[erlang-questions] Error building R12B-4 on SPARC T1

[Mikael Pettersson]

Toby Thain writes:
 > 
 > On 21-Jun-09, at 3:23 PM, Mikael Pettersson wrote:
 > 
 > > Vance Shipley writes:
 > >> New T1000, new Solaris 10 5/09, new Erlang R13B01, same problem.
 > >>
 > >> I can't be the only one experiencing this.
 > >>
 > >> 	-Vance
 > >>
 > >> On Sun, Sep 14, 2008 at 10:31:11AM +0200, Mikael Pettersson wrote:
 > >> }  Vance Shipley writes:
 > >> }   > Building R12B-4 is failing for me on a Sun T1000 server:
 > >> }   >
 > >> }   >    gcc  -g  -O3 -fomit-frame-pointer
 > >> }   >    ...
 > >> }   >    gcc: Internal error: Segmentation Fault (program cg)
 > >> }   >    Please submit a full bug report to
 > >> }   >     <URL:http://forum.java.sun.com/forum.jspa?forumID=905>.
 > >> }   >    make[3]: *** [obj/sparc-sun-solaris2.10/opt/smp/ 
 > >> erl_bif_info.o] Error 1
 > >> }   >    make[3]: Leaving directory `/usr/local/src/otp_src_R12B-4/ 
 > >> erts/emulator'
 > >> }   >    make[2]: *** [opt] Error 2
 > >> }   >    make[2]: Leaving directory `/usr/local/src/otp_src_R12B-4/ 
 > >> erts/emulator'
 > >> }   >    make[1]: *** [smp] Error 2
 > >> }   >    make[1]: Leaving directory `/usr/local/src/otp_src_R12B-4/ 
 > >> erts'
 > >> }   >    gmake: *** [emulator] Error 2
 > >> }   >
 > >> }   > This is using SUNWgccfss v4.2.0 although it failed  
 > >> identically with v4.0.3.
 > >> }   > I did submit a bug report to Sun as suggested.
 > >
 > > An ICE in gcc is _always_ a bug in that gcc (or in rare cases a HW  
 > > or OS bug).
 > > Report it to Sun.
 > 
 > A segfault however could be bad RAM.

That's one of the things my "(or in rare cases a HW ... bug)" referred to.
That and overheating. And dodgy PSUs.

 > It looks as if it is being  
 > caught and labelled an "internal error" (which imho usually describes  
 > a "can't happen", or self detected error; while a segfault would  
 > normally be a symptom of an *un*detected inconsistency).

GCC deliberately catches fatal signals like SIGSEGV and SIGBUS in order
to output "please report this bug" type messages.

In this case however it's an external program "cg" that got the SIGSEGV.
Proper GCCs don't have any "cg" component, so that's most likely one of
Sun's alterations.