Something went wrong in Linux compatibility in the previous days.

- vm_space of Linux process is sometime set to NULL. This caused a panic
in linux_sys_brk, because I added some code that relied on vm_space.
Itojun fixed it by checking vm_space in this function. Anyway I think
vm_space should never be NULL. IMHO there is still a problem somewhere.

- kern/12455 tells the story of recent problems with WordPerfect (they
did not occur with 1.5S)

- On NetBSD/macppc, I now panic on a java test. This is also fairly
recent:
trap type 300 at 2d2f9c
Stopped in pid 190 (java) at    linux_sys_execve+0xf8:  lwz r0, r9,
0xac,
db> trace
at linux_sys_execve+188
at fork1+3dc
at linux_sys_clone+f8
at trap+35c
at ddblow+e0
at end+4fc3d6e8
at end+4fc27868
at end+4fc27184
at end+4fdc3d74

I can see there was a recent commit for kernel threads in the Linux
directory. Could these problem be related to it?

-- 
Emmanuel Dreyfus.
NetBSD, parceque je le vaux bien.
p99dreyf@criens.u-psud.fr