Subject: 2.0/current toolchain bootstrap problem on ppc
To: None <firstname.lastname@example.org, email@example.com>
From: John Klos <firstname.lastname@example.org>
Date: 01/09/2005 12:37:16
While using either the 2.0 or the current source trees to bootstrap the
toolchain on macppc, I'm seeing some problems with powerpc--netbsd-lint.
In summary, when I'm compiling on an older machine (300 MHz G3 with 512k
of L2 cache), I see no problems.
When I compile on a 1.3 GHz 7455 with 256k of CPU speed L2 cache and 2
megs of L3, I occasionally see powerpc--netbsd-lint get a signal 11. If I
try again, it happens in another place.
When I use a 1.4 GHz 7457 with 512k of CPU speed L2 and 2 megs of L3, I
see signal 11s quite often. If I keep restarting (with -u, of course), it
will eventually get the whole way through building the source tree, but it
can take up to a dozen times.
The systems used for testing have been used to serve and compile for many
months, so problems with the CPUs or the systems themselves seems very
So - assuming that powerpc--netbsd-lint has some sort of issue with dirty
cache, what's the best way to debug this?
If our gods and our loves are nothing but scientific phenomena,
then it must be said that our love is scientfic as well.
- L'Eve Future