[Mingw-users] More progress with building gcc 8.1.0

Back to archive index

David Gressett DGres****@amli-*****
Wed Jul 11 02:14:42 JST 2018


I used Eli's suggestion to fix the ENOTSUP problem and found another #define deficiency

This time, the build crashed wen compiling gcc8.1.0/gcc/pretty-print.c.

There is a new Windows-specific addition her, and it has two undefined constant problems:

COMMON_LVB_UNDERSCORE

COMMON_LVB_REVERSE_VIDEO

A bit of web searching quickly revealed that these are documented in Microsoft's online documentation.

The actual numeric values are documented. They should be in wincon.h:

For now, I have hacked the definitions into the pretty-print.c source code using Eli's ENOTSUP fix as a model.

That seems to be tha last of the definition failures. 

The build crashed at the  point where it compared  the stage 2 and stage 3 binaries:

Comparing stages 2 and 3
warning: gcc/cc1obj-checksum.o differs
warning: gcc/cc1objplus-checksum.o differs

The 7.3.0 build produced the same warning, but the 8.1.0 build crashed immediately after that warning:

make[1]: *** [stage3-bubble] Error 58

I'm still investigating that.

But, back to wincon.h.

There is a lot of detailed documentation on the Microsoft documentation web site, complete with values for constants that can go into wincon.h,
and it is all publically documented, so I am now working on a rebuild of wincon.h, which is the best solution for the gcc build failure.




More information about the MinGW-Users mailing list
Back to archive index