avr-gcc-list
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[avr-gcc-list] RE: tool chain buglist


From: Weddington, Eric
Subject: [avr-gcc-list] RE: tool chain buglist
Date: Mon, 18 Feb 2008 13:59:25 -0700

 

> -----Original Message-----
> From: Andrew Hutchinson [mailto:address@hidden 
> Sent: Sunday, February 17, 2008 6:29 AM
> To: Weddington, Eric
> Cc: Paulo Marques; address@hidden
> Subject: tool chain buglist
> 
> Eric,
> 
> can you get toolchain list updated?

I can try, but I'm under some deadlines...
 
> There are various patches posted againt gcc problems but are not yet 
> listed. Also the following are  missing:
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34916

Actually this was already in the list.
 
> This is nasty bug that lies inside gcc. it's status needs to 
> be elevated.

That gets complicated.

The GCC project has a list of ports that have "primary" status, or
"secondary" status as part of its release criteria per version. For 4.3
it is this:
<http://gcc.gnu.org/gcc-4.3/criteria.html>

The AVR port is neither primary, nor secondary. What this means is that
any regressions for the AVR will automatically be marked down to P5 when
it comes time to a release. It also means that it is up to interested
people to make sure that the port continues to work. In contrast, a port
that has at least secondary status *must* at least build/bootstrap in
order for GCC to release, i.e. it can block a release. This is why those
ports that are not secondary will not block a release, hence those
regressions get marked down to P5.

I have a plan to get the AVR port moved up to secondary status. Last
year I went to the GCC Summit and talked to the Release Manager and
other people on the GCC project about the AVR port. Many people were
very surprised to hear how popular it is. In order for the AVR port to
make secondary status these things need to happen:
- Regular testing of the GCC Regression Test Suite
- Regular reporting of test results to the gcc-testsuite mailing list
- Test failures fixed so a "substantial majority of the tests pass" per
the criteria.
- Regular bug fixing (which I think is happening now) in a timely manner
- The above things need to be done to demonstrate the "quality" of the
port.
- Then the GCC Steering Committee can be petitioned to have the AVR port
be listed as a secondary platform. But no guarantees, of course.

This is why it has been so critical to have the avrtest program working
so the GCC Regression Test Suite can be executed. Eventually it will be
good to have a benchmark test suite running so we can keep up the
quality of the toolchain.

In talking with the Release Manager (Mark Mitchell) he suggested that I
petition the Steering Committee for the 4.4 release. Since it's coming
up on a 4.3 release, then I'll be petitioning the Steering Committee
soon. There is another GCC Summit coming up in June, which I'll be
attending. Depending on how everything's going, I may wait until then so
I can do it face-to-face to drive the point home.

> 
> Also:
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32871
> 
> I have not confirmed this one.

Ok, that one has been added. Thanks for pointing that one out.

Eric Weddington




reply via email to

[Prev in Thread] Current Thread [Next in Thread]