On Wed, Sep 19, 2012 at 11:50:53AM -0500, Sonasath, Moiz wrote:

Felipe,

On Wed, Sep 19, 2012 at 11:04 AM, Felipe Balbi [off-list ref] wrote:

quoted

Hi,

On Wed, Sep 19, 2012 at 10:02:48AM -0500, Sonasath, Moiz wrote:

quoted

Felipe,

On Wed, Sep 19, 2012 at 6:53 AM, Felipe Balbi [off-list ref] wrote:

quoted

Hi,

On Wed, Sep 19, 2012 at 05:00:27PM +0530, Kishon Vijay Abraham I wrote:

quoted

From: Moiz Sonasath <redacted>

For the gadget pullup functionality to work in
SS mode it requires a particular sequence of
toggling the run-stop bit. Here is the required
sequence:

- Set DCTL[31]
- Clear DCTL[31]
- Clear OMAP5430_CONTROL_CORE__PHY_POWER_USB[14]
- Clear DCTL[8:5] = 0x00
- Set DCTL[8:5] = 0x05
- Wait 25 Ms
- Set DCTL[31]
- Set OMAP5430_CONTROL_CORE__PHY_POWER_USB[14]

Tested rigourously the gadget pull-up functionality
in bot HS and SS modes.

Signed-off-by: Moiz Sonasath <redacted>
Signed-off-by: Kishon Vijay Abraham I <redacted>

this needs to split into three patches:

add new poweron field, implement it on omap-usb3, use it on
dwc3/gadget.c

btw, I don't think the changes to run_stop bit are necessary and if

they

quoted

quoted

are, that'd either be a silicon errata or it would've been mentioned on
the databook. I don't remember seeing that on the databook so I'm
assuming that this is caused by a bad use of the PHY.

Why that mdelay(25) ? why 25 ms ? That's quite a long time, actually.

Felipe, This is infact a HW bug that the Si-Val team did accept and gave

us

quoted

this workaround sequence with the precise delay :-)

Supposedly this will be fixed in ES 2.0.

in that case this doesn't have to go to mainline since we're not
supporting ES1.0 in mainline :-)

at minimum this should've come with a proper revision check anyway.

Actually most of it is under a rev check :)

fair enough.

Perhaps the last: usb_phy_shutdown(dwc->usb3_phy);
in the else part should be in
if (dwc->revision <= DWC3_REVISION_187A) check

Is this an OMAP errata or Synopsys errata ? If it's a Synopsys errata we
need to make sure to add the comment above the workaround, if it's an
OMAP errata, we can't apply the workaround to all users since they might
not need it, so we need a more clever scheme.

On top of that, if it's an ES1 errata, I rather not have this in
mainline since we won't support ES1@all in mainline, which means this
workaround will be useless in a mainline kernel tree.

-- 
balbi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20120919/4f21c229/attachment.sig>