[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: nptl 0.5

Hash: SHA1

Hong Zhang wrote:

> Please do NOT implement pthread_kill_other_threads_np(). It was
> used to workaround problems with abort(2). Under linuxthreads,
> abort() can not dump core under heavy multi-threaded environment.
> Therefore, we should fix abort(2).

Exactl.y I have no intention to add this function.

> JVM can be implemented using strict pthread interface. There are
> a few extra things that will be really helpful. The most important
> one is to make pthread_self() and pthread_getspecific() work inside
> signal handlers, including nested signal handlers.

It already works.  No platform without some form of "thread register"
will be supported and the thread register allows access to the thread
handle and TSD regardless of the state of stack or interrupt handling.

> Most people love to have pthread_suspend_np(), pthread_resume_np(),
> pthread_getcontext_np(), pthread_setcontext_np(). In current jvm
> implementation, these features are implemented using pthread_kill(3)
> and sigsuspend(2). As long as NPTL has very fast pthread_kill(),
> I don't see much, if any, benefits to have these xxx_np().

- From what I gathered these functions are not really the interface to
use.  And yes,  pthread_kill is very fast.

- -- 
- --------------.                        ,-.            444 Castro Street
Ulrich Drepper \    ,-----------------'   \ Mountain View, CA 94041 USA
Red Hat         `--' drepper at redhat.com `---------------------------
Version: GnuPG v1.0.7 (GNU/Linux)


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]