[v2] PCI: mt7621: Sleep a bit after power on the PCIs phy ports

  Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
to sleep a bit after call to mt7621_pcie_init_port() driver function to get
into reliable boots for both warm and hard resets. The needed time for these
devices to always detect the ports seems to be from 75 to 100 milliseconds.
There is no datasheet or something similar to really understand why this
extra time is needed in these devices but not in most of the boards which
use mt7621 SoC. This issue has been reported by openWRT community and the
complete discussion is in [0]. The selected time of 100 milliseconds has
been also tested in these devices ending up in an always working platform.
Hence, properly add the extra 100 milliseconds msleep() function call to make
also these devices work.

[0]: https://github.com/openwrt/openwrt/pull/11220

Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
---
Hi Bjorn / Lorenzo,

As per Lorenzo comments in v1[0] here it is the patch with changes in commit
message and introducing a new definition for this needed extra delay time.
I wish you the best new year for you both.

Changes in v2:
- Add a new define 'INIT_PORTS_DELAY_MS' avoiding to reuse 'PERST_DELAY_MS'.
- Rewrite commit message and add a link to openWRT discussion.

Previous patch lore link:
[0]: https://lore.kernel.org/lkml/20221209071703.2891714-1-sergio.paracuellos@gmail.com/T/

Thanks,
    Sergio Paracuellos

 drivers/pci/controller/pcie-mt7621.c | 2 ++
 1 file changed, 2 insertions(+)
  

Hi!

On Sat, Dec 31, 2022 at 8:40 AM Sergio Paracuellos
<sergio.paracuellos@gmail.com> wrote:
>
> Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
> to sleep a bit after call to mt7621_pcie_init_port() driver function to get
> into reliable boots for both warm and hard resets. The needed time for these
> devices to always detect the ports seems to be from 75 to 100 milliseconds.
> There is no datasheet or something similar to really understand why this
> extra time is needed in these devices but not in most of the boards which
> use mt7621 SoC. This issue has been reported by openWRT community and the
> complete discussion is in [0]. The selected time of 100 milliseconds has
> been also tested in these devices ending up in an always working platform.
> Hence, properly add the extra 100 milliseconds msleep() function call to make
> also these devices work.
>
> [0]: https://github.com/openwrt/openwrt/pull/11220
>
> Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
> ---
> Hi Bjorn / Lorenzo,
>
> As per Lorenzo comments in v1[0] here it is the patch with changes in commit
> message and introducing a new definition for this needed extra delay time.
> I wish you the best new year for you both.
>
> Changes in v2:
> - Add a new define 'INIT_PORTS_DELAY_MS' avoiding to reuse 'PERST_DELAY_MS'.
> - Rewrite commit message and add a link to openWRT discussion.
>
> Previous patch lore link:
> [0]: https://lore.kernel.org/lkml/20221209071703.2891714-1-sergio.paracuellos@gmail.com/T/
>
> Thanks,
>     Sergio Paracuellos
>
>  drivers/pci/controller/pcie-mt7621.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/pci/controller/pcie-mt7621.c b/drivers/pci/controller/pcie-mt7621.c
> index ee7aad09d627..63a5f4463a9f 100644
> --- a/drivers/pci/controller/pcie-mt7621.c
> +++ b/drivers/pci/controller/pcie-mt7621.c
> @@ -60,6 +60,7 @@
>  #define PCIE_PORT_LINKUP               BIT(0)
>  #define PCIE_PORT_CNT                  3
>
> +#define INIT_PORTS_DELAY_MS            100
>  #define PERST_DELAY_MS                 100
>
>  /**
> @@ -369,6 +370,7 @@ static int mt7621_pcie_init_ports(struct mt7621_pcie *pcie)
>                 }
>         }
>
> +       msleep(INIT_PORTS_DELAY_MS);
>         mt7621_pcie_reset_ep_deassert(pcie);
>
>         tmp = NULL;
> --
> 2.25.1
>

Gentle ping on this patch :-).

