Bug 166930

What do you mean exactly? The boot completely fails? Are you trying to boot an installed system or are you trying to perform an installation? Does hitting keys and pressing enter do anything or is the system frozen?

Trying to do an installation. It starts booting from the first cd and I see about five lines of text flash really quickly, like is normally seen as it starts booting from the cd. When it gets to the point where it starts the new animated welcome screen, instead I get the blank screen.

The system does seem frozen, as I thought perhaps it was just my monitor, but pressing the key sequence that would normally start the installer does nothing. Also, Ctrl-Alt-Del does not restart the system at this point.

Again, I now the cd is fine because I have used it to install successfully on three other systems.

This means you see the CD's boot menu, and after selecting "Installation" and pressing return, the screen is black?

Does changing the resolution in the bootmenu (maybe hidden in the "more options") to a lower value help?

This happens before the boot menu is shown

Another idea: does booting from CD2 work?
(I previous SUSE releases, it always had a more failsave bootloader - I hope this hasn't changed ;-)

I tried this. It booted from CD2 and asked me to insert CD1 or DVD, which I did and then immediatly I am back to the blank screen.

I never see the boot menu and is preventing an installation of RC1 on this system.

Well this will be hard to debug. What graphics card do you have?
Steffen: Is there a way to force syslinux into text mode or a way to debug this? Please add a comment.

Uh, that's extremely bad. Sounds like bug 129724 is haunting us again. :-(

Joshua, make sure CAPS-Lock is on when the boot loader starts. This will
make it skip the welcome-screen. Does it help?

Do you have time to test a few test-ISOs?

Activating the CAPS-LOCK did help with me. I can now see the Boot Menu, 
but selecting a different language with F2 also hangs the system. F3 and all other screens doesn't hang the system.
It's a HP Compaq NW8240 with a ATI FireGL V5000 wich is practically the same as X700 Pro wich mensioned before.

I will not be able to check this until I am home this evening and will report then. 

I would be happy to help resolve this and test some ISOs for you.

> I haven't tried the 64bit cd, as I have to use 32bit to run the Linux Novell
> client.
Could you give it a try nevertheless. BTW, it's possible to run 32bit apps on Linux x86_64 ...

Created attachment 78984 [details]
test iso

Please give this iso a try. It will boot and enter debug mode (shows debug info in the upper left corner). If you see that window, press TAB to continue in debug mode or press ENTER to single-step through the program (use ESC to continue normally).

What I'd like to know is the value after 'ip' (the instruction pointer) when it hangs.

Note that there's no initrd or kernel on that iso, so you can't actually start anything.

1) The capslock on worked for me and the installation boot menu was shown

2) I also had trouble once I tried to start the installation. First, the console screen stated the its resolution setting was invalid (don't have the exact message) and to either select a different console setting or hit space to continue. I hit space and it proceeded to load, but once it went to start X, my screen switched to another resolution but was blank. The system was not locked, as I could hit Ctrl-Alt-F1 and return to the console.

I tried all the resolutions available via the F3 key and all failed to load X properly.

Video card - ATI X700 Pro
Monitor - Gateway VX900

The Beta9 cd worked fine for the entire install process on this system.

3) Ran the test iso this morning, it stops at ip acd: 86.7

Created attachment 79190 [details]
test iso 2

Uhm, stops at a rather unexpected point.

Please try this new iso. Does it work for you?

Anyway, this indicates it is not the ATI issue. Lowering severity.

For the not existing video mode: in the boot loader menu, press
three times Shift+F10 (increasing debug level to 3), this will show
you later the complete command line before the kernel is loaded.
What 'vga' parameter did the boot loader select?

Tried test iso 2

In the syslinux debug window, shown was:

err 8
ip 9ad: f.7

I then hit enter and was presented a nongraphical text boot menu. I selected Installation and hit enter, which of course did not work, but I wanted to check that it hadn't locked up.

