WelcomeWelcome | FAQFAQ | DownloadsDownloads | WikiWiki

Author Topic: Core v7.0beta1  (Read 5613 times)

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #15 on: January 09, 2016, 07:23:35 PM »
The mouse doesn't work in a VMware VM when booting from the ISO image. This appears to be an issue with the Xvesa extension. I can verify that it is available to the system by 'cat'ing /dev/input/event2 in the console, but specifiying the path using startx -- -mouse doesn't have any effect. I have tested this with VMware versions 9 with Nvidia host graphics and version 12 with ATI host graphics. I can't tell if the keyboard works while Xvesa is running, but it does work in the console. There is no X log to give any hints. I ran tc-install.sh from the command line and after rebooting ran tce-load -iw Xorg-7.7-3d to get to a working system so this confirms that the mouse does work with the Xorg extension in the same VM.

Offline Juanito

  • Administrator
  • Hero Member
  • *****
  • Posts: 9009
Re: Core v7.0beta1
« Reply #16 on: January 09, 2016, 09:47:23 PM »
qemu-common.tcz.dep references libcap-ng.tcz which is not present
libcap-ng copied to tc-7.x repo - thanks.

Quote
SDL.tcs.dep references libmad.tcz which is not present.
Maybe I'm missing something, but libmad looks to be present in the tc-7.x repo to me?

Quote
Looks like xautolock-2.2 from the 4.x repo works in 7.x.  I suppose it would work in 5.x and 6.x, too.
xautolock-2.2 copied to tc-5.x, 6.x and 7.x repos - thanks

Offline Juanito

  • Administrator
  • Hero Member
  • *****
  • Posts: 9009
Re: Core v7.0beta1
« Reply #17 on: January 09, 2016, 09:49:07 PM »
The mouse doesn't work in a VMware VM when booting from the ISO image. This appears to be an issue with the Xvesa extension.

Is vmware meant to work with Xvesa?

Does vmware work with Xvesa in tc-6.x?

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #18 on: January 09, 2016, 10:20:32 PM »
Since I don't recall having this same problem with TC 6.4.1, I'll say yes it worked. But now I have another problem. Open-vm-tools won't compile with gcc 5. Something about the new C++ ABI. I've tried setting the "old behavior" macros without success. In the process of building extensions for this new version, did this problem come up and was it successfully resolved?

Offline Juanito

  • Administrator
  • Hero Member
  • *****
  • Posts: 9009
Re: Core v7.0beta1
« Reply #19 on: January 09, 2016, 10:25:24 PM »
gcc was compiled to support the old and new libstdc++ abi - I suppose the default is to compile for the new abi, but I haven't compiled any extensions yet where this came up...

Offline curaga

  • Administrator
  • Hero Member
  • *****
  • Posts: 9390
Re: Core v7.0beta1
« Reply #20 on: January 10, 2016, 02:38:46 AM »
The only barriers that can stop you are the ones you create yourself.

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #21 on: January 10, 2016, 12:22:37 PM »
I'm building open-vm-tools 10.x but the patch is for 9.x. Still there were a few hints in the patch that I could use so now I have the extension built and mostly working so thanks for the link. The problem at hand is that open-vm-tools 10.x now uses fuse for hgfs instead of a kernel module. This is good in that now open-vm-tools won't have any modules and should be a little more TC version independent. The bad part is that for a user to mount a shared folder /dev/fuse needs to have something other than 1600 permissions, like 1666. What can I do different (other than sudo) to mount the share without chmod'ing /dev/fuse?

Offline curaga

  • Administrator
  • Hero Member
  • *****
  • Posts: 9390
Re: Core v7.0beta1
« Reply #22 on: January 10, 2016, 12:51:48 PM »
The only other way is to have udev do the chmod, which requires remastering the udev rules (backup happens too late).
The only barriers that can stop you are the ones you create yourself.

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #23 on: January 10, 2016, 01:09:16 PM »
Maybe that's something for beta 2?  ;D

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #24 on: January 10, 2016, 01:34:49 PM »
The udev rules for fuse have it as 0666 but the file comes from the fuse extension. It looks like the fuse module is built into the kernel? If so shouldn't the udev rule already be in the master core.gz?

Offline curaga

  • Administrator
  • Hero Member
  • *****
  • Posts: 9390
Re: Core v7.0beta1
« Reply #25 on: January 11, 2016, 01:48:29 AM »
There is a udev rule for fuse in the base? Perhaps it needs to be removed from the extension then. I don't use fuse, would you like to investigate?
The only barriers that can stop you are the ones you create yourself.

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #26 on: January 11, 2016, 05:25:41 AM »
The udev rule is a link to the extension. I think it should be in the base if fuse is compiled into the kernel. I can't find the kernel config file on this website and there is no /proc/config.gz or module in an extension to create it so I can't say for sure. My hunch is based on fuse being in /proc/filesystems but there is no module in /proc/modules for it.

Offline Juanito

  • Administrator
  • Hero Member
  • *****
  • Posts: 9009
Re: Core v7.0beta1
« Reply #27 on: January 11, 2016, 05:40:15 AM »
The relevant rules in the base are:
Code: [Select]
$ grep fuse /etc/udev/rules.d/*rules
/etc/udev/rules.d/50-udev-default.rules:KERNEL=="fuse", ACTION=="add",  MODE="0666", OPTIONS+="static_node=fuse", \
/etc/udev/rules.d/50-udev-default.rules:  RUN+="/bin/mount -t fusectl fusectl /sys/fs/fuse/connections"

The rule in the fuse extension is:
Code: [Select]
$ cat /usr/local/share/fuse/files/*fuse*
KERNEL=="fuse", MODE="0666"

The kernel config is at:

http://tinycorelinux.net/7.x/x86/release/src/kernel/

..not sure what you mean by "the udev rule is a link to the extension?

Offline andyj

  • Sr. Member
  • ****
  • Posts: 375
Re: Core v7.0beta1
« Reply #28 on: January 11, 2016, 06:13:48 AM »
Thanks for the kernel config. It's post time is 6 and a half hours ahead of my current local time, so it's like looking into the future!  ;D  I'll need to spin up a clean VM to investigate the fuse issue more.

Offline Juanito

  • Administrator
  • Hero Member
  • *****
  • Posts: 9009
Re: Core v7.0beta1
« Reply #29 on: January 13, 2016, 11:34:37 PM »
gold linker functionality added to binutils/gcc

To use:
Code: [Select]
CC="gcc -fuse-ld=gold -flto -fuse-linker-plugin -march=i486 -mtune=i686 -Os -pipe" CXX="g++ -fuse-ld=gold -flto -fuse-linker-plugin -march=i486 -mtune=i686 -Os -pipe -fno-exceptions -fno-rtti" ./configure --blah-blah