Hi Tony,
I am just answering about things I already tested first
and already written several times.
On Tue, 6 Nov 2018 07:43:49 -0800
Tony Lindgren <tony@xxxxxxxxxxx> wrote:
> > Maybe I am wrong. Do you have a cite from the trm?
>
> Hmm I don't think there's much in the TRM about that.
> You should be able to verify it yourself easily with
> grep wake /proc/interrupts though.
>
> > > > > Care to try something like this in the in omap3-gta04.dtsi:
> > > > >
> > > > > &mmc2 {
> > > > > interrupts-extended = <&intc 86 &omap3_pmx_core 0x12e>;
> > > > > interrupt-names = "irq", "wakeup";
> > > > > vmmc-supply = <&vaux4>;
> > > > > bus-width = <4>;
> > > > > ti,non-removable;
> > > > > cap-power-off-card;
> > > > > mmc-pwrseq = <&wifi_pwrseq>;
> > > > > };
> > > > >
> > > > > Untested..
> > > > No, tested,
> > > > see my previous mails in the thread.
> > > >
> > > > But I retested it in case of some PEBCAK issues.
> > > > plain 4.19.0-rc6 + dt patches
> > >
> > > Hmm it should just work(TM). Care to post your patch for that?
> > >
> > https://misc.andi.de1.cc/0001-arm-omap-dts-add-wakeirq-for-mmc2-of-gta04.patch
>
> Yup that's all there really is to it for other devices. So I
> guess the only difference is that you also need to change
> to sdio 1-bit mode here?
>
As already tested, 1 bit +pinctl works.
4bit irq + gpio remap irq also works but gives many irqs when wifi is
powered off.
DAT1 seems to be low then.
If not runtime suspended omap_hsmmc also gets its sdio irq and wifi
works fast and
reliably with 4 bits.
> > Now it is tested on several devices. So it is not my device
> > somehow strange.
>
> OK
>
> > > And do you see interrupts with grep wake /proc/interrupts?
> > >
> > The additional interrupt is there, but count stays at 1-2.
>
> Hmm that's interesting. Maybe check the 16-bit padconf
> register at offset 0x12e and see if what value it has?
>
Must recheck.
> Also maybe add a printk to the sdio runtime idle path to
> make sure it's idling and waking properly at that point?
>
I added a printk to tell when an irq arrives. With omap_hsmmc
runtime suspend disabled, (echo on >power/control)
I get every 100ms an interrupt ( if (status & CIRQ_EN) ). If there is
traffic, of course more.
When I am using 1bit and runtime suspend, the picture about irq is the
same, except that mmc runtime goes to suspend for a few ms one or two
times per seconds and than waking up by the wakeup irq. Count
in /proc/interrupts raises by 1-2 irqs per second. Well,
autosuspend_delay is also set to 100ms, so that pattern is logical.
So having a wakeup irq here does not lead to powersaving in the soc
easily but to increased performance and powersaving in the wifi chip
because I am forced to have that irq here to have sdio_irq handling at
all without changing omap_hsmmc.c.
Mainline kernel gives only about 100KByte/s, with sdio irqs I get >
1000KByte/s.
And that is the first goal here to bring that performance to mainline
kernel, powersaving in the wifi chip by about 80%. SoC powersaving
seems to be a hard to earn fruit here.
Regards,
Andreas
Attachment:
pgp6qdGEVnG_7.pgp
Description: OpenPGP digital signature
