Chuck Remes: cremes' OpenDarwin Blog

Go to new blog

Sun, 27 Feb 2005 19:57:14 GMT

This blog is going bye-bye. Head on over to desperatepundit.com and check out the new digs. This page should auto-redirect you within 10 seconds or so. If not, click on the link in the previous sentence.

Don't read too much into the new name. I just like the word "pundit," that's all. Th new blog will continue to focus on OpenDarwin related things with an occasional look at other things.

Thu, 03 Feb 2005 04:38:05 GMT

I plan to have a new ODTulipDriver release (the darwin-tulip project) within the next several weeks. I took a look at the freebsd tulip driver directory a few days ago and noticed that several manufacturers have released cards based on chipsets already supported by this driver. If I add the appropriate vendor-id and device-id codes, the driver will know to "match" against them and load.

Also, at some point I intend to put some serious work into supporting the 21040A chipset which drives the ethernet on the original PowerBook G3 "Kanga" model. It's a deficiency listed by the XPostFacto project as needing a fix; I would like to contribute to this project. The author, Ryan Rempel, has done a tremendous amount of great work and deserves the support.

Hiatus is over!

Wed, 02 Feb 2005 03:35:47 GMT

Sorry for the long hiatus. I'm back and ready to kick some ass.

My first project is to get opendarwin-x86 booting from HFS+ partitions. My other to-do's on the list are all driver projects. Those projects inevitably lead to lots of kernel panics and hard crashes. When testing them under OS X on a PPC box, all is well since the journaled filesystem completes its integrity checks and boots up very quickly. This is not true at all on the x86 systems; they currently boot from UFS. There is a long fsck even on small partitions (I keep them under 3 GB).

I have a vague idea how to accomplish this. I posted a note to the opendarwin hackers list and am awaiting a hint or two before jumping in with both feet.

Kernel panic resolved

Thu, 11 Nov 2004 23:43:17 GMT

The kernel panic I posted about last was being caused by a hardware problem. Though I ran the Apple Hardware Test on a 16 hour loop, it failed to detect any problem with the hardware.

I took it in to a local Apple store and waited about 10 days for CPU1 to get replaced. Everything is back to normal now.

Kernel semi-panic?

Sun, 03 Oct 2004 16:28:08 GMT

Started getting weird errors in my system.log this morning. They look like: Oct 3 10:23:32 localhost kernel: Machine check (1) - corrected - pc = 00000000909F4FD0, msr = 100000000200F030, dsisr = 42000000, dar = 00000000F02CFEF4 Oct 3 10:23:32 localhost kernel: Machine check (1) - AsyncSrc = 4000000000000000, CoreFIR = 0200000000000000 Oct 3 10:23:32 localhost kernel: Machine check (1) - L2FIR = 0000000000000000, BusFir = 0000000000000000 Oct 3 10:23:34 localhost kernel: Machine check (1) - corrected - pc = 00000000FFFF8ABC, msr = 100000000200F030, dsisr = 42000000, dar = 00000000F029FEF4 Oct 3 10:23:34 localhost kernel: Machine check (1) - AsyncSrc = 4000000000000000, CoreFIR = 0200000000000000 Oct 3 10:23:34 localhost kernel: Machine check (1) - L2FIR = 0000000000000000, BusFir = 0000000000000000 Oct 3 10:25:03 localhost kernel: Machine check (1) - corrected - pc = 0000000090003788, msr = 100000000200F030, dsisr = 40000000, dar = 0000000006B6811C Oct 3 10:25:03 localhost kernel: Machine check (1) - AsyncSrc = 4000000000000000, CoreFIR = 0200000000000000 Oct 3 10:25:03 localhost kernel: Machine check (1) - L2FIR = 0000000000000000, BusFir = 0000000000000000 I googled for this and found nothing except for "uncorrectable" machine checks as part of a panic log. I think I might be developing a hardware problem. This is on a PowerMac G5 dual 2.0 running OSX 10.3.5. Hints anyone?

VMware and OpenDarwin

Sat, 14 Aug 2004 00:17:17 GMT

Got a note from an employee at VMware stating that he successfully installed OD 7.2.1 on the latest release. It would be really helpful if other people could give this a try and report their results to me.

OD 7.2.1 and an emulated environment

Mon, 19 Jul 2004 12:52:07 GMT

Since the release of OpenDarwin 7.2.1 hit the news on slashdot, there's been an increase in interest trying to get this thing running. It looks like we have a few people trying to run it under VirtualPC or vmware.

Just a few notes about that...

VMWare doesn't properly emulate a VESA 2.0 compliant video card. OpenDarwin needs that, so it will likely kernel panic during boot. Getting the folks behind VMWare to fix their video card emulation is one way to resolve this. The second way is for somebody to dig into the OD framebuffer code and remove this restriction.

