Alpha hasn't bootstrapped for 4 weeks (libio problem)

Brad Lucier lucier@math.purdue.edu
Tue Dec 14 16:40:00 GMT 1999


The alpha port of the mainline compiler hasn't bootstrapped since
this change:

Revision 1.102 / (download) - annotate - [select for diffs] , Tue Nov 16 04:47:55 1999 UTC (4 weeks ago) by rth 
Branch: MAIN 
CVS Tags: gcc_ss_19991201, gcc_ss_19991122 
Changes since 1.101: +6 -2 lines
Diff to previous 1.101 (colored) 

        * alpha.c (alpha_build_va_list): Use make_lang_type and
        initialize TYPE_NAME for the va_list record.


                        3160: 
1.91          rth       3161: tree
                        3162: alpha_build_va_list ()
1.1           law       3163: {
1.102        |rth       3164:   tree base, ofs, record, type_decl;
1.1           law       3165: 
                        3166:   if (TARGET_OPEN_VMS)
1.91          rth       3167:     return ptr_type_node;
                        3168: 
1.102        |rth       3169:   record = make_lang_type (RECORD_TYPE);
             |          3170:   type_decl = build_decl (TYPE_DECL, get_identifier ("__va_list_tag"), record);
             |          3171:   TREE_CHAIN (record) = type_decl;
             |          3172:   TYPE_NAME (record) = type_decl;
             |          3173: 
1.91          rth       3174:   /* C++? SET_IS_AGGR_TYPE (record, 1); */
  
It seems that something in libio doesn't like it:

/export/u11/lucier/programs/egcs/objdir-test/gcc/xgcc -B/export/u11/lucier/programs/egcs/objdir-test/gcc/ -B/export/u10/egcs-test/alphaev6-unknown-linux-gnu/bin/ -c -g -O2 -fvtable-thunks -D_GNU_SOURCE -fno-implicit-templates -I. -I../../../libio -nostdinc++ -D_IO_MTSAFE_IO ../../../libio/isscan.cc
../../../libio/isscan.cc: In method `istream &istream::scan (const char *, ...)':
../../../libio/isscan.cc:36: could not convert `ap' to `__va_list_tag'
../../../libio/isscan.cc: In method `istream &istream::vscan (const char *, __va_list_tag)':
../../../libio/isscan.cc:50: could not convert `args' to `__va_list_tag'

Brad Lucier


More information about the Gcc-bugs mailing list