Hi Mike,
Today's linux-next merge of the clk tree got a conflict in
drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
dedicated spinlock for the SWDT") and 765b7d4c4cb3
("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
flag") from the clk tree.
I fixed it up (see below and in a couple of places I chose
CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
be wrong) and can carry the fix as necessary (no action is required).
--
Cheers,
Stephen Rothwell [email protected]
diff --cc drivers/clk/zynq/clkc.c
index 089d3e3,e05c9e3..0000000
--- a/drivers/clk/zynq/clkc.c
+++ b/drivers/clk/zynq/clkc.c
@@@ -293,8 -294,9 +295,9 @@@ static void __init zynq_clk_setup(struc
swdt_ext_clk_mux_parents[i + 1] = dummy_nm;
}
clks[swdt] = clk_register_mux(NULL, clk_output_name[swdt],
- swdt_ext_clk_mux_parents, 2, CLK_SET_RATE_PARENT,
- SLCR_SWDT_CLK_SEL, 0, 1, 0, &swdtclk_lock);
+ swdt_ext_clk_mux_parents, 2, CLK_SET_RATE_PARENT |
+ CLK_SET_RATE_NO_REPARENT, SLCR_SWDT_CLK_SEL, 0, 1, 0,
- &gem0clk_lock);
++ &swdtclk_lock);
/* DDR clocks */
clk = clk_register_divider(NULL, "ddr2x_div", "ddrpll", 0,
On 27/08/13 10:03, Stephen Rothwell wrote:
> Hi Mike,
>
> Today's linux-next merge of the clk tree got a conflict in
> drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> dedicated spinlock for the SWDT") and 765b7d4c4cb3
> ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> flag") from the clk tree.
>
> I fixed it up (see below and in a couple of places I chose
> CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> be wrong) and can carry the fix as necessary (no action is required).
The case you mentioned looks correct to me.
I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
are orthogonal.
Thanks
James
On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
> On 27/08/13 10:03, Stephen Rothwell wrote:
> > Hi Mike,
> >
> > Today's linux-next merge of the clk tree got a conflict in
> > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> > dedicated spinlock for the SWDT") and 765b7d4c4cb3
> > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> > flag") from the clk tree.
> >
> > I fixed it up (see below and in a couple of places I chose
> > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> > be wrong) and can carry the fix as necessary (no action is required).
>
> The case you mentioned looks correct to me.
>
> I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
> over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
> clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
> only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
> are orthogonal.
I can just agree, the case included in the mail looks correct, but in
case of other conflicts both flags should be set. Just like in the case
shown here.
Sören
Quoting Sören Brinkmann (2013-08-27 08:44:11)
> On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
> > On 27/08/13 10:03, Stephen Rothwell wrote:
> > > Hi Mike,
> > >
> > > Today's linux-next merge of the clk tree got a conflict in
> > > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> > > dedicated spinlock for the SWDT") and 765b7d4c4cb3
> > > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> > > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> > > flag") from the clk tree.
> > >
> > > I fixed it up (see below and in a couple of places I chose
> > > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> > > be wrong) and can carry the fix as necessary (no action is required).
> >
> > The case you mentioned looks correct to me.
> >
> > I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
> > over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
> > clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
> > only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
> > are orthogonal.
>
> I can just agree, the case included in the mail looks correct, but in
> case of other conflicts both flags should be set. Just like in the case
> shown here.
Stephen's fix is correct. The Zynq patches came in as fixes so I think
this will be a rare event.
Regards,
Mike
>
> Sören
Hi all,
On Tue, 27 Aug 2013 09:53:19 -0700 Mike Turquette <[email protected]> wrote:
>
> Quoting Sören Brinkmann (2013-08-27 08:44:11)
> > On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
> > > On 27/08/13 10:03, Stephen Rothwell wrote:
> > > > Hi Mike,
> > > >
> > > > Today's linux-next merge of the clk tree got a conflict in
> > > > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> > > > dedicated spinlock for the SWDT") and 765b7d4c4cb3
> > > > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> > > > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> > > > flag") from the clk tree.
