TazPuppy 5.0 rc2

Message

mistfire · #796 Post by **mistfire** » Tue 12 Feb 2019, 02:11

@thinkpadfreak
On my latest experiment I modified both initrd and sfs_load. It can now track the kernel modules from other sfs modules other than zdrv sfs file. If there are external kernel modules for a specific kernel version found. It will trigger the depmod. The boot time is now much faster on subsequent boot

mistfire · #797 Post by **mistfire** » Tue 12 Feb 2019, 10:53

I sucessfully modified the rc.update and sfs_load scripts which I add auditing of gdk-pixbuf loaders, gconv modules, glib schemas, fonts, hi-color icons, gio modules, gtk immodules, pango modules, mimetypes (The audited info always stored at /var/filecount) This will determine when they are need to generate cache file by comparing the number of files on last session compared to current session instead of generating them in every boot. This will make the boot time even faster.

mistfire · #798 Post by **mistfire** » Tue 12 Feb 2019, 11:00

I also created a tool named tazcook-sfs which creates sfs modules for an app very easily. It uses the capabilities of tazpkg, chroot, and aufs.

All it need was an internet connection. run the script then install apps using tazpkg interactive mode (tazpkg will run at sandbox environment) after installing apps just exit to the tazpkg shell and it will packed it the installed program as SFS modules. This will create an SFS module with high accuracy.

thinkpadfreak · #799 Post by **thinkpadfreak** » Tue 12 Feb 2019, 11:14

I noticed that no sfses are listed in the right pane of bootmanager though the extra sfses are listed in sfs_load. There seems to be some inconsistency in beta 20.

I am looking forward to improvements in the next release.

mistfire · #800 Post by **mistfire** » Tue 12 Feb 2019, 13:17

@thinkpad I studied the bootmanager script. The problems lies only on Xdialog program. The script actually works on puppy. By the way I will try to redirect it on sfs_load script

mistfire · #801 Post by **mistfire** » Wed 13 Feb 2019, 02:11

TazPuppy Beta 21 released

Changes:
* Improved modular design
* Faster boot when savefile is loaded subsequently
* Some fixes

New tool added in TazPuppy: tazcook-sfs
(Located at Menu>System Tools>Make SFS module package)
Create SFS module version of an app easily and accurately (requires internet connection) Just specify where to save the sfs modules, install packages using tazpkg shell. Exit the tazpkg shell and it will start creating sfs modules. It allows to install many apps as long the user does not quit tazpkg shell

NOTE: When SFS module created by tazcook-sfs is loaded in TazPuppy. The tazpkg sees it as installed an package but not visible for updating package.

Download: https://drive.google.com/file/d/1ndMJCN ... sp=sharing
MD5 Checksum: 89ff398b77575e5f2c19454d76798888

Build kit: https://drive.google.com/file/d/1hdGmNf ... sp=sharing

thinkpadfreak · #802 Post by **thinkpadfreak** » Wed 13 Feb 2019, 16:58

Hello.
I have installed beta 21. I have tried loading 5 extra sfses as I did before.

Part of boot.log

Code: Select all

zram started
Starting pup_sfs_load...
sfs_load: Loading '/initrd/mnt/dev_save/OpenOffice415-ja-i386.sfs'...
umount: can't unmount /initrd/pup_z: Device or resource busy
losetup-FULL: /initrd/mnt/dev_save/OpenOffice415-ja-i386.sfs: failed to set up loop device: Device or resource busy

sfs_load: Do not assosiate /initrd/mnt/dev_save/OpenOffice415-ja-i386.sfs to /dev/loop4 but /dev/loop2.
Starting httpd...

After loading openoffice (sfs), I see the message "next boot will be faster!" at every boot.
Indeed, the message disappears sooner than before, but sfs_load is changing the mount point of openoffice (sfs) from /dev/loop4 to /dev/loop2. The boot.log says, "losetup-FULL."
I am afraid that this means the 5th extra sfs (openoffice) exceeds the number of loadable sfses.

So I have to conclude that it is safer to limit the number of extra sfses to 4 at most, in which case the message "next boot will be faster!" does not appear.
I think that there is nothing wrong with the openoffice (sfs), because it does not cause any trouble on other Puppies.

mistfire · #803 Post by **mistfire** » Thu 14 Feb 2019, 02:20

@thinkpadfreak losetup-FULL is just command. It means that it was a full version losetup is used (compiled executable) not losetup from busybox. By the way does your open office sfs sucessfully loaded upon boot?

thinkpadfreak · #804 Post by **thinkpadfreak** » Thu 14 Feb 2019, 02:45

mistfire wrote:
> By the way does your open office sfs sucessfully loaded upon boot?

What I did is as follows:
First, I loaded openoffice sfs as the 5th extra sfs by means of sfs_load.
(It took much longer than when I loaded other sfses.)
Openoffice started normally.
I rebooted tazpuppy. The message "next boot will be faster!" appeared.
(After that, I see the message at every boot.)

So openoffice itself works fine. But the behavior of tazpuppy when it loads extra sfses is different from those of other Puppies.

I wonder if other users of tazpuppy do not load as many as 5 or more extra sfses. It seems that some more tests are needed about this issue.
(As I stated before, it is my understanding that modern Puppies can load up to 6 extra sfses safely.)

