[kernel] r37084 broke booting on Zotac nForce 630i-ITX

[diver]

Using binary search I was able to find exact revision which broke booting on my motherboard Zotac nForce 630i-ITX.

hrev37083 and earlier boots fine.
hrev37084 and later hangs at boot logo before 1st icon is lit.

[anevilyak]

Something seems to have gone wrong with your sysinfo attachment, it appears to contain HTML as well. Looks like that board/CPU support PSE36 though. How much RAM do you have in there?

[anevilyak]

On a related note it's somewhat odd that hrev37084 specifically triggers the hang since all it does is add a function definition that isn't actually called anywhere yet.

[diver]

Ah, yeah, thanks. I've reattached it.

[bonefish]

Replying to anevilyak:

On a related note it's somewhat odd that hrev37084 specifically triggers the hang since all it does is add a function definition that isn't actually called anywhere yet.

Indeed. And being inline there isn't even code generated for it. So the Haiku builds from hrev37083 and hrev37084 should be utterly identical save for the revision number.

Please verify with clean builds. Also a syslog would be nice.

[anevilyak]

Replying to bonefish:

Please verify with clean builds. Also a syslog would be nice.

There wouldn't be any syslog output in this instance would there? If it hangs before even the first icon's lit up I'd imagine the syslog daemon's not initialized yet, though I suppose there's still the in-memory buffer that the USB stick could potentially capture/write.

[bonefish]

Replying to anevilyak:

There wouldn't be any syslog output in this instance would there? If it hangs before even the first icon's lit up I'd imagine the syslog daemon's not initialized yet, though I suppose there's still the in-memory buffer that the USB stick could potentially capture/write.

That's the one I'm looking for. Since the hang happens so early, it is probably not a hang, but a kernel panic with the blue screen support not initialized yet. I haven't checked the code, but the KDL output might be captured in the debug syslog. Even if it isn't, some info is better than none at all. The serial output would definitely be best. The board doesn't seem to have a serial port, but at least a COM header, so a port could be connected.

[diver]

Unfortunately, there is no syslog. COM header? Can't see it.

[bonefish]

Replying to diver:

Unfortunately, there is no syslog.

You mean in the boot loader? That would be weird, since it is already created there.

COM header? Can't see it.

I only did a quick google, which turned up ​this site listing an internal COM interface. Might of course be incorrect info.

[diver]

Now I see it, but I don't know how to connect to it.

[anevilyak]

If you can find one in a store over there, something like this would do the trick:

​http://www.newegg.com/Product/Product.aspx?Item=N82E16812200466&cm_re=serial_port_header-_-12-200-466-_-Product

[diver]

Aha, thanks!

[diver]

I tried to boot from real partition in qemu and in vbox and they both hung before the 1st icon. I captured serial output from vbox. As emulators refused to work with images I prepared I thought that it must be something with my build environment. So I had reformatted my hdd and installed Ubuntu 10.04, re-downloaded sources and now current revisions boots fine both natively and in emus. Feel free to close this one and sorry for the noise.

[bonefish]

Thanks for the update. Still weird, but anyway...