terry@eecea.eece.ksu.edu (Terry Hull) (05/18/89)
Has anyone tried to get X11R3 with the Purdue speedups compiled with
GCC 1.35 on a color Sun 3/60 running SunOS 3.5? I've seen postings of
people getting it up with 1.34 but not 1.35. Anyway, I started with
the vanilla R3 distribution, applied the 2.0 and 2.1 Purdue patches
and compiled with 1.35 of GCC.
Now when I try to start X, I get:
XIO fatal IO error 32 (Broken pipe on X server unix:0.0)
After 39 requests (29 known prcoessed) with 0 events remaining.
The connection was probably broken by a server shutdown or
KillClient.
The only thing compiled with cc was oscolor.c. When using GCC, it
complained about a recursive definition of the fetch macro.
Any help that you could provide would be greatly appreciated. Thanks
in advance for your time.
--
Terry Hull
Department of Electrical and Computer Engineering, Kansas State University
Work: terry@eecea.eece.ksu.edu, rutgers!ksuvax1!eecea!terry
Play: tah386!terry@eecea.eece.ksu.edu, rutgers!ksuvax1!eecea!tah386!terryterry@eecea.eece.ksu.edu (Terry Hull) (05/18/89)
In article <645@eecea.eece.ksu.edu> terry@eecea.eece.ksu.edu (Terry Hull) writes: >Has anyone tried to get X11R3 with the Purdue speedups compiled with >GCC 1.35 on a color Sun 3/60 running SunOS 3.5? I've seen postings of >people getting it up with 1.34 but not 1.35. Anyway, I started with >the vanilla R3 distribution, applied the 2.0 and 2.1 Purdue patches >and compiled with 1.35 of GCC. I received a suggestion via e-mail from rusty@garnet.berkeley.edu to reduce the options in GCC to -O -traditional -fstrength-reduce and it works! I had been trying to use -finline-functions in addition to these, and it sure seems to break the server. -- Terry Hull Department of Electrical and Computer Engineering, Kansas State University Work: terry@eecea.eece.ksu.edu, rutgers!ksuvax1!eecea!terry Play: tah386!terry@eecea.eece.ksu.edu, rutgers!ksuvax1!eecea!tah386!terry
keith@EXPO.LCS.MIT.EDU (05/21/89)
The X server on the sun should build fine with GCC 1.35; rms gave me a preliminary version (1.34.92) to test before he shipped 1.35. I compiled our entire core sources (R3++) -O -g -fstrength-reduce -fpcc-struct-return and it ran without trouble. We are now regularly building the X server with GCC (as it nearly doubles the speed in monochrome) so you should expect fewer troubles with the next release of X (we just need to keep making sure it compiles with the sun compiler as well. Sigh). You should probably not use -finline-functions; gcc has a funny idea of "reasonable" inlinable functions; the sample server uses enough macros where needed for performance to make the speedup from -finline-functions negligable, while the code bloat is quite amazing. Keith Packard MIT X Consortium