[PATCH] dt-bindings: ufs: Fix the compatible string definition

[PATCH] dt-bindings: ufs: Fix the compatible string definition

[Douglas Anderson]

If you look at the bindings for the UFS Host Controller it says:

- compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
              also list one or more of the following:
                 "qcom,msm8994-ufshc"
                 "qcom,msm8996-ufshc"
                 "qcom,ufshc"

My reading of that is that it's fine to just have either of these:
1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
2. "qcom,ufshc", "jedec,ufs-2.0"

As far as I can tell neither of the above is actually a good idea.

For #1 it turns out that the driver currently only keys off the
compatible string "qcom,ufshc" so it won't actually probe.

For #2 the driver won't probe but it's not a good idea to keep the SoC
name out of the compatible string.

Let's update the compatible string to make it really explicit.  We'll
include a nod to the existing driver and the old binding and say that
we should always include the "qcom,ufshc" string in addition to the
SoC compatible string.

While we're at it we'll also include another example SoC known to have
UFS: sdm845.

Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
---

 .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
index 2df00524bd21..69a06a1b732e 100644
--- a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
+++ b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
@@ -4,11 +4,14 @@ UFSHC nodes are defined to describe on-chip UFS host controllers.
 Each UFS controller instance should have its own node.
 
 Required properties:
-- compatible		: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
-			  also list one or more of the following:
-					  "qcom,msm8994-ufshc"
-					  "qcom,msm8996-ufshc"
-					  "qcom,ufshc"
+- compatible		: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0"
+
+			  For Qualcomm SoCs must contain, as below, an
+			  SoC-specific compatible along with "qcom,ufshc" and
+			  the appropriate jedec string:
+			    "qcom,msm8994-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
+			    "qcom,msm8996-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
+			    "qcom,sdm845-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
 - interrupts        : <interrupt mapping for UFS host controller IRQ>
 - reg               : <registers mapping>
 
-- 
2.19.0.605.g01d371f741-goog

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Vivek Gautam]

Hi Doug,

On Sat, Oct 13, 2018 at 3:09 AM Douglas Anderson <dianders@chromium.org> wrote:
>
> If you look at the bindings for the UFS Host Controller it says:
>
> - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
>               also list one or more of the following:
>                  "qcom,msm8994-ufshc"
>                  "qcom,msm8996-ufshc"
>                  "qcom,ufshc"
>
> My reading of that is that it's fine to just have either of these:
> 1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
> 2. "qcom,ufshc", "jedec,ufs-2.0"
>
> As far as I can tell neither of the above is actually a good idea.
>
> For #1 it turns out that the driver currently only keys off the
> compatible string "qcom,ufshc" so it won't actually probe.
>
> For #2 the driver won't probe but it's not a good idea to keep the SoC
> name out of the compatible string.
>
> Let's update the compatible string to make it really explicit.  We'll
> include a nod to the existing driver and the old binding and say that
> we should always include the "qcom,ufshc" string in addition to the
> SoC compatible string.
>
> While we're at it we'll also include another example SoC known to have
> UFS: sdm845.
>
> Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> ---
>
>  .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> index 2df00524bd21..69a06a1b732e 100644
> --- a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> +++ b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> @@ -4,11 +4,14 @@ UFSHC nodes are defined to describe on-chip UFS host controllers.
>  Each UFS controller instance should have its own node.
>
>  Required properties:
> -- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> -                         also list one or more of the following:
> -                                         "qcom,msm8994-ufshc"
> -                                         "qcom,msm8996-ufshc"
> -                                         "qcom,ufshc"
> +- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0"
> +
> +                         For Qualcomm SoCs must contain, as below, an
> +                         SoC-specific compatible along with "qcom,ufshc" and
> +                         the appropriate jedec string:
> +                           "qcom,msm8994-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> +                           "qcom,msm8996-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> +                           "qcom,sdm845-ufshc", "qcom,ufshc", "jedec,ufs-2.0"

Thanks for the patch. It looks good to me.
Reviewed-by: Vivek Gautam <vivek.gautam@codeaurora.org>

P.S.: While you are at it, can you please move 'ufs-qcom.txt'
to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
The current name and file location is misleading.

