------- Additional Comments From bredelin at ucla dot edu 2005-03-20 15:57 ------- Subject: Re: ICE in cgraph.C with C++
belyshev at depni dot sinp dot msu dot ru wrote: > ------- Additional Comments From belyshev at depni dot sinp dot msu dot ru > 2005-03-20 11:25 ------- > >>This is with today's (Mar 17, 2005) CVS, with Richard's Guenther's patch to >>modify inlining heuristics. > > > which one? please add references to all patches you installed or, better, try > to > reproduce ICE with clean tree by adding appropriate --param options. > 1. For now, I will just include the patch: diff -r1.170 tree-inline.c 1249a1250 > case TARGET_EXPR: 1250a1252,1253 > if (DECL_P (x) && DECL_IGNORED_P (x)) > break; 1252d1254 < case TARGET_EXPR: 2. However, it seems that as of Mar 18 CVS, the crash no longer occurs. This is probably because of this change: Fri Mar 18 10:00:16 2005 UTC by hubicka PR middle-end/20225 * cgraph.c (cgraph_mark_reachable_node): Assert that it is not called too late. * varasm.c (find_decl_and_mark_needed): Mark needed only when not called too late. 3. This patch might be reversed: http://gcc.gnu.org/ml/gcc-patches/2005-03/msg01869.html If the regression returns, I will try to find --param options that cause the ICE on a clean tree. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20521