Muli Ben-Yehuda's journal

Henceforth, 2.6.19-rc1 shall be known as the Kernel of Pain.

A few days ago Gteko, an Israeli company, was bought by Microsoft. I had a dim recollection of having heard of them before, but could not recall when or why. Now I know. A few months ago, Gteko invested in Gilad’s company Codefidence, an “Open Source Development and Support” company. So right now Codefidence’s biggest investor is Microsoft – which is not something many open source companies can boast of…

If you’re building a Linux initramfs into your kernel image via CONFIG_INITRAMFS_SOURCE=”/path/to/some/directory” and make dies with “target pattern contains no `%'”, check if /path/to/some/directory has a file with ‘:’ in the name. If it does, get rid of it (assuming it’s not needed…). What’s happening is that we use the list of files in the initramfs as dependencies to make to know when to rebuild the initramfs, and the ‘:’ in the middle of the dependency list freaks make out.

We now return you to our regularly scheduled torture of getting aic94xx to load its firmware from the initramfs.

Its been a while since I’ve updated this thing. I keep meaning to, and
then something more urgent comes up, and poof! another month has
passed.

Looking back over the last few entries, the major events have been
covered, albeit rather succinctly: we got married, war broke out and
we bought a
house. Just the facts, ma’am!

There’s something gigantic,
huge, life-changing going on, but I’m not at liberty to discuss it
yet. Y’all will know all about it soon enough.

On the work front I’m still pleasantly hacking away on the Calgary
IOMMU support in Linux and Xen. The Linux kernel status is pretty
good: the code is merged and in good condition, and now that
2.6.18 has opened a bunch of outstanding patches should make their way
in from Andi Kleen’s x86-64 tree. There are two issues I’m aware of:
kernel.org
bugzilla #7180 and RH
bugzilla #203971, and it looks like Jon figured out the RH
one. There’s also an outstanding locking improvement (don’t lock the
bitmap when building an entire sg list, lock and unlock it for each
entry to increase parallelism – suggested by jejb) but I need to
finish thinking through the ramifications.

On the Xen front, I finally finished the monster merge from hell with
the platform ops changes (dom0 ops have been broken up into dom0 ops,
platform ops and domctrl ops) and am now getting back to moving all
Calgary hardware frobbing bits currently done by dom0 into Xen. The
code is available on xenbits.xensource.com as
usual.

More later.

We moved into the new place on Thursday. It is starting to feel like being home again.

Greg Kroah-Hartman‘s OLS 2006 Keynote “Myths, Lies, and Truths about the Linux kernel” is definitely worth a read.

I was supposed to be at OLS now, but instead I stayed in Haifa to play “dodge-the-rocket.” Such is life in the middle east.

To all of you OLS-goers, enjoy the conference, and don’t forget to wildly cheer Jon when he gives our Using IOMMUs for Virtualization talk during Wednesday’s opening session!

It has come to my attention that I’m annoyed by anonymous comments. Anonymous commenters, you don’t have to be logged in or have a livejournal user-id, but please sign your comments with your name. Anonymous comments will be deleted at my sole discretion.

Rockets are falling in Haifa. It certainly gives a new meaning to “debugging under fire!”

The dollar is sky-rocketing due to the escalation in the north and the stock market is plummeting. Normally I would be in a quivering ball of panic right now, but we bought the house on Sunday, and our money is safely tucked away. Talk about timing!

Pictures of the house (and the wedding) will be forthcoming … eventually.

We now return you back to your regularly scheduled Calgary debugging.