly timer2. On some devies also possibly timer12.

Quoted 6 lines
Quoted 3 lines

Maybe we should not show the error for timers, or change it to dev_info if EBUSY and timer?

Well, I am not sure yet whether I understand that -ENXIO vs. -EBUSY business there fully.
I want to really have a checkmark behind that issue in my head...

Seems we should change it to dev_info saying something like "timer already in use as a system timer".

Regards,

Tony

#4 lore raw

Hi Tony,

Am Tue, 28 Nov 2023 12:04:09 +0200
schrieb Tony Lindgren <[email protected]>:

Quoted 5 lines
Quoted 5 lines

hmm, it is about this I think:

       timer1_target: target-module@8000 {     /* 0x4a318000, ap 9

1c.0 */ compatible = "ti,sysc-omap2-timer", "ti,sysc";

Quoted 9 lines
Quoted 4 lines

but I see no ti,no-reset-on-init or ti,no-idle as checked by sysc_check_active_timer(). It is a bit strange. Well, we have some alwon below.
It sysc-omap2-timer here right instead of sysc-omap4-timer?

For timers in the wkup domain yes they are "ti,sysc-omap2-timer", that's typically timer1 and possibly timer2. On some devies also possibly timer12.

ok, then at least omap2 vs. omap4 timer is no problem. I will check what happens in check_active_timer.
The background of all this is that I am scrutinizing everything in that area due to my bt200 mainline spl xs. ancient xloader trouble. I think now with my success in GPS I will have enough mental power to start it doing something there again.

Quoted 9 lines
Quoted 6 lines
Quoted 3 lines

Maybe we should not show the error for timers, or change it to dev_info if EBUSY and timer?

Well, I am not sure yet whether I understand that -ENXIO vs. -EBUSY business there fully.
I want to really have a checkmark behind that issue in my head...

Seems we should change it to dev_info saying something like "timer already in use as a system timer".

Yes, if the cause of -EBUSY is really an active timer.

Regards,
Andreas