Re: [Patch v4 03/18] x86/speculation: Reorganize cpu_show_common()

[Tim Chen]

On 11/03/2018 11:07 AM, Thomas Gleixner wrote:
> Tim,
> 
> On Tue, 30 Oct 2018, Tim Chen wrote:
>> Extract the logic to show IBPB, STIBP usages in cpu_show_common()
>> into helper functions.
>>
>> Later patches will add other userspace Spectre v2 mitigation modes.
>> This patch makes it easy to show IBPB and STIBP
>> usage scenario according to the mitigation mode.
> 
> First of all, I asked you before to do:
> 
> # git grep 'This patch' Documentation/process
> 
> This leads you to:
> 
>  "Describe your changes in imperative mood, e.g. "make xyzzy do frotz"
>   instead of "[This patch] makes xyzzy do frotz" or "[I] changed xyzzy
>   to do frotz", as if you are giving orders to the codebase to change
>   its behaviour."
> 
> Documentation is there for a reason.
> 
> Aside of that, I'd really have a hard time to figure out what you are
> trying to say, if I didn't have the context already. Change logs need to
> make sense on their own. So something like this:
> 
>   The Spectre V2 printout in cpu_show_common() handles conditionals for the
>   various mitigation methods directly in the sprintf() argument list. That's
>   hard to read and will become unreadable if more complex decisions need to
>   be made for a particular method.
> 
>   Move the conditionals for STIBP and IBPB string selection into helper
>   functions, so they can be extended later on.
> 
> follows the obvious ordering for change logs:
> 
>   1) Describe context and problem
>   
>   2) Describe the solution
> 
> and is understandable without needing to know about the context in which
> this change was developed.
> 
> Hmm? This is a suggestion, feel free to rewrite it in you own words. The
> same applies to other change logs as well. I won't comment on those.

Thanks for the suggestion.  Will update.

>  
>>  static ssize_t cpu_show_common(struct device *dev, struct device_attribute *attr,
>>  			       char *buf, unsigned int bug)
>>  {
>> @@ -872,9 +888,8 @@ static ssize_t cpu_show_common(struct device *dev, struct device_attribute *attr
>>  
>>  	case X86_BUG_SPECTRE_V2:
>>  		return sprintf(buf, "%s%s%s%s%s%s\n", spectre_v2_strings[spectre_v2_enabled],
>> -			       boot_cpu_has(X86_FEATURE_USE_IBPB) ? ", IBPB" : "",
>>  			       boot_cpu_has(X86_FEATURE_USE_IBRS_FW) ? ", IBRS_FW" : "",
>> -			       (x86_spec_ctrl_base & SPEC_CTRL_STIBP) ? ", STIBP" : "",
>> +			       ibpb_state(), stibp_state(),
>>  			       boot_cpu_has(X86_FEATURE_RSB_CTXSW) ? ", RSB filling" : "",
>>  			       spectre_v2_module_string());
> 
> Any particular reason for changing the output ordering here? If yes, then
> the changelog should mention it. If no, why?
> 

I was putting the features related to user application protection together. It
was not necessary and I can leave it at the same place.

Tim