Watching Arm failures

Information Source:

[http://buildd.debian.org/~jeroen/status/architecture.php?a=arm Arm buildd]

Feel free to:

gcc4 ICEs

cp_expr_size, at cp/cp-objcp-common.c:101

This is filed as #323133 in bugzilla against gcc-4.0. Gcc issue 21123. Now fixed. When we get a corresponding gcc4 upload then these can all be requeued.

Action: walk through the maybe-failed/building list and file bugs against all.

Packages in this list need to use g++3.4 on arm/armeb/hppa/m68k archs.

unclassified ICE's

These need proper bugreports generating and filing upstream. Comparing against cvs-snapshot on an unstable system if you can.

{{{fitsrw.c:596: error: insn does not satisfy its constraints: (insn:HI 2186 2184 3064 219 fitsrw.c:1358 (set (reg/v:SF 20 f4 [orig:186 minval ] [186])

fitsrw.c:596: internal compiler error: in reload_cse_simplify_operands, at postreload.c:391}}}

{{{pattern.c: In function 'piece_to_pattern_distance': pattern.c:506: fatal error: internal consistency failure compilation terminated.}}}

{{{ugsc.c: In function 'phaser1set': ugsc.c:334: fatal error: internal consistency failure compilation terminated.}}}

{{{cc -g -Wall -O2 -fPIC -rdynamic -Wl,-export-dynamic -L. main.o -lruby1.9 -lpthread -ldl -lcrypt -lm -o ruby1.9 make[2]: Leaving directory `/build/buildd/ruby1.9-1.9.0+20050921' lib/fileutils.rb:449: [BUG] Segmentation fault ruby 1.9.0 (2005-09-21)[arm-linux] }}}

Normal issues

missing closure support in libffi

This causes libffi-using programs to blow up with [http://buildd.debian.org/~jeroen/status/package.php?p=ctypes&a=arm#fail-arm obscure errors] when they try to declare an "ffi_closure" object.

move to failed - bugs filed

depwait

needs retrying

needs urgent research - regressions

{{{g++ -c -pipe -DQT_NO_CODECS -DQT_LITE_UNICODE -Wall -W -fno-exceptions -O2 -I. -o ../objects/qbuffer.o qbuffer.cpp qstring.h: In member function 'uchar& QChar::cell()': qstring.h:166: error: cannot bind packed field '((QChar*)this)->QChar::cl' to 'uchar&'}}}

needs research - never worked

java

This is pretty bad, since db* want java. gcj-4.0 (as opposed to gcj-3.3 and gcj-3.4) does work fine on ARM, but gij-4.0 does not. See #337263. Using a different VM such as sablevm might be a solution here.

not-for-us

These packages are not for arch=arm so are not expected to build. They still appear on the buildd status pages as failures.

fixed/past issues