From: erik quanstrom <quanstro@quanstro.net>
To: 9fans@9fans.net
Subject: [9fans] fp crash
Date: Mon, 28 Jan 2013 18:37:32 -0500 [thread overview]
Message-ID: <4deb4818bb26c316406212a8d1146e14@ladd.quanstro.net> (raw)
i've been looking at the program below, which calculates
some bad values on arm and amd64 and thus crashes.
(i added an a few asserts to make the crashes more reliable.)
386 does not have this issue.
#include <u.h>
#include <libc.h>
#include <stdio.h>
void
main(void)
{
printf("%10.4lf\n", 9306112.0000);
exits("");
}
the "problem" boils down to this
#include <u.h>
#include <libc.h>
#define fpword0(x) ((FPdbleword*)&x)->hi
#define Exp_11 0x3ff00000
#define Frac_mask1 0xfffff
void
badcalc(double d)
{
double d2;
d2 = d;
fpword0(d2) &= Frac_mask1;
fpword0(d2) |= Exp_11;
print("%g %g\n", d, d2);
}
void
main(void)
{
badcalc(9306112.0000);
exits("");
}
so the question is, is the compiler free to assume that
d2 has not been modified? (it does not do this if uvlong
is subsituted)
- erik
---
; 5c -S fh.c
TEXT badcalc+0(SB),0,$28
MOVD d+0(FP),F3
MOVW d2-4(SP),R3
AND $1048575,R3
MOVW R3,d2-4(SP)
MOVW d2-4(SP),R1
ORR $1072693248,R1
MOVW R1,d2-4(SP)
MOVW $.string<>+0(SB),R0
MOVD F3,8(R13)
MOVD F3,16(R13)
BL ,print+0(SB)
RET ,
TEXT main+0(SB),0,$8
MOVD $9.30611200000000000e+06,F0
MOVD F0,4(R13)
BL ,badcalc+0(SB)
DATA .string<>+0(SB)/8,$"%g %g\n\z\z"
MOVW $.string<>+7(SB),R0
BL ,exits+0(SB)
RET ,
GLOBL .string<>+0(SB),$8
END ,
next reply other threads:[~2013-01-28 23:37 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-28 23:37 erik quanstrom [this message]
2013-01-28 23:50 ` Charles Forsyth
2013-01-29 0:08 ` erik quanstrom
2013-01-29 1:41 ` erik quanstrom
2013-01-29 3:27 ` Charles Forsyth
2013-01-29 3:34 ` erik quanstrom
2013-01-29 17:17 ` Charles Forsyth
2013-01-29 17:19 ` erik quanstrom
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=4deb4818bb26c316406212a8d1146e14@ladd.quanstro.net \
--to=quanstro@quanstro.net \
--cc=9fans@9fans.net \
/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;
as well as URLs for NNTP newsgroup(s).