Thanks & Regards
Vivek

-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Doug Anderson]

Vivek,

On Mon, Oct 15, 2018 at 8:23 AM Vivek Gautam
<vivek.gautam@codeaurora.org> wrote:
>
> Hi Doug,
>
> On Sat, Oct 13, 2018 at 3:09 AM Douglas Anderson <dianders@chromium.org> wrote:
> >
> > If you look at the bindings for the UFS Host Controller it says:
> >
> > - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> >               also list one or more of the following:
> >                  "qcom,msm8994-ufshc"
> >                  "qcom,msm8996-ufshc"
> >                  "qcom,ufshc"
> >
> > My reading of that is that it's fine to just have either of these:
> > 1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
> > 2. "qcom,ufshc", "jedec,ufs-2.0"
> >
> > As far as I can tell neither of the above is actually a good idea.
> >
> > For #1 it turns out that the driver currently only keys off the
> > compatible string "qcom,ufshc" so it won't actually probe.
> >
> > For #2 the driver won't probe but it's not a good idea to keep the SoC
> > name out of the compatible string.
> >
> > Let's update the compatible string to make it really explicit.  We'll
> > include a nod to the existing driver and the old binding and say that
> > we should always include the "qcom,ufshc" string in addition to the
> > SoC compatible string.
> >
> > While we're at it we'll also include another example SoC known to have
> > UFS: sdm845.
> >
> > Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
> > Signed-off-by: Douglas Anderson <dianders@chromium.org>
> > ---
> >
> >  .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
> >  1 file changed, 8 insertions(+), 5 deletions(-)
> >
> > diff --git a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > index 2df00524bd21..69a06a1b732e 100644
> > --- a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > +++ b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > @@ -4,11 +4,14 @@ UFSHC nodes are defined to describe on-chip UFS host controllers.
> >  Each UFS controller instance should have its own node.
> >
> >  Required properties:
> > -- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> > -                         also list one or more of the following:
> > -                                         "qcom,msm8994-ufshc"
> > -                                         "qcom,msm8996-ufshc"
> > -                                         "qcom,ufshc"
> > +- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0"
> > +
> > +                         For Qualcomm SoCs must contain, as below, an
> > +                         SoC-specific compatible along with "qcom,ufshc" and
> > +                         the appropriate jedec string:
> > +                           "qcom,msm8994-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> > +                           "qcom,msm8996-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> > +                           "qcom,sdm845-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
>
> Thanks for the patch. It looks good to me.
> Reviewed-by: Vivek Gautam <vivek.gautam@codeaurora.org>

Thanks for the review!


> P.S.: While you are at it, can you please move 'ufs-qcom.txt'
> to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
> The current name and file location is misleading.

I'd rather someone at Qualcomm do this.  Do you have a suggested
person?  The reason I feel that Qualcomm needs to get involved is that
I see that when I look at the file you refer to says it's for:

  "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
  "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
  "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
  present on MSM8996 chipset.

...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
handles the compatible string:

       "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.

So I'm a little confused.  Should the SDM845 UFS PHY been handled by
the older UFS PHY driver?  ...or should all the older UFS PHYs be
moved to be handled by the newer QMP PHY driver?  ...or are they
really different hardware blocks, in which case how would you describe
the difference (both are described as UFS QMP PHYs I think).

BTW: I have a patch attempting to fixup the QMP PHY bindings at
<https://lkml.kernel.org/r/20181012213632.252346-1-dianders@chromium.org>.