> > > >
> > > > I fixed it up (see below and in a couple of places I chose
> > > > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> > > > be wrong) and can carry the fix as necessary (no action is required).
> > >
> > > The case you mentioned looks correct to me.
> > >
> > > I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
> > > over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
> > > clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
> > > only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
> > > are orthogonal.
> >
> > I can just agree, the case included in the mail looks correct, but in
> > case of other conflicts both flags should be set. Just like in the case
> > shown here.
>
> Stephen's fix is correct. The Zynq patches came in as fixes so I think
> this will be a rare event.
Can you guys discuss this and come up with a single answer. I read the above as:
(for the two places I used CLK_SET_RATE_NO_REPARENT where the two
branches each added that and CLK_SET_RATE_NOREPARENT respectively)
"Stephen was wrong"
"Stephen should have taken both"
"Stephen was right"
:-)
I can fix up my merge resolution if you tell me the correct fix. Also,
you will need to know so that you can tell Linus (or whoever else has to
resolve these conflicts).
--
Cheers,
Stephen Rothwell [email protected]
Hi all,
On Wed, 28 Aug 2013 10:04:31 +1000 Stephen Rothwell <[email protected]> wrote:
>
> On Tue, 27 Aug 2013 09:53:19 -0700 Mike Turquette <[email protected]> wrote:
> >
> > Quoting Sören Brinkmann (2013-08-27 08:44:11)
> > > On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
> > > > On 27/08/13 10:03, Stephen Rothwell wrote:
> > > > > Hi Mike,
> > > > >
> > > > > Today's linux-next merge of the clk tree got a conflict in
> > > > > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> > > > > dedicated spinlock for the SWDT") and 765b7d4c4cb3
> > > > > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> > > > > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> > > > > flag") from the clk tree.
> > > > >
> > > > > I fixed it up (see below and in a couple of places I chose
> > > > > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> > > > > be wrong) and can carry the fix as necessary (no action is required).
> > > >
> > > > The case you mentioned looks correct to me.
> > > >
> > > > I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
> > > > over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
> > > > clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
> > > > only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
> > > > are orthogonal.
> > >
> > > I can just agree, the case included in the mail looks correct, but in
> > > case of other conflicts both flags should be set. Just like in the case
> > > shown here.
> >
> > Stephen's fix is correct. The Zynq patches came in as fixes so I think
> > this will be a rare event.
>
> Can you guys discuss this and come up with a single answer. I read the above as:
>
> (for the two places I used CLK_SET_RATE_NO_REPARENT where the two
> branches each added that and CLK_SET_RATE_NOREPARENT respectively)
>
> "Stephen was wrong"
> "Stephen should have taken both"
> "Stephen was right"
>
> :-)
>
> I can fix up my merge resolution if you tell me the correct fix. Also,
> you will need to know so that you can tell Linus (or whoever else has to
> resolve these conflicts).
OK, I thought about it some more and the resolution now looks like
below. Is this correct/better?
