Stuck at tc@box:$

[Juanito]

When I said "startx", I was told: "-sh: startx: not found"

This happens because the Xprogs extension is not loaded.

To start a minimum gui, you need the Xvesa, Xprogs, flwm, aterm and wbar extensions, and the extensions that they depend on, to be loaded.

You can check which extensions are loaded with "tce-status -i".

[curaga]

The command "fdisk -l" just refused to give me any answer.

This would point to a missing or buggy ide driver. Try to find out which ide chipset the machine has, which may admittedly be difficult, since you need to get lspci there somehow.

You might also try "dmesg | grep -i ide" on BasicLinux, or some other floppy distro that sees your hd. Also gloss over the dmesg just in case there's ide messages without the ide string.

edit: Maybe also try the last 3.x version, TC 3.8.4. It was the last version to use old-style ide drivers, and so might see your hd. Of course apps will be old, and the package search may not work anymore, but it could be worth a try.

[Gerolf]

Thank you very much for your answers, Juanito and Curaga! I wasn't quite aware of the problem that a Linux system may not be able to access the hard drive it has been started from, but in fact it is the boot manager which accesses the hard drive, and Linux takes over control from RAM drive only after the initial filesystem has been copied into it by the boot manager. So the absence of the required stuff for starting the GUI is a consequence of the operating system's inability to read the IDE partition.

As expected by now, the command "tce-status -i" too didn't give any answer, and "lspci" wasn't found. According to the documentation, the Thinkpad 310E has an "ALI Aladdin III M1521/M1523" chipset. "dmesg | grep -i" reported four lines about a "legacy IDE quirk". Under Basic Linux, the relevant "dmesg" lines containing "ide" or "hda" were:

PCI_IDE: unknown IDE controller on PCI bus 00 device 11, VID=10b9, DID=5219
PCI_IDE: not 100% native mode: will probe irqs later
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hda: IBM-DDLA-21620, ATA DISK drive
hda: IBM-DDLA-21620, 1551MB w/96kB Cache, CHS=788/64/63
hda: hda1 hda2 < hda5 hda6 >

As proposed by you, I copied multicore_3.8.4.iso to hda1, booted from it, and it worked out of the box, starting X server and desktop. My first success with Tiny Core Linux! However, to my confusion "fdisk -l" only reported the attached USB pendrive as sda1, and "sudo mount /dev/hda1 /mnt" failed again saying "No such device or address" though I got the following information by "dmesg":

Uniform Multi-Platform E-IDE driver
ide_generic: please use "probe_mask=0x3f" module parameter for probing all legacy ISA IDE ports
ide-gd driver 1.18
ide-cd driver 5.00

So I got an old Basic Linux that only sees the harddisk and an old Tiny Core Linux that only sees the pendrive. I am unexperienced with Linux drivers and don't know how to fix this. Also, I'm looking for an up-to-date operating system for this old computer. I understand that Robert doesn't want to develop a new Linux distribution just to support old hardware, but somehow it would be a pity if this ability got lost due to decisions like removal of IDE drivers.

Finally I tried a workaround starting a frugally installed Tiny Core Linux 6.4.1 from pendrive via loadlin because it turned out to be easy to access that device under FreeDos. I downloaded usbaspi.sys (USB controller driver) and di1000dd.sys (mass storage driver) from the web. Then I added "device=c:\usbaspi.sys /w /v" and "device=di1000dd.sys" to fdconfig.sys. Now the memory stick appeared as drive D: under FreeDos (I never tried this before, it would have been much easier to install Windows 98 that way). Then I could boot Tiny Core Linux 6.4.1 with the command "loadlin d:\tcore641\boot\vmlinuz root=/dev/sda1/tcore641 initrd=d:\tcore641\boot\core.gz ro". However, the startup process again ended up at the "tc@box:~$" command prompt, where "startx" could not be found. So this problem has nothing to do with an invisible harddrive, has it?