-Doug

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Vivek Gautam]

On Mon, Oct 15, 2018 at 10:54 PM Doug Anderson <dianders@chromium.org> wrote:
>
> Vivek,
>
> On Mon, Oct 15, 2018 at 8:23 AM Vivek Gautam
> <vivek.gautam@codeaurora.org> wrote:
> >
> > Hi Doug,
> >
> > On Sat, Oct 13, 2018 at 3:09 AM Douglas Anderson <dianders@chromium.org> wrote:
> > >
> > > If you look at the bindings for the UFS Host Controller it says:
> > >
> > > - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> > >               also list one or more of the following:
> > >                  "qcom,msm8994-ufshc"
> > >                  "qcom,msm8996-ufshc"
> > >                  "qcom,ufshc"
> > >
> > > My reading of that is that it's fine to just have either of these:
> > > 1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
> > > 2. "qcom,ufshc", "jedec,ufs-2.0"
> > >
> > > As far as I can tell neither of the above is actually a good idea.
> > >
> > > For #1 it turns out that the driver currently only keys off the
> > > compatible string "qcom,ufshc" so it won't actually probe.
> > >
> > > For #2 the driver won't probe but it's not a good idea to keep the SoC
> > > name out of the compatible string.
> > >
> > > Let's update the compatible string to make it really explicit.  We'll
> > > include a nod to the existing driver and the old binding and say that
> > > we should always include the "qcom,ufshc" string in addition to the
> > > SoC compatible string.
> > >
> > > While we're at it we'll also include another example SoC known to have
> > > UFS: sdm845.
> > >
> > > Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
> > > Signed-off-by: Douglas Anderson <dianders@chromium.org>
> > > ---
> > >
> > >  .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
> > >  1 file changed, 8 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > > index 2df00524bd21..69a06a1b732e 100644
> > > --- a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > > +++ b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> > > @@ -4,11 +4,14 @@ UFSHC nodes are defined to describe on-chip UFS host controllers.
> > >  Each UFS controller instance should have its own node.
> > >
> > >  Required properties:
> > > -- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> > > -                         also list one or more of the following:
> > > -                                         "qcom,msm8994-ufshc"
> > > -                                         "qcom,msm8996-ufshc"
> > > -                                         "qcom,ufshc"
> > > +- compatible           : must contain "jedec,ufs-1.1" or "jedec,ufs-2.0"
> > > +
> > > +                         For Qualcomm SoCs must contain, as below, an
> > > +                         SoC-specific compatible along with "qcom,ufshc" and
> > > +                         the appropriate jedec string:
> > > +                           "qcom,msm8994-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> > > +                           "qcom,msm8996-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> > > +                           "qcom,sdm845-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> >
> > Thanks for the patch. It looks good to me.
> > Reviewed-by: Vivek Gautam <vivek.gautam@codeaurora.org>
>
> Thanks for the review!
>
>
> > P.S.: While you are at it, can you please move 'ufs-qcom.txt'
> > to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
> > The current name and file location is misleading.
>
> I'd rather someone at Qualcomm do this.  Do you have a suggested
> person?  The reason I feel that Qualcomm needs to get involved is that
> I see that when I look at the file you refer to says it's for:
>
>   "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
>   "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
>   "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
>   present on MSM8996 chipset.
>
> ...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
> handles the compatible string:
>
>        "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.
>
> So I'm a little confused.  Should the SDM845 UFS PHY been handled by
> the older UFS PHY driver?  ...or should all the older UFS PHYs be
> moved to be handled by the newer QMP PHY driver?  ...or are they
> really different hardware blocks, in which case how would you describe
> the difference (both are described as UFS QMP PHYs I think).

As you rightly said "ufs/ufs-qcom.txt" describes the bindings for
14nm, and 20nm ufs phy. These phys are however handled by the older
ufs phy driver present at:
drivers/phy/qualcomm/phy-qcom-ufs-qmp-{14nm,20nm}.c
The sdm845 UFS phy driver is handled by the new consolidated qmp phy
driver: drivers/phy/qualcomm/phy-qcom-qmp.c whose bindings are
described by 'qcom-qmp-phy.txt'.
We didn't attempt to move the 14nm phy to new driver as we already had
8996 using the bindings.

So, really these are two separate drivers with different bindings. I
believe it should be okay to move the file. If you are fine, I can
attempt to post a small patch to do that.
Thanks.

>
> BTW: I have a patch attempting to fixup the QMP PHY bindings at
> <https://lkml.kernel.org/r/20181012213632.252346-1-dianders@chromium.org>.

