[tree-ssa] Bootstrap fails with enourmous memory usage

[Diego Novillo]

On Sun, 24 Nov 2002, Andreas Jaeger wrote:

>   PID USER     PRI  NI  SIZE SWAP SHARE STAT %CPU %MEM   TIME COMMAND
> 12244 aj        25   0 1591M 1872   884 R    97.8 78.7  43:21 stage1/cc1 -quiet
> 
Interesting.  If it fails with the RTL checker enabled
(misc, tree, gc and rtlflag are defaults), then it may be due to
the excessive amount of RTL that is generated by the
gimplification process that we still don't remove with the tree
optimizers.  I don't think it fails on b-i-b, right?

If it doesn't fail on b-i-b and you can get it to pass without
'rtl' in the check string, then all I can suggest is to not use
it for now.  Although, making sure that it's due to the size of
the RTL structures would be a good thing.


Diego.