Richard Biener <rguent...@suse.de> writes: >> * wide-int.h no longer includes tm.h. >> >> This is done by adding a new MAX_BITS_PER_UNIT to machmode.def, >> so that the definition of MAX_BITSIZE_MODE_ANY_MODE no longer relies on >> BITS_PER_UNIT. Although I think we usually assume that BITS_PER_UNIT >> is a constant, that wouldn't necessarily be true if we ever did support >> multi-target compilers in future. MAX_BITS_PER_UNIT is logically >> the maximum value of BITS_PER_UNIT for any compiled-in target and must >> be a constant. > > It's not necessary to include gmp.h either - that is included from > system.h now.
Ah, yeah. I've removed that too. > I've looked at the resulting wide-int.h and like it a lot > compared to what is on the branch (less code duplication for one). > > I think we should go ahead with this change (keeping the double-int > changes out for now, I didn't yet look at that patch). We can > iterate on the details on the branch. Thanks. Kenny also asked me to commit it on IRC, so I was going to go ahead. But I just tried boostrapping the patch again after Mike's recent merge with trunk, and it now fails during stage 2 with: _ZN2wi3hwiElPK9tree_node/4312 (wi::hwi_with_prec wi::hwi(long int, const_tree)) @0x2b5e15a81980 Type: function definition analyzed Visibility: public weak comdat comdat_group:_ZN2wi3hwiElPK9tree_node one_only References: tree_code_type/903 (read) Referring: Function wi::hwi_with_prec wi::hwi(long int, const_tree)/4312 is inline copy in tree_node* build_int_cst(tree, long int)/941 Clone of _ZN2wi3hwiElPK9tree_node/281 Availability: local Function flags: body local Called by: Calls: _Z23tree_class_check_failedPK9tree_node15tree_code_classPKciS4_/1199 _Z23tree_class_check_failedPK9tree_node15tree_code_classPKciS4_/1199 ./gt-tree.h:271:2: internal compiler error: verify_cgraph_node failed (where wi::hwi is one of my new functions). I'll try to have a look tomorrow. Thanks, Richard
