Bug#132990: Test of latest boot floppies
Greetings: I tried out the new 2.4 boot floppies from ~blade/boot-floppies and had no problems. This is the Thinkpad machine that originally had the 'white screen of death' problem and later, the crash in the base install. I went through the whole process including base install from the network and everything worked fine. Also, the problem I mentioned once about being unable to use a serial console was caused by the fact that Thinkpads come with the serial port disabled in firmware by default. This has been fixed and the serial console works fine. Thanks, -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
/etc/messages.trm
I compiled dbootstrap with debug turned on, copied it onto my partially installed system, and ran it. The immediate problem was: An error occurred while loading application messages from /etc/messages.trm I looked, and there is no /etc/messages.trm; apparently the boot floppies I used didn't have one. I also tried copying /etc/messages.en to /etc/messages.trm but I got the same result. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
dbootstrap source
Greetings: Sorry for pummeling the list with so many stupid questions, but I have looked in all of the places that are obvious (to me) and I can't find the source code for dbootstrap. I would like to build it myself in such a way as to allow it to generate a proper core file (much more useful to me than a register dump). Can I do an anonymous CVS checkout from somewhere? -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#132990: this is two different bugs
On Fri, Feb 15, 2002 at 11:49:51PM -0800, David Kimdon wrote: > There appear to be two different bugs here. > > 1. white screen of death, at boot right as the installer is trying to > start it keeps segfaulting. This happens on some machine and not on > others, with some disks and not others. > > 2. dbootstrap segfaults during the base system install. It isn't > clear that this bug is related to the white screen at boot. > > -David > This seems likely to me also. I have seen and reported both problems but they occurred in boot disks that I downloaded from different places. 1) came from debian/dists/woody/main/disks-i386/current/images-1.44/bf2.4 2) came from people.debian.org/~blade/bf2.4 on Feb 2. The same error occurs with either a network install or a local partition install. (Identical register dumps) Also, (1) re-occurred on disks from ~blade downloaded on Feb 10. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Why does dbootstrap not make a core file?
On Thu, Feb 14, 2002 at 08:43:03PM -0700, Chris Tillman wrote: > On Thu, Feb 14, 2002 at 06:39:35PM -0700, Mark Zimmerman wrote: > > So, that leaves me ready to ftp in a copy of the base system .tgz > > file. Has anyone built such a thing for woody? I couldn't find one. > > Alternately, (or in addition) if anyone has built a version of > > dbootstrap with debug symbols I'll try it out. > > > > Hmmm. I don't know why you couldn't have used wget, on the installer > disk to get that. IWFM. > It probably would have worked fine if I had thought to look for it. Just another Senior Moment, I guess. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Why does dbootstrap not make a core file?
On Thu, Feb 14, 2002 at 04:31:44PM -0500, Matt Zimmerman wrote: > On Thu, Feb 14, 2002 at 01:58:22PM -0700, Mark Zimmerman wrote: > > > On Thu, Feb 14, 2002 at 11:44:22AM -0500, Matt Zimmerman wrote: > > > On Thu, Feb 14, 2002 at 09:29:09AM -0700, Mark Zimmerman wrote: > > > > > > > connection and plenty of mounted partitions. I tranferred an ftp > > > > client and the curses and readline libraries to the system via floppy > > > > and tried to run ftp. Unfortunately, it said: > > > > > > > > ftp: ftp/tcp: unknown service > > > > > > Copying in /etc/services should fix that. > > > > > > > I didn't mention it before but that was the first thing I tried. > > It's probably an NSS problem then. Do you have /etc/nsswitch.conf and > /lib/libnss*? > It's working now. These are the libraries I ended up needing: -rwxr-xr-x1 root root 1170812 Feb 15 00:52 libc.so.6 -rwxr-xr-x1 root root 248144 Feb 15 00:47 libncurses.so.5 -rwxr-xr-x1 root root12372 Feb 15 00:47 libnss_dns.so.2 -rwxr-xr-x1 root root34144 Feb 15 00:47 libnss_files.so.2 -rwxr-xr-x1 root root 146320 Feb 15 00:47 libreadline.so.4 plus /etc services, /etc/nsswitch.conf, and the ftp client program. libc.so.6 had to be copied since the version in the boot floppies couldn't resolve getlogin. So, that leaves me ready to ftp in a copy of the base system .tgz file. Has anyone built such a thing for woody? I couldn't find one. Alternately, (or in addition) if anyone has built a version of dbootstrap with debug symbols I'll try it out. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Why does dbootstrap not make a core file?
