Software :: Referencing Non Exported Kernel Symbol In LKM
Aug 21, 2010
So after much intense reading of linux documentation and posts I can't find an answer to my question. Maybe someone can help.I am trying to access a kernel symbol that isn't exported (the symbol exists in /proc/kallsyms but not in Modules.symvers). Is there a non intrusive way (by avoiding compiling the kernel with CONFIG_KALLSYMS and parsing /proc/kallsyms in my LKM) to reference kernel symbols that are in /proc/kallsyms and not in Modules.symvers i.e. not exported symbols.
I want to install BLCR (URL...) from source. The problem is when I run ./configure, I get this error:Code: checking for Linux kernel symbol table... failed.configure: error: Failed to locate kernel symbol table. Try using --with-system-map or --with-vmlinux.What should I do next?
opensuse v11.2linux 2.6.31.12-0.1-desktop x86_64The kernel symbol file is still at 2.6.31.3. Apparently the symbol file did not get updated with the rest of the kernel development files. At least one program does not build because of that
Apache doesn't start with libedit.so.0.0.34 ( from libedit0-3.0.snap20090923-4.1.x86_64 ).
Code:
[Tue Dec 15 16:20:51 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] PHP Warning: Directive 'register_long_arrays' is deprecated in PHP 5.3 and greater in Unknown on line 0 /usr/sbin/httpd2-prefork: symbol lookup error: /usr/lib64/libedit.so.0: undefined symbol: tgetent Solution: I copied /usr/lib64/libedit.so.0.0.27 from a previous install and symlinked /usr/lib64/libedit.so.0 to it.
ive made a slight mistake on my Ubuntu 9.10 64bit. I was trying to get Voodoo Motion Tracking software to work, and it was having trouble locating some Qt Libs, so I decided it was a good idea to move its included libs into /user/lib32 directory. Voodoo now works, but Skype instead comes up with:
Code:
skype: symbol lookup error: /usr/lib32/libQtDBus.so.4: undefined symbol: _ZN14QObjectPrivate15checkWindowRoleEv Skype was installed using their own provided 64bit .deb - it was previously working fine.
The most recent update impacted my thunderbird-3.0. When launched I get these error messages
/usr/lib/thunderbird-3.0.3pre/thunderbird-bin: Symbol `SSL_ImplementedCiphers' has different size in shared object, consider re-linking /usr/lib/thunderbird-3.0.3pre/thunderbird-bin: symbol lookup error: /usr/lib/libnssutil3.so.1d: undefined symbol: PL_ClearArenaPool
I can only assume the libnssutil3 library was upgraded and this broke the library call in the thunderbird-3.0.
I performed a yum update and now my yum is not working and I am unable to start my apache processes (in a dmz configuration).This is the error I am getting from yum.
yum clean Traceback (most recent call last): File "/usr/bin/yum", line 4, in ? import yum
I'm running Ubuntu 10.10 and was looking for software to mix mp3's. I want to try mixxx (found here) I installed the most recent package by using the ppa of the project (as described here) Install seems to be ok (no error messages)But when I try to run mixxx from the menu nothing happens. Nothing. Running it from the terminal with the command
Code:
mixxx
I get this error message:
Quote:
mixxx: symbol lookup error: /usr/lib/libQtNetwork.so.4: undefined symbol: _ZN16QIODevicePrivate4peekEPcx
Downloaded the stable kernel 2.6.36.4 from kernel.org/ (Current Kernel:2.6.18-92.el5) However i have extracted the source code .ran make mrproper copied the old configuration file and selected it by using make menuconfig .However after selecting it and clicking ok i m getting following error.
Code: .
config:656:warning: symbol value 'm' invalid for IP_DCCP_CCID3 .config:657:warning: symbol value 'm' invalid for IP_DCCP_TFRC_LIB #config:1353:warning: symbol value 'm' invalid for FIXED_PHY # configuration written to .config 'm' invalid for IWL4965
Previously after doing aptitude upgrade, my pidgin seems broken with no gui shown. So I decided to download the latest source from pidgin website and build it from source, but I think I made situation worse because it then complained that ssl lib was needed. Then I removed the libpurple (e.g. aptitude purge libpurple0 libpurple-bin libpurple-dev) and reinstalled pidgin (aptitude install pidgin). Now it shows the error
pidgin: symbol lookup error: pidgin: undefined symbol: purple_media_element_info_get_type
I searched on the internet and can not find a solution. The clues on the internet says that's because the piding I use is the older version of libpurple. But I think I've removed all with purge and reinstalled it. Maybe some legacy binary is referenced. What or where it might be? Or where there may contain related information.
I have a filesystem mount at /A. I can export that filesystem via nfs with no issues.
I then mount an ISO image in a folder under /A: mount -o loop fname.ISO /A/isomnt. The original mount /A had the nohide option.
The files in the ISO image are available on the local machine, but the submount doesn't get exported by nfs to any other systems (FWIW: it does show up via samba). I have even run exportfs -a after the ISO mount, but that did nothing.
I have several ISO files that I would like to make available this way, without having to make several nfs exports for all of these, and then make mount commands on all of the remote systems.
Is this possible? If so, what do I need to do to make this work?