Sure, I will review this as well. Thanks for reminiding.

Best regards
Vivek


-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Doug Anderson]

Hi,

On Mon, Oct 15, 2018 at 10:51 PM Vivek Gautam
<vivek.gautam@codeaurora.org> wrote:
> > > P.S.: While you are at it, can you please move 'ufs-qcom.txt'
> > > to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
> > > The current name and file location is misleading.
> >
> > I'd rather someone at Qualcomm do this.  Do you have a suggested
> > person?  The reason I feel that Qualcomm needs to get involved is that
> > I see that when I look at the file you refer to says it's for:
> >
> >   "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
> >   "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
> >   "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
> >   present on MSM8996 chipset.
> >
> > ...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
> > handles the compatible string:
> >
> >        "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.
> >
> > So I'm a little confused.  Should the SDM845 UFS PHY been handled by
> > the older UFS PHY driver?  ...or should all the older UFS PHYs be
> > moved to be handled by the newer QMP PHY driver?  ...or are they
> > really different hardware blocks, in which case how would you describe
> > the difference (both are described as UFS QMP PHYs I think).
>
> As you rightly said "ufs/ufs-qcom.txt" describes the bindings for
> 14nm, and 20nm ufs phy. These phys are however handled by the older
> ufs phy driver present at:
> drivers/phy/qualcomm/phy-qcom-ufs-qmp-{14nm,20nm}.c
> The sdm845 UFS phy driver is handled by the new consolidated qmp phy
> driver: drivers/phy/qualcomm/phy-qcom-qmp.c whose bindings are
> described by 'qcom-qmp-phy.txt'.
> We didn't attempt to move the 14nm phy to new driver as we already had
> 8996 using the bindings.
>
> So, really these are two separate drivers with different bindings. I
> believe it should be okay to move the file. If you are fine, I can
> attempt to post a small patch to do that.

I guess what I should have said was that the new name you're proposing
"qcom,ufs-phy.txt" is confusing and opening the file doesn't help
clarify things.  The name and the binding make it sound like this is
_the_ file to look at for Qualcomm UFS PHYs.  ...and then you look in
the examples in the file and it seems that this even includes Qualcomm
QMP PHYs for UFS.

...so while I agree that the file "ufs-qcom.txt" needs to be moved to
the "PHY" directory, I think at the same time we need to change the
name of the file and maybe the contents to disambiguate which things
belong in this file vs. the "qcom-qmp-phy.txt".  ...and I feel like
someone at Qualcomm will have the most information to properly do
that.

For instance, you could call the older bindings
"qcom-qmp-phy-14nm-20nm.txt" or something like that.

One point of clarification I'd like to know is if there's really a
good reason to have two drivers here.  Certainly if the hardware is
really different then a new driver can make sense, but if there are
two drivers for arbitrary reasons then maybe they should be combined
into one eventually?


-Doug

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Vivek Gautam]

