[Gta04-owner] suspend current, USB-PHY, EHCI, module...

NeilBrown neilb at suse.de
Wed Nov 27 12:41:54 CET 2013


On Tue, 26 Nov 2013 17:42:32 +0100 "Dr. H. Nikolaus Schaller"
<hns at goldelico.com> wrote:

> Hi Neil,
> I stumbled over this page
> 
> http://code.metager.de/source/history/linux/stable/drivers/usb/host/ehci-omap.c
> 
> where it appears to show the commit log of the OMAP specific EHCI drivers.

Looks like the same as you would get from
   git log drivers/usb/host/ehci-omap.c
though formatted differently of course.

> 
> There are interesting things about fixes for suspend/resume and even hidden

Would that be:
   USB: ehci-omap: Tweak PHY initialization sequence
??  Comment says it is only relevant since 3.10, so presumably it was broken
recently and is now fixed.  That patch is in 3.11, so you would only need if
it you have a 3.10 kernel.

> and undocumented control registers on the SoC that may play a role in
> the power demand of the PHY and the EHCI and how they suspend/wake up.

That would be USB: ehci-omap: Fix resume failures after bus suspend
??
That patch dates back to 2.6.35, so it is in all the kernels we have been
testing.

> 
> Since this reminds me of the symptom of applying a RESET to the PHY chip
> making power consumption raise. Maybe, the EHCI is powering up.
>  
> Maybe going through it may revel some other idea.

There isn't much there since 3.7, and it all seems to be general code
improvements with no minimal function change.


> 
> In parallel I have checked the options to improve the hardware in the GTA04A5
> board. I have found these chips to replace the SMSC USB3322
> 
> Candidates (unfortunately none is small or has REFCLK):
> * Cypress CY7C68003-24LQXI
> * TI TUSB2210/2211
> * NXP ISP1703
> 
> Unfortunately the TI and the NXP chips are at least twice as large so that we
> can't fit them in place of the USB3322. The Cypress chip is as small as the
> SMSC but lacks the REFCLK line and needs a 26MHz clock input so that I
> don't know yet if it can be used with a DM3730 at all. Finally the ISP1703
> (although even tested and recommended by TI) is no longer in production.
> An ISP1705 may also fit.
> 
> So it looks as if we have to live with the USB3322 in the GTA04A5 as well.

That's very sad, given that it is declared by TI to be incompatible :-(

I suspect there is some way to work around the problem, but without a full
understand of what they are, it is very time consuming making any progress....


On a largely unrelated note, I just tried an experiment with rather
surprising results.

1/ Place GTA04 on desk, screen up, for half and hour and then check average
   power usage during this time.
2/ Place a small weight on the screen and leave of another half an hour.
   Check power usage again.

I get very different numbers - much higher with the weight present.

The tsc2007 shouldn't be using more than a few micro-amps (1.8V through
50Kohm+ is 36uAmps for the penirq line).
But maybe the penirq line is noisy and it is bothering the OMAP3.  I might
try to more thoroughly neutralise that input pin during suspend.

I noticed this because my phone uses much more power when in my pocket, than
when on the desk.

NeilBrown

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20131127/a41f9fc5/attachment.bin>


More information about the Gta04-owner mailing list