Re: [tree-ssa] mainline->branch merge status

[Jan Hubicka <hubicka at ucw dot cz>]

> 
> This has been one of the most horrendous merges I've had to go through. 
> Thanks to Andrew Haley for helping me with merging the java bits.  The
> merge is still not fully functional, though.  There are several new
> regressions (attached).
> 
> Most of them seem related to branch prediction.  Jan, maybe this is due
> to the drift between mainline and the branch in the CFG bits?

Zdenek already sent the cfg hook patches for mainline merge, I plan to
do the same with cgraph code soon.
With the predictions, I am not sure how much of them I can merge -
probably the cfg hooks for it, but that is about all.  I will take a
look.
> 
> Andrew tells me that he will deal with the new java failures after the
> merge is committed.
> 
> A heads-up for those of us following SPEC results.  The new merge brings
> in the removal of assignments to non-lvalues.  This produces a compile
> error in 176.gcc:
> 
> reorg.c: In function `find_end_label':
> reorg.c:831: error: invalid lvalue in increment
> reorg.c: In function `delete_from_delay_slot':
> reorg.c:1033: error: invalid lvalue in increment
> reorg.c: In function `make_return_insns':
> reorg.c:4157: error: invalid lvalue in increment
> reorg.c: In function `dbr_schedule':
> reorg.c:4237: error: invalid lvalue in increment
> specmake: *** [reorg.o] Error 1
> 
> This comes from the old obstack.h file.  I fixed the file, but cc1
> miscompares at runtime.  Maybe my fix is wrong, or maybe we've inherited
> a bug from mainline.  I still don't know.
> 
> The merge bootstraps on x86 and x86-64, but I don't want to commit it
> yet, as there are too many regressions for my taste.  I'd appreciate it
> if folks can take a look and help me fix them.  Jan, maybe you could
> merge the mainline and branch cgraph stuff?

I am working on it, first part is already in testing :)

Honza