[Rich]

Hi Gerolf

As proposed by you, I copied multicore_3.8.4.iso to hda1, booted from it, and it worked out of the box, starting X server and desktop. My first success with Tiny Core Linux! However, to my confusion "fdisk -l" only reported the attached USB pendrive as sda1, and "sudo mount /dev/hda1 /mnt" failed again saying "No such device or address" though I got the following information by "dmesg":

Uniform Multi-Platform E-IDE driver
ide_generic: please use "probe_mask=0x3f" module parameter for probing all legacy ISA IDE ports
ide-gd driver 1.18
ide-cd driver 5.00

What happens if you add the following kernel parameter to your boot loader:

Code: [Select]

ide_generic.probe_mask=0x3f

[Juanito]

I copied multicore_3.8.4.iso to hda1, booted from it, and it worked out of the box, starting X server and desktop. My first success with Tiny Core Linux! However, to my confusion "fdisk -l" only reported the attached USB pendrive as sda1, and "sudo mount /dev/hda1 /mnt" failed again saying "No such device or address" though I got the following information by "dmesg"

If tc-3.8.4 booted to a gui desktop, then it must have been able to access your hd in order to able to load the required extensions (or it got them from the usb stick).

What does this give:

Code: [Select]

$ cat /etc/fstab
$ ls -l /etc/sysconfig/tcedir

[curaga]

@Juanito

In TC 3 days, Xvesa and Xlibs were included in the initrd.

@Gerolf

Linux has never supported your IDE chipset, neither in the ALI IDE driver (which TC 3 had) nor in the ALI ATA driver (which current TC has). Even in Basic Linux it used a generic IDE driver, that the current kernel for some reason did not use. Please try the bootcode on current TC as Rich suggested, and if it does not work, then these ones:
ide-pci-generic.all-generic-ide
ata_generic.all_generic_ide=1

Your usb boot most likely needs the waitusb=5 boot parameter, since the usb controller can be slow to wake up. Do try the IDE first though, a freedos -> usb boot is a little cumbersome.

[Juanito]

In TC 3 days, Xvesa and Xlibs were included in the initrd.

..so much for my memory 

[Gerolf]

When I said: "boot: tc vga=771 ide_generic.probe_mask=0x3f" at the TC 3.8.4 boot prompt, then the desktop showed up, but the mouse pointer was frozen. With "boot: tci vga=771", the startup hang at "Loading extensions...". "boot: mc vga=771 ide_generic.probe_mask=0x3f" took me to the command prompt, and now "fdisk -l" and "cat /etc/fstab" showed me all partitions on hda (harddisk) and sda (pendrive), but "startx" wasn't found, and "/etc/sysconfig/tcedir" didn't exist. Now, "boot: tc vga=771 ide-pci-generic.all-generic-ide" solved the mouse problem: It was movable, the harddisk was readable, but "/etc/sysconfig/tcedir" still didn't exist.

Then I tried the same at the TC 6.4.1 prompt, booting the ISO file from harddisk. The prompt looks different here, there is no "boot:". After "/boot/vmlinuz initrd=/boot/core.gz loglevel=3 cde showapps desktop=flwm_topside", I added: "vga=771 ide-pci-generic.all-generic-ide", and then I stuck again at the "tc@box:~$" command prompt, with "fdisk -l" only mentioning sda1. So I set all my hope on the "ata_generic.all_generic_ide=1" kernel parameter. To my surprise, "fdisk -l" now listed the harddisk partitions as sda1, sda2, sda5, sda6, and the pendrive as sdb1. I could mount and read sda1, where the "CorePlus-6.4.1.iso" file lies. However again, there was no desktop, no "startx", just a link pointing from "/etc/sysconfig/tcedir" to "/tmp/tce", which contained an empty "optional" subdirectory. "fromISOfile" said: "Need root privileges", "sudo fromISOfile" said nothing.

In my attempt to chain boot TC 6.4.1 via FreeDos from USB, I could pass the required kernel parameters to loadlin on the DOS command line. When I redirected loadlin's output to a log file, I read an error message that I didn't notice yesterday: "can't open image file for initrd". However, when I simply moved the "boot" and "cde" folders up one level to the top directory and changed the command line accordingly to "loadlin d:\boot\vmlinuz root=/dev/sdb1 initrd=d:\boot\core.gz ro cde vga=771 ata_generic.all_generic_ide=1", only a loadlin warning remained: "loading above 16MB using INT15". The "waitusb=5" parameter seemed not to be required. "tce-status -i" now listed "Xlibs Xprogs Xvesa aterm firmware fluxbox flwm icewm iw jwm kmaps openbox wbar wifi". Now "startx" was present, but "failed in waitforX".

