Re: [PATCH]: Fixing blksz_bits calling inside drivers/mmc/mmc.c and drivers/mmc/omap.c

[Felipe Balbi <ext-felipe.lima@nokia.com>]

I just saw an error in the patch generation...

(...)

--- linux-omap/drivers/mmc/mmc.c	2006-10-26 06:42:31.000000000 -0400
+++ linux-omap-dev/drivers/mmc/mmc.c	2006-10-26 09:53:24.000000000 -0400
@@ -996,7 +996,7 @@
 
 		mmc_set_data_timeout(&data, card, 0);
 
-		data.blksz = 1 << 3;
+		data.blksz_bits = 1 << 3;
 		data.blocks = 1;
 		data.flags = MMC_DATA_READ;
 		data.sg = &sg;

The module should actually be inverted here... just like:

(...)

--- linux-omap/drivers/mmc/mmc.c	2006-10-26 06:42:31.000000000 -0400
+++ linux-omap-dev/drivers/mmc/mmc.c	2006-10-26 09:53:24.000000000 -0400
@@ -996,7 +996,7 @@
 
 		mmc_set_data_timeout(&data, card, 0);

-		data.blksz_bits = 1 << 3;
+		data.blksz = 1 << 3;
 		data.blocks = 1;
 		data.flags = MMC_DATA_READ;
 		data.sg = &sg;


I look at my quilt "repository" and the patch is ok... Don't know how 
that happened.

ext Felipe Balbi wrote:
> ext Dirk Behme wrote:
>> Felipe Balbi wrote:
>>> This small patch solves the problem ;-)
>>>
>>> Here's the error dump:
>>> drivers/mmc/omap.c: In function 'mmc_omap_prepare_dma':
>>> drivers/mmc/omap.c:636: error: 'struct mmc_data' has no member named 
>>> 'blksz_bits'
>>> drivers/mmc/omap.c:639: error: 'struct mmc_data' has no member named 
>>> 'blksz_bits'
>>> drivers/mmc/omap.c: In function 'mmc_omap_prepare_data':
>>> drivers/mmc/omap.c:828: error: 'struct mmc_data' has no member named 
>>> 'blksz_bits'
>>>
>>> Obs: The same occurs for drivers/mmc/mmc.c file
>>>
>>>
>>> ------------------------------------------------------------------------ 
>>>
>>>
>>> Index: linux-omap/drivers/mmc/omap.c
>>> ===================================================================
>>> --- linux-omap.orig/drivers/mmc/omap.c    2006-10-27 
>>> 09:18:34.000000000 -0400
>>> +++ linux-omap/drivers/mmc/omap.c    2006-10-27 11:03:14.000000000 
>>> -0400
>>> @@ -633,10 +633,10 @@ mmc_omap_prepare_dma(struct mmc_omap_hos
>>>      int sync_dev = 0;
>>>  
>>>      data_addr = host->phys_base + OMAP_MMC_REG_DATA;
>>> -    frame = 1 << data->blksz_bits;
>>> +    frame = 1 << data->blksz;
>>>      count = sg_dma_len(sg);
>>>  
>>> -    if ((data->blocks == 1) && (count > (1 << data->blksz_bits)))
>>> +    if ((data->blocks == 1) && (count > (1 << data->blksz)))
>>>          count = frame;
>>>  
>>>      host->dma_len = count;
>>> @@ -825,7 +825,7 @@ mmc_omap_prepare_data(struct mmc_omap_ho
>>>      }
>>>  
>>>  
>>> -    block_size = 1 << data->blksz_bits;
>>> +    block_size = 1 << data->blksz;
>>>  
>>>      OMAP_MMC_WRITE(host, NBLK, data->blocks - 1);
>>>      OMAP_MMC_WRITE(host, BLEN, block_size - 1);
>>>
>>> --- linux-omap/drivers/mmc/mmc.c    2006-10-26 06:42:31.000000000 -0400
>>> +++ linux-omap-dev/drivers/mmc/mmc.c    2006-10-26 
>>> 09:53:24.000000000 -0400
>>> @@ -996,7 +996,7 @@
>>>  
>>>          mmc_set_data_timeout(&data, card, 0);
>>>  
>>> -        data.blksz = 1 << 3;
>>> +        data.blksz_bits = 1 << 3;
>>>          data.blocks = 1;
>>>          data.flags = MMC_DATA_READ;
>>>          data.sg = &sg;
>>
>> This doesn't look correct to me (and compilation still
>> fails). Why do you add data.blksz_bits in mmc.c again? It
>> was removed by RMK at
>>
>> http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=blobdiff;h=5b9caa7978d34311839840243e38921255f6da4b;hp=74eaaee66de0e59532ba5655fddac5aaeea5ba08;hb=132919ba80ad207755fe271277bfefff865a54fe;f=drivers/mmc/mmc.c 
>>
>>
>> Whats about the attachment instead? It additionally
>> removes warning
>>
>> drivers/mmc/omap.c:587: warning: 'mmc_omap_switch_callback'
>> defined but not used
>>
>> too.
>>
>> Signed-off-by: Dirk Behme <dirk.behme_at_gmail.com>
>> ------------------------------------------------------------------------
>>
>> Index: linux-osk/drivers/mmc/omap.c
>> ===================================================================
>> --- linux-osk.orig/drivers/mmc/omap.c
>> +++ linux-osk/drivers/mmc/omap.c
>> @@ -581,12 +581,6 @@ static void mmc_omap_switch_timer(unsign
>>      schedule_work(&host->switch_work);
>>  }
>>  
>> -/* FIXME: Handle card insertion and removal properly. Maybe use a mask
>> - * for MMC state? */
>> -static void mmc_omap_switch_callback(unsigned long data, u8 mmc_mask)
>> -{
>> -}
>> -
>>  static void mmc_omap_switch_handler(void *data)
>>  {
>>      struct mmc_omap_host *host = (struct mmc_omap_host *) data;
>> @@ -633,10 +627,10 @@ mmc_omap_prepare_dma(struct mmc_omap_hos
>>      int sync_dev = 0;
>>  
>>      data_addr = host->phys_base + OMAP_MMC_REG_DATA;
>> -    frame = 1 << data->blksz_bits;
>> +    frame = data->blksz;
>>      count = sg_dma_len(sg);
>>  
>> -    if ((data->blocks == 1) && (count > (1 << data->blksz_bits)))
>> +    if ((data->blocks == 1) && (count > (data->blksz)))
>>          count = frame;
>>  
>>      host->dma_len = count;
>> @@ -825,7 +819,7 @@ mmc_omap_prepare_data(struct mmc_omap_ho
>>      }
>>  
>>  
>> -    block_size = 1 << data->blksz_bits;
>> +    block_size = data->blksz;
>>  
>>      OMAP_MMC_WRITE(host, NBLK, data->blocks - 1);
>>      OMAP_MMC_WRITE(host, BLEN, block_size - 1);
>>
>>
>>   
> But then you will need to modify the struct mmc_data... and that 
> didn't seem correct for me 'cause every module using this structure 
> (and probable future modules) would need to modify the same way you 
> did...
>

-- 
Regards,

Felipe Balbi
ext-felipe.lima@nokia.com
OSMRC