1 |
On 09/17/14 13:50, Markos Chandras wrote: |
2 |
> -----BEGIN PGP SIGNED MESSAGE----- |
3 |
> Hash: SHA512 |
4 |
> |
5 |
> On 09/17/2014 02:38 PM, Ian Stakenvicius wrote: |
6 |
>> On 17/09/14 04:31 AM, Micha³ Górny wrote: |
7 |
>>> Dnia 2014-09-13, o godz. 10:47:49 Markos Chandras |
8 |
>>> <hwoarang@g.o> napisa³(a): |
9 |
>> |
10 |
>>>> Here is some weirdness with eg mips64/n32 multilib profile |
11 |
>>>> when trying a world update |
12 |
>>>> |
13 |
>>>> [ebuild U ] sys-devel/libtool-2.4.2-r1:2 [2.4.2:2] |
14 |
>>>> USE="-static-libs {-test} -vanilla" ABI_MIPS="(n32%*) o32%* |
15 |
>>>> -n64%" 0 kB |
16 |
>>>> |
17 |
>>>> As you can see n32 and o32 are enabled but n64 is not. |
18 |
>>>> Obviously this is not full mips64 multilib. This is probably |
19 |
>>>> due the portage profile stacking/inheritance problems on |
20 |
>>>> mips64, where the mips64/multilib profiles inherit the default |
21 |
>>>> o32 one. Michal (multilib CC'd) can provide more information on |
22 |
>>>> what exactly goes wrong since he understands the problem better |
23 |
>>>> than me. Michal also said that on amd64, the multilib profiles |
24 |
>>>> defaults to 64-bit only. I believe this contradicts with what |
25 |
>>>> someone expects from MIPS64 where all three ABIs need to be |
26 |
>>>> present *by default* unless you override the ABI_MIPS variable |
27 |
>>>> in make.conf. Correct? |
28 |
>> |
29 |
>>> Well, long story short we inherit from 'top-level' profile that |
30 |
>>> has some o32 settings inside. I believe that it could be saner |
31 |
>>> to move those from arch/mips/mips64 -> arch/mips/mips64/o32 (like |
32 |
>>> we have /n32 and /n64 there), so that instead of having to unset |
33 |
>>> them, we'd just have them set for the relevant real profiles. |
34 |
>> |
35 |
>>> However, I'm not sure if this doesn't come with some pitfalls. |
36 |
>> |
37 |
>> |
38 |
>> Blueness and I talked about this (proper n32 / n64 / o32 defaults |
39 |
>> and forces/masks) in #gentoo-dev two or three weeks ago; I thought |
40 |
>> we worked out the correct modifications to profiles to get it right |
41 |
>> and he had already pushed the fixes... ?? |
42 |
> |
43 |
> I can't see anything. Did you actually push them? What was decided as |
44 |
> the plan for action? |
45 |
|
46 |
I did not have time to get to them. I was going to play with two |
47 |
different approaches. One is simply turning off o32 at multilib/../n32 |
48 |
level, the other was to restructure the profiles entirely and put o32, |
49 |
n32 and 64 on par, not inherit from a hire level "default" profile. I'm |
50 |
leaning towards that approach, but I'm worried it might play havoc on |
51 |
our users. |
52 |
|
53 |
> |
54 |
>> |
55 |
>> Is it just a matter of documenting the full map of exactly what |
56 |
>> multilib profile should be forced-on and default-on in each, and |
57 |
>> then either adjusting profiles if they don't match up OR allowing |
58 |
>> users to select the correct profile for what they want? |
59 |
>> |
60 |
>> For example, i'm not understanding why n64 -should- be enabled by |
61 |
>> default on the mips64/n32 profile?? If you wanted more than |
62 |
>> {n,o}32 shouldn't you be choosing the base mips64 profile? |
63 |
>> |
64 |
>> |
65 |
> Perhaps it should not but neither should o32 |
66 |
> |
67 |
> - -- |
68 |
> Regards, |
69 |
> Markos Chandras |
70 |
> -----BEGIN PGP SIGNATURE----- |
71 |
> Version: GnuPG v2 |
72 |
> |
73 |
> iQF8BAEBCgBmBQJUGcn9XxSAAAAAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w |
74 |
> ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXRGRDlGMzA4MUI2MzBDODQ4RDBGOEYxMjQx |
75 |
> RjEwRUQ0QjgxREVCRjE5AAoJEB8Q7UuB3r8ZjSIH/3g0y2CTrZrJtmo7n9XvUIBK |
76 |
> F35WepF3zkkKE6BFxNF1qbLE7OrIUbhERZzn3bDdEz/9vBlE/oMfrV+M2tbdMX77 |
77 |
> rohS5jvAQN5F2TYhdtJBWrDV1PRidfTIx+qAL1IFIw+xNA96Wzyy+TveqRYb6BG5 |
78 |
> jqIHvyNdFu8yZKGp5OEbJgz9Jk/d4bB4cDPkIKbZODWl2aD6iuVO6hvSveB9WQrd |
79 |
> Rmair7kUi27Drrr8aNd0pt9tF0PFoFM1+Tt1axLXWDHkx3gLsqCEg7fXSKHsoUGO |
80 |
> kJH8WcWhJiYhK4nN+NJuaBm35UKFpZMY1ac5iA8UoR92QN00NqAeiMa+vMt9u7s= |
81 |
> =wYWx |
82 |
> -----END PGP SIGNATURE----- |
83 |
> |
84 |
|
85 |
|
86 |
-- |
87 |
Anthony G. Basile, Ph. D. |
88 |
Chair of Information Technology |
89 |
D'Youville College |
90 |
Buffalo, NY 14201 |
91 |
(716) 829-8197 |