I have a machine which had a hardware fault and I want to read the disks by plugging them into a SATA-USB docking station in another machineWhen I do thisI can see the partitions on the old disk, but cannot read the LVM volumes on one of these physical partitions.Commands like pvscan do not find anything on this disk. Apparently you are supposed to export the volumes before trying to read them on a new system, but this was not performed. Does anyone know how to access the data on this disk? Surely it is possible.
I have some strange behavior in Latex. I have added labels to all of my tables (currently there are 3) and my figure (currently there is only 1). So far I only notice this behavior with tables, though it may happen with figures as well...I will have to add another figure to test that.However, what is happening is that my list of tables is correctly generating the numbers of the tables (which are 3.1, 3.2, and 4.1). In the resulting PDF document, the tables have the correct numbers in the captions as well. The odd thing is that when I attempt to reference them using ef, it is printing the Chapter.Section[.Subsection] number the table is located in instead of the table's number itself.Right now I have two tables inside of Chapter 3 Section 3. They are as follows:
I have an exported (via NFS) file system that contains other sub-mounted file systems (eg, DVD drives, ISO files, etc). While the sub-mounts show up locally, they aren't visible when the parent mount is exported.
In /etc/exportfs, the nohide option is specified for that export.
I am running into a small bit of difficulty with images I want to use in a Java program, keeping them bundled with the program in a JAR file and accessing them within the package.For those who are thinking it, this is definitely not homework.In my program, I have several images I would like to use on buttons on a toolbar.I have Googled, and I found something that comes close, but it deals with fetching images from a URL - nothing that deals with local files, more specifically, local images kept inside of a JARchive.
I am using OpenSuSE 11.3 (64 bit) and am trying to force my new server use NFSv4. Later, GSSAPI, sould be added, but for beginning, I want to configure without it for now. If I try using NFSv3, this works good, but I need NFSv4. I have exported my directory:
Code: omega:/ # cat /etc/exports /distrib *(ro,no_root_squash,sync,no_subtree_check) After mount attempt I have a error: Code: omega:/ # exportfs -var
firstly it took a very long time to display the browser.After the browser was working on my localmachine it took really long time to show the response.I am doing all this on internet and the server has bandwidth as 6Mbps and my internet connection has a speed 1Mbps what could be the reason for such a slow response.The same exists in case of Chromium is there a way to make it fast.Ubuntu 10.04 server edition and ssh2.
I have /home on its own partition.I reinstalled ubuntu Intrepid with a different user name.Now when I use F-Spot for instance i get a error "Could not open /home/olduser/.file".But how can i remove all references to /home/oldusername in all applications and point them to my current /home/username?
This is supposed to be possible right and better than naming each single cell?How do you do it?Also, maybe my approach is wrong as it's going to lead to a very big nested IF statement.There is a list of numbers which are price ceilings and another list of numbers containing the fees applied to these. I want to a formula that checks the item price against the ceilings and if it's less than or equal to that ceiling uses the fee in the adjacent row (which is the other data range).The only way I know how to do this is a big nested IF statement but does anyone know better?
I've got a situation. I'm having GNU bash version 3.00.16(1) on Solaris 10. I need to declare an array say arr1 which will be populated by an output of a command.
declare -a arr1 arr1=( $(/some/command) )
Supposing it will eventually (after executing the command) have element values as -
arr1[0]=1234 arr1[1]=5678 arr1[2]=7890
Now, I need to declare another set of arrays, one for each of the element values above - e.g.
declare -a arr1_1234 declare -a arr1_5678 declare -a arr1_7890
And I also need to populate elements of each of above 3 arrays with output of another command in a loop. So, these arrays will hold values something like -
I'm able to declare and populate arr1[*]. My question is how do I declare, populate and print the subsequent arrays and their elements?I am feeling rather thick to get this working.
I can reach other hosts by means of their global addresses by either the IP address or hostname (that has the global address). What I want to (also) do is have a hostname that references the IPv6 link local IP address (an AAAA record in DNS, or just the fe80::<whatever> address in /etc/hosts) and use that host name in commands to access that host. The problem is, an interface ID is needed when making such a reference.
It sure looks like the programs just pass the host name string on to the resolver library, which does not understand the significance of the '%' even though it could find and see that the name preceding the '%' is consistent with that being an IPv6 link local address (e.g. the logic could have been "split at first % and see if preceeding name is found as a link local address and accept that if so, or ignore the split otherwise" ... but it isn't). Is there a different syntax for this ... or was it overlooked in the design of programming around IPv6?I want to be able to address a host by its link local address, while still using a mnemonic instead of having to type the IPv6 address.
I always thought that Linux was immune to this sort of thing, but I've gotten an infestation of the Clicksor malware. When I bring up a page in Firefox from some websites (not this one, fortunately), it highlights words that it recognizes and then trots out a popup referencing that word and leading to a junk website. I've also had this problem in Chrome. How can I make that nasty beast go away?
When i ssh to server using -X, i always confuse about which display number i should export. It seems to me sometimes the display number has been used by something, so what i can do is only
Anyone have better documentation or an update to the this version of the file Tomcat HOWTO openSUSE as that document is referencing 10.2. Or a document for use with SLED.
I'm trying to remap my Nostromo N52 using pystromo, but can't figure out how assign the key with the ~ tilde on it. Everyone knows and refers to this key as the tilde key, but I can't for the life of me find the real name of it, since the primary symbol on it is not tilde but [URL]...
