An issue with booting SeaTools

[Smasher]

Hi everybody,

the name of the topic may confuse you but I think I am posting in the right forum. I am trying to diagnose my Seagate disk drive using SeaTools Bootable. Booting into it gave me a bare Linux console while a GUI was promised. That was my starting point for investigations, which lead me to learn about Tiny Core Linux. I figured out SeaTools Bootable is actually just a bootable Tiny Core Linux with some extensions. But something seems to be misconfigured...

Let's look together at their sysconfig.cfg. The most interesting part is the following:

Code: [Select]

waitusb=5:LABEL=SEABOOT tce=LABEL=SEABOOT mydata=mydata_STB lst=onboot_STB_Xorg.lst
After reading FAQ I understood that the loader should wait for a USB-device labeled SEABOOT and it is defined to contain Tiny Core Extensions. I am not sure what parameter mydata does, but lst seems to define the list of extensions to install.

P.s. I am also hanging around on IRC if anyone wants to chat me

Anyway from what I can tell: mydata_STB doesn't get unpacked after boot and onboot_STB_Xorg.lst isn't installed. My pendrive is labeled correctly as SEABOOT, the folder structure looks correct to me as well (shortened):

Code: [Select]

/f
├── boot
│   └── syslinux
│       ├── syslinux.cfg
│       └── ...
└── tce
    ├── mydata_STB.tgz
    ├── onboot_STB_Xorg.lst
    └── optional
        ├── acl.tcz
        ├── acl.tcz.dep
        ├── acl.tcz.md5.txt
        ├── asian-fonts.tcz
        ├── aterm.tcz
        ├── aterm.tcz.md5.txt
        ├── ... 485 files here

And I still can't quite figure it out why it doesn't work. Maybe some of the experienced eyes on this forum can help me to get it running?

P.s.: I am also hanging around in IRC if anyone wants to chat me

[Smasher]

Figured that out, no need to clarify:

* New mydata= boot code to support alternate name for backup. Defaults to mydata.tgz.

From http://forum.tinycorelinux.net/index.php?topic=9290.0

Maybe that could be added to FAQ and [F2]: http://tinycorelinux.net/faq.html#bootcodes

Also, xexec= boot code seems to be supported but not documented anywhere

[curaga]

Maybe your usb stick is slow, taking longer than 5 seconds to register? The "noacpi" bootcode may also hinder.

[Smasher]

Yep, changing the delay from 5 to 15 changed things. Now the extensions get loaded during the boot, but when the loading is finished, my display turns black and I cannot do anything. Here is the list of extensions in onboot_STB_Xorg.lst

Code: [Select]

Xfbdev.tcz
Xorg-7.7.tcz
xf86-video-modesetting.tcz
xf86-input-synaptics.tcz
kmaps.tcz
libEGL.tcz
libepoxy.tcz
xfwm4.tcz
aterm.tcz
harfbuzz.tcz
imlib2-bin.tcz
pcre.tcz
sqlite3.tcz
wbar.tcz
seagate-font.tcz
asian-fonts.tcz
ncurses-terminfo.tcz
ncurses-utils.tcz
slocate.tcz
findutils.tcz
bash.tcz
nano.tcz
coreutils.tcz
util-linux.tcz
grep.tcz
pci-utils.tcz
expat2.tcz
openssl.tcz
pixman.tcz
mylocale.tcz
curl.tcz
SeaChest_noAuto.tcz
SeaTools212_noAuto.tcz

Might one of them cause such behavior?

[Juanito]

Enter “version” at a terminal prompt to see which version of tinycore you have.

You could load a different window manager (flwm) to see if that works?

Note also you should load Xfbdev or Xorg-7.7, but not both.

[Smasher]

The version is 8.2.1.

I removed Xfbdev.tcz from onboot_STB_Xorg.lst.

I tried to set desktop=jwm and desktop=flwm.

Nothing helped so far, the displays goes black. Feels like X is missing configurations or just misconfigured...

I noticed now, that if I add "pause" to kernel command line, the screen still goes black. And I think X gets loaded after that pause, so X is not the issue, I guess?

[Juanito]

You can use the “text” boot code to stop tinycore from trying to start X.

If you only load one window manager (I suggest you use flwm for troubleshooting), you don’t need the “desktop” boot code.

Once at the console prompt, execute the first line of ~/.xsession to troubleshoot.

[Smasher]

My bootcode list now looks like this:

Code: [Select]

initrd=/boot/64bit/modules_4817_64.gz,/boot/32bit/rootfs_4817_32.gz lang=en_US.UTF-8 kmap=qwerty/us noutc loglevel=2 noacpi noswap libata.allow_tpm=1 waitusb=15:LABEL=SEABOOT tce=LABEL=SEABOOT mydata=mydata_STB lst=onboot_STB_Xorg.lst vga=791 xexec="ST4L --exitPowerOff" showapps text
The screen goes black right after the last extension gets loaded. I don't enter the text mode. Looks like bootloader stops execution at this point.

My screen looks like this a second before going black: https://i.snag.gy/eD0Aln.jpg (sorry for that very bad focus)

[Juanito]

You can stop the extensions being loaded with the “base” boot code and stop the backup being restored with “norestore”.