Since this was new, I booted again the "CorePlus-6.4.1.iso" file from harddisk with "vga=771 ata_generic.all_generic_ide=1" kernel parameters. "tce-status -i" then listed the same 14 items, but "startx" failed because the "Requested flwm_topside could not be loaded". So I booted again and selected "FLWM Classic Window Manager". This again "failed in waitforX". And it was quite the same when I added those kernel parameters in "menu.lst" to boot the frugal installation on the sixth harddisk partition with EXT3 filesystem. In short, I can access all drives by now, but I still can't get up the desktop in current Tiny Core Linux!

[Juanito]

Then I tried the same at the TC 6.4.1 prompt, booting the ISO file from harddisk. The prompt looks different here, there is no "boot:". After "/boot/vmlinuz initrd=/boot/core.gz loglevel=3 cde showapps desktop=flwm_topside"...

You need to remove "cde" - this is for booting from cd/dvd

[curaga]

OK, please use the frugal install.

Your laptop also has a very old graphics card, it might not support the Vesa standard properly. You already added a framebuffer bootcode, so please try removing Xvesa and installing Xfbdev. If you have network access, this can be done with:

tce-load -wi Xfbdev
# edit onboot.lst to remove Xvesa.tcz
vi /etc/sysconfig/tcedir/onboot.lst

[Gerolf]

When I simply omitted the "cde" boot parameter for my frugal installation on EXT3 partition sda6, then "startx" was "not found", and "tce-status -i" did not list anything. It was the same when I replaced "cde" with "tc". So the "cde" seemed to be required. Now I omitted the "vga=771" instruction. I had used it because according to the Thinkpad 310E documentation, 800x600x256 is the highest resolution of the SVGA laptop display, whereas the "C&T 65550 HiQV32 64-bit SVGA accelerator" graphics controller also handles 1024x768 resolution with 256 colors. Nevertheless, this "failed in waitforX", though a "Tiny Core" logo had shown up in graphics mode just before the "Booting Core 6.4.1" message appeared. "xsetup" just gave me two empty lines.

Since the Thinkpad 310E doesn't have a network card, I attached a USB-to-RJ45 adapter with a "Realtek RTL8152B" chip. While Windows 98 supposed this thing to be an external CDR0M drive, it was recognized by TC Linux 6.4.1, as "dmesg" reported the following lines:
r8152 3-2:1.0 eth0: v1.06.0 (2014/03/03)
usbcore: registered new interface driver r8152
usbcore: registered new interface driver cde_ether

So to my surprise, "tce-load -wi Xfbdev" immediately downloaded the requested "Xfbdev.tcx" file to "/etc/sysconfig/tcedir/optional". (This encourages me not to give up hope though for me, this is already the hardest setup ever.) I also found an "onboot.lst" file in "/etc/sysconfig/tcedir" which wasn't there before. As it contained just a "Xfbdev.tcx" line, there seemed to be no need for editing. This time, "xsetup" offered me the "Tny Core Linux - Mouse Setup" (sic), but no video mode selection as I would have expected. According to the FAQ, I added "tinycore tce=sda6" to the kernel parameters in GRUB4DOS's "menu.lst" to make the system changes permanent. This worked, but "startx" still "failed in waitforX", with or without the "vga=771" boot parameter.

[Rich]

Hi Gerolf
Copy all the files from  cde/optional  to  tce/optional. Edit your  onboot.lst  file to include Xvesa, Xprogs, flwm_topside,
aterm and wbar with the  .tcz  appended to those names. Reboot.

[gerald_clark]

Forget it.  64M is not enough.
The book gives installation methods, as well as the wiki.
You need to be able to boot from another media and use the available install methods to install it to the hard drive.
But, with 64M,  you can forget X.

[curaga]

I disagree, depends on the apps he wants to run. I just tried the latest beta on Qemu with 64mb, after a normal X boot (+ cache-clear for measurement) there's 19mb of RAM free.

That's quite enough for playing music, reading, dillo, etc.