Thanks,
    Sergio Paracuellos
  

On Mon, Jan 23, 2023 at 09:55:20AM +0100, Sergio Paracuellos wrote:
> Hi!
> 
> On Sat, Dec 31, 2022 at 8:40 AM Sergio Paracuellos
> <sergio.paracuellos@gmail.com> wrote:
> >
> > Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
> > to sleep a bit after call to mt7621_pcie_init_port() driver function to get
> > into reliable boots for both warm and hard resets. The needed time for these
> > devices to always detect the ports seems to be from 75 to 100 milliseconds.
> > There is no datasheet or something similar to really understand why this
> > extra time is needed in these devices but not in most of the boards which
> > use mt7621 SoC. This issue has been reported by openWRT community and the
> > complete discussion is in [0]. The selected time of 100 milliseconds has
> > been also tested in these devices ending up in an always working platform.
> > Hence, properly add the extra 100 milliseconds msleep() function call to make
> > also these devices work.
> >
> > [0]: https://github.com/openwrt/openwrt/pull/11220
> >
> > Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
> > ---
> > Hi Bjorn / Lorenzo,
> >
> > As per Lorenzo comments in v1[0] here it is the patch with changes in commit
> > message and introducing a new definition for this needed extra delay time.
> > I wish you the best new year for you both.
> >
> > Changes in v2:
> > - Add a new define 'INIT_PORTS_DELAY_MS' avoiding to reuse 'PERST_DELAY_MS'.
> > - Rewrite commit message and add a link to openWRT discussion.
> >
> > Previous patch lore link:
> > [0]: https://lore.kernel.org/lkml/20221209071703.2891714-1-sergio.paracuellos@gmail.com/T/
> >
> > Thanks,
> >     Sergio Paracuellos
> >
> >  drivers/pci/controller/pcie-mt7621.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/pci/controller/pcie-mt7621.c b/drivers/pci/controller/pcie-mt7621.c
> > index ee7aad09d627..63a5f4463a9f 100644
> > --- a/drivers/pci/controller/pcie-mt7621.c
> > +++ b/drivers/pci/controller/pcie-mt7621.c
> > @@ -60,6 +60,7 @@
> >  #define PCIE_PORT_LINKUP               BIT(0)
> >  #define PCIE_PORT_CNT                  3
> >
> > +#define INIT_PORTS_DELAY_MS            100
> >  #define PERST_DELAY_MS                 100
> >
> >  /**
> > @@ -369,6 +370,7 @@ static int mt7621_pcie_init_ports(struct mt7621_pcie *pcie)
> >                 }
> >         }
> >
> > +       msleep(INIT_PORTS_DELAY_MS);
> >         mt7621_pcie_reset_ep_deassert(pcie);
> >
> >         tmp = NULL;
> > --
> > 2.25.1
> >
> 
> Gentle ping on this patch :-).

I was about to merge it - there are a couple of things that I don't
like.

First one is the comment, "Sleep a bit" does not mean anything. I'd
rather say "delay port initialization" or something like that, be
precise please.

More importantly, this is a fix but it is not clear from the commit
log. Please report what's wrong in the commit log.

I will rework the commit log and merge it then (if you want to avoid
another version just post an updated log here and I will merge it).

Thanks,
Lorenzo
  

Hi Lorenzo,

Thanks for your comments.

On Thu, Feb 2, 2023 at 5:27 PM Lorenzo Pieralisi <lpieralisi@kernel.org> wrote:
>
> On Mon, Jan 23, 2023 at 09:55:20AM +0100, Sergio Paracuellos wrote:
> > Hi!
> >
> > On Sat, Dec 31, 2022 at 8:40 AM Sergio Paracuellos
> > <sergio.paracuellos@gmail.com> wrote:
> > >
> > > Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
> > > to sleep a bit after call to mt7621_pcie_init_port() driver function to get
> > > into reliable boots for both warm and hard resets. The needed time for these
> > > devices to always detect the ports seems to be from 75 to 100 milliseconds.
> > > There is no datasheet or something similar to really understand why this
> > > extra time is needed in these devices but not in most of the boards which
> > > use mt7621 SoC. This issue has been reported by openWRT community and the
> > > complete discussion is in [0]. The selected time of 100 milliseconds has
> > > been also tested in these devices ending up in an always working platform.
> > > Hence, properly add the extra 100 milliseconds msleep() function call to make
> > > also these devices work.
> > >
> > > [0]: https://github.com/openwrt/openwrt/pull/11220
> > >
> > > Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
> > > ---
> > > Hi Bjorn / Lorenzo,
> > >
> > > As per Lorenzo comments in v1[0] here it is the patch with changes in commit
> > > message and introducing a new definition for this needed extra delay time.
> > > I wish you the best new year for you both.
> > >
> > > Changes in v2:
> > > - Add a new define 'INIT_PORTS_DELAY_MS' avoiding to reuse 'PERST_DELAY_MS'.
> > > - Rewrite commit message and add a link to openWRT discussion.
> > >
> > > Previous patch lore link:
> > > [0]: https://lore.kernel.org/lkml/20221209071703.2891714-1-sergio.paracuellos@gmail.com/T/
> > >
> > > Thanks,
> > >     Sergio Paracuellos
> > >
> > >  drivers/pci/controller/pcie-mt7621.c | 2 ++
> > >  1 file changed, 2 insertions(+)
> > >
> > > diff --git a/drivers/pci/controller/pcie-mt7621.c b/drivers/pci/controller/pcie-mt7621.c
> > > index ee7aad09d627..63a5f4463a9f 100644
> > > --- a/drivers/pci/controller/pcie-mt7621.c
> > > +++ b/drivers/pci/controller/pcie-mt7621.c
> > > @@ -60,6 +60,7 @@
> > >  #define PCIE_PORT_LINKUP               BIT(0)
> > >  #define PCIE_PORT_CNT                  3
> > >
> > > +#define INIT_PORTS_DELAY_MS            100
> > >  #define PERST_DELAY_MS                 100
> > >
> > >  /**
> > > @@ -369,6 +370,7 @@ static int mt7621_pcie_init_ports(struct mt7621_pcie *pcie)
> > >                 }
> > >         }
> > >
> > > +       msleep(INIT_PORTS_DELAY_MS);
> > >         mt7621_pcie_reset_ep_deassert(pcie);
> > >
> > >         tmp = NULL;
> > > --
> > > 2.25.1
> > >
> >
> > Gentle ping on this patch :-).
>
> I was about to merge it - there are a couple of things that I don't
> like.
>
> First one is the comment, "Sleep a bit" does not mean anything. I'd
> rather say "delay port initialization" or something like that, be
> precise please.

