I have a binary MP kernel for amd64 that I am willing to share with people who want to test and make sure that we all test the same thing.
Contact me off list. On Wed, Feb 25, 2009 at 10:10:34PM -0600, Marco Peereboom wrote: > I can comment from experience that the apmd changes only made it happen > quicker; it happened nonetheless just less frequent. I haven't tried D > yet on the laptop but I will do that right now. > > On Wed, Feb 25, 2009 at 09:48:34PM -0500, Dan Harnett wrote: > > On Wed, Feb 25, 2009 at 09:07:52PM -0500, Ted Unangst wrote: > > > Ok, there are several code points here, and I'm having trouble keeping > > > track of them all and who's machine worked how when. > > > > > > First, there's A. This is the code that shipped in 4.4. > > > Then we have B. This was the code that went in ~2 months ago. > > > Then there was C, part of B backed out, but much like A, (call it A') > > > for a few days. > > > Then we went back to B for a brief time. > > > Now we are at D, all of B backed out. I'll call this A''. > > > > > > So there are basically 2 varieties of the code, A and B. > > > > > > A and A'' should be identical, except for other unrelated changes in > > > the kernel. The first question is, did anyone experience these > > > problems with 4.4? If you are having trouble, is your system stable > > > with B? > > > > I'm not sure that this is a new bug recently introduced or if it has > > always been there, but never triggered. I haven't had the X61s long > > enough to comment on it's stability. The T61 I've had for close to a > > year now and it appeared to be completely stable. It's been following > > amd64 snapshots just shy of 4.3-beta. Even with the acpicpu.c and est.c > > changes that were in 4.3-current, backed out for 4.4, reintroduced in > > 4.4-current, and now backed out once again, it appeared stable. The > > changes to apmd appear to trigger the bug. Now, with 1.50 and 1.51 of > > apmd.c reverted, I can no longer reproduce the bug. With my recent luck > > (or misfortune), I'd bet this is an old bug that has been lingering for > > a while. I also haven't been able to reproduce it on i386.