Second, the tulip driver shipping with OD 7.2.1 doesn't yet work with all DEC chipsets. I'm awaiting some data verification from some people running in this configuration, but I believe vmware (and VirtualPC) both emulate a DEC 21040A chipset. That's next on my list for support since I want to get the original "Kanga" PowerBook G3 working under OpenDarwin and it also needs that chipset support.

If you are available to beta test this under vmware, VirtualPC or on a Kanga PowerBook, send me an email.

Hey

Sat, 17 Jul 2004 21:44:51 GMT

Slashdotters, welcome to my humble blog.

Floating Point Problems Fixed

Wed, 14 Jul 2004 12:31:26 GMT

Shantonu figured out the problem. He's a very smart guy and just continues to prove that over and over.

Tested a new kernel he built and it successfully corrects the issue. Without getting into too much detail, opendarwin was not recognizing the features of the chipset. It wasn't correctly saving & restoring some registers (during a context switch?) which caused it to loop forever when a looping structure used a double/float in its boolean test. This only affected Athlon-based systems, so he added code to recognize the chipset and treat it like an Intel chip.

Felix has tagged the fix and will provide probably one more Release Candidate. We're close!

FLOP'ed

Thu, 08 Jul 2004 04:07:20 GMT

Having a very odd problem with opendarwin-x86 that appears to be a potential gcc bug. Whenever I try to to a rpm install or use darwinports to do a 'port install' on my x86 box, either gnumake (rpm) or tcl (port) will spin in userland at near 100% cpu time.

Compiling symboled versions (-g) of these programs and installing them yields a bit more information. Using gdb to break into them, they are always stuck on a line doing some kind of floating point operation. It is quite peculiar.

I'm working with another opendarwin user on irc (irc.freenode.net, #opendarwin) to try and track this down. I'll report back what I find.

As of this writing, opendarwin-x86 7.2.1 RC3 is of little use since so many programs need floating point to work!

Things I'm working on

Mon, 05 Jul 2004 17:24:37 GMT

Felix and I talked alot during WWDC about the OpenDarwin roadmap. I'm not going to talk about that here, but I will list several projects that will get my attention in the coming months.

In no particular order...

sablevm - get this freeware java environment working under opendarwin (ppc and x86) and add it to darwinports
gnustep - get all of gnustep compiled and working under opendarwin and add it to darwinports
modify the src/build/rc.cdrom script so it can properly create HFS+ partitions under x86
create a simple "bless" utility for x86 so opendarwin 7.2x and later can boot more intelligently
continue improving the tulip driver to cover more of the original DEC chips
finish port of myrinet OSX drivers to IOKit, and test them under opendarwin-x86
create a driver shim for IOKit so we can quickly port free/open/netbsd network drivers to opendarwin
create some IOKit examples (template code) for common devices like USB or FireWire
refactor tulip driver so it can use an IOPCCard nub just as easily as an IOPCIDevice

I'll create a separate page at some point to contain my open projects. That will be a better reference than a blog post, but I wanted to get this out there for comment.

Blog boo boo

Sun, 04 Jul 2004 05:23:08 GMT

I made a big mistake when configuring this blog to receive remote updates while I was away at WWDC. It prevented me from posting anything, so I apologize if you didn't get a proper WWDC news "fix" this past week.

I'll be posting a few things in the upcoming days that I think are of interest.

See ya!

Sat, 26 Jun 2004 14:09:14 GMT

Off to WWDC today. Plane leaves at 5:30PM from O'Hare and gets me to SF around 8:30PM or so.

Watch this space during the week for some darwin-related blogging goodness.

WWDC 2004

Tue, 22 Jun 2004 04:05:23 GMT

I'll be at WWDC 2004 this year and plan to blog about non-NDA details that I pick up. Since pretty much the whole show is under NDA, this means I'll blog about things I hear from other show-goers that aren't covered. I plan to discuss OpenDarwin related topics with various interested parties and try to drum up some more support for it. How? Don't know...

Looking forward to it.

it boots!

Mon, 31 May 2004 15:48:39 GMT

Felix was able to build a bootable CD for OD 7.2.1. It only boots x86 right now, but PPC should be close behind.

The "fix" was to mark the kernel as 7.2.0 instead of 7.2.1. There is apparently a version incompatibility with some pieces of IOKit. We'll have to track those down and update the xml files to reference 7.2.1.

Slouching towards OD 7.2.1

Mon, 31 May 2004 14:25:22 GMT

Working this holiday weekend with my buddy Felix to get OpenDarwin 7.2.1 in shape for release by WWDC.

I have successfully compiled and built rpm's for most of the broken projects listed in an earlier entry.

Build Correctly
AppleMacRISC2PE
AppleSCCSerial
AppleGMACEthernet
AppleOnboardAudio

Broken because of missing opensource parts
Apple16X50Serial
DSNSLPlugins

Broken for some other reason
am_utils
cddafs
curl

Felix has been burning a lot of coasters trying to get a bootable CD... just a minor problem with xnu (the kernel).

