[klibc] [PATCH] klibc: reproducible builds

Mike Waychison mikew at google.com
Thu Oct 21 15:54:56 PDT 2010


H. Peter Anvin wrote:
> On 10/20/2010 03:05 AM, Maciej Żenczykowski wrote:
>> Let's try that again, maybe this time the mail will be delivered.
>>
>>>>  KLIBCAR          := $(AR)
>>>> -KLIBCRANLIB      := $(RANLIB)
>>>> +KLIBCAR_cq       := $(KLIBCAR) $(if $(KBUILD_REPRODUCIBLE),Dcq,cq)
>>>> +KLIBCAR_crs      := $(KLIBCAR) $(if $(KBUILD_REPRODUCIBLE),Dcrs,crs)
>>>> +KLIBCAR_cru      := $(KLIBCAR) $(if $(KBUILD_REPRODUCIBLE),Dcr,cru)
>>> Did you mean to drop 'u' here?
>> Yeah, Dcru is not valid.
>>
>> D means skip (don't generate) timestamps, u means use them (ie.
>> add/replace only if newer then existing copy), as such you can't
>> combine them...
>>
>> This makes using a macro, like you suggested, hard...
>>
> 
> Not really, it just means the macro need to specify both directions:
> 
> 	$(call klibc-ar,cru,Dcr)
> 
> (and yes, klibc-ar would be a lot nicer a name than $(invoke-KLIBCAR).
> 

Yup, this would look a lot better.



More information about the klibc mailing list