[Open64-user] RE: [Ipf-orc-support] Problems with compiled source code
Brought to you by:
ributzka,
suneeljain
From: Chan, S. C <sun...@in...> - 2002-03-08 22:27:20
|
Apparently its dying in global scheduler. Did you get the same problem with the "released" compiler binary? Sun > -----Original Message----- > From: Gautam Chakrabarti [mailto:cha...@pi...] > Sent: Friday, March 08, 2002 2:21 PM > To: ope...@li... > Cc: ipf...@so... > Subject: [Ipf-orc-support] Problems with compiled source code > > > Hi, > > I had earlier installed ORC from the binaries, the ORC FAQ suggests > installing from the binaries. Then I compiled the source code as given > in README.src using makefiles Make.cross and Make.NUE.lib, > and then did > INSTALL. Please let me know if this process is wrong, this process was > mentioned for sgicc. > > After this installation, compilation is failing in BE after > giving some > warnings in INLINE and BE (when I use O2 or O3). I am giving part of > the errors for the benchmark Equake :::: > > --------------------------------------------------------- > /nue/usr/lib/gcc-lib/ia64-orc-linux/orcc-1.0/inline > -LIST:source=off:notes=off > -PHASE:l:w:c -TENV:PIC -m1 -INTERNAL:return_val=on > -INTERNAL:mldid_mstid=on > -INTERNAL:return_info=on -m0 -show -O3 -TARG:abi=i64 > -fB,/tmp/ccB.GvyZ96 > -fI,/tmp/ccI.dSPj0L equake.c > !!! DevWarn during Reading WHIRL file: Should use ST_pu_type instead > !!! DevWarn during Reading WHIRL file: Count limit reached on > the following > DevWarn: > !!! DevWarn during Reading WHIRL file: Should use ST_pu_type instead > !!! DevWarn during Reading WHIRL file: TODO: implement *skip* option > !!! DevWarn during Reading WHIRL file: IPA_NODE::Skip is not > yet implemented > > /nue/usr/lib/gcc-lib/ia64-orc-linux/orcc-1.0/be > -LIST:source=off:notes=off > -PHASE:l:w:c -G8 -TENV:PIC -m1 -INTERNAL:return_val=on > -INTERNAL:mldid_mstid=on > -INTERNAL:return_info=on -m0 -show -O3 -TARG:abi=i64 -LANG:=ansi_c > -fB,/tmp/ccI.dSPj0L -s -fs,/tmp/ccs.MGyQpY equake.c > Compiling equake.c (/tmp/ccI.dSPj0L) -- Back End > Compiling main(0) > !!! DevWarn during Loop Nest Optimizer: Count limit reached > on the following > DevWarn: > !!! DevWarn during Loop Nest Optimizer: TODO: Tune LNO > machine model parameters > for IA-64 > Signal: Segmentation fault in Global code motion phase. > Error: Signal Segmentation fault in phase Global code motion > -- processing > aborted > orcc ERROR: /nue/usr/lib/gcc-lib/ia64-orc-linux/orcc-1.0/be > died due to signal > 4 > make: *** [equake.o] Error 32 > -------------------------------------------------------------- > ---------- > > I found its giving the segmentation fault in the call to > > CG_Generate_Code(pu, alias_mgr, PU_Info_pu_dst(current_pu), FALSE); > > inside Post_Process_Backend() > > The warnings from inline are coming through the function > ST_type() and it > asks that the function ST_pu_type() should be used instead. I > am wondering > if it should have got into this...I think CG_Generate_Code_p > is not getting a > proper value in the call to CG_Generate_Code().... > > Please let me know what I need to do. > > Thanks, > Gautam > > ------------------------------------------- > Gautam Chakrabarti > MS Student > Computer Sc. & Engineering Dept. > Michigan State University > > Email: cha...@ms... > Home Page: http://www.cse.msu.edu/~chakra10 > -------------------------------------------- > > _______________________________________________ > ipf-orc-support mailing list > ipf...@li... > https://lists.sourceforge.net/lists/listinfo/ipf-orc-support > |