From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00 autolearn=ham
	autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,43da1b63a7c8cf39
X-Google-Attributes: gid103376,public
X-Google-ArrivalTime: 2004-03-26 02:00:14 PST
Path: 
 archiver1.google.com!news1.google.com!news.glorb.com!newsgate.cistron.nl!news.tele.dk!news.tele.dk!small.news.tele.dk!news-fra1.dfn.de!news-nue1.dfn.de!newsfeed.r-kom.de!news.m-online.net!news-FFM2.ecrc.net!news.iks-jena.de!not-for-mail
From: Lutz Donnerhacke <lutz@iks-jena.de>
Newsgroups: comp.lang.ada
Subject: Re: Signal handler blocks proteced type
Date: Fri, 26 Mar 2004 10:00:13 +0000 (UTC)
Organization: IKS GmbH Jena
Message-ID: <slrnc67vpd.m2.lutz@taranis.iks-jena.de>
References: <slrnc664b5.nt.lutz@taranis.iks-jena.de>
NNTP-Posting-Host: taranis.iks-jena.de
X-Trace: branwen.iks-jena.de 1080295213 26359 217.17.192.37 (26 Mar 2004
 10:00:13 GMT)
X-Complaints-To: usenet@iks-jena.de
NNTP-Posting-Date: Fri, 26 Mar 2004 10:00:13 +0000 (UTC)
User-Agent: slrn/0.9.8.0 (Linux)
Xref: archiver1.google.com comp.lang.ada:6546
Date: 2004-03-26T10:00:13+00:00
List-Id: <comp.lang.ada>

* Lutz Donnerhacke wrote:
> A protected type with a signal handler is instantiated in order to catch a
> signal. If the signal is not catched, everything works fine. If the signal
> was delivered, the inner scope will not be terminated, because the protected
> type can not be finalized.
>
> Why?

Ada.Interrupts.Detach_Handler is broken. It does not return if the signal
was delivered.

Changing the example to 'pragma Interrupt_Handler' and an explicit
Attach_Handler works fine (termination of the enviroment task succeeds).

Any call to Detach_Handler after a signal delivery results in a blocking
runtime call.

 select
    delay 2.0;
 else abort
    Detach_Handler(SIGHUP);
 end select;

does not work. Even an Abort_Task call from an other task has no effect.

The runtime (Linux-Threads) seems to miss an SIGCHILD and waits forever for
the wrong clone.