| 1 |
On 01/19/2016 01:47 PM, Alex McWhirter wrote: |
| 2 |
> In many places im seeing LD output this in the configure script of a |
| 3 |
> multilib glibc ebuild. This is configuring 32bit glibc on a 64bit host. |
| 4 |
> |
| 5 |
> sparc:v9 architecture of input file `somefile' is incompatible with |
| 6 |
> sparc output |
| 7 |
> |
| 8 |
> |
| 9 |
> |
| 10 |
> So it looks like LD is working right, but GCC is still compiling 64bit. |
| 11 |
> Or it could be G++ compiling 64bit as i don't see a -m32 flag set for |
| 12 |
> it? I seem some references to G++. |
| 13 |
> |
| 14 |
|
| 15 |
I found the issue here, but i'm not sure how to handle it. The logic in |
| 16 |
common.eblit is wrong for sparc64. See below |
| 17 |
|
| 18 |
<snip> |
| 19 |
local cpu |
| 20 |
case ${CTARGET} in |
| 21 |
sparc64-*) |
| 22 |
case $(get-flag mcpu) in |
| 23 |
niagara[234]) |
| 24 |
if version_is_at_least 2.8 ; then |
| 25 |
cpu="sparc64v2" |
| 26 |
elif version_is_at_least 2.4 ; then |
| 27 |
cpu="sparc64v" |
| 28 |
elif version_is_at_least 2.2.3 ; |
| 29 |
then |
| 30 |
cpu="sparc64b" |
| 31 |
fi |
| 32 |
;; |
| 33 |
niagara) |
| 34 |
if version_is_at_least 2.4 ; then |
| 35 |
cpu="sparc64v" |
| 36 |
elif version_is_at_least 2.2.3 ; |
| 37 |
then |
| 38 |
cpu="sparc64b" |
| 39 |
fi |
| 40 |
;; |
| 41 |
ultrasparc3) |
| 42 |
cpu="sparc64b" |
| 43 |
;; |
| 44 |
*) |
| 45 |
# We need to force at least v9a |
| 46 |
because the base build doesn't |
| 47 |
# work with just v9. |
| 48 |
# |
| 49 |
https://sourceware.org/bugzilla/show_bug.cgi?id=19477 |
| 50 |
[[ -z ${cpu} ]] && append-flags |
| 51 |
"-Wa,-xarch=v9a" |
| 52 |
;; |
| 53 |
esac |
| 54 |
;; |
| 55 |
</snip> |
| 56 |
|
| 57 |
Pulling from CHOST and basing CTARGET off of that is probably not the |
| 58 |
best idea for sparc64 because the result will always be 64 bit. Hence |
| 59 |
why we have build errors when attempting 32bit multi lib. |
| 60 |
|
| 61 |
Also, mcpu=ultrasparc will set the target to v9a and all sparv9 boxes |
| 62 |
are backwards compatible to v9a. Ultrasparc being the lowest common |
| 63 |
denominator for 64 bit means we can clean up the wild card a bit. That's |
| 64 |
not a big deal really, forcing v9a should have a similar effect. |
| 65 |
|
| 66 |
I need to research the specific targets a bit and maybe i can come up |
| 67 |
with an answer. Do we have {$ABI} here? |
Archives