[LIP] Re: Re: A few Queries
Arun Sharma <[email protected]>
| Newsgroups | gmane.user-groups.linux.india.programmers |
|---|---|
| Message-ID | <[email protected]> |
On Sun, Jun 13, 2004 at 11:00:01AM +0530, suresh kumar wrote: > > TASKLET_SOFTIRQ is not a type of tasklet. It's a type of softirq. If > > you search for tasklet_init() in the kernel sources, you'll find about > > 60 hits as of 2.6.6. Each one of them is a type of tasklet. > Does it mean, during tasklet_init, we register a callback function which > will be called when ever a tasklet is scheduled for execution and based on > this function we are defining the type, and if we raise a more tasklets > which calls this function, then no tasklets of this same type can be > processed simultaneously. Sounds good so far. > So for network processing > one should never use tasklets. Is that correct? Depends on the what you're trying to do in the tasklet. In terms of parallelism, softirq > tasklet > bottom half - tasklets are built on top of softirqs. - if you use tasklets you don't need to lock - bottom halves have been removed in 2.6.x and replaced by tasklets - network processing generally uses softirqs -Arun ------------------------------------------------------- This SF.Net email is sponsored by the new InstallShield X. From Windows to Linux, servers to mobile, InstallShield X is the one installation-authoring solution that does it all. Learn more and evaluate today! http://www.installshield.com/Dev2Dev/0504