Really weird. Together with comment 13 this might indicate a hardware
problem.

Might of course be a random memory corruption bug in the boot loader code,
but that would not be enough to account for comment 13, item 2).

Joshua, what about the vga option in comment 15?

Mehmet, could you give the test isos a try?

I will test the iso's today. I've opened up a new bugreport. forgot all about this one
https://bugzilla.novell.com/show_bug.cgi?id=168586 (duplicate)

Maybe some info to make it easier for you. 

SuSE 10 boot DVD had the same problem.
I use a wide-screen laptop.

Tried the iso's

Iso 1: 86.7 same as Joshua
Iso 2: 9ad f.7 also same as Joshua

*** Bug 168586 has been marked as a duplicate of this bug. ***

Created attachment 79674 [details]
test iso 3

Please give this iso a try. It will look a bit weird but the point is: Apart from the look, does it work?

Another question: did the video mode menu
look right when you did _not_ switch the language?

I finally had a chance last night to check the vga settings.

In RC1 & RC2: vga=0x3365

In Beta9: vga=0x317

I used the Caps Lock and Shift+F10 on the RC2 to bypass the welcome splash screen and to manually change 0x3365 to 0x317 and was able to successfully install in graphical mode.

In testing last night, I had forgotten that Beta9 already had the welcome screen and it worked properly on this system. Something changed in RC1.

The last CD is the same as iso2 gives the same error 

Err 8
9ad f.7
:wq!

Same here, iso3 has same results as iso2

Created attachment 79690 [details]
test iso 4

Ok, assuming beta9 was ok, here is an iso with only some changes after beta9. Hopefully we
can narrow the bug down with a binary search. :-/

iso 4 worked perfectly

Created attachment 79863 [details]
test iso 5

Thanks!

Please give this one a try.

Created attachment 79976 [details]
test iso 6

To speed things up a bit: in case test iso 5 still worked (and only then), please try this one, too.

I'm unable to boot from both CD's without turning CAPS-LOCK on. If I enable caps-lock the boot menu appears. But if I change the language the screen freezes. 

And the F3 button gives me the following choices:
Text Mode, VESA, 1024x168, 1280x1024, 1280x1024, 1280x1024, 1280x1024, 1400x1050. I have to select the 3th 1280x1024 for me to continue. If I select a different 1280 it comes with an screensize error. 

Created attachment 80286 [details]
test iso 7

Thanks! We're closing in. :-)
Please try this one. Does it work?

ISO 7 almost works

The welcome screen appears. I can change the language and the F3 with resolutions also works. I see all the resolutions but only I see 1280x1024 twice:

The first one links to vga=0x31a
The second one links to vga=0x365

Might still be ok - some BIOSes have some modes twice. If you have
a SUSE Linux running on this machine, just run 'hwinfo --framebuffer'.
Is mode 0x365 listed there?

Created attachment 80303 [details]
test iso 8

Does this iso still work?

That 1280x1024 is listed twice is ok, I just checked the mode numbers
with bug 145749, and they are the same there.

Yes, ISO 8 also works. 
And to answer your question. Yes 0x365 is listed on hwinfo.

Thanks.

Presently I've run out if ideas. The change that breaks things appears
to be the addition of some new translations. Unfortunately the code dealing
with translations works unchanged since years. :-(

So it _looks_ as if there's a random memory corruption bug that surfaces
just with your very memory layout. I've no idea how to debug this further
without physical access to an affected machine.

It seems not to be, though, an issue with the mentioned ATI cards (I've
reports that rc2 works fine from some such machines). And I've got no other
negative reports than your two.

For 10.2 (or current opensuse factory tree), there has been a nearly complete
rewrite of the graphics code. Please give it a try.

I've tried it and it works, but actually the final release of 10.1 also worked. It was only the last few beta's that were giving these errors.

I did actually fix some memory corruption bug in 10.1 final; but I didn't
thing it would be relevant in this case.