Mailing list for all users of the OCaml language and system.
 help / color / mirror / Atom feed
From: Dmitry Bely <dmitry.bely@gmail.com>
To: caml-list@inria.fr
Subject: [Caml-list] Compiler bug?
Date: Mon, 6 Aug 2012 14:04:27 +0400	[thread overview]
Message-ID: <CAMT7qiTJD4hFd4Jr0tKCzhmR=ngN0yV9Pw6CCgEfRPP6a6DYvg@mail.gmail.com> (raw)

Trying to catch some very rare floating-point related bug I've
realized that I probably don't understand some basic things about
OCaml execution model. Let's consider the simplest function:

let add1 x = x +. 1.

For x86 it is compiled as

	.CODE
	ALIGN	4
	PUBLIC	_camlTest__add1_1008
_camlTest__add1_1008:
	sub	esp, 8
L100:
	mov	ebx, eax
L101:	mov	eax, _caml_young_ptr
	sub	eax, 12
	mov	_caml_young_ptr, eax
	cmp	eax, _caml_young_limit
	jb	L102
	lea	eax, [eax+4]
	mov	DWORD PTR [eax-4],2301
	fld1
	fadd	REAL8 PTR [ebx]
	fstp	REAL8 PTR [eax]
	add    esp, 8
	ret
L102:	call	_caml_call_gc
L103:	jmp	L101

X parameter is passed as a pointer to float value ([eax] which is
immediately saved to [ebx]). But If GC happens (L102), float value can
be moved around the OCaml heap and [ebx] become invalid, right? So the
generated code is just wrong. Am I missing something?

- Dmitry Bely

             reply	other threads:[~2012-08-06 10:04 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-08-06 10:04 Dmitry Bely [this message]
2012-08-06 10:11 ` Alain Frisch
2012-08-06 10:20   ` Dmitry Bely
2012-08-06 10:34     ` Alain Frisch
2012-08-06 11:03       ` Dmitry Bely
2012-08-06 11:32         ` Alain Frisch
2012-08-06 12:16           ` Dmitry Bely
2012-08-07  1:35           ` Cedric Cellier
2012-08-08 16:03           ` Dmitry Bely
2012-08-08 18:03             ` Alain Frisch
2012-08-08 18:22               ` Jesper Louis Andersen
2012-08-08 18:40                 ` Dmitry Bely
2012-08-08 19:29                   ` Fabrice Le Fessant
2012-08-08 23:34                 ` Anil Madhavapeddy
2012-08-09  0:53                 ` Francois Berenger
  -- strict thread matches above, loose matches on Subject: below --
2006-05-17 23:14 compiler bug? Dan Koppel
2006-05-17 23:33 ` [Caml-list] " John Carr
2006-05-18 17:15 ` Xavier Leroy
2006-05-18 17:34   ` Jacques Carette
2006-05-18 17:46     ` Xavier Leroy
2006-05-18 19:31       ` Jacques Carette
2006-05-18 20:07         ` David Brown
2006-05-18 20:15           ` Jacques Carette
2006-05-18 20:20           ` Alain Frisch
2006-05-18 18:19     ` skaller
2006-05-18 18:53       ` Jacques Carette
2006-05-19  1:47         ` skaller
2006-05-19  2:17           ` Brian Hurt
2006-05-19  3:11             ` skaller
2006-05-19 16:48           ` Jacques Carette
2006-05-19 19:10             ` skaller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAMT7qiTJD4hFd4Jr0tKCzhmR=ngN0yV9Pw6CCgEfRPP6a6DYvg@mail.gmail.com' \
    --to=dmitry.bely@gmail.com \
    --cc=caml-list@inria.fr \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox