Hardinfo crashes during processing the Benchmarks.

Bug #178267 reported by moma
10
Affects Status Importance Assigned to Milestone
hardinfo (Debian)
Fix Released
Unknown
hardinfo (Ubuntu)
Fix Released
Medium
Unassigned
Gutsy
Won't Fix
Medium
Unassigned
Hardy
Fix Released
Medium
Unassigned

Bug Description

Hello,

This installation is: 32 bits Ubuntu 7.10.

$ lsb_release -a
Distributor ID: Ubuntu
Description: Ubuntu 7.10
Release: 7.10
Codename: gutsy
----------------------

I have installed the "hardinfo" program from the Ubuntu's repository.

$ apt-cache search hardinfo
hardinfo - Displays system information

$ apt-cache show hardinfo
Package: hardinfo
Priority: optional
Section: universe/x11
Installed-Size: 616
Maintainer: Ubuntu MOTU Developers <email address hidden>
Original-Maintainer: Agney Lopes Roth Ferraz <email address hidden>
Architecture: i386
Version: 0.4.2.1.svn30042007-1
------------

1) Start the hardinfo application.

2) Select Benchmarks -> CPU Fibonacci (or CPU MD5 or CPU SHA1 ). I do not touch the other options in the tree list.
I move the hardinfo's window a lot while it calculates the benchmarks.
The program crashes.

The window movement has probably nothing todo with the crash. Sometimes it crashes while not moving the window. I run Compiz with the Nvidias latest (newest) driver 169.7. (manually installed driver).

------------
An output from the hardinfo on a terminal window at the crash.
See: http://www.edbl.no/tmp/hardinfo_test1.txt

-------------
I've tried to run it via Valgrind but I cannot re-create the crash by that. It won't abort.
$ G_SLICE=always-malloc G_DEBUG=gc-friendly valgrind -v --tool=memcheck --leak-check=full --num-callers=40 --log-file=valgrind.log hardinfo

-------------
This machine's system spesification:
See: http://www.edbl.no/tmp/system_speck1.txt

Revision history for this message
moma (osmoma) wrote :

I have now done gdb backtrace as instructed on https://wiki.ubuntu.com/Backtrace

Revision history for this message
moma (osmoma) wrote :

Second backtrace

(gdb) Program received signal SIGABRT, Aborted.
[Switching to Thread -1221794128 (LWP 6139)]
0xffffe410 in __kernel_vsyscall ()

(gdb) backtrace full
(gdb) info registers
(gdb) thread apply all backtrace
(gdb) quit

Revision history for this message
Cesare Tirabassi (norsetto) wrote :

Looks very similar to this bug reported in Debian:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=445585

I can see that there are a lot of crashes reported in this version of hardinfo, and I could reproduce some myself with my machine.
I tried the version in hardy and it seems that most of them have been solved.

Revision history for this message
Cesare Tirabassi (norsetto) wrote :

As I said, there are quite a number of crashes for the gutsy version of hardinfo:

https://bugs.launchpad.net/ubuntu/+source/hardinfo

Most of these have been resolved by the issue currently in hardy:

http://bugs.debian.org/cgi-bin/pkgreport.cgi?which=pkg&data=hardinfo&archive=no&version=&dist=unstable

It therefore makes a lot of sense to me to backport this to gutsy. To this end, I have built two test packages in my ppa which are available for testing. They are available at the following links:

i386: http://ppa.launchpad.net/norsetto/ubuntu/pool/main/h/hardinfo/hardinfo_0.4.2.3-1~ppa1_i386.deb
amd64: http://ppa.launchpad.net/norsetto/ubuntu/pool/main/h/hardinfo/hardinfo_0.4.2.3-1~ppa1_amd64.deb

You can download and then install them with gdebi or dpkg -i.

Please report here if they solve your problem or if you have any other problem so that we can decide if to do this backport.

Changed in hardinfo:
importance: Undecided → Medium
status: New → Confirmed
Changed in hardinfo:
status: Unknown → Fix Released
Revision history for this message
moma (osmoma) wrote :

Thanks for your reply.

I have now done new tests with the following results.

1) Test of hardinfo on Hardy Heron Alpha 2:
The program works generally very well. I cannot make it crash.