Hi Doug,


On 10/16/2018 10:29 PM, Doug Anderson wrote:
> Hi,
>
> On Mon, Oct 15, 2018 at 10:51 PM Vivek Gautam
> <vivek.gautam@codeaurora.org> wrote:
>>>> P.S.: While you are at it, can you please move 'ufs-qcom.txt'
>>>> to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
>>>> The current name and file location is misleading.
>>> I'd rather someone at Qualcomm do this.  Do you have a suggested
>>> person?  The reason I feel that Qualcomm needs to get involved is that
>>> I see that when I look at the file you refer to says it's for:
>>>
>>>    "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
>>>    "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
>>>    "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
>>>    present on MSM8996 chipset.
>>>
>>> ...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
>>> handles the compatible string:
>>>
>>>         "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.
>>>
>>> So I'm a little confused.  Should the SDM845 UFS PHY been handled by
>>> the older UFS PHY driver?  ...or should all the older UFS PHYs be
>>> moved to be handled by the newer QMP PHY driver?  ...or are they
>>> really different hardware blocks, in which case how would you describe
>>> the difference (both are described as UFS QMP PHYs I think).
>> As you rightly said "ufs/ufs-qcom.txt" describes the bindings for
>> 14nm, and 20nm ufs phy. These phys are however handled by the older
>> ufs phy driver present at:
>> drivers/phy/qualcomm/phy-qcom-ufs-qmp-{14nm,20nm}.c
>> The sdm845 UFS phy driver is handled by the new consolidated qmp phy
>> driver: drivers/phy/qualcomm/phy-qcom-qmp.c whose bindings are
>> described by 'qcom-qmp-phy.txt'.
>> We didn't attempt to move the 14nm phy to new driver as we already had
>> 8996 using the bindings.
>>
>> So, really these are two separate drivers with different bindings. I
>> believe it should be okay to move the file. If you are fine, I can
>> attempt to post a small patch to do that.
> I guess what I should have said was that the new name you're proposing
> "qcom,ufs-phy.txt" is confusing and opening the file doesn't help
> clarify things.  The name and the binding make it sound like this is
> _the_ file to look at for Qualcomm UFS PHYs.  ...and then you look in
> the examples in the file and it seems that this even includes Qualcomm
> QMP PHYs for UFS.
>
> ...so while I agree that the file "ufs-qcom.txt" needs to be moved to
> the "PHY" directory, I think at the same time we need to change the
> name of the file and maybe the contents to disambiguate which things
> belong in this file vs. the "qcom-qmp-phy.txt".  ...and I feel like
> someone at Qualcomm will have the most information to properly do
> that.
>
> For instance, you could call the older bindings
> "qcom-qmp-phy-14nm-20nm.txt" or something like that.

Sure, I get your point. I will propose something that removes the confusion.

>
> One point of clarification I'd like to know is if there's really a
> good reason to have two drivers here.  Certainly if the hardware is
> really different then a new driver can make sense, but if there are
> two drivers for arbitrary reasons then maybe they should be combined
> into one eventually?

Right, the 14nm phy driver can be happily merged into the new qmp-phy 
driver.
But we should take care of older bindings. Removing the driver will break
things on targets with older bindings, precisely 8996.

20nm is bit tricky as it exported few APIs directly to ufs host 
controller, and
that's the reason we have declared that as BROKEN after the ufs cleanup.
So, until we are really in a kill mode, the old ufs-phy driver will have 
to live.

Regards
Vivek
>
>
> -Doug

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Doug Anderson]

