April 21, 2009 Archives

Tue Apr 21 20:17:33 CEST 2009

Where it's at

I know, I know, I should post more often. I have no excuse and as such will cut to the chase: where we are alpha-wise at the moment.

Xorg 1.5+

The patch from Ivan has finally made it into the mainline kernel. The earliest vanilla version having it is 2.6.30-rc2. I'm currently running that kernel and xorg-1.5 on my dev/testing machine and it's looking good so far. There still is a problem with the glint driver, but I suspect that is independent of the PCI access patches. The "nv" driver works perfectly.

Originally, I had hoped I could get them included in .29 or 29.N, but it was not to be. In the course of working for that I asked myself how much pestering a kernel developer is willing to put up with (and what a sign of annoyance would look like with Ivan). I would have loved to see more/quicker progress, but when people stop reacting to your mails at all, you start to second-guess yourself.

AGP on the UP1500

Matt has done a superb job of finding out email addresses and other contact info regarding the PALCode sources. For those don't know: the PALCode is a sort-of BIOS the Alpha architecture uses. As such it is one of the key reasons why the AGP port on the UP1500 is little more than a glorified PCI slot currently: AGP 4x does not work at all because there are bugs in the PALCode. Since the source to that was never published and binary-patching such a crucial piece of software (without knowing what you're doing) is an exceedingly bad idea, getting the PALCode for the UP1500 is just about the only way this will ever get fixed. Unfortunately, there are several different owners of the IP in the PALCode source. We'll keep you posted on the progress regarding that.

In a related note, Matt keeps trying to get Hewlett-DEC-Compaq and subsidiaries to release any piece of source code from the CCC (C compiler) and related libraries (especially the math library). No luck yet, but Matt can be very persistent, I have learned. ;)

Moreover, Matt has created a patch page on the AlphaLinux wiki. Right now, there are patches for the kernel futex implementation (what caused the tst-robust1 freezes in the glibc test suite), a patch for several math functions in the glibc (have been broken since forever) and a patch for ed which fixes a certain set of compile-time optimizations.

General package status and workforce/load

Armin76 and I are able to stem the package work tide (testing, keywording, stabilizing) as it is now. But if one of us can't commit the necessary time (several hours a week, I'd guess), things will start to pile up. Fortunately, Matt has sent in his archtester's quiz and as soon as I find the time, I will give it a thorough look and work with him to get him up to speed. Even if he hasn't commit rights, not having to watch stuff compile all the time will bring down our workload.

Security bugs

Nothing terribly newsworthy, I'm trying to get to them quickly. If you see a slow sec bug that we might have overlooked, feel free to bug me by mail or us by IRC (Freenode/#gentoo-alpha).

Hardware and Performance

I recently benchmarked the UP1500s IDE controller (ALi M5229) with a Maxtor 120G disk against a SymBIOS Logic 53C895 with a 33G Seagate Barracuda. Even through the SCSI disk is quite a ways older than the IDE disk, it outperformed the IDE disk in all disciplines (except seek times, but the IDE disk won by a hair there). So I went and looked up pricing for U2W/LVD160/LVD320 disks on the major hardware retailer websites. The prices are simply insane. 73GB cost about 160 Euros, bigger disks have a better price/GB ratio, but are still too expensive. 300GB would set me back 320 Euros. Not going to happen. I have a few other ideas up my sleeve, but I will need to work on them a bit more. Oh, and on ebay.de you only find small U2W disks - the big ones are SCA only.


Posted by klausman | Permanent Link | Categories: Software, Community
Comment by mail to blog@ this domain. Why?