[haiku-3rdparty-dev] Re: Multi-Threaded Code

  • From: "Earl Pottinger" <dmarc-noreply@xxxxxxxxxxxxx> (Redacted sender "earl_colby_pottinger" for DMARC)
  • To: "haiku-3rdparty-dev@xxxxxxxxxxxxx" <haiku-3rdparty-dev@xxxxxxxxxxxxx>
  • Date: Tue, 19 Apr 2016 18:13:55 +0000 (UTC)

I also found in ...kernel/thread.h
status_t thread_block()

void thread_unblock()

status_t _user_suspend_thread()
status_t _user_resume_thread()
This makes me wonder if I should be using these calls instead.
      From: Dario Casalinuovo <b.vitruvio@xxxxxxxxx>
 To: haiku-3rdparty-dev <haiku-3rdparty-dev@xxxxxxxxxxxxx> 
 Sent: Tuesday, April 19, 2016 8:01 AM
 Subject: [haiku-3rdparty-dev] Re: Multi-Threaded Code
   
Hello,

On Tue, Apr 19, 2016 at 8:47 AM, Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> wrote:

Am 18/04/2016 um 02:53 schrieb Earl Pottinger (Redacted sender 
earl_colby_pottinger for DMARC):

What I wanted to asked before doing a major rewrite, does anyone know if
resume_thread() and suspend_thread() are very fast calls?


In addition to what the others said, they both have syscall overhead, but are 
otherwise very simple and short calls.


I've questioned myself a similar question recently, how it compares to 
acquiring a lock to a semaphore? Is there any reason I should use resume_thread 
over releasing a semaphore?
-- 
Best Regards,Daro

  

Other related posts: