[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 --- Comment #9 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-03 08:29:07 UTC --- Author: jakub Date: Thu Feb 3 08:29:03 2011 New Revision: 169784 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=169784 Log: PR target/47564 * toplev.c (target_reinit): Save and restore *crtl and regno_reg_rtx around backend_init_target and lang_dependent_init_target calls. * cgraphunit.c (cgraph_debug_gimple_stmt): New function. (verify_cgraph_node): Don't call set_cfun here. Use cgraph_debug_gimple_stmt instead of debug_gimple_stmt. Set error_found for incorrectly represented calls to thunks. * gcc.target/i386/pr47564.c: New test. Added: trunk/gcc/testsuite/gcc.target/i386/pr47564.c Modified: trunk/gcc/ChangeLog trunk/gcc/cgraphunit.c trunk/gcc/testsuite/ChangeLog trunk/gcc/toplev.c
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #10 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-03 08:32:44 UTC --- Fixed.
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |jakub at gcc dot gnu.org |gnu.org | --- Comment #8 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-02 10:00:11 UTC --- Created attachment 23211 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=23211 gcc46-pr47564.patch Updated fix. For 4.7 I'd say we want to split this reinitialization up, into reinitializations that matter for tree optimizations (ideally those could be saved into (and restored from) some save area pointed to from TARGET_OPTION_NODE, e.g. init_set_costs can be saved into target_cfgloop structure), and initializations that only matter from expand_gimple_cfg entry till free_after_compilation (there we could remember the last TARGET_OPTION_NODE (or global) and if current TARGET_OPTION_NODE (or global) is different from it, call the init_* calls at the beginning of expand_gimple_cfg.
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Summary|internal compiler error in |[4.6 Regression] internal |memory_address_addr_space, |compiler error in |at explow.c:504 |memory_address_addr_space, ||at explow.c:504 --- Comment #5 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-01 12:32:51 UTC --- I get the ICE on x86_64-linux -m64 -O2 too, so that's not it. Started failing with http://gcc.gnu.org/viewcvs?root=gccview=revrev=167964
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 Jakub Jelinek jakub at gcc dot gnu.org changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2011.02.01 14:18:55 Target Milestone|--- |4.6.0 Ever Confirmed|0 |1
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 --- Comment #6 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-01 14:40:50 UTC --- Created attachment 23198 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=23198 gcc46-pr47564.patch Ugh, this is ugly. The problem is that tree_rest_of_compilation calls init_emit and lots of other init* functions and assumes x_rtl will stay initialized until expansion. But if during any of the following optimization passes set_cfun is called to some other function which has different target options, target_reinit is called, which calls free_after_compilation which will clear x_rtl. Among other things it clears reg_rtx_no, which means when expanding E pseudo allocation starts from 0 instead of first pseudo (i.e. allocates various hard/virtual registers first, which of course can't lead to anything sane). Most of the set_cfun calls actually came from cgraph verification, where it is not really needed - IMHO we can set_cfun there just when we are about to diagnose an issue and when we do that, we will internal_error at the end of function anyway and thus don't have to worry about restoring it at all. The attached patch fixes that. On the testcase there is another set_cfun call after tree_rest_of_compilation has been called though, in particular from cgraph_release_function_body during inlining (called from cgraph_release_node). I guess in such a case we don't actually need target reinit, perhaps we could just change cfun after saving the previous value, instead of using push_cfun/pop_cfun. Or perhaps target_reinit could save x_rtl copy before it and restore it afterwards, so that it doesn't clear it if it wasn't all 0's before.
[Bug target/47564] [4.6 Regression] internal compiler error in memory_address_addr_space, at explow.c:504
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47564 --- Comment #7 from Jakub Jelinek jakub at gcc dot gnu.org 2011-02-01 15:47:07 UTC --- Following works by not doing target_reinit once tree_rest_of_compilation calls init_function_start, until final. --- function.c.jj 2011-01-25 18:40:08.0 +0100 +++ function.c 2011-02-01 16:33:48.0 +0100 @@ -4294,7 +4294,10 @@ invoke_set_current_function_hook (tree f cl_optimization_restore (global_options, TREE_OPTIMIZATION (opts)); } - targetm.set_current_function (fndecl); + /* Don't call targetm.set_current_function in between + prepare_function_start and following free_after_compilation. */ + if (regno_reg_rtx == NULL) +targetm.set_current_function (fndecl); } }
