Mageia forum

Hello

Tried to install Symphony 3.0.1 rpm on Mageia 2, but it aborts in the middle of the install process, due a memory corruption.

Since rpm install didn't work, I copied all the Symphony program files from a Mageia 1 partition. Trying to load the program crashes inmediately with the same memory corruption message:

*** glibc detected *** /opt/ibm/lotus/Symphony/framework/rcp/eclipse/plugins/com.ibm.rcp.base_6.2.3.20110915-1350/linux/x86/symphony: malloc(): memory corruption: 0x096f32d8 **

Anyone have an idea what's going on?

I would suspect that mageia 2 has jumped a few library versions and is now incompatible with your application.

It does happen, both ways in fact, i was trying to get the new version on bino to compile on mga1 but would not have it, to many dependancies, but thats a different subject.

what you need to do is see what is failing so ill talk your through manualy stepping through the rpm install.

first unbundle the rpm into a tempory directory with the following.

cd /var/tmp
rpm2cpio symphony.rpm|cpio -id

now extract the spec file from the rpm.

rpm --scripts -qp symphony.rpm > symphony.spec

have a look at the spec file.

it should have an entry like "postinstall scriptlet (using /bin/sh):"

thats what happens once the files are put down.

before you run any of those commands , if they are binaries (file filename) will tell you that

then check what librarys they depend upon.

ldd binaryfilename.

do the libraries all exist.

if so try to run it.

does it give the error that you have seen in the rpm -ivh command.

if so try "strace -f binarycommand"

theres going to be alot of stuff comming from this command, system calls, file open commands, etc.

but look around the time it fails.

any clues there. I can only guess, but if you paste the strace output of the lines around when the error happens we may be able to give a reaspn, or even google may help, it may be a generic library incompatibility issue.

regards peter

To be more pragmatic, why do you want to use Symphony if you can have LibreOffice or Apache OpenOffice?

Well, we include IBM Lotus Symphony in our customized Mageia distro for elementary schools to give them more options. Symphony is liked by many students because it has a more elegant interface than LibreOffice/OpenOffice, it has a good clipart (not as big as openclipart, but a useful one nevertheless) and it has an easier way to use data to produce merged documents. Free Software is also about having choices, right?

wintpe, I'm on travel right now, but on the weekend I'll do your suggested procedures, to see what's going on. Thanks in advance.

Wintpe, tried to do your suggested procedure. But the section postinstall scriptlet of Symphony is a very complex one, composed of hundreds of lines, with a lot of func procedures, args and params. I wouldn't know how to manually run the stuff.

Since I copied the files from a working Symphony installation in my Mageia 1 partition, did ldd on the symphony executable

ldd /opt/ibm/lotus/Symphony/framework/shared/eclipse/plugins/com.ibm.symphony.standard.launcher.linux.x86_3.0.1.20120110-2000/IBM\ Lotus\ Symphony
linux-gate.so.1 => (0xffffe000)
libc.so.6 => /lib/i686/libc.so.6 (0xb7637000)
/lib/ld-linux.so.2 (0xb77d0000)

So, nothing out of the common here.... who knows...

so now you have that installed, what happens when you try and run stuff,

if that crashes straight away, try the strace i suggested.

if that works then its going to be something in that postinstall thats at fault/

you could copy the post install, completly to a file, make it executable, and make the shell startup

(ie the #!/bin/sh part) have -x at the end

so the top line of the file is

#!/bin/sh -x

this will put the install script into debug mode and its output may shed a clue or two even if its a bit over your head, clues lead us to the problem, by observing the point at which it goes wrong.

regards peter

I did the strace on the Symphony executable, but around the memory corruption line I did not found anything that could be a clue on what's going on... And yes, it crashes straight away.