Hi,

On Tue, Oct 16, 2018 at 11:28 PM Vivek Gautam
<vivek.gautam@codeaurora.org> wrote:
>
> Hi Doug,
>
>
> On 10/16/2018 10:29 PM, Doug Anderson wrote:
> > Hi,
> >
> > On Mon, Oct 15, 2018 at 10:51 PM Vivek Gautam
> > <vivek.gautam@codeaurora.org> wrote:
> >>>> P.S.: While you are at it, can you please move 'ufs-qcom.txt'
> >>>> to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
> >>>> The current name and file location is misleading.
> >>> I'd rather someone at Qualcomm do this.  Do you have a suggested
> >>> person?  The reason I feel that Qualcomm needs to get involved is that
> >>> I see that when I look at the file you refer to says it's for:
> >>>
> >>>    "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
> >>>    "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
> >>>    "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
> >>>    present on MSM8996 chipset.
> >>>
> >>> ...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
> >>> handles the compatible string:
> >>>
> >>>         "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.
> >>>
> >>> So I'm a little confused.  Should the SDM845 UFS PHY been handled by
> >>> the older UFS PHY driver?  ...or should all the older UFS PHYs be
> >>> moved to be handled by the newer QMP PHY driver?  ...or are they
> >>> really different hardware blocks, in which case how would you describe
> >>> the difference (both are described as UFS QMP PHYs I think).
> >> As you rightly said "ufs/ufs-qcom.txt" describes the bindings for
> >> 14nm, and 20nm ufs phy. These phys are however handled by the older
> >> ufs phy driver present at:
> >> drivers/phy/qualcomm/phy-qcom-ufs-qmp-{14nm,20nm}.c
> >> The sdm845 UFS phy driver is handled by the new consolidated qmp phy
> >> driver: drivers/phy/qualcomm/phy-qcom-qmp.c whose bindings are
> >> described by 'qcom-qmp-phy.txt'.
> >> We didn't attempt to move the 14nm phy to new driver as we already had
> >> 8996 using the bindings.
> >>
> >> So, really these are two separate drivers with different bindings. I
> >> believe it should be okay to move the file. If you are fine, I can
> >> attempt to post a small patch to do that.
> > I guess what I should have said was that the new name you're proposing
> > "qcom,ufs-phy.txt" is confusing and opening the file doesn't help
> > clarify things.  The name and the binding make it sound like this is
> > _the_ file to look at for Qualcomm UFS PHYs.  ...and then you look in
> > the examples in the file and it seems that this even includes Qualcomm
> > QMP PHYs for UFS.
> >
> > ...so while I agree that the file "ufs-qcom.txt" needs to be moved to
> > the "PHY" directory, I think at the same time we need to change the
> > name of the file and maybe the contents to disambiguate which things
> > belong in this file vs. the "qcom-qmp-phy.txt".  ...and I feel like
> > someone at Qualcomm will have the most information to properly do
> > that.
> >
> > For instance, you could call the older bindings
> > "qcom-qmp-phy-14nm-20nm.txt" or something like that.
>
> Sure, I get your point. I will propose something that removes the confusion.
>
> >
> > One point of clarification I'd like to know is if there's really a
> > good reason to have two drivers here.  Certainly if the hardware is
> > really different then a new driver can make sense, but if there are
> > two drivers for arbitrary reasons then maybe they should be combined
> > into one eventually?
>
> Right, the 14nm phy driver can be happily merged into the new qmp-phy
> driver.
> But we should take care of older bindings. Removing the driver will break
> things on targets with older bindings, precisely 8996.
>
> 20nm is bit tricky as it exported few APIs directly to ufs host
> controller, and
> that's the reason we have declared that as BROKEN after the ufs cleanup.
> So, until we are really in a kill mode, the old ufs-phy driver will have
> to live.

OK, sounds like a plan.  I'll assume you're posting the patch to move
the old PHY bindings and add some of the above information to them so
people aren't confused.

...all this is sort off the original subject, though.  ;-)  Just a
quick summary here is that nothing suggests ${SUBJECT} patch shouldn't
land and all the additional discussion has been about making further
improvements to the bindings situation for UFS on Qualcomm.

-Doug

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Rob Herring]

On Fri, 12 Oct 2018 14:39:26 -0700, Douglas Anderson wrote:
> If you look at the bindings for the UFS Host Controller it says:
> 
> - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
>               also list one or more of the following:
>                  "qcom,msm8994-ufshc"
>                  "qcom,msm8996-ufshc"
>                  "qcom,ufshc"
> 
> My reading of that is that it's fine to just have either of these:
> 1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
> 2. "qcom,ufshc", "jedec,ufs-2.0"
> 
> As far as I can tell neither of the above is actually a good idea.
> 
> For #1 it turns out that the driver currently only keys off the
> compatible string "qcom,ufshc" so it won't actually probe.
> 
> For #2 the driver won't probe but it's not a good idea to keep the SoC
> name out of the compatible string.
> 
> Let's update the compatible string to make it really explicit.  We'll
> include a nod to the existing driver and the old binding and say that
> we should always include the "qcom,ufshc" string in addition to the
> SoC compatible string.
> 
> While we're at it we'll also include another example SoC known to have
> UFS: sdm845.
> 
> Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> ---
> 
>  .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
> 

Reviewed-by: Rob Herring <robh@kernel.org>

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Vivek Gautam]

