Tiny Core 14.0 Alpha 1 Testing

[Juanito]

The update from tc-13 -> tc-14 is about updating the base libs - i.e. glibc, gcc, zlib, etc - and the kernel.

It doesn't look like the kernel graphics drivers are causing your problem, so it must be some combination of mesa, xorg-server and llvm.

Due to problems with intel hd4400 graphics, in tc-14 mesa was updated from 21.3.4 - > 22.3.3 and xorg-server to the latest version, but this did not fix the problem and the problem differs between 32bit/64bit.

The release notes for mesa-22.3.4 show one radeon fix, but it doesn't look like your issue.

You could try using tc-13 mesa extensions in tc-14 or try recompiling mesa-22.3.3 with debug flags to see if you learn more - I don't believe any of us has the hardware to test.

[nick65go]

I  tried using tc-13 extensions in tc-14, and it works! I am lucky.
The only thing that I replaced (in the end, after many try and errors) was Xorg--7.7-3d.tcz
so the problem is on /tmp/tcloop/Xorg-7.7-3d/usr/local/lib/dri/radeonsi_dri.so
It seems that llvm-lib, xf86-video-ati, firmware-radeon, all are OK.

[nick65go]

Compiling / debuging is over my "pay grade". I could grab Alpine/Archlinux and I am done. But I still love tinycore.
So, for any good samaritean, here is the info need:

[ANNOUNCE] mesa 22.3.3 https://lists.freedesktop.org/archives/mesa-dev/2023-January/225908.html
 radeonsi: rewrite si_update_ps_colorbuf0_slot to fix crashes and recursions
 radeonsi/vcn: av1 film_grain output fix

or [ANNOUNCE] mesa 22.3.4 https://lists.freedesktop.org/archives/mesa-dev/2023-January/225916.html
 radeonsi: respect smoothing_enabled
 radeonsi/gfx11: fix ge_cntl programming
 radeonsi: report 0 block size for Polaris HEVC encoding

[Rich]

Hi nick65go

... PS: I am not picky, just say that filetools, shell versus gui, has different output text colors (see attach).

There are a couple of things going on here. The blue background of the GUI contrasts differently
than the black background of the terminal. The color commands from tc-functions also force
the text to bold, which fltk reacts to differently than regular text. I made a fake filetool.sh
script to show how the colors react for bold and regular text, with blue, white, and black
backgrounds:


I attached the fake filetool.sh script in case anyone wants to play with it. Copy it to /usr/bin/
and use the Dry Run option in the filetool GUI.

[nick65go]

@Rich: Thank you for the clear visual comparison. My main preference is for strong contrast foreground vs. background, instead of bold text. Because I can not see very well fade colors (ex: grey on back, green on blue, etc). In the end, I could modify the colors myself, but I try to do be synchronized with original tinycore.

[alphons]

Was a few days off, building linux from scratch.
Today i had the pleasure to do some testing on the TinyCore 14 Alpha.

Results: https://github.com/alphons/TinyCore/blob/main/TestingTinyCore14Alpha.md

Summary:

• starting samba (shares / to everyone as root for testing)
• starting ssh (ssh login works)
• compiling kernel Linux-6.1.2 (making new iso and booting)
• making and executing dotnet console program
• starting mongo database

By compiling the same kernel again on the just booted 14.alpha involves all the steps of compiling kernel, making modules, making new iso etc. So i think it was a good test 

When samba works, a whole lot of extensions must work also, because it is a beast to run.

cheers.

[FlyingDutchman]

Hi,

Congrats on the new release. I tested TCL pure64 release 14 on my installation and it works fine. I also did a quick check on the dependency files and found some files listed as dependency, that do not exist in the package folder (http://tinycorelinux.net/14.x/x86_64/tcz/). I don't use these packages, so no issue for me, but maybe someone else may benefit if these issues are resolved. Please note that these issues are in the x86_64 packages. I didn't check the x86 packages.

