------- Comment #29 from rguenth at gcc dot gnu dot org 2006-08-28 08:42 ------- I completely agree with Dan's analysis. Note that for the patch in comment #20 we can hoist the loop adding all addressable vars to the call clobbering list out of the loop iterating over all analyzed pointers by only remembering if one had pt_anything set.
My preferred solution would be to develop the correct fix for 4.3 and eventually allow backporting that to the release branch(es) and not hold the release of 4.2 for this bug. In fact, the usual try of adding "-fno-strict-aliasing" if one faces a subtle bug will fix it, so it's not harder to detect than a strict aliasing violation by the user. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28778
