Re: GCC 3.4.1 Status Report

[Jan Hubicka <jh at suse dot cz>]

> Jan Hubicka wrote:
> 
> >>The observant among you have noticed that June 15th came and went 
> >>without a release.
> >>
> >>However, I intend to create a prerelease in the very near future -- 
> >>perhaps as early as tomorrow.  After that point, the branch will go 
> >>slushy; I'll need to approve all patches.
> >>
> >>There are still 56 bugs targeted at 3.4.1; I plan to postpone nearly all 
> >>to a later release.
> >>
> >>There are a few PRs that do look critical to me, including PR 14782, 
> >>which refers to an unaligned access on PA, and PR 16026 which is a 
> >>problem in assign_parms.
> >>
> >>There are are also a large number of unit-at-a-time and/or 
> >>no-unit-at-a-time problems and discrepancies.  Jan, you've still not 
> >>made any notations on those PRs about whether or not you think they can 
> >>be fixed.  Gaby, you've not commented on 13560; I'd hoped you would try 
> >>   
> >>
> >
> >I am just going trought those bugreports.  There are few marked as
> >unit-at-a-time (or non-unit-at-a-time) but they are not closely related:
> >
> >target/14798 seems to be SH backend bug with patch being disucssed there...
> >
> >C++/14865  THis looks like C++ frontend bug in the legacy way of doing
> >reachability using SYMBOL_REFERENCED.  I am not quite sure what to do
> >here. 
> >
> >middle-end/15114  I used to have patch for this, but Richard had better
> >ideas to fix, but I am unsure what is the status right now...
> >
> >I am testing fixes for remaining two problems and will try into those 3
> >too later...
> > 
> >
> What is the status now?

I've commited patch for pr14950 and I am just testing patch for the
TImode problems on x86-64.  15114 has been moved to 3.4.2, 13000 to
3.5.0 (assuming that we manage to get cfg-before-inlining patch in
during stage1) and I've proposed to mark 15345 as wontfix that
apparently happent.  14798 is apparently fixed now.

There remains the 14865 I've spent some time on, but I don't quite
understand why it broke in 3.3 (ie before unit-at-a-time went in)
Perhaps you know better than I do?

There is new bug 16035 that appears to be attributed to Jason's patch.

I am leaving to Hamburg today for a week but I am taking 3.4 tree with
me, so if there are remaining stoppers I should look at, I will do so.
Otherwise I plan to iterate over the open mainline bugreports during the
stay there.

Sorry for recent delays, hope to be back in usual schedule at 30th in
the worst case.
Honza
> 
> -- 
> Mark Mitchell
> CodeSourcery, LLC
> (916) 791-8304
> mark@codesourcery.com