On 10/17/2018 9:41 PM, Doug Anderson wrote:
> Hi,
>
> On Tue, Oct 16, 2018 at 11:28 PM Vivek Gautam
> <vivek.gautam@codeaurora.org> wrote:
>> Hi Doug,
>>
>>
>> On 10/16/2018 10:29 PM, Doug Anderson wrote:
>>> Hi,
>>>
>>> On Mon, Oct 15, 2018 at 10:51 PM Vivek Gautam
>>> <vivek.gautam@codeaurora.org> wrote:
>>>>>> P.S.: While you are at it, can you please move 'ufs-qcom.txt'
>>>>>> to Documentation/devicetree/bindings/phy/qcom,ufs-phy.txt.
>>>>>> The current name and file location is misleading.
>>>>> I'd rather someone at Qualcomm do this.  Do you have a suggested
>>>>> person?  The reason I feel that Qualcomm needs to get involved is that
>>>>> I see that when I look at the file you refer to says it's for:
>>>>>
>>>>>     "qcom,ufs-phy-qmp-20nm" for 20nm ufs phy,
>>>>>     "qcom,ufs-phy-qmp-14nm" for legacy 14nm ufs phy,
>>>>>     "qcom,msm8996-ufs-phy-qmp-14nm" for 14nm ufs phy
>>>>>     present on MSM8996 chipset.
>>>>>
>>>>> ...but there's another Qualcomm file, 'qcom-qmp-phy.txt'.  That
>>>>> handles the compatible string:
>>>>>
>>>>>          "qcom,sdm845-qmp-ufs-phy" for UFS QMP phy on sdm845.
>>>>>
>>>>> So I'm a little confused.  Should the SDM845 UFS PHY been handled by
>>>>> the older UFS PHY driver?  ...or should all the older UFS PHYs be
>>>>> moved to be handled by the newer QMP PHY driver?  ...or are they
>>>>> really different hardware blocks, in which case how would you describe
>>>>> the difference (both are described as UFS QMP PHYs I think).
>>>> As you rightly said "ufs/ufs-qcom.txt" describes the bindings for
>>>> 14nm, and 20nm ufs phy. These phys are however handled by the older
>>>> ufs phy driver present at:
>>>> drivers/phy/qualcomm/phy-qcom-ufs-qmp-{14nm,20nm}.c
>>>> The sdm845 UFS phy driver is handled by the new consolidated qmp phy
>>>> driver: drivers/phy/qualcomm/phy-qcom-qmp.c whose bindings are
>>>> described by 'qcom-qmp-phy.txt'.
>>>> We didn't attempt to move the 14nm phy to new driver as we already had
>>>> 8996 using the bindings.
>>>>
>>>> So, really these are two separate drivers with different bindings. I
>>>> believe it should be okay to move the file. If you are fine, I can
>>>> attempt to post a small patch to do that.
>>> I guess what I should have said was that the new name you're proposing
>>> "qcom,ufs-phy.txt" is confusing and opening the file doesn't help
>>> clarify things.  The name and the binding make it sound like this is
>>> _the_ file to look at for Qualcomm UFS PHYs.  ...and then you look in
>>> the examples in the file and it seems that this even includes Qualcomm
>>> QMP PHYs for UFS.
>>>
>>> ...so while I agree that the file "ufs-qcom.txt" needs to be moved to
>>> the "PHY" directory, I think at the same time we need to change the
>>> name of the file and maybe the contents to disambiguate which things
>>> belong in this file vs. the "qcom-qmp-phy.txt".  ...and I feel like
>>> someone at Qualcomm will have the most information to properly do
>>> that.
>>>
>>> For instance, you could call the older bindings
>>> "qcom-qmp-phy-14nm-20nm.txt" or something like that.
>> Sure, I get your point. I will propose something that removes the confusion.
>>
>>> One point of clarification I'd like to know is if there's really a
>>> good reason to have two drivers here.  Certainly if the hardware is
>>> really different then a new driver can make sense, but if there are
>>> two drivers for arbitrary reasons then maybe they should be combined
>>> into one eventually?
>> Right, the 14nm phy driver can be happily merged into the new qmp-phy
>> driver.
>> But we should take care of older bindings. Removing the driver will break
>> things on targets with older bindings, precisely 8996.
>>
>> 20nm is bit tricky as it exported few APIs directly to ufs host
>> controller, and
>> that's the reason we have declared that as BROKEN after the ufs cleanup.
>> So, until we are really in a kill mode, the old ufs-phy driver will have
>> to live.
> OK, sounds like a plan.  I'll assume you're posting the patch to move
> the old PHY bindings and add some of the above information to them so
> people aren't confused.
>
> ...all this is sort off the original subject, though.  ;-)  Just a
> quick summary here is that nothing suggests ${SUBJECT} patch shouldn't
> land and all the additional discussion has been about making further
> improvements to the bindings situation for UFS on Qualcomm.

