Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Michael Braun]

Hi,

I'm running OpenWRT Kernel 3.8.3 (which already has f66dea709cd9309b2ee9f715697818001fb518de and 5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN (QorlQ, PPC) device.
Before updating the kernel from 3.3.0, USB host support was working fine. Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and the lsusb output is empty, so USB host support is not working. When I apply the following patch, USB host support starts working again, so I guess 3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
Do you have an idea how to fix it more appropriately?

Thanks,
  M. Braun

--- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077 +0200
+++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838 +0200
@@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
                 if (!spin_event_timeout(in_be32(non_ehci + FSL_SOC_USB_CTRL) &
                                 PHY_CLK_VALID, FSL_USB_PHY_CLK_TIMEOUT, 0)) {
                         printk(KERN_WARNING "fsl-ehci: USB PHY clock invalid\n");
-                       return -EINVAL;
                 }
         }

RE: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Liu Shengzhou-B36685]

Hi Braun,

It seems the duplicated tdi_reset caused the PHY_CLK_VALID bit unstable,
introduced by patch "EHCI: centralize controller initialization".
I submitted a patch to fix it. 
Please review the patch http://patchwork.ozlabs.org/patch/237201/

Regards,
Shengzhou


> -----Original Message-----
> From: Michael Braun [mailto:michael-dev@fami-braun.de]
> Sent: Wednesday, April 17, 2013 6:08 PM
> To: Liu Shengzhou-B36685
> Cc: Alan Stern; projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman;
> linux-usb@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with
> Freescale P1020
> 
> Hi,
> 
> I'm running OpenWRT Kernel 3.8.3 (which already has
> f66dea709cd9309b2ee9f715697818001fb518de and
> 5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN (QorlQ,
> PPC) device.
> Before updating the kernel from 3.3.0, USB host support was working fine.
> Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and the
> lsusb output is empty, so USB host support is not working. When I apply
> the following patch, USB host support starts working again, so I guess
> 3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
> Do you have an idea how to fix it more appropriately?
> 
> Thanks,
>   M. Braun
> 
> --- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077
> +0200
> +++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838
> +0200
> @@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
>                  if (!spin_event_timeout(in_be32(non_ehci +
> FSL_SOC_USB_CTRL) &
>                                  PHY_CLK_VALID, FSL_USB_PHY_CLK_TIMEOUT,
> 0)) {
>                          printk(KERN_WARNING "fsl-ehci: USB PHY clock
> invalid\n");
> -                       return -EINVAL;
>                  }
>          }
> 

Re: [Projekt-wlan] Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[michael-dev]

Hi,

thanks for the quick reply.

> Please review the patch http://patchwork.ozlabs.org/patch/237201/
I applied it, but it does not make any difference on my platform.

Regards,
  M. Braun

Am 17.04.2013 12:53, schrieb Liu Shengzhou-B36685:
> Hi Braun,
> 
> It seems the duplicated tdi_reset caused the PHY_CLK_VALID bit 
> unstable,
> introduced by patch "EHCI: centralize controller initialization".
> I submitted a patch to fix it.
> Please review the patch http://patchwork.ozlabs.org/patch/237201/
> 
> Regards,
> Shengzhou
> 
> 
>> -----Original Message-----
>> From: Michael Braun [mailto:michael-dev@fami-braun.de]
>> Sent: Wednesday, April 17, 2013 6:08 PM
>> To: Liu Shengzhou-B36685
>> Cc: Alan Stern; projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman;
>> linux-usb@vger.kernel.org; linux-kernel@vger.kernel.org
>> Subject: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with
>> Freescale P1020
>> 
>> Hi,
>> 
>> I'm running OpenWRT Kernel 3.8.3 (which already has
>> f66dea709cd9309b2ee9f715697818001fb518de and
>> 5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN 
>> (QorlQ,
>> PPC) device.
>> Before updating the kernel from 3.3.0, USB host support was working 
>> fine.
>> Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and the
>> lsusb output is empty, so USB host support is not working. When I 
>> apply
>> the following patch, USB host support starts working again, so I 
>> guess
>> 3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
>> Do you have an idea how to fix it more appropriately?
>> 
>> Thanks,
>>   M. Braun
>> 
>> --- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077
>> +0200
>> +++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838
>> +0200
>> @@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
>>                  if (!spin_event_timeout(in_be32(non_ehci +
>> FSL_SOC_USB_CTRL) &
>>                                  PHY_CLK_VALID, 
>> FSL_USB_PHY_CLK_TIMEOUT,
>> 0)) {
>>                          printk(KERN_WARNING "fsl-ehci: USB PHY clock
>> invalid\n");
>> -                       return -EINVAL;
>>                  }
>>          }
>> 

