Kasper Dupont wrote: > > In 2.2.21-pre1 some vm86 bugs have been attempted fixed. > The bug did become smaller, but unfortunately didn't get > completely fixed. > > [snip] I will be > back with a patch for 2.4.18 once I have resolved > the mentioned problems. > > Any comments? Here is the patch, which I has also made available by http: http://www.daimi.au.dk/~kasperd/linux_kernel/vm86.2.4.18.patch Somebody suggested me that Alan might be interesting in this particular path. I have done some testing, of the new patch, and it seems to work. Though userspace programs using vm86 might need to be changed slightly to handle the situation. Anyway that situation should be better than both the orignial Oops bug, and the incorrect IP register value caused by the first patch. Now the userspace code is no longer left in an imposible to handle situation. About the other bug I mentioned here is a hexdump of a small DOS program demonstrating the difference. This prints 0 on a real PC but 2 on an emulator using vm86, my patch also fixes this problem. [kasperd:pts/1:~] od -h