| 1 |
2016-06-17 19:25 GMT+08:00 Luca Barbato <lu_zero@g.o>: |
| 2 |
> On 17/06/16 13:12, Lei Zhang wrote: |
| 3 |
>> 2016-06-16 21:42 GMT+08:00 Lei Zhang <zhanglei.april@×××××.com>: |
| 4 |
>>> 2016-06-16 21:13 GMT+08:00 Luca Barbato <lu_zero@g.o>: |
| 5 |
>>>> On 16/06/16 14:47, Lei Zhang wrote: |
| 6 |
>>>>> 2016-06-16 17:43 GMT+08:00 Luca Barbato <lu_zero@g.o>: |
| 7 |
>>>>>> On 16/06/16 05:36, Lei Zhang wrote: |
| 8 |
>>>>>>> Attached is what I've got so far. |
| 9 |
>>>>>>> |
| 10 |
>>>>>>> Some of them are not documented in man pages, so I checked their |
| 11 |
>>>>>>> declarations in musl's headers. Functions strtold_l, strtoll_l and |
| 12 |
>>>>>>> strtoull_l are not available in musl, but libc++ is smart enough to |
| 13 |
>>>>>>> not use them when linked against musl. |
| 14 |
>>>>> |
| 15 |
>>>>> I made a mistake. Actually strtold_l *is* available in musl, but |
| 16 |
>>>>> hidden by _GNU_SOURCE. |
| 17 |
>>>> |
| 18 |
>>>> Not hidden by _BSD_SOURCE as well? |
| 19 |
>>> |
| 20 |
>>> No, only by _GNU_SOURCE. |
| 21 |
>>> |
| 22 |
>>>>>> beside vasprintf, the rest is covered by _XOPEN_SOURCE 700 |
| 23 |
>>>>>> |
| 24 |
>>>>>> Might be worth trying to do -U_GNU_SOURCES -D_XOPEN_SOURCE=700 and see |
| 25 |
>>>>>> how it goes. |
| 26 |
>>>>> |
| 27 |
>>>>> So far vasprintf and strtold_l are not covered. vasprintf can be |
| 28 |
>>>>> covered with -D_BSD_SOURCE. But _BSD_SOURCE is deprecated since glibc |
| 29 |
>>>>> 2.20; I don't know if this has any impact on musl. |
| 30 |
>>>> |
| 31 |
>>>> It should not and -D_BSD_SOURCE should be fine for musl. |
| 32 |
>>> |
| 33 |
>>> I just found that vasprintf is only protected by _GNU_SOURCE in glibc, |
| 34 |
>>> which means in this case _BSD_SOURCE only works for musl. |
| 35 |
>>> |
| 36 |
>>>>>> How vasprintf is used? |
| 37 |
>>>>> |
| 38 |
>>>>> It's used by <locale>. The code is kinda hard to read; I can't tell |
| 39 |
>>>>> exactly what it's doing... |
| 40 |
>>>> |
| 41 |
>>>> same for the _l variants I guess. |
| 42 |
>>> |
| 43 |
>>> So what now? So far strtold_l still needs _GNU_SOURCE in musl; |
| 44 |
>>> strtold_l, strtoll_l, strtoull_l and vasprintf all need _GNU_SOURCE in |
| 45 |
>>> glibc. |
| 46 |
>> |
| 47 |
>> To sum it up: |
| 48 |
>> |
| 49 |
>> * As for musl, "-U_GNU_SOURCES -U_GNU_SOURCES=700 -D_BSD_SOURCE" |
| 50 |
>> covers every symbol needed by libc++ except for strtold_l. |
| 51 |
>> |
| 52 |
>> * As for glibc, "-U_GNU_SOURCE -D_ISOC11_SOURCE -D_DEFAULT_SOURCE" |
| 53 |
>> covers everything except for strtold_l, strtoll_l, strtoull_l and |
| 54 |
>> vasprintf. |
| 55 |
> |
| 56 |
> |
| 57 |
> Sounds almost good, now some configure/cmake patches to put that in and |
| 58 |
> some #define _GNU_SOURCES /#undef in selected places might get it |
| 59 |
> working completely... |
| 60 |
|
| 61 |
After putting about eight "#define _GNU_SOURCE"s in various *.cpp |
| 62 |
files, it seems to work now :) |
| 63 |
|
| 64 |
I'll see if I can reduce the number of "#define"s needed. |
| 65 |
|
| 66 |
Lei |
Archives