Yes, this patch is good to go.

Thanks
Vivek
>
> -Doug

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Evan Green]

On Thu, Oct 18, 2018 at 12:52 AM Vivek Gautam
<vivek.gautam@codeaurora.org> wrote:
>
...
>
> Yes, this patch is good to go.
>

I don't think this patch ever got picked up, despite being reviewed
and good to go. Can it be applied somewhere?
-Evan

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Marc Gonzalez]

On 12/10/2018 23:39, Douglas Anderson wrote:

> If you look at the bindings for the UFS Host Controller it says:
> 
> - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
>               also list one or more of the following:
>                  "qcom,msm8994-ufshc"
>                  "qcom,msm8996-ufshc"
>                  "qcom,ufshc"
> 
> My reading of that is that it's fine to just have either of these:
> 1. "qcom,msm8996-ufshc", "jedec,ufs-2.0"
> 2. "qcom,ufshc", "jedec,ufs-2.0"
> 
> As far as I can tell neither of the above is actually a good idea.
> 
> For #1 it turns out that the driver currently only keys off the
> compatible string "qcom,ufshc" so it won't actually probe.
> 
> For #2 the driver won't probe but it's not a good idea to keep the SoC
> name out of the compatible string.
> 
> Let's update the compatible string to make it really explicit.  We'll
> include a nod to the existing driver and the old binding and say that
> we should always include the "qcom,ufshc" string in addition to the
> SoC compatible string.
> 
> While we're at it we'll also include another example SoC known to have
> UFS: sdm845.
> 
> Fixes: 47555a5c8a11 ("scsi: ufs: make the UFS variant a platform device")
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> ---
> 
>  .../devicetree/bindings/ufs/ufshcd-pltfrm.txt       | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> index 2df00524bd21..69a06a1b732e 100644
> --- a/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> +++ b/Documentation/devicetree/bindings/ufs/ufshcd-pltfrm.txt
> @@ -4,11 +4,14 @@ UFSHC nodes are defined to describe on-chip UFS host controllers.
>  Each UFS controller instance should have its own node.
>  
>  Required properties:
> -- compatible		: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
> -			  also list one or more of the following:
> -					  "qcom,msm8994-ufshc"
> -					  "qcom,msm8996-ufshc"
> -					  "qcom,ufshc"
> +- compatible		: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0"
> +
> +			  For Qualcomm SoCs must contain, as below, an
> +			  SoC-specific compatible along with "qcom,ufshc" and
> +			  the appropriate jedec string:
> +			    "qcom,msm8994-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> +			    "qcom,msm8996-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
> +			    "qcom,sdm845-ufshc", "qcom,ufshc", "jedec,ufs-2.0"
>  - interrupts        : <interrupt mapping for UFS host controller IRQ>
>  - reg               : <registers mapping>
>  

Tweaking CC list.

Martin, this one's for you, according to Rob ;-)

https://lore.kernel.org/patchwork/patch/999046/

FWIW, it has already been:

Reviewed-by: Vivek Gautam <vivek.gautam@codeaurora.org>
Reviewed-by: Rob Herring <robh@kernel.org>

Regards.

Re: [PATCH] dt-bindings: ufs: Fix the compatible string definition

[Martin K. Petersen]

Marc,

>> - compatible: must contain "jedec,ufs-1.1" or "jedec,ufs-2.0", may
>>               also list one or more of the following:
>>                  "qcom,msm8994-ufshc"
>>                  "qcom,msm8996-ufshc"
>>                  "qcom,ufshc"

Applied to 5.1/scsi-queue, thanks!

-- 
Martin K. Petersen	Oracle Linux Engineering