I am trying to make some changes in the kernel of Linux and then want to print some debugging statements and also want to count some parameters in compile time of the kernel.
I want to store the content in /proc file how can I achieve this task and store the info in /proc file without using LKM.
I would recommend you to read this article which helps you in creating, removing entries in /proc filesystem
http://www.thegeekstuff.com/2012/04/create-proc-files/
Using LKM approach is the best way to create proc entries and store the requisite info as "proc" itself is a filesystem provided by linux kernel for debugging purposes from user space.So, writing a module of your own using the proc related system calls would be easier and better for your purpose.Moreover, rather than building it as static module, you can compile it as a dynamic module to load and unload it at run-time.
If you don't want to use LKM, all the basic operations like proc_registration, creating entries,implementing read/write has to be done separately from scratch like writing your own system calls, registering it in kernel space and using them in your codes.
Related
I work on Linux kernel loadable module and I look for proper way to configure it and way to load/store binary data when module is loaded/unloaded. The module needs to read configuration data in time of loading, but it may change data and save them in run-time. I read on many places that reading/writing to file is not recommended and I also read that sysfs can be used for such purpose. Can be binary configuration data stored in sysfs ? Or exist more suitable solution ? Can you provide link to some example or doc where I can found some details about how to load/save persistent configuration by module ?
Peter
Maybe you can use the firmware interface in the module.
It should be able to load a binary file which include your settings with request_firmware(..)
I think it isn't required to write this binary file to the hardware...?
There are probably a lot of ways to do this. The way that leaps to my mind is with a character device and a script running in userspace. You could arrange that the script loads your module and then writes the binary data to the character device file created by the module. The script could proceed to loop on a read of the device file. When the module wants to export new state, it unblocks the read. The script then copies out the data, and repeats.
I'm writing a Linux kernel driver for a device that has quite a bit going on and a bunch of interaction from user-space. In order to debug and troubleshoot these interactions I'm storing them in a data array in the driver and need to export this data to user-space via some sort of file node. It'd be great if I could just export the binary data and let user-space parse it into human-readable text but the requirement from other parties involved is to have this done in-driver so one can just cat the file node from the command line and get something readable.
Currently I'm just using a sysfs node but I'm pretty sure that's the wrong place for that sort of thing and I need to able to write out more than PAGE_SIZE of data. What is the proper filesystem object I should be using for this type of thing? I looked at debugfs but it seems to be for more specific things much like sysfs is.
I think the best way is to use the debugfs as its developed just for debugging purpose and only people with the necessary permission which you can decide will be able to access it.
So go ahead with using debugfs
In Linux file system, there are files such as /dev/zero and /dev/random which are not real files on hard disk.
Is there any way that we can create a similar file and tell it to get ouput from executing a program?
For example, can I create file, say /tmp/tarfile, such that any program reading it actually gets the output from the execution of a different program (/usr/bin/tar ...)?
It is possible to create such a file/program, but it would require creation of a special filesystem in order to insert hooks into the VFS so that accesses can be detected and handled properly.
A kernel module I created is constantly generating information that I would like to store within the kernel and be readable from root. I was thinking of storing the information in some sort of log file with specific permissions, but I read that writing to files within the kernel space code is not good.
What is a good way to store information in the kernel that is fast and accessible by root?
~Thanks
If it's constantly generating new information, I would write it out with printk(). This way it will be seen by dmesg as well be written out to /var/log/kern.log.
Although, this is not for sensitive information, since dmesg could be used by any user.
Depending on the type of information, you can also provide it via /proc or /sys files, netlink, relayfs, /dev ioctl.
I'm writing a patch for VFS FAT implmentation on kernel 3.0
I want to add posix attributes to FAT files that are created in linux.
to achive that, I must save a file that contains all the relevant information on the mounted drive.
I know that reading and writing files from kernel space is something normally shouldn't be done, and I'm looking for another way to read/write the data.
I saw articles on the net that suggested to use /proc or to create a userspace daemon that will do the IO for me. I wanted to know if anyone saw or know where can I look at an implmentation of a thing like that,because I didn't find any examples for that over the net.
I'm not looking for a read/write to proc example, I want to see an entire solution for this issue.
Have a look at the quota implementation; this is a mechanism (ok, presumably not available on vfat) which reads/writes files from the kernel.
Additionally, the "loop" block device is another example of a kernel facility which does file IO.