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=unavailable autolearn_force=no version=3.4.4
Path: 
 eternal-september.org!reader01.eternal-september.org!reader02.eternal-september.org!feeder.eternal-september.org!.POSTED!not-for-mail
From: Simon Wright <simon@pushface.org>
Newsgroups: comp.lang.ada
Subject: Re: How D.15 "Timing Events" are implemented? (SIGALRM)
Date: Thu, 16 Nov 2017 16:22:55 +0000
Organization: A noiseless patient Spider
Message-ID: <lyk1yqdwgw.fsf@pushface.org>
References: <ouiug6$ldd$1@gioia.aioe.org>
Mime-Version: 1.0
Content-Type: text/plain
Injection-Info: reader02.eternal-september.org;
 posting-host="4a40b7fae8b5c6e88c4b9a9a737094bf";
	logging-data="1426"; mail-complaints-to="abuse@eternal-september.org";
	posting-account="U2FsdGVkX1+pNTxfVBlsRMNShMNxuSTw/B/2xDRmekY="
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/25.3 (darwin)
Cancel-Lock: sha1:/7q0L7S4OGQ8wZfITu4XI+AGqss=
	sha1:XsSkfxaV+V5ORpHnd4z4/l/4jP0=
Xref: feeder.eternal-september.org comp.lang.ada:48934
Date: 2017-11-16T16:22:55+00:00
List-Id: <comp.lang.ada>

Victor Porton <porton@narod.ru> writes:

> Signal handlers (including SIGALRM) on Unix must not call non-reentrant 
> functions.
>
> So how D.15 Timing Events works in Ada? (I am interested mostly in GNAT 
> implementation.)

You can look at the implementation just as well as I can.

The plain GCC source runs a highest-priority task, which as implemented
loops over the queued events every 100 ms.

My Cortex GNAT RTS version[1] does the same but every 10 ms (plus a
what was meant to be a slight optimisation if there is a next event, but
which I see now I look at it is wrong if the next event is more than 10
ms away).

AdaCore's bare-board implementations run timing events off timer
interrupts (note, *not* SIGALRMs).

[1] https://github.com/simonjwright/cortex-gnat-rts/blob/master/common/a-rttiev.adb

> Because Ada RM does not require the handler to be "reentrant", this
> cannot be implemented straight as a SIGALRM handler. What is the
> implementation?

If this were relevant, I'd say that the ARM may not require the handler
to be reentrant, but people who buy (or use!) Ada compilers expect them
to work, so the implementers have to deal with "problems" like this all
the time.