This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa] Java byte compiler status


Diego Novillo writes:
 > On Tue, 2003-07-22 at 13:34, Andrew Haley wrote:
 > 
 > > Perhaps if I check in the front end changes, some tree-SSA guru can
 > > have a look at the trees produced for the failing test cases and tell
 > > me if I'm doing something wrong.
 > > 
 > Could you try disabling tree optimizers individually?  The ones that
 > have been giving me a few failures lately are DCE, must alias and the
 > dominator optimizers.  Try -fno-tree-dce, -fno-tree-must-alias and
 > -fno-tree-dominator-opts.

Fascinating.  Using -fno-tree-dce fixes some of the failures but
(wonderfully!) breaks the compiler with

G19990302_02.java: In class `G19990302_02':

G19990302_02.java: In method `G19990302_02.main(java.lang.String[])':

G19990302_02.java:12: internal compiler error: in coalesce_ssa_name, at tree-ssa.c:1538

Please submit a full bug report,

with preprocessed source if appropriate.

See <URL:http://gcc.gnu.org/bugs.html> for instructions.


i.e. the test case passes with -O but fails with -O -fno-tree-dce.

Anyway, with -O -fno-tree-dce I have 13 failures.  4 of these are new
ICEs like the one above.

With -O -fno-tree-dce -fno-tree-must-alias -fno-tree-dominator-opts I
have 7 failures.  No ICEs.

With -O -fdisable-tree-ssa I have 1 execution failure.  I guess that
one is probably a front end bug.

Andrew.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]