[U-Boot] [PATCH V4] I2C: mxc_i2c rework

Marek Vasut marek.vasut at gmail.com
Sat Jul 30 12:52:53 CEST 2011


On Saturday, July 30, 2011 08:42:19 AM Jason Hui wrote:
> Hi, Marek,
> 
> On Sat, Jul 30, 2011 at 1:09 AM, Marek Vasut <marek.vasut at gmail.com> wrote:
> > On Friday, July 29, 2011 01:24:49 PM Jason Hui wrote:
> >> Hi, Marek,
> >> 
> >> On Fri, Jul 29, 2011 at 5:32 PM, Marek Vasut <marek.vasut at gmail.com> wrote:
> >> > Rewrite the mxc_i2c driver.
> >> >  * This version is much closer to Linux implementation.
> >> >  * Fixes IPG_PERCLK being incorrectly used as clock source
> >> >  * Fixes behaviour of the driver on iMX51
> >> >  * Clean up coding style a bit ;-)
> >> 
> >> I don't think you did the right thing by chaning IPG_PERCLK to IPG_CLK
> >> As I said, the IPG_CLK is for IP register clock, while IPG_PERCLK is for
> >> i2c function clock.
> >> 
> >> if you run clock command from mx51evk, you will get:
> >> ...
> >> ipg clock     : 66500000Hz
> >> ipg per clock : 665000000Hz
> >> MX51EVK U-Boot >
> >> 
> >> It will give you that ipg per clock is 665M, which seems too big. It
> >> is due to we configure
> >> the pre-divider/pos-devider for perclk to zero, which leads to ipg_per
> >> clock to be same as pll2 clock.
> >> But I don't think this will have some issue.
> > 
> > Yes it will, the divider will be computed to be maximum in all cases ...
> > you won't be able it divide 665MHz by anything from the table 40-7 in
> > MX51RM to achieve any reasonable frequency.
> > 
> > On the contrary, 66.5MHz does give fine results.
> 
> it that, we can change the ipg_perclk to low freq, but we should not change
> the i2c clock to ipg_clk, this is not correct.
> 
> >> BTW, I have applied your patch and test on mx53evk board, it seems the
> >> i2c does not work correctly.
> > 
> > Great. The clock used by the I2C module for this task are likely the
> > module_clock, which are -- like on MX51 -- 66.5MHz. What do you get when
> > you run the "clock" command on MX53EVK ?
> 
> I'm not in the office for the whole next week, thus, I can't give you
> the clock output for
> mx53evk, but I'm sure the ipg_perclk on mx53evk is not set at 66.5Mhz.

Yes, it's 33.3MHz on iMX53. I got a (remote) hand of a MX53 board. The IPG 
clock, on the other hand, are 66.5MHz on both MX51 and MX53. Maybe the issue 
with PMIC you're seeing is something else?

btw. how did you obtain these results ? Can you check the FDR divider value with 
and without this patch?

Thanks
> 
> Jason
> 
> > Thanks
> > 
> >> After apply your patch:
> >> MX53EVK U-Boot > pmic dump  30
> >> PMIC ID: 0x0000ffff [Rev: unknown]
> >> 
> >> 0x00: 0001ffff 00ffffff 0039ffff 0000ffff 00ffffff 0000ffff 0000ffff
> >> 0000ffff 0x08: 0000ffff 0000ffff 0000ffff 0000ffff 0000ffff 0000ffff
> >> 0000ffff 0040ffff 0x10: 0000ffff 0000ffff 0000ffff 0000ffff 0000ffff
> >> 0001ffff 0000ffff 0000ffff 0x18: 0045ffff 0045ffff 0000ffff 0080ffff
> >> 0021ffff 0000ffff 0002ffff 0000ffff 0x20: 0004ffff 0000ffff 0021ffff
> >> 0000ffff 0000ffff 0000ffff 0000ffff 0000ffff 0x28: 0000ffff 0000ffff
> >> 0000ffff 0000ffff 0000ffff 0004ffff 0000ffff 00aeffff
> >> 
> >> The old:
> >> MX53EVK U-Boot > pmic dump 30
> >> PMIC ID: 0x000045d0 [Rev: 2.0]
> >> 
> >> 0x00: 00015088 00ffffff 00395208 00000081 00fff7ff 0000401c 00000418
> >> 000045d0 0x08: 00000000 00000000 00000001 00000000 00000000 00000040
> >> 00000000 00400000 0x10: 00000000 00000000 00000000 00000000 00000011
> >> 0001ffff 00000000 00007fff 0x18: 00454a52 00456739 0000631a 0080739c
> >> 0021284a 00000a0a 00024fd0 000001d8 0x20: 00049208 00000000 00218000
> >> 00000000 00000000 00000000 00000000 00000000 0x28: 00000000 00000000
> >> 00000000 00008000 00000000 00046046 000001c0 00aeeaee
> >> 
> >> Jason


More information about the U-Boot mailing list