Messages in this thread | | | Date | Fri, 21 Oct 2005 16:41:04 +0200 | From | Yitzchak Eidus <> | Subject | Re: question about code from the linux kernel development ( se ) book |
| |
First Steve I want to thanks you for answering me your answered all the questions that i had I can understand your complains , and i will try to make it better next time ( i almost never used mailing lists... , so sorry about the white space striping , and the spelling... )
Jesper - yes it was a typo
Nick - thanks :)
On 10/21/05, Steven Rostedt <rostedt@goodmis.org> wrote: > > Oh God, please indent code examples, and if your email client strips white > space, change your email client. > > On Fri, 21 Oct 2005, Yitzchak Eidus wrote: > > > first i am very sorry if it isnt the place to ask questions like this > > but i didnt know where else to ask ( i tryed irc channels and i was > > send from there to this list ) > > anyway: > > does this following code look buggy? : > > [ Indention added ] > > > DECLARE_WAITQUEUE ( wait , current ); > > add_wait_queue ( q , &wait ); > > while ( !condition ) { > > set_current_stat ( TASK_INTERRUPTABLE ); i > > if ( signal_pending ( current ) ) > > /* handle signal */ > > I assume that the signal_pending is the if result and not the schedule. > Since there was no indentation I couldn't tell. > > > schedule ( ); > /* Moved brace down added */ > } > > set_current_state ( TASK_RUNNING ); > > remove_wait_queue ( q , &wait ); > > Before I go to your questions, I'll first answer that this _is_ buggy > code. If the condition is checked and you are woken up between the > while (!condition) and the set_current_state, then you will end up > sleeping forever or until someone sends you a signal. > > > first:doesnt in the way from checking the !condition to > > set_current_state the condition can be changed no? > > Yes, and then put it again after schedule. > > > > > second:why not putting the schedule ( ); right after the > > set_current_state ( ) , what the point in checking the if ( > > signal_pending ( ) first, if the proccess doesnt started to sleep yet? > > Yes, I would put the signal_pending check after the schedule (as most of > the kernel does this). > > > third: in the cleaning in the way from putting the set_current_state ( > > TASK_RUNNING ) into remove_wait_queue , cant the queue wait list ( q ) > > wake up again the wait procsess? > > Yeah, so? There's no harm in that, except for an extra cpu cycles that > are done to wake it up. That, I wouldn't change. > > > ( thnks for the help , please if it can be done answer quickly i am > > tanker in the idf and need to come back to the army soon , ( no > > internet there... ) ) > > BTW, this is not an IRC, we use normal capitalization and normal spelling > (when we know how to spell a word ;-). So the next time you send to the > list, send it as if you were writing a serious letter, or you may just be > ignored. (as you might have been if I didn't respond). > > -- Steve > - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |