[PATCH v1 1/1] mtd: cfi_cmdset_0001: Factor out do_write_buffer_locked() to reduce stack frame

Raghavendra, Vignesh vigneshr at ti.com
Sun Feb 1 01:59:59 PST 2026


Hi,

On 1/29/2026 11:53 PM, Miquel Raynal wrote:
> Hi Andy,
> 
>>> Fix this by factoring out do_write_buffer_locked().
>>
>> ...
>>
>>>  	XIP_INVAL_CACHED_RANGE(map, initial_adr, initial_len);
>>>  	ENABLE_VPP(map);
>>
>> It seems more logical to leave these two in the original call.
>>
>> ...
> 
> [...]
> 
>>> +	DISABLE_VPP(map);
>>
>> Otherwise this will seem dangling here.
>>
>>>  	put_chip(map, chip, cmd_adr);
>>>  	mutex_unlock(&chip->mutex);
>>>  	return ret;
>>
>> ...
>>
>> Another approach is to leave goto as is in the _locked() and move DISABLE_VPP()
>> there.
>>
>> Tell me what do you prefer?
> 
> While I also find more logical to keep the ENABLE_VPP/DISABLE_VPP
> together, I do not mind to see them in one side or the other. I would by
> default let them in the main caller and suffi the inner function with
> "_locked()" as you did, but I'm fine either ways.
> 

Keeping ENABLE_VPP/DISABLE_VPP together along with _locked() suffix for
inner function is cleaner to read.

Regards
Vignesh



More information about the linux-mtd mailing list