[Rich]

Hi Juanito
He's running 64 bit TC 8 (kernel 4.8.17) with 32 bit extensions (Core64). I noticed this in his  onboot.lst  file:

Code: [Select]

xf86-video-modesetting.tczThe only place I can find this is in the TC 5 repository:
http://tinycorelinux.net/5.x/x86/tcz/xf86-video-modesetting.tcz.info
I wonder if this might be part of the problem?

[Smasher]

What I've found out after a long trial and error makes absolutely no sense to me but here it is, the following diff made it work:

Code: [Select]

diff --git a/tce/onboot_STB_Xorg.lst b/tce/onboot_STB_Xorg.lst
index 9c2c74e..ec87eeb 100644
--- a/tce/onboot_STB_Xorg.lst
+++ b/tce/onboot_STB_Xorg.lst
@@ -1,6 +1,5 @@
 Xfbdev.tcz
 Xorg-7.7.tcz
-xf86-video-modesetting.tcz
 xf86-input-synaptics.tcz
 kmaps.tcz
 libEGL.tcz
@@ -29,5 +28,4 @@ openssl.tcz
 pixman.tcz
 mylocale.tcz
 curl.tcz
-SeaChest_noAuto.tcz
 SeaTools212_noAuto.tcz
As soon as I add one of the extensions, either xf86-video-modesetting.tcz or SeaChest_noAuto.tcz, the boot process breaks. The first one contains usr\local\lib\xorg\modules\drivers\modesetting_drv.so and the second one 10 binaries with the path scheme usr\bin\SeaChest_* Then I tried something, what confused me even more. If I add SeaChest.tcz instead of SeaChest_noAuto.tcz to the boot list, it works! That extension, in addition to the binaries named above, has a file usr\local\tce.installed\SeaChest and some PNG images in usr\local\include\seachest\. In search of commonalities, I examined the dependencies of these extensions. Both of xf86-video-modesetting.tcz.dep and SeaChest_noAuto.tcz.dep include graphics-KERNEL.tcz and I think this is the key. I just still don't understand yet the reason behind it.

I made the extensions available for download here: https://filetransfer.io/data-package/sapqcvCm

[Juanito]

Once in the GUI, does /var/log/Xorg.0.log exist (this will show whether Xfbdev or Xorg-7.7 is being used)?

[Rich]

Hi Smasher
Try adding  graphics-KERNEL.tcz  as the first entry in your onboot.lst file in the non-working version and see if that fixes it.

[Smasher]

Hey Rich,

I tried adding graphics-KERNEL.tcz to tce/onboot.lst and to the top of tce/onboot_STB_Xorg.lst and it didn't help. I can tell even more: if I omit the problematic extensions (SeaChest_noAuto and xf86-video-modesetting) but add graphics-KERNEL.tcz to a list, it breaks as well.

I learned that if I wait a while after screen turning black I get access to a console. And from the log, I learned that X cannot start because of the lack of hardware rendering support and a missing library for software rendering. Here is the relevant part of Xorg.0.log

Code: [Select]

[    30.233] (II) AIGLX: Screen 0 is not DRI2 capable
[    30.233] (EE) AIGLX: reverting to software rendering
[    30.233] (EE) AIGLX error: dlopen of /usr/local/lib/dri/swrast_dri.so failed (/usr/local/lib/dri/swrast_dri.so: cannot open shared object file: No such file or directory)
[    30.233] (EE) GLX: could not load software renderer
[    30.233] (II) GLX: no usable GL providers found for screen 0
[    30.233] (II) modeset(G0): Damage tracking initialized
[    30.233] (EE)
[    30.233] (EE) Backtrace:
[    30.233] (EE) 0: /usr/local/lib/xorg/Xorg (xorg_backtrace+0x49) [0x80dd9e5]
[    30.233] (EE) 1: /usr/local/lib/xorg/Xorg (0x8048000+0x139d5e) [0x8181d5e]
[    30.233] (EE) 2: linux-gate.so.1 (__kernel_rt_sigreturn+0x0) [0xf77d1b80]
[    30.233] (EE) 3: /usr/local/lib/xorg/Xorg (RRScreenSetSizeRange+0x85) [0x810ff0b]
[    30.233] (EE) 4: /usr/local/lib/xorg/Xorg (xf86RandR12CreateScreenResources+0x2ac) [0x8175c78]
[    30.233] (EE) 5: /usr/local/lib/xorg/Xorg (0x8048000+0xfe1a2) [0x81461a2]
[    30.233] (EE) 6: /usr/local/lib/xorg/Xorg (0x8048000+0x1ff77) [0x8067f77]
[    30.233] (EE) 7: /lib/libc.so.6 (__libc_start_main+0x157) [0xf73a723f]
[    30.233] (EE) 8: /usr/local/lib/xorg/Xorg (0x8048000+0x2076c) [0x806876c]
[    30.233] (EE)
[    30.233] (EE) Segmentation fault at address 0x58
The whole log: https://pastebin.com/raw/fqHVYQdv

@Juanito: on a working version Xorg.0.log exists as well

[Juanito]

What graphics hardware do you have?

Near the beginning of the xorg log it will show which drivers it is trying to load - it sounds like yours is missing?