Re: [PATCH 2/2] ghes_edac: refactor error status fields decoding

[Shuai Xue <xueshuai@linux.alibaba.com>]

Hi, Robert,

Thank you for your quick comments!

On 2021/12/7 PM7:47, Robert Richter wrote:
> On 07.12.21 11:19:05, Shuai Xue wrote:
> 
>> @@ -285,6 +285,48 @@ int cper_dimm_err_location(struct cper_mem_err_compact *mem, char *msg)
>>  	return n;
>>  }
>>  
>> +const char *cper_mem_err_status_str(u64 status)
> 
> [...]
> 
> Same here, add an EXPORT_SYMBOL_GPL for the function.
Will add it in next version.


>> --- a/include/linux/cper.h
>> +++ b/include/linux/cper.h
>> @@ -568,7 +568,8 @@ void cper_print_proc_arm(const char *pfx,
>>  			 const struct cper_sec_proc_arm *proc);
>>  void cper_print_proc_ia(const char *pfx,
>>  			const struct cper_sec_proc_ia *proc);
>> -int cper_mem_err_location(struct cper_mem_err_compact *mem, char *msg);
>> -int cper_dimm_err_location(struct cper_mem_err_compact *mem, char *msg);
>> +int cper_mem_err_location(const struct cper_mem_err_compact *mem, char *msg);
>> +int cper_dimm_err_location(const struct cper_mem_err_compact *mem, char *msg);
> 
> Do we really need that 'const' here?
I think we do. It is read only and should not be modified in these functions,
just as cper_print_proc_arm' style.


>> +const char *cper_mem_err_status_str(u64 status);
> 
> The function i/f is different compared to the others, though the
> purpose is the same. Let's use same style:
> 
>  int cper_mem_err_status(const struct cper_mem_err_compact *mem, char *msg);
Sorry, I don't catch it. cper_mem_err_status_str() decodes the error status and return
a string, the same style as cper_severity_str and cper_mem_err_type_str do. May
we need to move the declaration ahead with cper_severity_str?

Best Regards,
Shuai