This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Getting past gcc-3.2 20000611 breakage.
- From: Matt Thomas <matt at 3am-software dot com>
- To: gcc at gcc dot gnu dot org
- Date: Sun, 16 Jun 2002 11:26:24 -0700
- Subject: Getting past gcc-3.2 20000611 breakage.
I'm currently working on getting the VAX ELF support I've done for
gcc-2.95.3 in NetBSD applied to the gcc mainline branch.
Progress was going well until I cvs updated my sources yesterday.
gcc-3.2 started breaking on very simple sources which worked fine
before:
GNU C version 3.2 20020611 (experimental) (vax-netbsdelf)
compiled by GNU C version 2.95.3 20010315 (release) (NetBSD nb2).
<stdin>: In function `x':
<stdin>:9: error: missing barrier after block 0
<stdin>:9: internal compiler error: verify_flow_info failed
void x(int a) {
switch (a) {
case 1: d(); break;
default: abort();
case 3: e(); break;
case 7: e(); break;
case 15: e(); break;
}
}
Now I suspect the problem is related to the following change:
2002-06-05 David S. Miller <davem@redhat.com>
Delete SEQUENCE rtl usage outside of reorg and ssa passes.
but fixing it is beyond my abilties to fix. By doing dated cvs
updates I was about to verify -D'2002/06/11 UTC' sources work while
-D'2002/06/12 UTC' sources fail. The above Change appeared in the
ChangeLog between the two update.
I've prepared my first set of changes for VAX ELF but I can't test
them (expect for compiling) on the latest sources since they fail.
They do work with the 20000610 sources (I'm running a NetBSD/vax
userland with them) and they should be submitted as a patch real soon.
My question is what to do now for my further development? Back up to
the 2000/06/11 source and do development from there? Hold off and wait
for the breakage to be fixed?
--
Matt Thomas Internet: matt@3am-software.com
3am Software Foundry WWW URL: http://www.3am-software.com/bio/matt/
Cupertino, CA Disclaimer: I avow all knowledge of this message