--
Cheers,
Stephen Rothwell [email protected]
6c92037c7a4574c16bc86873fe603b39b7fbe98a
diff --cc drivers/clk/zynq/clkc.c
index 089d3e3,e05c9e3..6f104db
--- a/drivers/clk/zynq/clkc.c
+++ b/drivers/clk/zynq/clkc.c
@@@ -293,8 -294,9 +295,9 @@@ static void __init zynq_clk_setup(struc
swdt_ext_clk_mux_parents[i + 1] = dummy_nm;
}
clks[swdt] = clk_register_mux(NULL, clk_output_name[swdt],
- swdt_ext_clk_mux_parents, 2, CLK_SET_RATE_PARENT,
- SLCR_SWDT_CLK_SEL, 0, 1, 0, &swdtclk_lock);
+ swdt_ext_clk_mux_parents, 2, CLK_SET_RATE_PARENT |
+ CLK_SET_RATE_NO_REPARENT, SLCR_SWDT_CLK_SEL, 0, 1, 0,
- &gem0clk_lock);
++ &swdtclk_lock);
/* DDR clocks */
clk = clk_register_divider(NULL, "ddr2x_div", "ddrpll", 0,
@@@ -366,8 -369,8 +370,8 @@@
CLK_DIVIDER_ONE_BASED | CLK_DIVIDER_ALLOW_ZERO,
&gem0clk_lock);
clk = clk_register_mux(NULL, "gem0_emio_mux", gem0_mux_parents, 2,
- CLK_SET_RATE_PARENT, SLCR_GEM0_CLK_CTRL, 6, 1, 0,
- CLK_SET_RATE_NO_REPARENT, SLCR_GEM0_CLK_CTRL, 6, 1, 0,
-- &gem0clk_lock);
++ CLK_SET_RATE_PARENT | CLK_SET_RATE_NO_REPARENT,
++ SLCR_GEM0_CLK_CTRL, 6, 1, 0, &gem0clk_lock);
clks[gem0] = clk_register_gate(NULL, clk_output_name[gem0],
"gem0_emio_mux", CLK_SET_RATE_PARENT,
SLCR_GEM0_CLK_CTRL, 0, 0, &gem0clk_lock);
@@@ -389,8 -393,8 +394,8 @@@
CLK_DIVIDER_ONE_BASED | CLK_DIVIDER_ALLOW_ZERO,
&gem1clk_lock);
clk = clk_register_mux(NULL, "gem1_emio_mux", gem1_mux_parents, 2,
- CLK_SET_RATE_PARENT, SLCR_GEM1_CLK_CTRL, 6, 1, 0,
- CLK_SET_RATE_NO_REPARENT, SLCR_GEM1_CLK_CTRL, 6, 1, 0,
-- &gem1clk_lock);
++ CLK_SET_RATE_PARENT | CLK_SET_RATE_NO_REPARENT,
++ SLCR_GEM1_CLK_CTRL, 6, 1, 0, &gem1clk_lock);
clks[gem1] = clk_register_gate(NULL, clk_output_name[gem1],
"gem1_emio_mux", CLK_SET_RATE_PARENT,
SLCR_GEM1_CLK_CTRL, 0, 0, &gem1clk_lock);
Hi Stephen
On 28/08/13 08:22, Stephen Rothwell wrote:
> On Wed, 28 Aug 2013 10:04:31 +1000 Stephen Rothwell <[email protected]> wrote:
>> >
>> > On Tue, 27 Aug 2013 09:53:19 -0700 Mike Turquette <[email protected]> wrote:
>>> > >
>>> > > Quoting Sören Brinkmann (2013-08-27 08:44:11)
>>>> > > > On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
>>>>> > > > > On 27/08/13 10:03, Stephen Rothwell wrote:
>>>>>> > > > > > Hi Mike,
>>>>>> > > > > >
>>>>>> > > > > > Today's linux-next merge of the clk tree got a conflict in
>>>>>> > > > > > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
>>>>>> > > > > > dedicated spinlock for the SWDT") and 765b7d4c4cb3
>>>>>> > > > > > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
>>>>>> > > > > > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
>>>>>> > > > > > flag") from the clk tree.
>>>>>> > > > > >
>>>>>> > > > > > I fixed it up (see below and in a couple of places I chose
>>>>>> > > > > > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
>>>>>> > > > > > be wrong) and can carry the fix as necessary (no action is required).
>>>>> > > > >
>>>>> > > > > The case you mentioned looks correct to me.
>>>>> > > > >
>>>>> > > > > I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
>>>>> > > > > over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
>>>>> > > > > clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
>>>>> > > > > only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
>>>>> > > > > are orthogonal.
>>>> > > >
>>>> > > > I can just agree, the case included in the mail looks correct, but in
>>>> > > > case of other conflicts both flags should be set. Just like in the case
>>>> > > > shown here.
>>> > >
>>> > > Stephen's fix is correct. The Zynq patches came in as fixes so I think
>>> > > this will be a rare event.
