[PATCH] setjmp: Replace jmp_buf-macros.h with jmp_buf-macros.sym
Adhemerval Zanella
adhemerval.zanella@linaro.org
Fri Nov 19 18:41:36 GMT 2021
On 19/11/2021 15:38, Florian Weimer wrote:
> * Adhemerval Zanella:
>
>> On 19/11/2021 15:33, Florian Weimer wrote:
>>> * Adhemerval Zanella:
>>>
>>>> On 19/11/2021 15:28, Florian Weimer wrote:
>>>>> * Adhemerval Zanella via Libc-alpha:
>>>>>
>>>>>> diff --git a/setjmp/__setjmp_check.c b/setjmp/__setjmp_check.c
>>>>>> new file mode 100644
>>>>>> index 0000000000..4be6e432b3
>>>>>> --- /dev/null
>>>>>> +++ b/setjmp/__setjmp_check.c
>>>>>> @@ -0,0 +1,34 @@
>>>>>> +/* Check jmp_buf sizes, alignments and offsets. */
>>>>>
>>>>> Shouldn't this be an internal test?
>>>>>
>>>>> I checked the generated constants against the removed files, and found
>>>>> no discrepancies, so I think this can go in.
>>>>
>>>> My understanding is the _Static_assert would be more useful at build time,
>>>> specially for cross-compilation.
>>>
>>> An internal will still have the same effect when cross-compiled.
>>
>> I don't have a strong preference in fact, I think an object consumes slight
>> less resources through (cpu time to build, disk size).
>
> It consumes space in the installed artifacts, though.
>
> $ ar t libc.a | grep setjmp
> setjmp.o
> bsd-setjmp.o
> bsd-_setjmp.o
> __setjmp_check.o
> $ readelf -sW libc.so.6 | grep setjmp
> 304: 0000000000041010 10 FUNC GLOBAL DEFAULT 14 setjmp@@GLIBC_2.2.5
> 546: 0000000000041020 7 FUNC GLOBAL DEFAULT 14 _setjmp@@GLIBC_2.2.5
> 1012: 0000000000040f80 87 FUNC GLOBAL DEFAULT 14 __sigsetjmp@@GLIBC_2.2.5
> 286: 0000000000000000 0 FILE LOCAL DEFAULT ABS __setjmp_check.c
> 4893: 0000000000040f80 87 FUNC LOCAL DEFAULT 14 __GI___sigsetjmp
> 5090: 0000000000041020 7 FUNC LOCAL DEFAULT 14 __GI__setjmp
> 7075: 0000000000040f80 87 FUNC GLOBAL DEFAULT 14 __sigsetjmp
> 9251: 0000000000041020 7 FUNC GLOBAL DEFAULT 14 _setjmp
> 9705: 0000000000041010 10 FUNC GLOBAL DEFAULT 14 setjmp
Hum I forgot to verify this, I will change to an internal test.
More information about the Libc-alpha
mailing list