Page 2 of 4

Re: Explorer's modification of CPUstress image

Posted: Fri Mar 01, 2013 4:40 am
by Explorer09
ady wrote: ......
, since Victor already uses a similar format in UBCD as I am suggesting here for your help files (it is not a direct copy from your help files). My suggestion was (is) for convenience and easier reading in your help files (there wouldn't be a need to clarify the artificial use of an extra line for 'append').
I'll update the help files for this, plus other code cleanups.

But, I don't know whether I should wait for BusyBox 1.21.1 bug fix release as well. I was hoping I could update BusyBox in the next version of CPUstress image, because it has an important bug fix for me :) .

Re: Explorer's modification of CPUstress image

Posted: Fri Mar 01, 2013 12:37 pm
by Victor Chew
Feedback from ady via PM:
I don't want to post as off-topic in Explorer's CPUStress dev. topic, and I have some doubts (not necessarilly reports) regarding 52b5.

In the blacklist.sed file, there are two consecutive lines with "$!N". Is this OK?

***

In the pmagic.sed file, there are two consecutive lines with "$!N". Is this OK?


***

In linux.sed, the example:
# LINUX /ubcd/boot/cpustres/bzImage
might be better as
# LINUX /pmagic/bzImage
to reflect the real path in use in UBCD? (I'm not sure which path is best as documented example.)

In linux.sed, are the new sed replacement lines (that start with "s|") requiring the key words to be the initial word of the line in the cfg files (without any other character before, not even spaces)? This requirement should be met. I am asking because I don't know whether this sed code actually is checking this or not. This requirement is the reason to put all “text help” with one initial space on every line.

***

I was not successful when I tried to run syslinux2grub4dos.cmd (Windows script). I am not sure if the problem is in my system or in the recent changes (I might try to find out in the next few days). This failure is one of the reasons, among others, for the above questions.

***
Please change the line endings of lst files back to Windows style, as they used to be.
***

Re: Explorer's modification of CPUstress image

Posted: Fri Mar 01, 2013 1:11 pm
by Victor Chew
In the blacklist.sed/pmagic.sed file, there are two consecutive lines with "$!N". Is this OK?
That was supposed to remove the consecutive newlines after masking out the blacklisted entries (by removing the newline after the target entry). But I have removed that after finding a better way of doing it in syslinux2grub4dos.sh.
In linux.sed, the example:
# LINUX /ubcd/boot/cpustres/bzImage
might be better as
# LINUX /pmagic/bzImage
to reflect the real path in use in UBCD? (I'm not sure which path is best as documented example.)
Changed.
In linux.sed, are the new sed replacement lines (that start with "s|") requiring the key words to be the initial word of the line in the cfg files (without any other character before, not even spaces)? This requirement should be met. I am asking because I don't know whether this sed code actually is checking this or not. This requirement is the reason to put all “text help” with one initial space on every line.
Fixed.
I was not successful when I tried to run syslinux2grub4dos.cmd (Windows script). I am not sure if the problem is in my system or in the recent changes (I might try to find out in the next few days). This failure is one of the reasons, among others, for the above questions.
What is the error? I run the scripts on Win7 without any problems. If there were problems, I wouldn't be able to build the ISO image at all!

Re: Explorer's modification of CPUstress image

Posted: Sat Mar 09, 2013 2:51 am
by Explorer09
It seems that I can't wait for the BusyBox release, so here's a minor release of my CPUstress image.

CPUstress v2.1expl7 (9 March 2013)
http://dl.dropbox.com/u/70170658/ubcd/c ... .1expl7.7z
MD5: 481afd31a3774f00f7ffad67d4698a57
SHA1: f6f3780cfd901abd2d6682e137194ceeb28a97c9
  • Changed "help" command into a wrapper script.
  • Improved build scripts.
  • Some cleanup to the code and documentation.
  • Use INITRD directive in isolinux entries in the documentation.
No programs are updated, but only scripts and documentation.

Re: Explorer's modification of CPUstress image

Posted: Mon Mar 11, 2013 2:43 pm
by Victor Chew
CPUstress v2.1expl7 (9 March 2013)
Noted. Thanks!

Re: Explorer's modification of CPUstress image

Posted: Sun Mar 31, 2013 1:34 am
by Explorer09
Hi everyone.
Here's a testing release for those who has a USB keyboard and multicore CPUs.
If there are no major problems, I'll make this one final.