1a) However, the CPU ZLib test fails with an error message. It says "Cannot load ZLib. /usr/lib/libz.so not found".
I suggest that when the hardinfo is installed from the repo it also pulls down the Zlib as dependency. I think the package name is zlib1g (is it right?).

Study the picture: http://www.edbl.no/tmp/hardinfo-error-2.png

1b) There are also other minor issues with the presented data that could be improved.
The issues are:

The Devices -> Sensors should give proper name to the sensors.
Now it just reports fan1, fan2,... and temp1, temp2, temp3 which are meaningless names.
Study the picture: http://www.edbl.no/tmp/hardinfo-improve-temperature-info.png

I know that lm-sensors (and CLI programs: sensors and sensors-detect) can distinguish between the various sensors and give them right names.
Also GNOMEs "Hardware Sensor Monitor" applet shows correct sensor names.
Study the picture: http://bildr.no/view/134268 (properties of "Hardware Sensor Monitor" applet in GNOME, taken from my Hardy Heron installation)
-----------------

2) Test of updgraded hardinfo on Ubuntu 7.10 gutsy.
I removed the old hardinfo (apt-get remove) and installed the upgraded version provided by Cesare Tirabassi
http://ppa.launchpad.net/norsetto/ubuntu/pool/main/h/hardinfo/hardinfo_0.4.2.3-1~ppa1_i386.deb

This hardinfo version works very well and it didn't crash anymore.

But note the matter listed in 1b).
----------

Revision history for this message
moma (osmoma) wrote :

This is an addition to point 1a) of my previous message.
The CPU ZLib test failed with an error message "Cannot load ZLib. /usr/lib/libz.so not found".

So what package provides libz.so?
$ dpkg -S libz.so
zlib1g: /usr/lib/libz.so.1
zlib1g: /usr/lib/libz.so.1.2.3.3

I had already installed the "zlib1g" package so something else had to be wrong. The symbolic link to "libz.so" was missing.
$ ls -l /usr/lib/libz.*
lrwxrwxrwx 1 root root 15 2007-12-22 21:24 /usr/lib/libz.so.1 -> libz.so.1.2.3.3
-rw-r--r-- 1 root root 81240 2007-11-15 13:08 /usr/lib/libz.so.1.2.3.3

This command fixed it.
$ sudo ln -s /usr/lib/libz.so.1 /usr/lib/libz.so

( In some rare cases I've seen that the lib links are not created until you install some additional packages such as dev packages).

And the CPU ZLib test completed just fine
My report: http://www.futuredesktop.org/tmp/hardinfo_report.html
----

Revision history for this message
Cesare Tirabassi (norsetto) wrote :

Thanks for the report.
Both issues 1a and 1b are correct but they should be reported in two other different and separated bug reports.

1a) The problem is that the wrong library is linked in arch/common/zlib.h:

libz = g_module_open("/usr/lib/libz.so", G_MODULE_BIND_LAZY);

the right soname is libz.so.1 so, unless you happen to have the development package installed (as myself), you get the error that you report.

1b) We can pass the suggestion upstream.

Can you please open two new bug reports for these?

Changed in hardinfo:
status: Confirmed → Fix Released
Revision history for this message
Cesare Tirabassi (norsetto) wrote :

Subscribing backport team.
Can we backport hardinfo 0.4.2.3-1 from hardy to gutsy?
It builds fine in a gutsy environment (both pbuilder and ppa) and has no reverse dependencies,

Changed in hardinfo:
importance: Undecided → Medium
status: New → Triaged
Revision history for this message
Hugo Nascimento (hadn) wrote :

How is the backporting of hardinfo from hardy to gutsy going? The current version of hardinfo in gutsy is useless.

Package http://ppa.launchpad.net/norsetto/ubuntu/pool/main/h/hardinfo/hardinfo_0.4.2.3-1~ppa1_i386.deb seems to be a good intermediate solution, before its final version is ready.

Revision history for this message
Cesare Tirabassi (norsetto) wrote :

Subscribing jdong since this was sitting in the backporters queue for the best of two months.

Revision history for this message
Sergio Zanchetta (primes2h) wrote :

The 18 month support period for Gutsy Gibbon 7.10 has reached its end of life -
http://www.ubuntu.com/news/ubuntu-7.10-eol . As a result, we are closing the
Gutsy task.

Changed in hardinfo (Ubuntu Gutsy):
status: Triaged → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.