Re: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Michael Braun]

Hi,

I've got hardware here to test with, so if there any changes to test, I'm willing to support.
Meanwhile, might it be a good idea to make that check optional - i.e. add a module parameter or something like this around it?

Regards,
 M. Braun

On Thu, Apr 18, 2013 at 05:13:39PM +0200, michael-dev wrote:
> Hi,
> 
> thanks for the quick reply.
> 
> >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> I applied it, but it does not make any difference on my platform.
> 
> Regards,
>  M. Braun
> 
> Am 17.04.2013 12:53, schrieb Liu Shengzhou-B36685:
> >Hi Braun,
> >
> >It seems the duplicated tdi_reset caused the PHY_CLK_VALID bit
> >unstable,
> >introduced by patch "EHCI: centralize controller initialization".
> >I submitted a patch to fix it.
> >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> >
> >Regards,
> >Shengzhou
> >
> >
> >>-----Original Message-----
> >>From: Michael Braun [mailto:michael-dev@fami-braun.de]
> >>Sent: Wednesday, April 17, 2013 6:08 PM
> >>To: Liu Shengzhou-B36685
> >>Cc: Alan Stern; projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman;
> >>linux-usb@vger.kernel.org; linux-kernel@vger.kernel.org
> >>Subject: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with
> >>Freescale P1020
> >>
> >>Hi,
> >>
> >>I'm running OpenWRT Kernel 3.8.3 (which already has
> >>f66dea709cd9309b2ee9f715697818001fb518de and
> >>5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN
> >>(QorlQ,
> >>PPC) device.
> >>Before updating the kernel from 3.3.0, USB host support was
> >>working fine.
> >>Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and the
> >>lsusb output is empty, so USB host support is not working. When
> >>I apply
> >>the following patch, USB host support starts working again, so I
> >>guess
> >>3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
> >>Do you have an idea how to fix it more appropriately?
> >>
> >>Thanks,
> >>  M. Braun
> >>
> >>--- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077
> >>+0200
> >>+++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838
> >>+0200
> >>@@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
> >>                 if (!spin_event_timeout(in_be32(non_ehci +
> >>FSL_SOC_USB_CTRL) &
> >>                                 PHY_CLK_VALID,
> >>FSL_USB_PHY_CLK_TIMEOUT,
> >>0)) {
> >>                         printk(KERN_WARNING "fsl-ehci: USB PHY clock
> >>invalid\n");
> >>-                       return -EINVAL;
> >>                 }
> >>         }
> >>

RE: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Liu Shengzhou-B36685]

Hi Braun,

At that time I had an P4080DS board which had the same issue and had been fixed with this patch.
I didn't test it on P1020 due to the absence of P1020. I think P1020 will need a new patch besides this one.
Later Ramneek took this issue on P1020 for more investigation.

Hello Ramneek, any update for the PHY_CLK_VALID issue?

Regards,
Shengzhou


> -----Original Message-----
> From: Michael Braun [mailto:michael.braun@fem.tu-ilmenau.de]
> Sent: Sunday, May 19, 2013 11:23 PM
> To: Liu Shengzhou-B36685
> Cc: projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman; linux-
> usb@vger.kernel.org; Alan Stern; linux-kernel@vger.kernel.org
> Subject: Re: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with
> Freescale P1020
> 
> Hi,
> 
> I've got hardware here to test with, so if there any changes to test, I'm
> willing to support.
> Meanwhile, might it be a good idea to make that check optional - i.e. add
> a module parameter or something like this around it?
> 
> Regards,
>  M. Braun
> 
> On Thu, Apr 18, 2013 at 05:13:39PM +0200, michael-dev wrote:
> > Hi,
> >
> > thanks for the quick reply.
> >
> > >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> > I applied it, but it does not make any difference on my platform.
> >
> > Regards,
> >  M. Braun
> >
> > Am 17.04.2013 12:53, schrieb Liu Shengzhou-B36685:
> > >Hi Braun,
> > >
> > >It seems the duplicated tdi_reset caused the PHY_CLK_VALID bit
> > >unstable, introduced by patch "EHCI: centralize controller
> > >initialization".
> > >I submitted a patch to fix it.
> > >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> > >
> > >Regards,
> > >Shengzhou
> > >
> > >
> > >>-----Original Message-----
> > >>From: Michael Braun [mailto:michael-dev@fami-braun.de]
> > >>Sent: Wednesday, April 17, 2013 6:08 PM
> > >>To: Liu Shengzhou-B36685
> > >>Cc: Alan Stern; projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman;
> > >>linux-usb@vger.kernel.org; linux-kernel@vger.kernel.org
> > >>Subject: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with
> > >>Freescale P1020
> > >>
> > >>Hi,
> > >>
> > >>I'm running OpenWRT Kernel 3.8.3 (which already has
> > >>f66dea709cd9309b2ee9f715697818001fb518de and
> > >>5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN
> > >>(QorlQ,
> > >>PPC) device.
> > >>Before updating the kernel from 3.3.0, USB host support was working
> > >>fine.
> > >>Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and
> > >>the lsusb output is empty, so USB host support is not working. When
> > >>I apply the following patch, USB host support starts working again,
> > >>so I guess
> > >>3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
> > >>Do you have an idea how to fix it more appropriately?
> > >>
> > >>Thanks,
> > >>  M. Braun
> > >>
> > >>--- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077
> > >>+0200
> > >>+++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838
> > >>+0200
> > >>@@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
> > >>                 if (!spin_event_timeout(in_be32(non_ehci +
> > >>FSL_SOC_USB_CTRL) &
> > >>                                 PHY_CLK_VALID,
> > >>FSL_USB_PHY_CLK_TIMEOUT,
> > >>0)) {
> > >>                         printk(KERN_WARNING "fsl-ehci: USB PHY
> > >>clock invalid\n");
> > >>-                       return -EINVAL;
> > >>                 }
> > >>         }
> > >>

RE: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Mehresh Ramneek-B31383]

Hi Shengzhou/Braun,

We changed the controller init sequence to make this work. I'll submit the patch upstream soon.

Regards,
Ramneek

-----Original Message-----
From: Liu Shengzhou-B36685 
Sent: Monday, May 20, 2013 9:07 AM
To: Michael Braun; Mehresh Ramneek-B31383
Cc: projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman; linux-usb@vger.kernel.org; Alan Stern; linux-kernel@vger.kernel.org
Subject: RE: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

Hi Braun,

At that time I had an P4080DS board which had the same issue and had been fixed with this patch.
I didn't test it on P1020 due to the absence of P1020. I think P1020 will need a new patch besides this one.
Later Ramneek took this issue on P1020 for more investigation.

Hello Ramneek, any update for the PHY_CLK_VALID issue?

