Created attachment 29392 [details] Test case ICE when the same member function pointer (to a base class's method) is used as a template parameter from two different methods of the same templated child class. What's a bit unusual here is that the bug is only triggered if the same instantiation occurs in two different methods. The reduced test case is short enough to include directly: -------- struct Base { void method() {} }; typedef void (Base::*MemPtr)(); // Template with a member function pointer "non-type parameter". template<MemPtr func> struct Wrapper {}; template<class C> struct Child : public Base { // Templated derived class instantiates the Wrapper with the same parameter // in two different virtual methods. void foo() { typedef Wrapper<&Base::method> W; } void bar() { typedef Wrapper<&Base::method> W; } }; // Instantiate Child with some type. template class Child<int>; -------- Here's the detailed gcc output: % /usr/local/bin/g++47 -v -save-temps -c bugtest.cc Using built-in specs. COLLECT_GCC=/usr/local/bin/g++47 Target: x86_64-portbld-freebsd8.2 Configured with: ./../gcc-4.7-20120818/configure --disable-nls --enable-languages=c,c++,objc,fortran --libdir=/usr/local/lib/gcc47 --libexecdir=/usr/local/libexec/gcc47 --program-suffix=47 --with-as=/usr/local/bin/as --with-gmp=/usr/local --with-gxx-include-dir=/usr/local/lib/gcc47/include/c++/ --with-ld=/usr/local/bin/ld --with-libiconv-prefix=/usr/local --with-pkgversion='FreeBSD Ports Collection' --with-system-zlib --enable-languages=c,c++,objc,fortran,java --prefix=/usr/local --mandir=/usr/local/man --infodir=/usr/local/info/gcc47 --build=x86_64-portbld-freebsd8.2 Thread model: posix gcc version 4.7.2 20120818 (prerelease) (FreeBSD Ports Collection) COLLECT_GCC_OPTIONS='-v' '-save-temps' '-c' '-shared-libgcc' '-mtune=generic' '-march=x86-64' /usr/local/libexec/gcc47/gcc/x86_64-portbld-freebsd8.2/4.7.2/cc1plus -E -quiet -v src/scratch/bugtest.cc -mtune=generic -march=x86-64 -fpch-preprocess -o bugtest.ii ignoring nonexistent directory "/usr/local/lib/gcc47/gcc/x86_64-portbld-freebsd8.2/4.7.2/../../../../../x86_64-portbld-freebsd8.2/include" #include "..." search starts here: #include <...> search starts here: /usr/local/lib/gcc47/include/c++/ /usr/local/lib/gcc47/include/c++//x86_64-portbld-freebsd8.2 /usr/local/lib/gcc47/include/c++//backward /usr/local/lib/gcc47/gcc/x86_64-portbld-freebsd8.2/4.7.2/include /usr/local/include /usr/local/lib/gcc47/gcc/x86_64-portbld-freebsd8.2/4.7.2/include-fixed /usr/include End of search list. COLLECT_GCC_OPTIONS='-v' '-save-temps' '-c' '-shared-libgcc' '-mtune=generic' '-march=x86-64' /usr/local/libexec/gcc47/gcc/x86_64-portbld-freebsd8.2/4.7.2/cc1plus -fpreprocessed bugtest.ii -quiet -dumpbase bugtest.cc -mtune=generic -march=x86-64 -auxbase bugtest -version -o bugtest.s GNU C++ (FreeBSD Ports Collection) version 4.7.2 20120818 (prerelease) (x86_64-portbld-freebsd8.2) compiled by GNU C version 4.7.2 20120818 (prerelease), GMP version 5.0.1, MPFR version 3.0.0, MPC version 0.8.2 GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072 GNU C++ (FreeBSD Ports Collection) version 4.7.2 20120818 (prerelease) (x86_64-portbld-freebsd8.2) compiled by GNU C version 4.7.2 20120818 (prerelease), GMP version 5.0.1, MPFR version 3.0.0, MPC version 0.8.2 GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072 Compiler executable checksum: 4c5936a9f1a6230b2e84e5f997572e48 src/scratch/bugtest.cc: In instantiation of 'void Child<C>::bar() [with C = int]': src/scratch/bugtest.cc:21:16: required from here src/scratch/bugtest.cc:17:49: internal compiler error: in tsubst_copy, at cp/pt.c:12131 Please submit a full bug report, with preprocessed source if appropriate. See <http://gcc.gnu.org/bugs.html> for instructions.
