Changes between Version 8 and Version 16 of Ticket #2121


Ignore:
Timestamp:
Dec 2, 2013, 4:35:37 AM (10 years ago)
Author:
Josh
Comment:

Updated the description to reflect current status. I'd like to make sure A15 will build on armhf and armel without patching.

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #2121

    • Property Status new → reopened
    • Property Component Core engine → Build & Packages
    • Property Owner set to Josh
    • Property Keywords arm added

  • Ticket #2121 – Description

    v8 v16  
    1 Recently after fixing nvtt package Debian also enabled building armel and armhf 0ad. Both builds fail with different issues. The latest build status and logs are available here: https://buildd.debian.org/status/package.php?p=0ad&suite=sid
     1Recently after fixing nvtt package Debian and Ubuntu also enabled building armel and armhf 0ad. Both builds fail with different issues. The latest build status and logs are available here: https://buildd.debian.org/status/package.php?p=0ad&suite=sid
    22
    33The arm specific build log are available at:
    … …  
    55* armhf: https://buildd.debian.org/status/logs.php?pkg=0ad&arch=armhf
    66
    7 == 1) armel
     7== 1) armel (fixed in SVN?)
    88fails with:
    99{{{
    … …  
    2525}}}
    2626
    27 == 2) armhf
    28 builds fine, but test fails with:
     27== 2) armhf on Ubuntu 12.04
     28fails with:
    2929{{{
    30 Illegal instruction
     30../../../source/simulation2/helpers/Geometry.cpp: In function 'bool Geometry::TestRaySquare(CFixedVector2D, CFixedVector2D, CFixedVector2D, CFixedVector2D, CFixedVector2D)':
     31../../../source/simulation2/helpers/Geometry.cpp:239:1: error: unable to find a register to spill in class 'LO_REGS'
     32../../../source/simulation2/helpers/Geometry.cpp:239:1: error: this is the insn:
     33(insn 75 20 17 2 (set (reg/v:DI 141 [ sum ])
     34        (plus:DI (mult:DI (sign_extend:DI (reg:SI 245 [ a$value ]))
     35                (sign_extend:DI (reg:SI 241 [ u$value ])))
     36            (reg/v:DI 141 [ sum ]))) ../../../source/maths/FixedVector2D.h:200 49 {*mulsidi3adddi_v6}
     37     (nil))
     38../../../source/simulation2/helpers/Geometry.cpp:239: confused by earlier errors, bailing out
    3139}}}
    32 As discussed with Philip on irc http://irclogs.wildfiregames.com/2013-09-07-QuakeNet-%230ad-dev.log :
    33 {{{
    34 16:51 < Philip`> We're building with "-mtune=cortex-a15 -mfpu=neon-vfpv4" so I suppose it's possible the compiler is using some VFPv4-only instructions
    35 16:54 < fabio4> it was changed recently: http://trac.wildfiregames.com/changeset?reponame=&new=13634%40ps%2Ftrunk%2Fbuild%2Fpremake%2Fpremake4.lua&old=13469%40ps%2Ftrunk%2Fbuild%2Fpremake%2Fpremake4.lua
    36 16:55 < Philip`> fabio4: Yeah, it was changed to match a Chromebook
    37 16:55 < Philip`> fabio4: (unnecessarily, I think)
    38 16:56 < Philip`> (or at least the -mfpu=neon-vfpv4 thing is probably unnecessary)
    39 16:57 < Philip`> (and tuning for Cortex-A15 is generally terrible if the code actually runs on any other kind of Cortex)
    40 }}}
    41 -mfloat-abi=hard is also probably already implied by the architecture.
    4240
    4341== 3) SSE warning:
    4442when running test on a proper architecture this is printed:
    4543{{{
    46 No SSE available. Slow fallback routines will be used.
     44No SSE available. Slow fallback routines will b used.
    4745}}}
    4846The message should be updated.
    49 
    50 Targetting alpha15 since the armhf issue is easy to fix, no idea about the armel one.