Path: csiph.com!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <kst-u@mib.org>
Newsgroups: comp.lang.c
Subject: Re: Unexpected and difficult to understand SIGSEGV
Date: Sat, 03 Oct 2015 19:13:40 -0700
Organization: None to speak of
Lines: 42
Message-ID: <lnzizzxtwr.fsf@kst-u.example.com>
References: <mujh9g$nui$1@dont-email.me> <mujj72$vjq$1@dont-email.me> <muk33b$3sv$1@dont-email.me> <8362e3dd-17c5-4d88-a88f-5116ea0f10e0@googlegroups.com> <muk7sk$n5h$1@dont-email.me> <9256d7fa-5b3e-4183-b071-d910254623ce@googlegroups.com> <muluoc$ui0$1@dont-email.me> <2da82a81-43ee-49eb-9357-6a7e06cfac03@googlegroups.com> <ln1tdc23h6.fsf@kst-u.example.com> <efbd4fa7-3d38-4a11-97b3-54acc72cee57@googlegroups.com> <h2h01bt6jip1misjmqimv8cnu9rd1pk5ft@4ax.com> <muph4b$8f0$1@dont-email.me> <jej01blpim3ebvoff7qpcp60i4j2a5jtpc@4ax.com> <mupke0$mld$1@dont-email.me> <mupm2q$tcv$1@dont-email.me> <f27745d6-b4a8-42c3-9d2f-0bc5ff370143@googlegroups.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Injection-Info: mx02.eternal-september.org; posting-host="945944de09706c9b4e29b53c9d2efdc2"; logging-data="1289"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+ayvPj/MlNbGRdO8eAo47H"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
Cancel-Lock: sha1:luHm1V3MQsSn4/KAmzJ0seY/oTI= sha1:PLRTJ+JsLmIdB3Fdpj5598+pLyo=
Xref: csiph.com comp.lang.c:71770

supercat@casperkitty.com writes:
> On Saturday, October 3, 2015 at 5:52:10 PM UTC-5, James Kuyper wrote:
>> There is no such requirement.
>
> What exactly would the Standard specify given:
>
>     int main(void)
>     {
>       int32_t x=-65537;
>       int16_t y=x;
>       printf("%ld\n",(long)y);
>       return 0;
>     }

Since the value of x is outside the range of int16_t, the implicit
conversion yields an implementation-defined result or raises an
implementation-defined signal.

> Would it be legitimate for an implementation to raise an implementation-
> defined signal or use some implementation-defined method of conversion
> that yields a value other than -1?

Certainly.  (The common implementation of 2's-complement arithmetic has
wraparound semantics, but the standard says nothing about that.)


>                                     The Standard clearly implies that
> a legitimate implementation must supply documentation sufficient to
> ascertain what the above code would do on that implementation, and
> most implementations would certainly yield -1, but the fact that an
> int16_t is required to use a two's-complement representation would not
> in and of itself preclude the possibility of an implementation defining
> some other means of doing the conversion (e.g. saturating to -32767 or
> -32768).

Correct.

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
Working, but not speaking, for JetHead Development, Inc.
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"