>> >
>> > Can you guys discuss this and come up with a single answer. I read the above as:
>> >
>> > (for the two places I used CLK_SET_RATE_NO_REPARENT where the two
>> > branches each added that and CLK_SET_RATE_NOREPARENT respectively)
>> >
>> > "Stephen was wrong"
>> > "Stephen should have taken both"
>> > "Stephen was right"
>> >
>> > :-)
:-)
I think the 3 way diff omitting hunks where once branch's changes are
discarded might have confused us, even though you mentioned that there
were other conflicts (I only twigged why I couldn't see them after
seeing your new resolution).
>> >
>> > I can fix up my merge resolution if you tell me the correct fix. Also,
>> > you will need to know so that you can tell Linus (or whoever else has to
>> > resolve these conflicts).
> OK, I thought about it some more and the resolution now looks like
> below. Is this correct/better?
It looks correct to me now.
Thanks
James
On Wed, Aug 28, 2013 at 05:22:25PM +1000, Stephen Rothwell wrote:
> Hi all,
>
> On Wed, 28 Aug 2013 10:04:31 +1000 Stephen Rothwell <[email protected]> wrote:
> >
> > On Tue, 27 Aug 2013 09:53:19 -0700 Mike Turquette <[email protected]> wrote:
> > >
> > > Quoting Sören Brinkmann (2013-08-27 08:44:11)
> > > > On Tue, Aug 27, 2013 at 11:09:52AM +0100, James Hogan wrote:
> > > > > On 27/08/13 10:03, Stephen Rothwell wrote:
> > > > > > Hi Mike,
> > > > > >
> > > > > > Today's linux-next merge of the clk tree got a conflict in
> > > > > > drivers/clk/zynq/clkc.c between commits 252957cc3a2d ("clk/zynq/clkc: Add
> > > > > > dedicated spinlock for the SWDT") and 765b7d4c4cb3
> > > > > > ("clk/zynq/clkc: Add CLK_SET_RATE_PARENT flag to ethernet muxes") from
> > > > > > Linus' tree and commit 819c1de344c5 ("clk: add CLK_SET_RATE_NO_REPARENT
> > > > > > flag") from the clk tree.
> > > > > >
> > > > > > I fixed it up (see below and in a couple of places I chose
> > > > > > CLK_SET_RATE_NO_REPARENT over CLK_SET_RATE_PARENT, which may, of course,
> > > > > > be wrong) and can carry the fix as necessary (no action is required).
> > > > >
> > > > > The case you mentioned looks correct to me.
> > > > >
> > > > > I can't see todays -next yet, but if by "choose CLK_SET_RATE_NO_REPARENT
> > > > > over CLK_SET_RATE_PARENT" you mean one branch adds CLK_SET_RATE_PARENT,
> > > > > clk-next adds CLK_SET_RATE_NO_REPARENT, and the resolution ends up with
> > > > > only CLK_SET_RATE_NOREPARENT then that sounds wrong, as the two flags
> > > > > are orthogonal.
> > > >
> > > > I can just agree, the case included in the mail looks correct, but in
> > > > case of other conflicts both flags should be set. Just like in the case
> > > > shown here.
> > >
> > > Stephen's fix is correct. The Zynq patches came in as fixes so I think
> > > this will be a rare event.
> >
> > Can you guys discuss this and come up with a single answer. I read the above as:
> >
> > (for the two places I used CLK_SET_RATE_NO_REPARENT where the two
> > branches each added that and CLK_SET_RATE_NOREPARENT respectively)
> >
> > "Stephen was wrong"
> > "Stephen should have taken both"
> > "Stephen was right"
> >
> > :-)
> >
> > I can fix up my merge resolution if you tell me the correct fix. Also,
> > you will need to know so that you can tell Linus (or whoever else has to
> > resolve these conflicts).
>
> OK, I thought about it some more and the resolution now looks like
> below. Is this correct/better?
Yes, looks correct to me.
Sören