1 |
Hi Ulrich, |
2 |
|
3 |
|
4 |
On 2021/8/24 16:46, Ulrich Mueller wrote: |
5 |
>>>>>> On Tue, 24 Aug 2021, WANG Xuerui wrote: |
6 |
>> It seems the discussion has gone quiet for a while now, so I take that |
7 |
>> we choose ARCH=loong over ARCH=loongarch according to GLEP 53? |
8 |
> LGTM |
9 |
> |
10 |
>> If that doesn't receive much objection, I'll prepare and send the |
11 |
>> first few eclass patches soon. |
12 |
> We also need to update the conditional in eselect: |
13 |
> https://gitweb.gentoo.org/proj/eselect.git/tree/libs/package-manager.bash.in#n70 |
14 |
Thanks for the reminder! |
15 |
> Does the GNU triplet (i.e. HOSTTYPE in bash) always use "loongarch" |
16 |
> literally, or can it have some suffix? |
17 |
|
18 |
According to their earlier reservation[1] and actual vendor system |
19 |
behavior, there are 3 possible values: |
20 |
|
21 |
- loongarch64 |
22 |
- loongarch32 |
23 |
- loongarchx32 |
24 |
|
25 |
Only the LP64 ABI is fully supported by the current upstream submission. |
26 |
The "loongarch32" thing might NOT be compatible with the LP64 ABI, |
27 |
instead it might be something embedded-oriented, even instruction |
28 |
subsets supported might differ. The "loongarchx32" is for an |
29 |
n32/x32-like ABI that doesn't exist yet, and probably will never get |
30 |
implemented. |
31 |
|
32 |
Accordingly, I think we only have to care about "loongarch64" for now. |
33 |
|
34 |
[1]: |
35 |
https://git.savannah.gnu.org/gitweb/?p=config.git;a=patch;h=c8ddc8472f8efcadafc1ef53ca1d863415fddd5f |