Pull latest 2.6.25-rcN git tree to refresh my physical PCI slot patches, hack-hack-hack, go to test them on my test machine, and ruh roh, something’s hanging.
Try a clean tree without any of my patches and still get the hang. Hmm…
Not sure when this bug crept in, but for some stupid reason, I decide to start with 2.6.23. git-bisect happily tells me that I only have 11,000 more commits to go. Yikes.
A few hours later, we have a smoking gun: pciehp regression hang.
Oh, and btw, death be to those who introduce non-bisectable commits. Even if your code is 100% bug-free, you still make life difficult for those trying to debug others’ code. Bleah!
Time to relax by reading the Baroque Cycle.