How about:

PCI: mt7621: delay phy ports initialization

>
> More importantly, this is a fix but it is not clear from the commit
> log. Please report what's wrong in the commit log.

The problem is that PCI ports are not detected properly without this
delay for those devices so plugged
cards on them do not work at all. I don't really know how to write the
commit log clearer but I'll try adding
this cards plugged not working stuff :-)

We can add maybe also if you want:

Fixes: 2bdd5238e756 ("PCI: mt7621: Add MediaTek MT7621 PCIe host
controller driver")

>
> I will rework the commit log and merge it then (if you want to avoid
> another version just post an updated log here and I will merge it).

Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
to sleep a bit after calling the mt7621_pcie_init_port() driver function to get
into reliable boots for both warm and hard resets. The needed time for these
devices to always detect the ports seems to be from 75 to 100 milliseconds.
Without ports properly detected no card plugged into the PCI's work at all.
There is no datasheet or something similar to really understand why this
extra time is needed in these devices but not in most of the boards which
use mt7621 SoC. This issue has been reported by openWRT community and the
complete discussion is in [0]. The selected time of 100 milliseconds has
been also tested in these devices ending up in an always working platform.
Hence, properly add the extra 100 milliseconds msleep() function call to make
also these devices work.

[0]: https://github.com/openwrt/openwrt/pull/11220

>
> Thanks,
> Lorenzo

Thanks,
    Sergio Paracuellos
  

On Sat, 31 Dec 2022 08:40:41 +0100, Sergio Paracuellos wrote:
> Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
> to sleep a bit after call to mt7621_pcie_init_port() driver function to get
> into reliable boots for both warm and hard resets. The needed time for these
> devices to always detect the ports seems to be from 75 to 100 milliseconds.
> There is no datasheet or something similar to really understand why this
> extra time is needed in these devices but not in most of the boards which
> use mt7621 SoC. This issue has been reported by openWRT community and the
> complete discussion is in [0]. The selected time of 100 milliseconds has
> been also tested in these devices ending up in an always working platform.
> Hence, properly add the extra 100 milliseconds msleep() function call to make
> also these devices work.
> 
> [...]

Applied to pci/controller/mt7621, thanks!

[1/1] PCI: mt7621: Sleep a bit after power on the PCIs phy ports
      https://git.kernel.org/pci/pci/c/0cb2a8f3456f

Thanks,
Lorenzo
  

On Fri, Feb 3, 2023 at 10:29 AM Lorenzo Pieralisi <lpieralisi@kernel.org> wrote:
>
> On Sat, 31 Dec 2022 08:40:41 +0100, Sergio Paracuellos wrote:
> > Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
> > to sleep a bit after call to mt7621_pcie_init_port() driver function to get
> > into reliable boots for both warm and hard resets. The needed time for these
> > devices to always detect the ports seems to be from 75 to 100 milliseconds.
> > There is no datasheet or something similar to really understand why this
> > extra time is needed in these devices but not in most of the boards which
> > use mt7621 SoC. This issue has been reported by openWRT community and the
> > complete discussion is in [0]. The selected time of 100 milliseconds has
> > been also tested in these devices ending up in an always working platform.
> > Hence, properly add the extra 100 milliseconds msleep() function call to make
> > also these devices work.
> >
> > [...]
>
> Applied to pci/controller/mt7621, thanks!
>
> [1/1] PCI: mt7621: Sleep a bit after power on the PCIs phy ports
>       https://git.kernel.org/pci/pci/c/0cb2a8f3456f
>
> Thanks,
> Lorenzo

Thanks!
   Sergio Paracuellos