On Thu, Feb 14, 2002 at 11:44:22AM -0500, Matt Zimmerman wrote: > On Thu, Feb 14, 2002 at 09:29:09AM -0700, Mark Zimmerman wrote: > > > connection and plenty of mounted partitions. I tranferred an ftp > > client and the curses and readline libraries to the system via floppy > > and tried to run ftp. Unfortunately, it said: > > > > ftp: ftp/tcp: unknown service > > Copying in /etc/services should fix that. > I didn't mention it before but that was the first thing I tried. --Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Why does dbootstrap not make a core file?
On Thu, Feb 14, 2002 at 11:09:15AM +, Phil Blundell wrote: > On Thu, 2002-02-14 at 04:06, Mark Zimmerman wrote: > > One of my tests of dbootstrap (in trying to get past the white screen > > of death) resulted in a register dump to the screen. I was wondering > > why it didn't dump core instead? Is this something related to how > > dbootstrap is built or is it caused somehow by the installation > > environment? I was hoping for a core file in case the builder could > > make use of it. > > Well, it would be easy enough to add an option to enable core dumps. > I'm not convinced it would be useful, though: > > Dbootstrap doesn't always have a filesystem available that it can > reasonably dump core onto. The only place where there is likely to be > enough space is the target install disk - this isn't accessible at all > during the early stages of installation, and the user might well find it > difficult to retrieve the core dump from there even if one was created - > the machine may not have a floppy drive, and the core may be too big to > fit on a disk anyway. I don't think the root disk includes enough tools > to copy the core away over the network either. > > The builder should be able to use the information in the register dump > to get a handle on the problem in most cases. Admittedly the user may > find transcribing it into an email a bit tedious, but at least it's > straightforward to do. > I recognize the truth of all of this in the general case, but I may be close to becoming a special case. Since it didn't crash until I got to the 'Install Base System' phase, I already have a working network connection and plenty of mounted partitions. I tranferred an ftp client and the curses and readline libraries to the system via floppy and tried to run ftp. Unfortunately, it said: ftp: ftp/tcp: unknown service and quit. When I get back to working on this tonight, I'll try copying libc also and/or look for an appropriate kernel module to load. Anyhow, if I can get ftp working, that will open up some possibilities: 1. Acquire or build a debug version of dbootstrap and see what happens. 2. ftp the base_system.tgz file to the hard drive and try to install from there. By the way, the current woody installation manual does not have a link to such a file. Is there one somewhere? -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Why does dbootstrap not make a core file?
One of my tests of dbootstrap (in trying to get past the white screen of death) resulted in a register dump to the screen. I was wondering why it didn't dump core instead? Is this something related to how dbootstrap is built or is it caused somehow by the installation environment? I was hoping for a core file in case the builder could make use of it. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bf2.4 dbootstrap white screen of death
On Sun, Feb 10, 2002 at 04:53:17PM +, Philip Blundell wrote: > On Sun, 2002-02-10 at 16:44, Mark Zimmerman wrote: > > Using the images from people.debian.org/~blade/bf2.4, I have gotten > > past the white screen problem and I am now at the 'Install the Base > > System' step. My network is configured and working (I can ping > > http.us.debian.org by name from VT2). When I select a network install > > and give it the OK, it immediately jumps back to the 'Choose the > > Language' screen. I seem to be in an infinite install loop... > > That indicates dbootstrap is crashing. Try booting with the "debug" > argument and see if that gives any extra information. > VT1 becomes this: *** Segmentation Fault Registers: EAX: 40068c00 EBX: 400e522c ECX: EDX: 0806470f EDI: 0070 ESI: 0001 EBP: b36c ESP: bfffed64 EIP: 40068dee EFLAGS: 00010246 Trap: 000e Code 0004 VT3 logs: Prior to exiting, dbootstrap said: Extracting base, Archive_Dir:netfetch Also, (this may be irrelevant) the logs showed several attempts to mount a cd drive. I don't have one on this machine and I never told dbootstrap that I did. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bf2.4 dbootstrap white screen of death
