Skip to contentSkip to conversation

flat tree clock setting

U-Boot-Users David Updegraff 4 participants 6 messages Nov 9, 2006

lore
Share & tools
lore
#1 lore raw
Quoted 4 lines · David Updegraff

Hi.

Might it be nice if FT could use the calculated -- more likely to be correct timebase value?

Wouldn't it be better to change get_tbclk() to look like this?

unsigned long get_tbclk(void)
{
#ifdef OF_TBCLK
return OF_TBCLK

#else
	ulong tbclk;

	tbclk = (gd->bus_clk + 3L) / 4L;

	return tbclk;

#endif
}

And then change the ft_build.c code to always call get_tbclk()?

I haven't tried compiling the above code, but I think you all get the idea.

Signature · 2 lines
Timur Tabi
Linux Kernel Developer @ Freescale
#2 lore raw
Quoted 12 lines · Timur Tabi

Wouldn't it be better to change get_tbclk() to look like this?

unsigned long get_tbclk(void)
{
#ifdef OF_TBCLK
return OF_TBCLK

#else
	ulong tbclk;

	tbclk = (gd->bus_clk + 3L) / 4L;

	return tbclk;
Quoted 4 lines

#endif
}

And then change the ft_build.c code to always call get_tbclk()?

Are there any boards on which OF_TBCLK is (properly) set to something other than what get_tbclk() returns (ignoring the difference in rounding)? I don't see any, nor can I think of a reason why such a difference should ever exist. Perhaps OF_TBCLK should just be removed?

-Scott

#3 lore raw
Quoted 3 lines · Scott Wood

Are there any boards on which OF_TBCLK is (properly) set to something other than what get_tbclk() returns (ignoring the difference in rounding)?

A search of the source code shows this:

File /temp/u-boot-83xx/include/configs/MPC8349EMDS.h
332 2:#define OF_TBCLK (bd->bi_busfreq / 4)
File /temp/u-boot-83xx/include/configs/MPC8349ITX.h
395 2:#define OF_TBCLK (bd->bi_busfreq / 4)
File /temp/u-boot-83xx/include/configs/MPC8360EMDS.h
327 2:#define OF_TBCLK (bd->bi_busfreq / 4)
File /temp/u-boot-83xx/include/configs/MPC8540ADS.h
309 2:#define OF_TBCLK (bd->bi_busfreq / 8)
File /temp/u-boot-83xx/include/configs/MPC8541CDS.h
320 2:#define OF_TBCLK (bd->bi_busfreq / 8)
File /temp/u-boot-83xx/include/configs/MPC8548CDS.h
326 2:#define OF_TBCLK (bd->bi_busfreq / 8)
File /temp/u-boot-83xx/include/configs/MPC8555CDS.h
320 2:#define OF_TBCLK (bd->bi_busfreq / 8)
File /temp/u-boot-83xx/include/configs/MPC8560ADS.h
302 2:#define OF_TBCLK (bd->bi_busfreq / 8)
File /temp/u-boot-83xx/include/configs/MPC8641HPCN.h
269 2:#define OF_TBCLK (bd->bi_busfreq / 4)
File /temp/u-boot-83xx/include/configs/stxxtc.h
587 2:#define OF_TBCLK (MPC8XX_HZ / 16)

Let's ignore the stxxtc for now, whatever that is.

On the 83xx, get_tbclk() returns "(gd->bus_clk + 3L) / 4L". On 85xx and 86xx, it does this:

	get_sys_info(&sys_info);
	return ((sys_info.freqSystemBus + 7L) / 8L);

sys_info.freqSystemBus can be any number of things (see get_sys_info).

Well, it looks like they COULD always contain the same value.

Quoted 2 lines

I don't see any, nor can I think of a reason why such a difference should ever exist. Perhaps OF_TBCLK should just be removed?

I think it's safe to say that *something* is redundant here, but I don't know yet what. My vote would be for OF_TBCLK, though, if that works.

The only thing I know for sure is that all of this clock stuff is making me cuckoo.

Signature · 2 lines
Timur Tabi
Linux Kernel Developer @ Freescale
#4 lore raw
In reply to Scott Wood
Quoted 4 lines · Scott Wood

Are there any boards on which OF_TBCLK is (properly) set to something other than what get_tbclk() returns (ignoring the difference in rounding)? I don't see any, nor can I think of a reason why such a difference should ever exist. Perhaps OF_TBCLK should just be removed?

Ok, I've done some more digging.

On PPC, bd->bi_busfreq is the same as gd->bus_clk. In fact, there are a bunch of clocks in bd that are equal to similar values in gd. The bd_t contains info that is passed to non-openfirmware versions of Linux, and the gd_t contains just general globally-defined data.

My guess is that anything which isn't directly used to boot non-OF Linux should use the gd structure. So get_tbclk() does the right thing (although the rounding could be wrong), but OF_TBCLK should probably use gd->bus_clk.

Signature · 2 lines
Timur Tabi
Linux Kernel Developer @ Freescale
#5 lore raw
In reply to David Updegraff
Quoted 3 lines · In message <[email protected]> you wrote:

Might it be nice if FT could use the calculated -- more likely to be correct timebase value?

Coding style problems: indentation not by TAB.

Please cleanup and resubmit.

Best regards,

Wolfgang Denk

Signature · 4 lines
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
I don't know if it's what you want, but it's what you get.  :-)
                      - Larry Wall in <[email protected]>