Re: lynx-dev LYMainLoop.c -- gcc signal 11, egcs-1.1.2

[Leonid Pauzner]

17-Oct-99 07:39 T.E.Dickey wrote:
>>
>> On Sat, Oct 16, 1999, T.E.Dickey wrote:
>> > >
>> > > Is anyone else having any problems compiling LYMainLoop.c in dev.12?
>> > >
>> > > gcc  -DHAVE_CONFIG_H  -DLOCALEDIR=\"/usr/local/share/locale\" -I. -I.. 
>> > > -Ichrtrans -I./chrtrans -I.. -I../src -I../WWW/Library/Implementation   
>> > > -O2 -DLINUX  -c ./LYMainLoop.c
>> > > gcc: Internal compiler error: program cpp got fatal signal 11

I also saw a compiler death on dev.11/dev.12  LYMainLoop.c
It was quite reproduceble, but gone away after rebooting
(the computer was turned off for several minutes, may be a hardware error?).
The simptom was like memory corruption due to buffer overrun
but the program in question is gcc 2.8.1 (I haven't seen the compiler
crash before)!

[The configuration is gcc 2.8.1 + DJGPP2.02 + cwsdpmi server,
the crashed executable was cc1.exe, i386 CPU].


>> > > make[1]: *** [LYMainLoop.o] Error 1
>> > > make[1]: Leaving directory `/home/pk/lynx/2.8.3dev.12/src'
>> > > make: *** [all] Error 2
>> > > address@hidden ~/lynx/2.8.3dev.12]$ gcc -v
>> > > Reading specs from 
>> > > /usr/local/lib/gcc-lib/i586-pc-linux-gnu/egcs-2.91.66/specs
>> > > gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release)

> I'm building with that version right now, to check (seems to work - but I'll
> try with your configuration settings).

>> > > address@hidden ~/lynx/2.8.3dev.12]$ uname -a
>> > > Linux odin 2.2.12 #3 Thu Sep 2 06:56:14 EDT 1999 i586 unknown
>> [...]
>> > > I can include my cfg_defs.h or lynx_cfg.h if anyone is curious.  I
>> > > doubt this will be debuggable, but figured it might be a data point
>> > > worth something.
>> >
>> > The only way I can see to debug it is to chop pieces out of LYMainLoop.c
>> > until the problem goes away - to see what's causing the problem.  I saw
>> > some comment recently that indicated egcs has some problems with code that
>> > compiles fine with the gcc versions.
>>
>> Any suggestions on which functions to start with?  Probably the main
>> loop.  Also, I just noticed in what I have up there -- it says "cpp"

> yes - or the functions "handle_LYK_xxx" that I split-out of the main loop.
> That would be possibly better (since chopping out the main loop would leave
> a lot of orphaned private functions, while chopping out the functions 
> themselves
> would just leave a lot of implicit declarations).

>> got signal 11, not gcc.  Maybe that's what usually comes up, I don't
>> have any old compile failures to compare to.
>>
>> I'll try that.
>>
>>
>> --
>> Hawkeye's Conclusion:
>>      It's not easy to play the clown
>>      when you've got to run the whole circus.


> --
> Thomas E. Dickey
> address@hidden
> http://www.clark.net/pub/dickey