CPUstress v2.2 beta2 (31 March 2013)
http://dl.dropbox.com/u/70170658/ubcd/c ... .2beta2.7z
Size: 4907602 bytes
MD5: 2b5b282eeedc32d366f5977437f659ef
SHA1: 629a93f8b7f1cbb99ee948849e3ac70c80d1715d

Updated programs:
  • CPUstress now uses a kernel built by myself rather than the one from Parted Magic. The custom kernel (3.8.5-cpustress) fixes two bugs below and is more friendly to old processors (i486). It supports SMP but not PAE.
  • eglibc updated to 2.17-r22665 (from 2.17-r22243).
  • BusyBox is rebuilt (same version). Now "modprobe", "lsmod", and "uname" are added.
Fixes:
  • The "USB keyboard not found" bug due to missing drivers. (viewtopic.php?t=14648)
  • Regression that only 1 CPU can be detected. (viewtopic.php?t=14646)
  • Edited "help" and "helpinfo" scripts to fix a recursive calling bug.
Changes:
  • The new eglibc now require Linux kernel 3.7.5 or above. (Parted Magic 2013_01_29 has such kernel.)
  • Add /etc/mdev.conf which allows loading modules on hotplug.
Notes:
  • I do not test the lockup problem (viewtopic.php?t=3390) that happened in the old kernel, so I'm not sure whether it was fixed or not. If anyone has the hardware and can test it, then I appreciate for your help.
  • Since Linux kernel 3.8, the i386 architecture is no longer supported. If you have 80386 you'll probably need to build a 3.7.x Linux kernel.

Re: Explorer's modification of CPUstress image

Posted: Mon Apr 01, 2013 8:10 pm
by Explorer09
Hope this one will be the last beta.

CPUstress v2.2 beta3 (2 April 2013)
http://dl.dropbox.com/u/70170658/ubcd/c ... .2beta3.7z
Size: 5127383 bytes
MD5: c307d5af5260c33a6866c9845c607418
SHA1: 0f668a78f5848f83f5f8669cd5d624de61df3fba
  • Linux kernel is rebuilt (3.8.5-cpustress2), which fixes a bug below and adds minor improvements.
  • stress is rebuilt (same version), but with hard-disk stressing disabled.
Fixes:
  • For some computers, the kernel might have detected only one CPU when ACPI support is disabled. (This happened on my Fujitsu laptop.) Now ACPI support is added in the kernel.
  • Fix the "Kernel args:" field output in the "/runonce" script.
Changes:
  • The 'ubcdcmd' boot option now accepts dmesg, lsmod, and uname for debugging use.
  • Adds the 'modprobe' boot option allowing the user to load modules.

Re: Explorer's modification of CPUstress image

Posted: Sun Apr 07, 2013 1:18 am
by Explorer09
Time to make this one final.

CPUstress v2.2 (6 April 2013)
http://dl.dropbox.com/u/70170658/ubcd/cpustress-2.2.7z
Size: 5123261 bytes
MD5: fb24b9911780ee4b84787ef8b1b6290e
SHA1: 63fbc62019acddbd0dcdd24ed53c55cde2d89bcd
  • Linux kernel updated to 3.8.6-cpustress. Build config is not changed.
  • Minor documentation changes.
@Victor Chew: Please update not only the kernel and initrd image, but also the text files. (That is, everything except build.txz) Because I've changed all references of '/pmagic/bzImage' to '/ubcd/boot/cpustress/bzImage' in them.

@Icecube: I'm requesting you permission to release this image under the name 'CPUstress'. If you're reading this, please answer me if you permit me or not. (If not, then I'll rename my project.) This question is to avoid copyright problems in the future.

Re: Explorer's modification of CPUstress image

Posted: Sun Apr 07, 2013 6:40 am
by Victor Chew
@Victor Chew: Please update not only the kernel and initrd image, but also the text files. (That is, everything except build.txz) Because I've changed all references of '/pmagic/bzImage' to '/ubcd/boot/cpustress/bzImage' in them.
Got it. Thanks!

