Hi,
I've built a driver that works as a block device, it's a ram disk. Basically, at this point I'm trying to modify it to support change notifications. Here is what I was thinking:
1) User passes a function pointer to a function that is called on change. However, can this lead to problems calling a user mode function from the kernel?
2) User passes a structure into the ioctl interface, and then blocks until a change, the structure is filled out and returned to the user when a change is detected.
What is the best way to go about this?
This Question has been solved and asker verified All Experts Exchange premium technology solutions are available to subscription members.
Experts Exchange has been collecting answers to technology questions since 1996…3 million and counting! If you have a question, chances are we already have your answer.
If you can't find the exact answer you're looking for, ask our exclusive community of 50,000 experts. You’ll get a personalized answer from a trusted professional.
Thousands of free tech tips, tricks, how-to’s and tutorials are available in our peer reviewed articles section. See for yourself how smart our experts are, no login required.
Save solutions to your questions, answers you’ve discovered through searching plus helpful articles in your personal knowledgebase for easy future access.
Access the answers to your technology questions today.
30-day free trial. Register in 60 seconds.
Members of the expert community talk about why the experience at Experts Exchange is different than what you will find anywhere else.
Try it out and discover for yourself.
30-day free trial. Register in 60 seconds.
Join the community of experts here and help other tech pros by answering question in your area of expertise. You can earn FREE access to all Experts Exchange's premium features and resources.
what level do you want to watch at? (block/byte/...)
would it be sufficient to monitor at the file system level?
(use inotify then: http://www.linuxjournal.co
Would also be interesting to know the reason behind it and the goal that you want to achieve by watching.
Ik you have a file storage device, dosn't inotify/dnotify handle that stuff for you?
incron daemon uses that too to pickup file changes and act upon it.
http://inotify.aiken.cz/?s
>>> to support change notifications
In order to implement a notify change at the block dev level, you could user the "struct nofifier_block" ,,, blocking_notifier_call_cha
However, I dint understand the purpose/objective of calling a user space function from the kernel space? If you are looking at just communicating from your kernel space (module) to user space, there are quite a few well known tricks for this.
>> ioclt
disadvantage is you would have to have a device node and a user space write program to access the node and pass ioctl commands to it.
Ok, not using a file system in a file system oriented OS that could be called a challange...
OTOH, even if mimimalistic anything that qualifies as 'managing the usage of storage devices' qualifies as a filesystem...
So if you have a an allocation bitmap or something looking like a directory you do have a filesystem. FAT32 doesn't have Inodes, iso9660 is also quite different... but there are filesystem drivers for them. So maybe you do need to rethink this.
Maybe it is easier to implement a fuse server for your '(notespesicaly-)filesyste
What I'm really trying to find out is, is there an advantage to trying to find a way to call a user space function from the kernel, or is that even possible, or do I have to use the method I already mentioned which is pass a struct into the kernel and use copy_to/from_user to copy the data back and forth after the process has blocked waiting for a notification.
If you want to call a user space function use call_usermodehelper() form either the kernel space or module.
on the other hand copy_to/from_user is just for copying bytes between user space and kernel space. (standard way of doing it). Having the application invoked each time depends on many other factors, like how/when the kernel notifies the applications.
Both these are used for diffirent purposes.
Alternatively, go to http://lxr.linux.no/#linux
and walk about the code of the several block drivers (search for the API's I have provided above and in the initial posts). if any have implemented in a similar way, or atlest this would get you pointers to more.
You will have to give more details on what you are trying to acheive with this method. It would be easier to understand the larger picture.
If you want to send notifications from kernel space to user space then another alternative is to use netlink socket.
AF_NETLINK - (netlink sockets) They are especially designed for the communication between the kernel space and the user space.
Here is some info on who to acheive this;
http://people.ee.ethz.ch/~
hope this helps!
How do NETLINK sockets differ from Signals, most importantly, why would I use that method over signals , but either way, the user mode application is blocking and awoken by a signal...right? It would see that by using netlink sockets, that I would have to implement a continuous polling approach, no?
in case of signal you will still have to pass the process pid into the kernel space. Only then you can find the task_stuct using the pid to send the signal. All this can be avoided in case of netlink sockets. if you want to pass large info from kernel space, can you still use only signals? the info you can send you be limited in case of signals.
Business Accounts
Answer for Membership
by: askbPosted on 2009-10-29 at 20:32:16ID: 25700036
Using a upcall method of the Linux kernel allows a kernel module to invoke a function in user space, but this requires your to set the PATH, envp which would be a headache unless you make sure you install you lib / binary in standard paths. Where as ioctl mechanism is implemented as a single system call that multiplexes the different commands to the appropriate kernel space function. From my understand
Did you at other implementations of other similar block devices under -> http://lxr.linux.no/#linux