Updated: Never — Philip's Blog

00:07 BST

Our developer summit here in Cambridge started off rather well today. Robert decided that a brisk half hour walk from King's to the computer lab was a good start of the day. I happen to agree.

During the "trying to get organized" phase of the early morning, I sent out a core report which spawned a little discussion and committed a trivial patch adding a hw.ata.ata_dma_check_80pin loader tunable to work around bugs in systems which don't set the "pin 34 is grounded" bit in the ATA config register to indicate that speeds faster than UDMA mode 2 (UDMA33) are okay.

This makes the Soekris net5501 SATA experience a much happier one. I'm told a number of laptops and even blade servers have issues like this too.

Just before committing the patch, I noticed that marck had submitted a virtually identical one as a PR about a year ago. I need to trawl the PR database more often...

In the afternoon, attended a weekly meeting of the "security group". Markus made an interesting case for providers to "sell distributed attacks", which reminded me of why I've always looked upon academia with a certain level of caution. "Warning, the real world may be very different from the one being discussed in this room". Nevertheless, it was good food for thought. Stuff to keep in mind when dealing with ISPs...

Dinner at a very nice Thai place. Dessert was something green of which I didn't quite catch the name. Tasty though. Unfortunately, the King's bar was closed when we returned from dinner, so a subgroup headed off to The Eagle.

England has not yet realized that World War I has been over for a good while, and pubs still close at 23:00 to ensure that factory workers building artillery are in bed in time. Grr...

I liked the "please keep this bar clear for returning glasses" sign. I'm sure the girls at Delirium would love one of those for the next FOSDEM beer event lest they get showered by returning beer glasses again.

Tomorrow, I'll be looking at unbreaking some of the damage my bridge patch from a couple of months ago caused. There seem to be some subtle failure modes in ARP and PFIL(9) which need to be addressed. I've received a patch for the ARP issue, but PFIL needs a bit more thought.

I love developer summits. Empties my laptop of patches!