Confirmed. 4.5.4 works for me.
Started with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=149223
Author: jason Date: Sat Feb 9 20:39:13 2013 New Revision: 195922 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195922 Log: PR c++/56247 * pt.c (eq_specializations): Set comparing_specializations. * tree.c (cp_tree_equal): Check it. * cp-tree.h: Declare it. Added: trunk/gcc/testsuite/g++.dg/template/ptrmem23.C Modified: trunk/gcc/cp/ChangeLog trunk/gcc/cp/cp-tree.h trunk/gcc/cp/pt.c trunk/gcc/cp/tree.c
Author: jason Date: Sat Feb 9 20:47:24 2013 New Revision: 195923 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195923 Log: PR c++/56247 * pt.c (eq_specializations): Set comparing_specializations. * tree.c (cp_tree_equal): Check it. * cp-tree.h: Declare it. Added: branches/gcc-4_7-branch/gcc/testsuite/g++.dg/template/ptrmem23.C Modified: branches/gcc-4_7-branch/gcc/cp/ChangeLog branches/gcc-4_7-branch/gcc/cp/cp-tree.h branches/gcc-4_7-branch/gcc/cp/pt.c branches/gcc-4_7-branch/gcc/cp/tree.c
Author: jason Date: Sat Feb 9 20:53:59 2013 New Revision: 195924 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195924 Log: PR c++/56247 * pt.c (eq_specializations): Set comparing_specializations. * tree.c (cp_tree_equal): Check it. * cp-tree.h: Declare it. Added: branches/gcc-4_6-branch/gcc/testsuite/g++.dg/template/ptrmem23.C Modified: branches/gcc-4_6-branch/gcc/cp/ChangeLog branches/gcc-4_6-branch/gcc/cp/cp-tree.h branches/gcc-4_6-branch/gcc/cp/pt.c branches/gcc-4_6-branch/gcc/cp/tree.c
Fixed.
Awesome quick work. Very impressive -- thank you!
The master branch has been updated by Patrick Palka <ppalka@gcc.gnu.org>: https://gcc.gnu.org/g:f83adb68ed9cef1fbd4c16447eb4e89676df9f62 commit r10-7787-gf83adb68ed9cef1fbd4c16447eb4e89676df9f62 Author: Patrick Palka <ppalka@redhat.com> Date: Sat Apr 18 06:22:21 2020 -0400 c++: spec_hasher::equal and PARM_DECLs [PR94632] In the testcase below, during specialization of c<int>::d, we build two identical specializations of the parameter type b<decltype(e)::k> -- one when substituting into c<int>::d's TYPE_ARG_TYPES and another when substituting into c<int>::d's DECL_ARGUMENTS. We don't reuse the first specialization the second time around as a consequence of the fix for PR c++/56247 which made PARM_DECLs always compare different from one another during spec_hasher::equal. As a result, when looking up existing specializations of 'b', spec_hasher::equal considers the template argument decltype(e')::k to be different from decltype(e'')::k, where e' and e'' are the result of two calls to tsubst_copy on the PARM_DECL e. Since the two specializations are considered different due to the mentioned fix, their TYPE_CANONICAL points to themselves even though they are otherwise identical types, and this triggers an ICE in maybe_rebuild_function_decl_type when comparing the TYPE_ARG_TYPES of c<int>::d to its DECL_ARGUMENTS. This patch fixes this issue at the spec_hasher::equal level by ignoring the 'comparing_specializations' flag in cp_tree_equal whenever the DECL_CONTEXTs of the two parameters are identical. This seems to be a sufficient condition to be able to correctly compare PARM_DECLs structurally. (This also subsumes the CONSTRAINT_VAR_P check since constraint variables all have empty, and therefore identical, DECL_CONTEXTs.) gcc/cp/ChangeLog: PR c++/94632 * tree.c (cp_tree_equal) <case PARM_DECL>: Ignore comparing_specializations if the parameters' contexts are identical. gcc/testsuite/ChangeLog: PR c++/94632 * g++.dg/template/canon-type-14.C: New test.