Code: [Select]

dependency perl_archive_zip.tcz in file auto-multiple-choice.tcz.dep not found
dependency gst-plugins-good-dev.tcz in file gstreamer-editing-services-dev.tcz.dep not found
dependency enchant-dev.tcz in file gtkhtml-dev.tcz.dep not found
dependency perl_xml_simple.tcz in file icon-naming-utils.tcz.dep not found
dependency perl_xml_simple.tcz in file icontool.tcz.dep not found
dependency icu65-dev.tcz in file js68-dev.tcz.dep not found
dependency python-olefile.tcz in file python-pillow.tcz.dep not found
dependency linux-5.10_api_headers.tcz in file tcc.tcz.dep not found
dependency xtables-addons-6.1.2-tinycore64.tcz in file xt_geoip_LE_IPv4.tcz.dep not found

Cheers.

[Rich]

Hi FlyingDutchman
Perl extensions are version specific and were not copied over because perl was recompiled.
gst-plugins-good-dev.tcz and gst-plugins-ugly-dev.tcz were both last seen in TC10-x86_64.
enchant-dev.tcz likely needs to be changed to enchant2-dev.tcz.
icu67-dev.tcz is listed in js52-dev.tcz.dep and js78-dev.tcz.dep.
python-olefile.tcz was last seen in TC11-x86_64.
linux-5.10_api_headers.tcz should now be linux-6.1_api_headers.tcz. FIXED
xtables-addons-5.10.3-tinycore64.tcz was last seen in TC12-x86_64.

Thank you for reporting this.

[Rich]

Hi nick65go
I did some more work on the ANSI color escape sequence patch.
I have not posted it to the repo yet.

The person that created the patch didn't do a good job on the
color handling, so I thought I'd try to fix that.

I displayed the pre-defined normal and bright colors in aterm, took
a screenshot, and used gimp2 to get their RGB values.

Then I wrote a new routine to handle escape sequences that
match a terminal as closely as possible.

The text colors now display correctly:


Text with default, white, and black background:


This one should be self explanatory:


Supported escape codes include:
8 normal colors for foreground and background.
Enable bold text, selects bright colors for foreground.
Disable bold text, selects normal colors for foreground.
Enable underlined text.
Disable underlined text.
Reset all settings to default values.

Differences between terminal and fltk behavior are:
terminal:
foreground defaults to white.
background defaults to black.
escape sequences remain in effect until another escape sequence changes something.

fltk:
foreground defaults to black.
background defaults to a light blue.
escape sequences get reset to default values when a newline character is encountered.

[adb014]

I noticed in the file http://tinycorelinux.net/14.x/x86/release/src/busybox/compile_busybox the comment "need to compile in tc-11.x as glibc >= 2.32 does not support rpc".. Wouldn't it be better to either

1) Remove support for NFS in tinycore and no longer need RPC at all, or
2) link busybox against libtirpc.so

Have the shell compiled on a different version of the glibc seems to me to be a horrible hack

[alphons]

1) Remove support for NFS in tinycore and no longer need RPC at all, or

I agree to remove the NFS mount option from busybox, also the more modern SMB (CIFS) option is enabled, should be a good alternative for most file sharing options.

Code: [Select]

sed -i 's/CONFIG_FEATURE_MOUNT_NFS=y/CONFIG_FEATURE_MOUNT_NFS=n/' busybox-1.36.0_config_suid


[Rich]

Hi alphons
That would break the  NFS  option in  /etc/init.d/tc-config.

[nick65go]

@alphons: You test TC in VMware (not free software), but I use poor-man Qemu (free software).
- what you MEASURABLLY gain (in speed, size) using SMB versus NFS ?
- I think qemu NBD (network block device) allows loops still fast enough. YMMV.

[adb014]

If NFS is needed, busybox can be linked against libtirpc.so.. If you want a self-contained binary link it against libtirpc.a

[adb014]

There is also at least 3 CVE with a CVSS of 7.5 in glibc (https://www.cvedetails.com/vulnerability-list/vendor_id-72/product_id-767/GNU-Glibc.html) impacting glibc prior to 2.32.. So this older glibc is already a problem, and a problem that is only going to get worse