Clock Regression in next-20150130 caused by cb75a8fcd14e

Clock Regression in next-20150130 caused by cb75a8fcd14e

[Tony Lindgren]

Hi all,

Looks like commit cb75a8fcd14e ("clk: Add rate constraints to clocks")
causes a regression on at least omaps where the serial console either
does not show anything, or just prints garbage.

Reverting cb75a8fcd14e makes things work again on next-20150130.

Any ideas?

Regards,

Tony

Clock Regression in next-20150130 caused by cb75a8fcd14e

[Guenter Roeck]

On Fri, Jan 30, 2015 at 05:04:44PM -0800, Tony Lindgren wrote:
> Hi all,
> 
> Looks like commit cb75a8fcd14e ("clk: Add rate constraints to clocks")
> causes a regression on at least omaps where the serial console either
> does not show anything, or just prints garbage.
> 
> Reverting cb75a8fcd14e makes things work again on next-20150130.
> 
> Any ideas?
> 
The patch seems to have some problems.
Also see http://www.spinics.net/lists/kernel/msg1916843.html.

Guenter

Clock Regression in next-20150130 caused by cb75a8fcd14e

[Mike Turquette]

Quoting Tony Lindgren (2015-01-30 17:04:44)
> Hi all,
> 
> Looks like commit cb75a8fcd14e ("clk: Add rate constraints to clocks")
> causes a regression on at least omaps where the serial console either
> does not show anything, or just prints garbage.
> 
> Reverting cb75a8fcd14e makes things work again on next-20150130.
> 
> Any ideas?

Stephen posted a patch[0] to fix this. I've squashed that into Tomeu's
commit that you reference above and my Panda board is booting fine once
again.

[0] http://lkml.kernel.org/r/<20150131013158.GA4323@codeaurora.org>

Please let me know if any other issues pop up. There are new WARNs for
OMAP3+ boards introduced by a different patch from Tomeu, but this is
really because the OMAP DPLL and FAPLL code are dereferencing struct clk
pointers when they should not be and is a separate issue from the
constraints patch (with a separate email thread).

Regards,
Mike

> 
> Regards,
> 
> Tony

Clock Regression in next-20150130 caused by cb75a8fcd14e

[Tony Lindgren]

* Mike Turquette <mturquette@linaro.org> [150201 15:29]:
> Quoting Tony Lindgren (2015-01-30 17:04:44)
> > Hi all,
> > 
> > Looks like commit cb75a8fcd14e ("clk: Add rate constraints to clocks")
> > causes a regression on at least omaps where the serial console either
> > does not show anything, or just prints garbage.
> > 
> > Reverting cb75a8fcd14e makes things work again on next-20150130.
> > 
> > Any ideas?
> 
> Stephen posted a patch[0] to fix this. I've squashed that into Tomeu's
> commit that you reference above and my Panda board is booting fine once
> again.
> 
> [0] http://lkml.kernel.org/r/<20150131013158.GA4323@codeaurora.org>
> 
> Please let me know if any other issues pop up. There are new WARNs for
> OMAP3+ boards introduced by a different patch from Tomeu, but this is
> really because the OMAP DPLL and FAPLL code are dereferencing struct clk
> pointers when they should not be and is a separate issue from the
> constraints patch (with a separate email thread).

Seems Linux next is broken again. Now we omaps get tons of:

clock: dpll_abe_ck failed transition to 'locked'
WARNING: CPU: 0 PID: 1 at drivers/clk/clk.c:925 clk_disable+0x28/0x34()

and 

WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/dpll3xxx.c:436
omap3_noncore_dpll_enable+0xdc/0x10c()

Regards,

Tony