This is the mail archive of the
mailing list for the GCC project.
Re: Global register variables (bugs #7871, #10684, and #11413)
- From: Jim Wilson <wilson at tuliptree dot org>
- To: Simon Marlow <simonmar at microsoft dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Sat, 06 Sep 2003 21:05:01 -0700
- Subject: Re: Global register variables (bugs #7871, #10684, and #11413)
- References: <3429668D0E777A499EE74A7952C382D1BE6C5D@EUR-MSG-01.europe.corp.microsoft.com>
Simon Marlow wrote:
There are various problems with global register variables in the 3.x
series, see for example bugs #7871, #10684, and #11413. Code with
global register variables gets miscompiled.
This is still on my to do list, but I have no idea when I will get back
to it. I solved one of my problems tonight though. I just bought a
PowerMac G5, so I can now look at the PowerPC testcases more easily, and
that will make it easier for me to work on this problem. Previously I
only had x86 machines, and there was a lack of good x86 testcases.
Fortunately it doesn't seem to affect x86 (although that might be
It is easy to write testcases that fail on x86. As a practical matter
though, the x86 has so few allocatable registers that if you use global
registers you will run into other problems. Thus they are rarely used,
and hence few x86 users have noticed that they are broken.
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com