Regards,
Shengzhou


> -----Original Message-----
> From: Michael Braun [mailto:michael.braun@fem.tu-ilmenau.de]
> Sent: Sunday, May 19, 2013 11:23 PM
> To: Liu Shengzhou-B36685
> Cc: projekt-wlan@fem.tu-ilmenau.de; Greg Kroah-Hartman; linux- 
> usb@vger.kernel.org; Alan Stern; linux-kernel@vger.kernel.org
> Subject: Re: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 
> with Freescale P1020
> 
> Hi,
> 
> I've got hardware here to test with, so if there any changes to test, 
> I'm willing to support.
> Meanwhile, might it be a good idea to make that check optional - i.e. 
> add a module parameter or something like this around it?
> 
> Regards,
>  M. Braun
> 
> On Thu, Apr 18, 2013 at 05:13:39PM +0200, michael-dev wrote:
> > Hi,
> >
> > thanks for the quick reply.
> >
> > >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> > I applied it, but it does not make any difference on my platform.
> >
> > Regards,
> >  M. Braun
> >
> > Am 17.04.2013 12:53, schrieb Liu Shengzhou-B36685:
> > >Hi Braun,
> > >
> > >It seems the duplicated tdi_reset caused the PHY_CLK_VALID bit 
> > >unstable, introduced by patch "EHCI: centralize controller 
> > >initialization".
> > >I submitted a patch to fix it.
> > >Please review the patch http://patchwork.ozlabs.org/patch/237201/
> > >
> > >Regards,
> > >Shengzhou
> > >
> > >
> > >>-----Original Message-----
> > >>From: Michael Braun [mailto:michael-dev@fami-braun.de]
> > >>Sent: Wednesday, April 17, 2013 6:08 PM
> > >>To: Liu Shengzhou-B36685
> > >>Cc: Alan Stern; projekt-wlan@fem.tu-ilmenau.de; Greg 
> > >>Kroah-Hartman; linux-usb@vger.kernel.org; 
> > >>linux-kernel@vger.kernel.org
> > >>Subject: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 
> > >>with Freescale P1020
> > >>
> > >>Hi,
> > >>
> > >>I'm running OpenWRT Kernel 3.8.3 (which already has 
> > >>f66dea709cd9309b2ee9f715697818001fb518de and
> > >>5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN 
> > >>(QorlQ,
> > >>PPC) device.
> > >>Before updating the kernel from 3.3.0, USB host support was 
> > >>working fine.
> > >>Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and 
> > >>the lsusb output is empty, so USB host support is not working. 
> > >>When I apply the following patch, USB host support starts working 
> > >>again, so I guess
> > >>3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.
> > >>Do you have an idea how to fix it more appropriately?
> > >>
> > >>Thanks,
> > >>  M. Braun
> > >>
> > >>--- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077
> > >>+0200
> > >>+++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838
> > >>+0200
> > >>@@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
> > >>                 if (!spin_event_timeout(in_be32(non_ehci +
> > >>FSL_SOC_USB_CTRL) &
> > >>                                 PHY_CLK_VALID, 
> > >>FSL_USB_PHY_CLK_TIMEOUT,
> > >>0)) {
> > >>                         printk(KERN_WARNING "fsl-ehci: USB PHY 
> > >>clock invalid\n");
> > >>-                       return -EINVAL;
> > >>                 }
> > >>         }
> > >>

Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Michael Braun]

Hi,

I'm running OpenWRT Kernel 3.8.3 (which already has 
f66dea709cd9309b2ee9f715697818001fb518de and 
5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN (QorlQ, 
PPC) device.
Before updating the kernel from 3.3.0, USB host support was working 
fine. Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and 
the lsusb output is empty, so USB host support is not working. When I 
apply the following patch, USB host support starts working again, so I 
guess 3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.

Regards,
  M. Braun

--- a/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:52.924403077 
+0200
+++ b/drivers/usb/host/ehci-fsl.c       2013-04-15 21:13:57.572410838 
+0200
@@ -273,7 +273,6 @@ static int ehci_fsl_setup_phy(struct usb
                 if (!spin_event_timeout(in_be32(non_ehci + 
FSL_SOC_USB_CTRL) &
                                 PHY_CLK_VALID, FSL_USB_PHY_CLK_TIMEOUT, 
0)) {
                         printk(KERN_WARNING "fsl-ehci: USB PHY clock 
invalid\n");
-                       return -EINVAL;
                 }
         }

Re: Regression in 3735ba8db8e6ea22ad3ff524328926d8d780a884 with Freescale P1020

[Alan Stern]

On Mon, 15 Apr 2013, Michael Braun wrote:

> Hi,
> 
> I'm running OpenWRT Kernel 3.8.3 (which already has 
> f66dea709cd9309b2ee9f715697818001fb518de and 
> 5ed338778f917a035f0f0a52327fc4f72e36f7a1 applied) on a P1020WLAN (QorlQ, 
> PPC) device.
> Before updating the kernel from 3.3.0, USB host support was working 
> fine. Now I get "fsl-ehci: USB PHY clock invalid" messages in dmesg and 
> the lsusb output is empty, so USB host support is not working. When I 
> apply the following patch, USB host support starts working again, so I 
> guess 3735ba8db8e6ea22ad3ff524328926d8d780a884 is the cause.

Why didn't you CC: the author of that commit?  Isn't he the person most 
likely to be able to fix it?

Alan Stern