Hi,
On Wed, May 07, 2014 at 05:45:32PM +0300, Mathias Nyman wrote:
> On 05/06/2014 11:43 PM, Felipe Balbi wrote:
> >Hi Mathias,
> >
> >I was using (up until today) v3.13 (on my desktop PC) to test LPM
> >transitions against OMAP5. Since I was having some issues with U1 exit
> >after some 200 iterations I decided to make sure v3.15-rc4 would have
> >the same behavior but, to my surprise, xHCI doesn't do any LPM
> >transitions anymore.
> >
> >What changed since v3.13 ? Why aren't LPM transitions enabled anymore ?
> >
> >I _do_ have an Intel chipset so XHCI_LPM_SUPPORT should be set.
> >
> >Any ideas ?
> >
>
> I think these patches could be related:
>
> commit 25cd2882e2fc8bd8ed7acaee0ec979f11feda6d7
> usb/xhci: Change how we indicate a host supports Link PM.
>
> commit 140e3026a57ab7d830dab2f2c57796c222db0ea9
> Revert "usbcore: set lpm_capable field for LPM capable root hubs"
>
> Both of them are added after 3.13, and do additional checking before setting
> udev->lpm_capable
last commit could be broken, looking at my root hub's BOS desc I get:
bU1DevExitLat 10 micro seconds
bU2DevExitLat 512 micro seconds
Likewise on the device that I'm connecting to that root hub:
bU1DevExitLat 1 micro seconds
bU2DevExitLat 500 micro seconds
So it should be enabled, and yet I see on dmesg:
[ 10.329092] usb 2-2.1.1: LPM exit latency is zeroed, disabling LPM.
But I can't see how this would ever fail:
154 if (udev->bos->ss_cap->bU1devExitLat == 0 &&
155 udev->bos->ss_cap->bU2DevExitLat == 0) {
156 if (udev->parent)
157 dev_info(&udev->dev, "LPM exit latency is zeroed, disabling LPM.\n");
158 else
159 dev_info(&udev->dev, "We don't know the algorithms for LPM for this host, disabling LPM.\n");
160 return 0;
161 }
weird
--
balbi
Attachment:
signature.asc
Description: Digital signature
