Latency Issue due to patch; ESDHC: Workaround for data crc error on p1010rdb

[Jain Priyanka-B32167 <B32167@xxxxxxxxxxxxx>]

Hi,

I am facing latency issue due to below implementation.
The function 'mmc_rescan' is getting called at regular intervals which is calling 'of_find_comaptible_node which is adding to the latency. Trace: 

=> of_find_compatible_node
=> esdhc_of_set_clock
=> sdhci_set_clock
=> sdhci_set_ios
=> mmc_power_up
=> mmc_rescan

'of_find_compatible_node' function call is required only once at Linux boot time. 


If I change the implementation to : 

static void esdhc_of_set_clock(struct sdhci_host *host, unsigned int clock)
{
        static int reduce_clock_flag, check_only_once_flag;
        /* Workaround to reduce the clock frequency for p1010 esdhc */
        if (check_only_once_flag == 0)
        {
                if (of_find_compatible_node(NULL, NULL, "fsl,p1010-esdhc"))
                        reduce_clock_flag = 1;
                check_only_once_flag = 1;
        }
        if (reduce_clock_flag == 1)
        {
                if (clock > 20000000)
                        clock -= 5000000;
                if (clock > 40000000)
                        clock -= 5000000;
        }
        /* Set the clock */
        esdhc_set_clock(host, clock);
}


It works fine. No more latency issue. Please review this approach. If there is a better way, please do suggest.

Thanks 
Priyanka Jain

-----Original Message-----
From: Huang Changming-R66093 
Sent: Tuesday, February 14, 2012 11:41 AM
To: Chris Ball
Cc: linux-mmc@xxxxxxxxxxxxxxx; Jain Priyanka-B32167
Subject: RE: [PATCH v7] ESDHC: Workaround for data crc error on p1010rdb



Thanks
Jerry Huang


> -----Original Message-----
> From: Chris Ball [mailto:cjb@xxxxxxxxxx]
> Sent: Tuesday, February 14, 2012 4:44 AM
> To: Huang Changming-R66093
> Cc: linux-mmc@xxxxxxxxxxxxxxx; Jain Priyanka-B32167
> Subject: Re: [PATCH v7] ESDHC: Workaround for data crc error on 
> p1010rdb
> 
> Hi,
> 
> On Mon, Feb 13 2012, Huang Changming-R66093 wrote:
> >> Actually, I think I'm still missing something.  Why doesn't this work?
> >>
> >> +static void esdhc_of_set_clock(struct sdhci_host *host, unsigned 
> >> +int
> >> +clock) {
> >> +       /* Workaround to reduce the clock frequency for p1010 esdhc */
> >> +       if (of_find_compatible_node(NULL, NULL, "fsl,p1010-esdhc")) {
> >> +               if (clock > 20000000)
> >> +                       clock -= 5000000;
> >> +               if (clock > 40000000)
> >> +                       clock -= 5000000;
> >> +       }
> >> +
> >> +       /* Set the clock */
> >> +       esdhc_set_clock(host, clock); }
> >>
> > These codes can work, too.
> > I forget the clock will be reduced 5M/10M only when the clock is 
> > greater than 20MHz/45MHz.
> > So I should not test the clock in this function.
> 
> Great, thanks -- please test and submit a patch with the code above, 
> and I'll merge it.


Thanks, Chris.
I have sent it again.


Jerry Huang

--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html