Re: Explorer's modification of CPUstress image

Posted: Wed Apr 10, 2013 11:37 am
by Icecube
Explorer09 wrote:@Icecube: I'm requesting you permission to release this image under the name 'CPUstress'. If you're reading this, please answer me if you permit me or not. (If not, then I'll rename my project.) This question is to avoid copyright problems in the future.
You can use the same name.

I don't have time to look closely at it, but a small remark:

Because now CPUstress doesn't use a filesystem image anymore as initial ramdisk (now: initramfs = cpio archive), we don't and shouldn't use "root=/dev/ram0 ramdisk_size=12000" as parameter anymore.

Another thing that might be nice, is to have a 64-bit version (kernel and programs).
In the 64-bit version I would exclude old programs like cpuburn and mprime v2.3.

I also seemed to remember that someone wanted to be able to read the CPU temperature.
This should be possible with lm-sensors:
http://www.lm-sensors.org/

Re: Explorer's modification of CPUstress image

Posted: Wed Apr 10, 2013 4:48 pm
by Explorer09
Just a note to Victor,
the CPU Burn-in website has been move to here: http://cpuburnin.com/
So you might update the URL on the UBCD homepage.
Icecube wrote: You can use the same name.
Thank you!
Icecube wrote: Because now CPUstress doesn't use a filesystem image anymore as initial ramdisk (now: initramfs = cpio archive), we don't and shouldn't use "root=/dev/ram0 ramdisk_size=12000" as parameter anymore.
Got it! They will be removed in the next version.
Icecube wrote: Another thing that might be nice, is to have a 64-bit version (kernel and programs).
In the 64-bit version I would exclude old programs like cpuburn and mprime v2.3.
64-bit version will bring me too much work. For now I can't do that. Any volunteers?
Icecube wrote: I also seemed to remember that someone wanted to be able to read the CPU temperature.
This should be possible with lm-sensors:
http://www.lm-sensors.org/
I might need to configure the kernel for this. But it's a good feature.

Re: Explorer's modification of CPUstress image

Posted: Wed Apr 10, 2013 10:22 pm
by Victor Chew
Just a note to Victor,
the CPU Burn-in website has been move to here: http://cpuburnin.com/
Will update now. Thanks!

Re: Explorer's modification of CPUstress image

Posted: Thu Apr 11, 2013 12:37 am
by Icecube
Some other CPU stress tools that might be worth to add are listed at: https://wiki.archlinux.org/index.php/Stress_Test

Re: Explorer's modification of CPUstress image

Posted: Thu Apr 11, 2013 5:24 pm
by Explorer09
Now I have these in my todo list.
Convert build scripts into makefiles.

Add these programs if possible:

lm-sensors
- http://www.lm-sensors.org/

cpuburn by redeim
- http://web.archive.org/web/201101100031 ... et/redelm/
- http://web.archive.org/web/201101100211 ... _4a_tar.gz

System Stability Tester (It has GUI, so I must compile it CLI-only.)
- http://systester.sourceforge.net/

Intel MKL LINPACK
- http://software.intel.com/en-us/article ... k-download

Re: Explorer's modification of CPUstress image

Posted: Thu Apr 11, 2013 6:28 pm
by Explorer09
CPUstress 2.2.1 (12 April 2013)
https://dl.dropboxusercontent.com/u/701 ... s-2.2.1.7z
Size: 5008158 bytes
MD5: c110bde94d8a0d8c5388e08646dd1fbc
SHA1: 360ab5b8cd749205b9e25db7453ad0519f32c02e

Programs updated or rebuilt: Changes:
  • Bug fixes regarding '--help' option in various scripts.
  • Edit various shell scripts to improve compatibility to POSIX sh. Removed as many bashisms as possible.
  • Removed "root=/dev/ram0 ramdisk_size=12000" from the isolinux command-line. Documentation is also updated for this. (Thanks to Icecube for the suggestion.)
  • 'cpuburn' and 'mprime' script improvements. They now also trap signals 1 and 15. 'mprime' now supports command-line arguments.
Note:
  • The command-line arguments of mprime rarely do something automatically. Most of the arguments will bring user to the interactive menu. So there's not much use when specifying at boot time.
@Victor: Again, please also update the text files in this release. Also please remove "root=/dev/ram0 ramdisk_size=12000" from the entries in ubcd/menus/syslinux/cpu.cfg as Icecube said.

