Muli Ben-Yehuda's journal

Usenix papers – submitted! next up – IOMMU paper for OLS.

Off to Austin for the Xen summit in 5 hours. See you on the other side!

Be careful what you wish for, you might just get it 🙂

Date: 		 Thu, 12 Jan 2006 00:08:15 -0800
From: Linux Kernel Mailing List [linux-kernel(at)vger.kernel.org]
To: git-commits-head(at)vger.kernel.org
Subject: [PATCH] x86_64: Use function pointers to call DMA mapping functions

tree b6b3b55318336adf769bf57141a01a9defbbb202
parent 8a6fdd3e912e0ce6f723431d66baf704bf8a1d26
author Muli Ben-Yehuda [mulix(at)mulix.org] Wed, 11 Jan 2006 22:44:42 +0100
committer Linus Torvalds [torvalds(at)g5.osdl.org] Thu, 12 Jan 2006 11:04:55 -0800

[PATCH] x86_64: Use function pointers to call DMA mapping functions

After many trials and tribulations, dma-ops has made it to 2.6.15-mm3.
Will it make it to 2.6.16? the game of patches continues!

Date:   Wed, 11 Jan 2006 04:21:35 -0800
From: Andrew Morton [akpm(at)osdl.org]
X-Mailer: Sylpheed version 1.0.4 (GTK+ 1.2.10; i386-redhat-linux-gnu)
To: linux-kernel@vger.kernel.org
Subject: 2.6.15-mm3

[major snippage]
+x86_64-dma-ops.patch
[snip]
 x86_64 tree updates

Following xavishayx‘s lead, here’s what I learned today:

When working on a patch to mercurial to show child changeset links in the web interface, make sure you’re testing it on a changeset that actually has child changesets. D’oh!

One of the reasons I write this journal is that sometimes the days just seem to slip by. Every day I run around, all busy-beaver-like, filled with a sense of Getting Things Done. Then a week goes by, and I have no idea what it was that I Got Done.

Weekend – lots of family obligations. A memorial service for my grandmother’s brother Pascu, and lunch at Orna’s parents with my parents.

Sunday – theoretically I was on vacation, but I did head into work for a brainstorming session on a new community project with my old alma mater. More details when we go public with it. Also had a short conversation with Andi Kleen with regards to the dmaops patch, which he substantially revised. I like his version much better and will continue future work on top of it. A copy of his revised patch is available at dma-mapping-ops-K7.

Monday – into the office again (vacation? what’s that?) for a bunch of meetings, albeit of the interesting kind, where we sit around a table, nursing our cups of coffee and argue about the right way to build the system. For hours at a time. This is exhausting, but also satisfying as you see the system take shape in your mind’s eye. The only thing I don’t like about this process is that sometimes the design is not fueled by an actual concrete prototype implementation. This can’t be helped when you’re designing hardware (which we are, to an extent), but for software, it’s inexcusable.

Lunch is ready, more later.

The trip to Shlomi Dolev’s group at BGU yesterday was much more enjoyable than I anticipated. We had a good discussion about Xen, IOMMUs, self stabilizing systems, IBM HRL, the Linux kernel and an assortment of other subjects of mutual interest and everyone was very nice. It was also more personal than I expected – I ended up telling them about my less than illustrious academic career…. They say a little humility is good for the soul.

A partial list of people I met and their research interests is available under the ‘self-stabilizing’ tag on del.icio.us. I hope we can collaborate in the future – a self-stabilizing hypervisor may be just the thing that’s needed to connect the theory and the practice in this area.

Let no one say that I’m not milking the Xen-IOMMU talk for all it’s worth… I’m giving it today yet again at BGU’s CS department.

Monday was hectic, Nick Bowen visited HRL and I attended had a bunch of interesting discussions with him. During the first one in the morning he asked if I was running Xen on my laptop; when I admitted that I wasn’t at the moment, he asked if I will be by the end of the year and said that he remembers these things. By the time of our second meeting after lunch, I was running Xen and showed it to him. He was suitably impressed.

In the evening I gave the Xen-IOMMU talk to Haifux. I was pretty beaten, leading to a somewhat lack-luster, albeit well-received, performance. Hopefully today will go better.

Yesterday was my first real vacation today. It was almost perfect, except for one pesky flame-war on hamakor-discuss in the morning. I really should develop an (even) thicker skin.

Today I’m going to travel to Be’er-Sheva and read some more of Essentials of Programming Languages.

By the way, this blog is now being channeled on Planet Offtopic2.

Also, Free OS Zoo is neat and useful.

I stayed up late last night rebasing the dma-mapping-ops patch against mainline and ak’s quilt patchset, then woke up late today. Spent a few hours dealing with urgent work matters (so much for vacationing…) and then a few more hours finishing the dma mapping-ops-patch. Tested and released dma-mapping-ops-K6 which ak will hopefully merge into mainline during the 2.6.16 merge cycle.

In the evening we were off to Tel Aviv for Hamakor‘s general assembly. It wasn’t particularly fun, but it had to be done. I’ll write more about the assembly and my thoughts on the state of Hamakor in general later.

Tomorrow I’m taking a vacation from vacationing and heading to the office in the morning for a bunch of VIP presentations. In the evening I’m giving the Xen-IOMMU talk at Haifux, and then we’re celebrating my parents wedding anniversary. So much for a quiet peaceful vacation 🙂