On Mon, Nov 13, 2017 at 1:02 PM, Marek Polacek <pola...@redhat.com> wrote: > On Mon, Nov 13, 2017 at 11:47:41AM +0100, Jakub Jelinek wrote: >> On Mon, Nov 13, 2017 at 11:44:38AM +0100, Richard Biener wrote: >> > > Unfortunately, it's not what I see. I ran two bootstraps, with and >> > > without the >> > > patch. Then I stripped all the .o files and ran cmp on them, but many >> > > of them >> > > differ. objdump -dr reveals why -- I see changes like >> > > >> > > mov $0xe02,%ecx >> > > vs >> > > mov $0xdf6,%ecx >> > > >> > > That's probably some sizeof changes? Any other ideas? :/ Somehow ignore >> > > these and only look if there are new pushes or similar? >> > >> > sizeof an empty type shouldn't change, no? Possibly the new target >> > hook "shifts" target hook offsets. Maybe you can arrange the new one >> > to be last... (just for testing). OTOH for example in target library >> > objects all the changes of this kind should have no effect. So, do >> > objects in target libraries differ similarly? >> >> If the changes are because of the patch changing stuff in bootstrapped >> compiler, the easiest way is to revert the patch and rebuild stage3 inside >> of the stage3 directory and then compare what you get with the unpatched >> bootstrapped compiler (or, if you first bootstrap patched tree and then >> unpatched, similarly, re-apply the patch and rebuild stage3 of the >> originally vanilla tree). > > In the end I did two bootstraps with the patch, but modifed one of them > to always return false for ix86_is_empty_record. Then I compared all the > *.o in both dirs. The result is attached. Then I looked at DW_AT_producer > for all these .o that differ; all of them are C++. Is this enough to > clear our concerns?
Hmm, a bunch of these are right at the beginning, bytes 41 and 65, in the header. Did you build them in the different trunk/trunk2 directories? I think Jakub was suggesting building them in the same directory. > And I also ran a bootstrap with --enable-cxx-flags=-Wabi=11, and didn't > see any warnings. If there's a codegen change, there ought to be a warning to go along with it. Jason