Code: Select all

5880  read(6, "Name:\tsymphony\nState:\tR (running"..., 1024) = 763
5880  close(6)                          = 0
5880  munmap(0xb76df000, 4096)          = 0
5880  sched_getaffinity(0, 512, {3})    = 4
5880  open("/sys/devices/system/cpu", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY|O_CLOEXEC) = 6
5880  getdents(6, /* 15 entries */, 32768) = 288
5880  getdents(6, /* 0 entries */, 32768) = 0
5880  close(6)                          = 0
5880  open("/sys/devices/system/cpu", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY|O_CLOEXEC) = 6
5880  getdents64(6, /* 15 entries */, 32768) = 448
5880  getdents64(6, /* 0 entries */, 32768) = 0
5880  close(6)                          = 0
5880  open("/proc/self/status", O_RDONLY) = 6
5880  fstat64(6, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0
5880  mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb76df000
5880  read(6, "Name:\tsymphony\nState:\tR (running"..., 1024) = 763
5880  read(6, "", 1024)                 = 0
5880  close(6)                          = 0
5880  munmap(0xb76df000, 4096)          = 0
5880  munmap(0xb70ac000, 106211)        = 0
5880  get_mempolicy(NULL, {}, 0, 0, 0)  = -1 ENOSYS (Function not implemented)
5880  open("/dev/tty", O_RDWR|O_NOCTTY|O_NONBLOCK) = 6
5880  writev(6, [{"*** glibc detected *** ", 23}, {"/opt/ibm/lotus/Symphony/framewor"..., 109}, {": ", 2}, {"malloc(): memory corruption", 27}, {": 0x", 4}, {"0980c2d8", 8}, {" ***\n", 5}], 7) = 178
5880  mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb76df000
5880  open("/opt/ibm/lotus/Symphony/framework/rcp/eclipse/plugins/com.ibm.rcp.j2se.linux.x86_1.6.0.20110713a-201108151128/jre/bin/classic/libgcc_s.so.1", O_RDONLY) = 7
5880  read(7, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0000\37\0\0004\0\0\0"..., 512) = 512
5880  futex(0xb76bd3e0, FUTEX_WAIT_PRIVATE, 2, NULL) = ? ERESTARTSYS (To be restarted)
5880  --- {si_signo=SIGINT, si_code=SI_KERNEL, si_value={int=4131212846, ptr=0xf63d4e2e}} (Interrupt) ---

Astroboy, would you please be so kind to place console-output between [CODE]-tags, to keep the forum readable.
You can do so be editing your post in "Full Editor"-mode.

Did the program produce a core file? at the point of the memory corruption?

Malloc is a c function call that allocates a chunk of memory in ram and returns a pointer to it.

your program then uses that memory to store data, a bit like an array,.

what may have happened here is an incompatibility with versions of glib, that have caused malloc to write over another area memory not allocated to it, hence the term memory corruption has occured.

I think the only way you are going to get this to perform in a safe way, is to sandbox it into a chroot environment, where it only has access to compatible libraries. That does not guarrentee that its not going to occur because theres a point at which malloc will make a call to the kernel routines, and its maybe here that it fails.
In fact the line from your strace supports this

futex(0xb76bd3e0, FUTEX_WAIT_PRIVATE, 2, NULL) = ? ERESTARTSYS (To be restarted)
5880 --- {si_signo=SIGINT, si_code=SI_KERNEL, si_value={int=4131212846, ptr=0xf63d4e2e}} (Interrupt) ---



It maybe thats theres a newer version of the product that may be compiled against newer libraries, if so it would be worth trying it.

regards peter

going back to doktor5000 suggestion, I must admit I ignored it as I thought lotus symphony was based on the defunct Corel suite.

I now see its based on openoffice.

Therefore I would suspect that you can instead port all the templates and art from it to the mageia version probably more easily then fixing this, as you are clearly already at its latest version.

regards peter

Or maybe first try Apache OpenOffice, as this is where IBM dedicated their symphony codebase to, as far as i know.

FWIW, please show the output of

Code: Select all

java -version

and the output of

Code: Select all

rpm -qa | grep java

Well, Apache OpenOffice 3.4 have minor features compared with the previous 3.3 version, and, in general, it still looks and feels like LibreOffice, making no sense to install both. As far as I know, Apache OpenOffice 4.0 will use the look and feel of what is now known as IBM Lotus Symphony. Of course, I'd like to use Mageia 2 and Symphony, but as there are some issue, maybe I'll have to wait for OpenOffice 4.0...

For the Java stuff, well, as you can see, I intentionally use the 1.6 version that comes with Mageia 1. The reason? it includes java-1.6.0-sun-plugin-1.6.0.26-0.2.mga1.nonfree (which lets me run java applets on a browser) and it is the one version that works just fine with other apps (including Symphony on Mageia 1).

Code: Select all

java -version
java version "1.6.0_26"
Java(TM) SE Runtime Environment (build 1.6.0_26-b03)
Java HotSpot(TM) Server VM (build 20.1-b02, mixed mode)

Code: Select all

rpm -qa | grep java
javamail-1.4.3-7.mga1
java-1.6.0-sun-alsa-1.6.0.26-0.2.mga1.nonfree
rootcerts-java-20110323.00-1.mga1
java-1.6.0-sun-1.6.0.26-0.2.mga1.nonfree
timezone-java-2011g-1.mga1
java-access-bridge-1.26.2-3.mga1
java-1.6.0-sun-plugin-1.6.0.26-0.2.mga1.nonfree
java-1.6.0-sun-devel-1.6.0.26-0.2.mga1.nonfree
java-1.6.0-sun-fonts-1.6.0.26-0.2.mga1.nonfree
java-1.6.0-sun-jdbc-1.6.0.26-0.2.mga1.nonfree