On Sat, Feb 09, 2002 at 06:34:54PM -0700, Mark Zimmerman wrote: > On Sat, Feb 09, 2002 at 11:22:51PM +0100, Eduard Bloch wrote: > > > > VT3 shows that dbootstrap is repeatedly starting and exiting. > > > > Could you try my CVS boot floppies from people.debian.org/~blade/bf2.4 ? > > The problem you describe sound like the old newt-utf8 breakage, but I > > cannot reproduce it. > > > Using the images from people.debian.org/~blade/bf2.4, I have gotten past the white screen problem and I am now at the 'Install the Base System' step. My network is configured and working (I can ping http.us.debian.org by name from VT2). When I select a network install and give it the OK, it immediately jumps back to the 'Choose the Language' screen. I seem to be in an infinite install loop... -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bf2.4 dbootstrap white screen of death
On Fri, Feb 08, 2002 at 05:45:07PM +0100, Eduard Bloch wrote: > #include > Mark Zimmerman wrote on Fri Feb 08, 2002 um 08:39:07AM: > > Greetings: > > > > I tried out a new installation on an IBM Thinkpad 560X with the bf2.4 > > Which kind of VGA chipset is used there? Anyways, if you come to the > floppy change prompt, framebuffer is allready working for you. > > > linux floppy=thinkpad > > linux floppy=thinkpad,mono > > > > Can anyone suggest a workaround for this? > > Please try with "video=vga16:off" and report the result. > I ran a bunch of new tests and here are the results: linux floppy=thinkpad,video=vga16:off results in the white screen as previously described. However, looking through dmesg yielded the clue that this is an ill-formatted set of options (comma should be space). So, I tried: linux floppy=thinkpad video=vga16:off and it booted to the point where it asks for the root floppy, but when I put it in I got I/O errors and a kernel panic. Apparently I ought not to use floppy=thinkpad on my thinkpad and the original badly formatted command kept it from using it. Next, I tried linux video=vga16:off and it loaded the root floppy successfully but things went astray after that. VT1 gets to the point where it says: init started: BusyBox ...(etc) VT2 repeatedly asks me to press Enter to activate this console but will not give me a shell. VT3 shows that dbootstrap is repeatedly starting and exiting. Finally, I tried booting with no options. This returned me to the original white screen problem. This time, I thought to look at VT3 for the logs and found that dbootstrap started and exited several times. The last log message shows that it exited and was scheduled for restart. A ps ax in VT2 shows it running though. I also tried (finally++) so see if I could get going with a serial console (console=ttyS0,9600). When I tried that, I got nothing on the screen and nothing out the serial port either. Same with ttyS1. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: bf2.4 dbootstrap white screen of death
On Fri, Feb 08, 2002 at 05:45:07PM +0100, Eduard Bloch wrote: > #include > Mark Zimmerman wrote on Fri Feb 08, 2002 um 08:39:07AM: > > Greetings: > > > > I tried out a new installation on an IBM Thinkpad 560X with the bf2.4 > > Which kind of VGA chipset is used there? Anyways, if you come to the > floppy change prompt, framebuffer is allready working for you. > Can I find out the VGA chipset from dmesg or something? Anyhow, video seems to work fine except for dbootstrap. I switched to the second VT and got the prompt and was able to do a 'ps ax' which showed dbootstrap running. I was hoping to be able to pass dbootstrap an argument that would make it all better. > > linux floppy=thinkpad > > linux floppy=thinkpad,mono > > > > Can anyone suggest a workaround for this? > > Please try with "video=vga16:off" and report the result. > Actually, I believe that I did try this one also, with the same effects; I just forgot to mention it. I will doublecheck it this evening. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
bf2.4 dbootstrap white screen of death
Greetings: I tried out a new installation on an IBM Thinkpad 560X with the bf2.4 boot floppies. It booted up and loaded root.bin without incident but dbootstrap gave me a solid white screen with a blinking cursor in the upper left corner. Hitting return caused the cursor to move down the screen so this does not appear to be a case of 'white on white' colors. I tried each of the following boot arguments but the symptoms were the same. linux floppy=thinkpad linux floppy=thinkpad,mono Can anyone suggest a workaround for this? I had previously booted successfully into dbootstrap with the reiserfs boot floppies and this problem did not exist. -- Mark -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]