| 1 |
On 12/29/2013 09:40 PM, Joshua Kinard wrote: |
| 2 |
> On 12/28/2013 5:58 PM, Michał Górny wrote: |
| 3 |
>> Hello, folks. |
| 4 |
>> |
| 5 |
>> I've noticed today that mips uses the same CHOST value for all three |
| 6 |
>> ABIs it supports: |
| 7 |
>> |
| 8 |
>> arch/mips/mips64/multilib/make.defaults:CHOST_o32="${CHOST}" |
| 9 |
>> arch/mips/mips64/multilib/make.defaults:CHOST_n32=${CHOST} |
| 10 |
>> arch/mips/mips64/multilib/make.defaults:CHOST_n64=${CHOST} |
| 11 |
>> arch/mips/mipsel/mips64el/multilib/make.defaults:CHOST_o32="${CHOST}" |
| 12 |
>> arch/mips/mipsel/mips64el/multilib/make.defaults:CHOST_n32="${CHOST}" |
| 13 |
>> arch/mips/mipsel/mips64el/multilib/make.defaults:CHOST_n64="${CHOST}" |
| 14 |
>> |
| 15 |
>> Long story short, this sucks and will cause trouble. |
| 16 |
>> |
| 17 |
>> In the multilib stuff, we're using CHOST for two purposes: |
| 18 |
>> |
| 19 |
>> 1. wrapped headers are put in /usr/include/$CHOST, |
| 20 |
>> |
| 21 |
>> 2. multilib executables are prefixed with $CHOST-. |
| 22 |
>> |
| 23 |
>> (1) here is not really a killer feature but I'd rather avoid changing |
| 24 |
>> this at this point. (2) is actually a killer feature, since the eclass |
| 25 |
>> sets CHOST properly and thanks to that AC_CHECK_TOOL and friends can |
| 26 |
>> find multilib *-config progs and stuff without any special hackery. |
| 27 |
>> |
| 28 |
>> And those are just the examples I can think of. I suspect that more stuff |
| 29 |
>> may actually expect CHOST to uniquely identify build, especially some |
| 30 |
>> tricky hidden features in autotools :). |
| 31 |
>> |
| 32 |
>> I'd suggest that you changed the CHOST values to uniquely identify ABI |
| 33 |
>> in use, at least in multilib profiles and preferably in all of them. |
| 34 |
> |
| 35 |
> Matt can probably vouch for this better, but the only two ABIs affected by |
| 36 |
> this are n32 and n64. mips[el]-unknown-linux-gnu implies a 32-bit |
| 37 |
> big/little endian CHOST, which means the o32 ABI. |
| 38 |
|
| 39 |
Why? mips64 can do o32 as well. |
| 40 |
|
| 41 |
> |
| 42 |
> According to this Debian wiki entry: |
| 43 |
> https://wiki.debian.org/Multiarch/Tuples |
| 44 |
|
| 45 |
Probably because they don't do mips64/o32. But we do, so we would also |
| 46 |
need *-linux-gnuabi32 |
| 47 |
|
| 48 |
-- |
| 49 |
Regards, |
| 50 |
Markos Chandras |
Archives