acpica-unix ftbfs on powerpc

see https://launchpad.net/ubuntu/+source/acpica-unix/20150717-1/+build/7764905

Reproducible by hand, after which the coredump from first failure shows:

Core was generated by `./aapits 1 1 ../tmp/aml'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 ___vfprintf_chk (fp=0x0, flag=flag@entry=1, format=0x100ba764 "ACPI Exception: %s, ", ap=0xff8bafa0, ap@entry=0xff8baf10) at vfprintf_chk.c:29
29 vfprintf_chk.c: No such file or directory.
(gdb) where
#0 ___vfprintf_chk (fp=0x0, flag=flag@entry=1, format=0x100ba764 "ACPI Exception: %s, ", ap=0xff8bafa0, ap@entry=0xff8baf10) at vfprintf_chk.c:29
#1 0x10032cc0 in vfprintf (__ap=0xff8baf10, __fmt=<optimized out>, __stream=<optimized out>) at /usr/include/powerpc-linux-gnu/bits/stdio2.h:127
#2 AcpiOsActualVprintf (Fmt=<optimized out>, Args=Args@entry=0xff8bafa0) at osunixxf.c:410
#3 0x1003028c in AcpiOsPrintf (Fmt=Fmt@entry=0x100ba764 "ACPI Exception: %s, ") at atosxfwrap.c:481
#4 0x100940e8 in AcpiException (ModuleName=ModuleName@entry=0x100ba7c8 <_AcpiModuleName> "utxfinit", LineNumber=LineNumber@entry=93, Status=Status@entry=1, Format=Format@entry=0x100ba818 "During OSL initialization")
    at ../../source/components/utilities/utxferror.c:136
#5 0x1009468c in AcpiInitializeSubsystem () at ../../source/components/utilities/utxfinit.c:93
#6 0x100049d8 in AtInitTermCommonTest (OsxfNum=AcpiOsInitializeC, ActFlag=AtActD_OneTime, ActCode=3, TFst=1, Check_Flags=<optimized out>, Benchmark=1) at atinit.c:1065
#7 0x100032c8 in ExecuteTest (test_case=test_case@entry=1, test_num=1) at atmain.c:365
#8 0x10000b88 in main (argc=4, argv=0xff8bb374) at atmain.c:480

So the segfault happens at _IO_acquire_lock_clear_flags2 (fp);
 in glibc.

AcpiGbl_OutputFile, is NULL.

Setting it to stdout gets us to the next segfault, an object being NULL. I get the feeling there is some initialization that is no longer being called. It doesn't stand out to me in the debdiff though.

Hi Al,

just temporarily assigning this to get your comment - do you have any idea why the above segfault woul dhappen during the aapit tests on powerpc? For some reason neither AcpiGbl_RootNode nor AcpiGbl_OutputFile is being initialized.

Discussing this in an upstream issue, some of this appears to be due to the debian automatic test-run being too stringent, while upstream diverges. For instance upstream now raises failures when called with unknown arguments, causing unexpected warnings.

Not yet clear how best to fix this. Upstream recommended simply not running the aapits tests, I'm working through the subsequent failures.

This bug was fixed in the package acpica-unix - 20150717-1ubuntu1

---------------
acpica-unix (20150717-1ubuntu1) wily; urgency=medium

  * Fix tests (LP: #1483836)
    - Don't run aapits tests
    - filter results for innocuous unexpected warning messages

acpica-unix (20150717-1) unstable; urgency=medium

  * Upgrade to latest upstream.
  * Refresh patches as needed.
  * Add new patch to correct AAPITS makefile after source file name changes

 -- Serge Hallyn <email address hidden> Thu, 13 Aug 2015 18:55:56 -0500