Details, details

Sun, 23 May 2004 23:01:59 GMT

Felix is putting a plan together so we can try and get OpenDarwin 7.2.1 out the door prior to WWDC this year. I'd like to bring a bunch of copies and hand them out at the conference.

If necessar, I'll take a day off from work to get any last minute items finished.

Fixing broken builds standing in the way of OD 7.2.1

Thu, 06 May 2004 00:35:03 GMT

Tonight I'm working on getting several projects to compile cleanly for the OD 7.2.1 release. The projects in question are...

Apple16X50Serial
AppleMacRISC2PE
AppleSCCSerial
AppleGMACEthernet
AppleOnboardAudio

DSNSLPlugins
am_utils
cddafs
curl
gcc_legacy
gcc_os
smb.log

I'm primarily focused on getting the Apple* projects compiling. I hope someone else is working on the remaining projects.

sourceforge account updated

Thu, 06 May 2004 00:31:39 GMT

Updated the account at sourceforge.net so the home page redirects to my opendarwin blog here. I should have done that a long time ago, but forgot. :)

Makefiles for Apple drivers

Thu, 22 Apr 2004 03:40:14 GMT

Building OpenDarwin is a pretty difficult chore. One of the obstacles is the reliance on xcodebuild and other Apple-proprietary tools. These are usually only available on darwin-ppc without any darwin-x86 analogue.

The AppleGMACEthernet project was broken in the darwin 7.2 release. The tarball made available by Apple wouldn't compile cleanly out of the box, but the recent 7.3 release tarball did. I backported a couple of lines from the header file that caused the older release to choke and got a clean compile.

Using this as my test-bed, I adapted the makefile environment from the tulip driver. It took about 15 minutes to get a clean compile by typing 'make' at the command line. I had to create an AppleGMACEthernet_info.c file that acts as the glue between IOKit and the kernel. Likewise, I had to copy in some of the bundle resources such as the Info.plist and InfoStrings.plist files. I put them in the base directory with the source files and "taught" the makefile how to build the whole mess into a kext. Works okay and looks pretty clean too.

I need to make a few more minor changes to get this to compile using the OpenDarwin CoreOS Makefile environment and RPMs.

After that, I'll tackle a few of the other broken projects. This puts us a bit closer to getting a genuine OpenDarwin 7.2.1 release out the door. And not a moment too soon... the rumor mills tell us OS X 10.3.4 (darwin 7.4) is being seeded. Yikes!

ODTulipDriver RPM

Tue, 13 Apr 2004 03:50:31 GMT

Just turned in the rpm package to Felix for the OD 7.2.1 release. Hopefully he'll have some time in the next few weeks to wrap up this release and get it out.

Here I am.

Fri, 02 Apr 2004 04:09:04 GMT

Still alive. Sorry for being quiet lately, but work has been particularly onerous as of late. Hope to do some updates soon.

driver crashes fixed!

Sun, 14 Mar 2004 15:00:42 GMT

After much hair-pulling and teeth-gnashing, I narrowed down the cause of the kernel panics on OpenDarwin 6.6.2. If the driver was compiled with optimization flag -O3 then it would panic every time. If I changed the optimization to -O0, no crashes.

Copying over a set of x86 binaries that had been compiled on darwin 7.x and running them on OD 6.x also worked fine. This points squarely at a bug in gcc that shipped with OD 6.x. It must be reordering some instructions that result in a null pointer being evaluated (or some other similar problem).

I'll see if I can't narrow it down to a single line of offending code and fix it programatically. If that doesn't work, then I'll just add a note to the README specifying this problem and its workaround.

driver tagged TLP-1-0-0 crashes OD 6.6.2

Sun, 14 Mar 2004 04:34:24 GMT

I have confirmed on my own that the darwin_tulip driver tagged TLP-1-0-0 (essentially the first release quality version) causes the OpenDarwin 6.6.2 kernel to panic. All testing of the driver was performed under darwin 7.01 where this problem never manifested itself.

Obviously I will do everything I can to track down this problem and fix it. I'll post an update tomorrow.

Sorry to everyone running the older OD release who got bitten by this.

developer release for darwin_tulip

Tue, 09 Mar 2004 05:18:45 GMT

Committed a bunch of code changes to CVS tonight. They primarily had to do with SYM/SIA media selection support for the DEC 21143 and PNIC2 cards.

If anyone is interested in checking this out, the cvs tag TLP-1-0-1a corresponds to this release. It outputs a lot of messages during media selection (e.g. autonegotiation) but then gets pretty quiet.

Looking for feedback and encouragement...

I'm creeping ever closer to supporting the DEC 21040A that shipped in the original 'Kanga' PowerBook G3.