The 5680/8890 Gentoo Thread - Page 8

Nandro: I'm in the process of doing a gentoo stage 1 install on my raided 8890. I'm not sure if it'll work yet, but I'll let you know. I'm currently progressing from stage 1 to stage 2, and the machine is happily compiling away.

I first installed Redhat using the driver from promise, and have been installing gentoo from within my redhat environment. That's the only way I could figure out how to have access to my array, since none of the live-cd images from gentoo or knoppix were able to see the array.

My plan is to merge the 'partial linux driver source' from the promise site into the kernel source of the gentoo install (following the docs from the driver readme) at the kernel config stage. From the looks of things, the 'partial driver source' is the full driver with the proprietary portion precompiled into a ftlib.o file, so I think it's the full driver with partial source.

If all goes well, I should end up with the raid (fasttrak.o) driver compiled into the kernel and all will be well.

I'll post in this thread as the install progresses.

Gah!
The promise driver won't compile. Followed their docs to the letter, but when I try to compile the driver I get an error 1.

It's really strange, since the offending lines look ok to me, but io_request_lock doesn't seem to be declared anywhere!

The offending lines are the spin_lock_irqsave and spin_lock_irqrestore.
Any ideas?

[edit] looked farther up in the compile output...looks like it can't find some header files. The headers it can't fine, however, are included from system headers that it DID find. Not sure why/how a system header would be trying to include non-existent headers tho. I'm going to try compiling to kernel without the fasttrak driver to see if I get similar errors...
Still working on it....

[edit 2] Ok...after some creative file copying, I've managed to get all the needed include files include properly. Still getting the error, except now it's the only error left.
If I get this fixed, and it works, I'll upload the modified makefile, and some docs on fixed the header problems.

[Edit 3] Too tired to continue tonight. This is seriously frustrating me. I looked at some other modules (aha 7xxx) and it uses exactly the same io_request_lock spinlocks, without any visible declaration of 'io_request_lock' and it compiles fine. I'll try to figure this out tomorrow, but for now, I need sleep. If anyone knows anything about spinlocks, I could sure use some advice...

Sent an email to Promise last night before I fell asleep asking if there were plans to release a kernel 2.6 driver, and got an unexpected response...
so it looks like I'll just have to be patient and wait for the drivers. Good news, I'd say, especially the open-source aspect. Maybe the binary drivers cause as much headache for Promise as they do for us.

Anyone heard an update from Promise yet? Thinking of poking at Gentoo soon, but want to get 2nd drive first & Raid 0. I cannot find the specific model used by 8890 on their site, only base product lines.

Also, I've heard about FireWire DV camcorders used as Netcams in Linux. Anyone try that yet?

Well, the update from promise is "sometime this summer". In computer terms, that's 2-4 years... I hope they're better than Valve when it comes to release dates... The alternative is you can blow away your current raid and use Linux software raid, but that's only for those who don't need to access windows stuff from linux...

Morning all,

As I mentioned further down this discussion group, I tried Solaris x86 on my 8890 and failed miserably. At that point I began to seriously look into Gentoo. The disk is ready to accept it, but I'm still doing the pre-install reading.

My notes so far:

First off, Dirtboy's intro link early in this thread changed due to talknotebooks.com no longer references here in DNS. The revised link (mmarkin, you may want to look over the other links in the thread for similar changes):

http://notebookforums.com/showthread...1369#post31369

Second, I don't think I've done this much reading on Linux before. It's not an eye-opener per se, since this process reminds me a lot like the SunOS 4.x install - slow, tedious, and detailed. Great info though, learning much just on this.

The initial plan for me is to get a recent LiveCD, get networking running (the newer ones should be 2.6.x kernel), and go from stage1. This should hopefully make like easier based on what I've read on this thread.

The disk is prepared already, I was able to use a 'system recovery' CD that's based on Gentoo /w the working NTFS code and resize the disk. So far so good.

The trick now is finding some time long enough to start this. Hopefully this week. I'm thinking of creating an updated summary for my own notes (common practice at work, and if this turns out OK on the notebook I'll do the same to the desktop).

Also, I've given up on following RAID for now, getting a 2nd 60@7200 is a bit much for me at the moment too. I'll revisit my search in Sept or so.

I'll let y'all know how things turn out,
--Tim

Hi all, an update...

Stage 1 Bootstrap succeeded, sort of: Had IDE/DMA errors on console while trying to unpack and setup gcc, though it finished anyway.

So I went forward on emerge system before getting some sleep (I've also been working on Solaris upgrade for major server at work, so I've been busy since early Sat morning ).

Well, woke up to find notebook hung. Last message says it was trying to download perl from one of the mirrors (which I'll probably remove from mirror list now). No network activity from the notebook, as per the router.

I should make note that I was doing the work in an ssh session from my desktop, so I could reference web pages etc. from Mozilla during the work. Both the ssh session AND the notebook's console were hung.

Should I restart back in Stage 1, with DMA turned off, or try emerge system again?

I reember seeing hdparm settings somewhere (probably here), I'll look for them later.

--Tim

I know one thing, if you're running RAID, you're out of luck...

No, its not RAID, not willing to spend any more money on that right now. I do have a second disk, but it's 40 Gig @ 5400 RPM with 16 Meg cache; It's used as Ghost, VCD, and AV storage.

As far as options, yes. I booted on the smp kernel just fine as-is, and set a basic USE group (in particular, 'dedicated alsa motif cdr dvd'), the mirror list (mirrorselect), and -pipe on the compile settings. Otherwise used the basic setup. I'm using the 2004.1 mimimal LiveCD (which found the USB mouse and network just fine, as well as the screen at higher text res then 80x24). I burned a Universal LiveCD as well, have not needed it yet.

My concern is my 60 Gig 7200 RPM disk (which is where this was going to reside). There may be a DMA issue with that so that's why I'm thinking of that off next boot. Another thing I'm doing right now is running it on memtest boot for a bit to see what pops up. Worst case is I'll just start with a Stage 2 or 3 and try again that way, but that kind of defeats the purpose don't it?

One interesting note: Boot to Windows, kick off MaxFan, and reboot into the LiveCD. The fan setting will stay on max (and will continue to do so until you shut it off and unplug it). There's not a heat issue in the apartment (it's kept no higher than 76 F), but who knows at this point...

--TSK

Update:

Attempt #2 is going well, it's rapidly heading towards Stage 3.

The main thing about this one is that I'm doing it from the 8890 directly and NOT from an ssh session on my desktop (apparently it would either compile or display too fast if done from ssh, not certain yet).

Also dropped make.conf to a simpler one: compile settings left alone, no USE changes, and the CHOST was set properly now & matches the CPU flag on compile settings. Plus dropped a couple mirrors out of the config.

Keeping my fingers crossed,
--TSK