Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Stefan Monnier <monnier@iro.umontreal.ca>
Newsgroups: comp.arch
Subject: Re: is Vax addressing sane today
Date: Mon, 23 Sep 2024 18:00:36 -0400
Organization: A noiseless patient Spider
Lines: 33
Message-ID: <jwvfrpqkpcq.fsf-monnier+comp.arch@gnu.org>
References: <vbd6b9$g147$1@dont-email.me>
	<09ce1622b872f0b0fa944e868a8c97be@www.novabbs.org>
	<vbnisc$2hb59$1@dont-email.me>
	<2024Sep10.094353@mips.complang.tuwien.ac.at>
	<vckf9d$178f2$1@dont-email.me> <O2DHO.184073$kxD8.113118@fx11.iad>
	<e290e18a59651f93e4b46f4839713b1c@www.novabbs.org>
MIME-Version: 1.0
Content-Type: text/plain
Injection-Date: Tue, 24 Sep 2024 00:00:37 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="7a2524499de57dc5fdc44c8e642a23e8";
	logging-data="3025851"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+Hpj/5ja7C3d427AfK+CEbUrcOdVXKt+I="
User-Agent: Gnus/5.13 (Gnus v5.13)
Cancel-Lock: sha1:2xEExspjHIGzNmbvjt5d7+QquPU=
	sha1:v/wmaIwwfJxGnPb5gmLp9bHpFBQ=
Bytes: 2888

> Some of the things that minimize the "badness" of taking an exception::
>
> a) deliver control to user signal handler without taking an
>   excursion through GuestOS. (think 10 cycles)
> b) when control arrives, receiving thread is already reentrant.
> c) when control arrives, the instruction (bits) and its operand
>   values are delivered to the exception handler. So, the exception
>   handler has what it needs to deal with the problem at hand.
> d) when control returns, the result (R0) is delivered back to the
>   destination register.
> e) (b, c, d) are performed without handler needing to understand
>   how. Handler is just a subroutine that receives arguments (c)
>   fixes the problem, and returns a non-excepting value, or abort.
> f) return has a way to re-execute the instruction or to skip the
>   instruction under control of handler without having access
>   to excepting-IP and without knowing the length of the
>   instruction.
> g) during (a..f) nobody ever has to disable interrupts or
>   exceptions or re-enable them later. Priority and privilege
>   are inherited automatically from excepting thread.

Note that in the case where you want the overflow exception to jump to
some alternate code path (a language-level exception handler, or a code
path that continues with a bigint instead of a register-sized integer),
(d) is useless because you don't want to return to the overflowing
instruction (nor to the immediately following instruction).  Instead you
usually want to lookup a side table indexed with the address of the
overflowing instruction to find the "exception handler" to "return" to.

(a) (b) and (c) are still very welcome, of course.


        Stefan