[patches] Fixes for the libjava failure

[Franz Sirl]

On Sunday 05 August 2001 15:10, Jan Hubicka wrote:
> > On Saturday 04 August 2001 22:23, Jan Hubicka wrote:
> > > > Re starting the checking on i686 (the continued checking passed
> > > > with this extra patch)
> > >
> > > Just to let you know, with the combination of both patches I finally
> > > got bootstrap and checking working again.
> >
> > I applied your patches and PPC still doesn't bootstrap, it stops in
> > libjava since about a week now. If I enable RTL checking it stops at the
> > end of stage2.
>
> It looks like another failure, as the failure I am tracking down is 2 days
> old.  I will try the PPC one at monday.

Jan, forget about that problem for now, it seems it is partly caused by 
Daniels powerpc64 merge. Daniel, can you investigate that?

The bootstrap fails here with the powerpc64 patch:

running mknumeric_limits
/home/fsirl/obj/gccm/gcc/xgcc -B/home/fsirl/obj/gccm/gcc/ 
-B/home/fsirl/gnubin/ppc-linux/bin/ -B/home/fsirl/gnubin/ppc-linux/lib/ 
-isystem /home/fsirl/gnubin/ppc-linux/include 
-I/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3 -I./ppc-linux/bits/.. 
-I/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3 -fno-exceptions     -o 
/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3/src/gen-num-limits 
/home/fsirl/cvsx/gccm/libstdc++-v3/src/gen-num-limits.cc
/home/fsirl/cvsx/gccm/libstdc++-v3/mknumeric_limits: line 273: 15240 Illegal 
instruction     (core dumped) "$BUILD_DIR/src/gen-num-limits" >>$OUT_H-t
gen-num-limits failed to execute, exiting.
mv: cannot stat 
`/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3/include/bits/std_limits.h': No 
such file or directory
make[4]: *** [ppc-linux/bits/std_limits.h] Error 1
make[4]: Leaving directory 
`/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3/include'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3'
make[2]: *** [all-recursive-am] Error 2
make[2]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/libstdc++-v3'
make[1]: *** [all-target-libstdc++-v3] Error 2
make[1]: Leaving directory `/home/fsirl/obj/gccm'
make: *** [bootstrap] Error 2
Mon Aug  6 12:27:29 MST 2001
0m0.000s 0m0.010s
85m50.930s 3m30.070s

Without the powerpc64 patch it fails here (for Jan again :-) ):

/home/fsirl/obj/gccm/gcc/gcj -B/home/fsirl/obj/gccm/ppc-linux/nof/libjava/ 
-B/home/fsirl/obj/gccm/gcc/ --encoding=UTF-8 -fassume-compiled 
-fclasspath=/home/fsirl/obj/gccm/ppc-linux/nof/libjava 
-L/home/fsirl/obj/gccm/ppc-linux/nof/libjava -g -O2 -msoft-float -fPIC 
-mstrict-align -MD -MT java/security/KeyPairGenerator.lo -MF 
java/security/KeyPairGenerator.d -c 
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java -fPIC -o 
java/security/.libs/KeyPairGenerator.o
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java: In class 
`java.security.KeyPairGenerator':
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java: In method 
`java.security.KeyPairGenerator.generateKeyPair()':
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java:255: 
verify_flow_info: Incorrect fallthru 3->4
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java:255: Wrong 
insn in the fallthru edge
(insn/i 32 316 239 (use (const:SI (plus:SI (symbol_ref:SI 
("@_CD_java_security_KeyPairGenerator"))
                (const_int 24 [0x18])))) -1 (nil)
    (nil))
/home/fsirl/cvsx/gccm/libjava/java/security/KeyPairGenerator.java:255: 
confused by earlier errors, bailing out
make[5]: *** [java/security/KeyPairGenerator.lo] Error 1
make[5]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/nof/libjava'
make[4]: *** [all-recursive] Error 1
make[4]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/nof/libjava'
make[3]: *** [multi-do] Error 1
make[3]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/libjava'
make[2]: *** [all-multi] Error 2
make[2]: Leaving directory `/home/fsirl/obj/gccm/ppc-linux/libjava'
make[1]: *** [all-target-libjava] Error 2
make[1]: Leaving directory `/home/fsirl/obj/gccm'
make: *** [bootstrap] Error 2
Mon Aug  6 14:49:16 MST 2001
0m0.000s 0m0.010s
120m42.240s 10m54.760s

> Why the regression tester didn't complain?

The regression-tester is a crosss-compiled PPC compiler, a native bootstrap 
is a much harder test.

Franz.