Re: Explorer's modification of CPUstress image

Posted: Sat Apr 13, 2013 12:49 am
by Victor Chew
Noted. Thanks!

Re: Explorer's modification of CPUstress image

Posted: Fri Apr 19, 2013 3:30 am
by Explorer09
CPUstress 2.3 (19 April 2013)
https://dl.dropboxusercontent.com/u/701 ... ess-2.3.7z
Size: 5913889 bytes
MD5: a47e6511c46c6f747592c734f87a0e39
SHA1: bf10e57d3069da3a9221891337a52708f130b2df

New programs: Programs updated or rebuilt:
  • Updated kernel to 3.8.8. Added drivers of I2C buses and hardware sensors. Disabled loadable module support in kernel.
  • BusyBox updated to 1.21.0. Added 'free', 'mpstat', 'ps', 'uptime', 'kill', 'date', 'cp', and 'vi'. 'echo' now becomes ash built-in. Removed 'modprobe' and 'lsmod'.
  • Added libgcc_s.so. This file is required by linpack.
  • Rebuilt eglibc (same version).
Changes:
  • Renamed 'cpuburn' wrapper to cpuburn-m.
  • The cpuburn-m and mprime wrappers are now conservative when killing processes. (They now kill by PID rather than by name.) They now also reap zombie processes.
  • Use "ubcdargs" instead of cmdhelpinfo, cmdstress, etc. in isolinux options.
  • Other minor bug fixes.
Note:
Please check your isolinux config files when you upgrade your CPUstress image to this version. There are 2 boot options incompatible with this new version:
  1. "ubcdcmd=cpuburn" no longer works. Use "ubcdcmd=cpuburn-in" instead.
  2. The options "cmdhelpinfo", "cmdstress", and "cmdstresscpu2" are obsolete. Use "ubcdargs" instead.

Re: Explorer's modification of CPUstress image

Posted: Fri Apr 19, 2013 3:54 am
by Explorer09
@Victor:
If you want to add syslinux menu entries for these new programs in UBCD, perhaps you could see this cpu.cfg as a sample.

And notice that the 'Stress' entry need to changed. If you read the note on my previous post, you'll know what I mean.
Before:

Code: Select all

APPEND noapic quiet ubcdcmd=stress cmdstress="--help"
After:

Code: Select all

APPEND noapic quiet ubcdcmd=stress ubcdargs="--help"
BTW, Stress and System Stability Tester are also included in Parted Magic.

Code: Select all

Name of Tool                       Version   License    Notes
----------------------------------+--------+----------+-----------------------------------------
Stress                            |  1.0.4 |   GPL    | Included in CPUstress and Parted Magic.
CPUburn                           |   1.4a |   GPL    | Included in CPUstress.
Intel Optimized LINPACK Benchmark | 11.0.3 | Freeware | Included in CPUstress.
System Stability Tester           |  1.5.1 |   GPL    | Included in CPUstress and Parted Magic.
----------------------------------+--------+----------+-----------------------------------------

Re: Explorer's modification of CPUstress image

Posted: Fri Apr 19, 2013 7:35 pm
by Explorer09
Oops, I left a small bug yesterday. So please use this one in UBCD instead.

CPUstress 2.3.1 (20 April 2013)
https://dl.dropboxusercontent.com/u/701 ... s-2.3.1.7z
Sizes: 5913923 bytes
MD5: 9bf60a2ecaec2feb0240fa3211722882
SHA1: 3b769a6d196866c5f02cfe58659d4257cbe88013

Changes:
Fix 'trap 1 2 3 15' bug in burn and cpuburn-m wrappers.

Re: Explorer's modification of CPUstress image

Posted: Tue Apr 23, 2013 7:07 pm
by Victor Chew
I have updated to CPUStress V2.3.1 as per your suggestions.

I didn't find an entry for Sensors, so I put one together.

Code: Select all

MENU LABEL Sensors V3.3.3
TEXT HELP
 Included in CPUstress. Provides essential tools for monitoring the hardware
 health of systems containing hardware health monitoring hardware such as LM78
 and LM75.
ENDTEXT
COM32 linux.c32 /ubcd/boot/cpustress/bzImage
INITRD /ubcd/boot/cpustress/initrd.gz
APPEND noapic quiet ubcdcmd=sensors
Please feel free to make corrections as you see fit.

Thanks!