Terry H · #805 Post by **Terry H** » Thu 14 Feb 2019, 04:55

thinkpadfreak wrote:mistfire wrote:
> if a certain loop device number is already use. It will automatically search for available loop device number which is not in use.

I wonder why zdrv must be mounted at /dev/loop4, instead of /dev/loop2 or dev/loop3. If /dev/loop4 is not in use, openoffice (sfs) will use it, so that depmod will not be run.

The fact that fdrv is listed in sfs_load is also strange. In case of UpupBB, for example, fdrv is not listed as an extra sfs.

(I do not know much about the mechanism of puppy, so what i am saying may be pointless.)

To eliminate the fdrv, have you thought about eliminating the fdrv and manually extracting the required firmware, which can be copied to /lib/firmware.

thinkpadfreak · #806 Post by **thinkpadfreak** » Thu 14 Feb 2019, 06:47

Terry H wrote:
> To eliminate the fdrv, have you thought about eliminating the fdrv and manually extracting the required firmware, which can be copied to /lib/firmware.

Thank you for the advice.
I tried eliminating fdrv by merging fdrv and zdrv to make a new zdrv.
Naturally, fdrv is not listed in sfs_load any more.

But still, the message "next boot will be faster!" appears at every boot, and the 5th extra sfs is mounted at /dev/loop2.
The extra sfses are mounted at /dev/loop6 ~ /devloop9 and /dev/loop2.
/dev/loop5 is not in use, probably because it is reserved for fdrv.

It is clear that tazpuppy behaves differently from other Puppies. If it is the way tazpuppy should be, I will give up further discussion about this issue.

mistfire · #807 Post by **mistfire** » Thu 14 Feb 2019, 07:23

@thinkpadfreak I reviewed to init code of initrd which was responsible of loading sfs files prior to execution of sfs_load. According to the comments of init script. 6 sfs files is the limit that can be only loaded when the puppy boots up because the performance degrades as the number of sfs file loaded increase.

thinkpadfreak · #808 Post by **thinkpadfreak** » Thu 14 Feb 2019, 08:34

mistfire wrote:
> According to the comments of init script. 6 sfs files is the limit that can be only loaded when the puppy boots up because the performance degrades as the number of sfs file loaded increase.

I wonder if the number 6 means the number including adrv, fdrv, etc.
UpupBB has adrv, fdrv, ydrv and zdrv. If we take main sfs and save file into account, they already make 6. But, of course, UpupBB can load extra sfses.

I also remember that older Puppies were able to load up to 3 extra sfses.

mistfire · #809 Post by **mistfire** » Thu 14 Feb 2019, 10:32

@thinkpad the number excludes adrv, fdrv, ydrv, and zdrv. yes its possible by editing the initrd. You need to modify the init script then create more pup_ro folders in initrd filesystem.

thinkpadfreak · #810 Post by **thinkpadfreak** » Thu 14 Feb 2019, 13:19

mistfire wrote:
> yes its possible by editing the initrd. You need to modify the init script then create more pup_ro folders in initrd filesystem.

If tazpuppy does not support 6 extra sfses, I will use it as it is. I am not going to go so far as to edit the initrd.

I thank you for making boot faster. However, I do not want to see the message "next boot will be faster!" at every boot.

I have unloaded devx sfs. As a result, the extra sfses are 4 in total, and tazpuppy boots smoothly.

AndresC2 · #811 Post by **AndresC2** » Fri 15 Feb 2019, 00:06

Hi ! mistfire

no work in with syslinux

booted from microsd with adapter

fatal error

puppy_tazpup_5.0.sfs not found

label TazPuppy
linux /Tazpuppy/vmlinuz
initrd /Tazpuppy/initrd.xz
append autologin pmedia=usbflash psubdir=/Tazpuppy
#append rootfstype=ramfs
menu label TazPuppy
text help
Start TazPuppy normally.
endtext

linux /Tazpuppy/vmlinuz load FINE
initrd /Tazpuppy/initrd.xz load FINE
but no puppy_tazpup_5.0.sfs

puppy have boot code like tiny core usbwait=5

thanks

mistfire · #812 Post by **mistfire** » Fri 15 Feb 2019, 05:18

@AndresC2 can you please send me the pupdebug file?

When puppy_tazpup_5.0.sfs not found fatal error message reached. You have option to save the temp file in your storage device. If you save it a tarball named pupdebug is created and you need to send it here.

mistfire · #813 Post by **mistfire** » Fri 15 Feb 2019, 05:20

@thinkpadfreak on my latest experiment I modified the initrd which enable to load up to 7 extra sfs modules. By default, only 4 extra sfs files was loaded by the initrd

AndresC2 · #814 Post by **AndresC2** » Fri 15 Feb 2019, 06:03

hi mistfire this is the file

mistfire · #815 Post by **mistfire** » Fri 15 Feb 2019, 07:31

@AndresC2 based on your log. Your bootable sd card on usb card adapter cant found. For the meantime try to use a usb flash drive instead of sd card adapter for boot

(old)Puppy Linux Discussion Forum

(old)Puppy Linux Discussion Forum

TazPuppy 5.0 rc2