* Patch: Assemblers do not handle minimum vlong value correctly.
@ 2017-03-22 4:48 Ori Bernstein
2017-03-22 5:05 ` [9front] " Benjamin Purcell
0 siblings, 1 reply; 2+ messages in thread
From: Ori Bernstein @ 2017-03-22 4:48 UTC (permalink / raw)
To: 9front
The Plan 9 assemblers use strtoll to parse the integer literals
in their input. It turns out that this is almost correct, but
VLONG_MIN is clamped. This patch changes to use strtoull
in order to allow the full range of integers.
Below is a program that demosntrates the problem:
term% cat test.s
GLOBL L2<>+0(SB),$8
DATA L2<>+0(SB)/8,$-9223372036854775808
TEXT main+0(SB),2,$0
MOVQ L2<>+0(SB),AX
RET
term% 6a test.s
term% 6l test.6 /amd64/lib/libc.a
term% acid 6.out
6.out:amd64 plan 9 executable
/sys/lib/acid/port
/sys/lib/acid/amd64
acid: new()
33489: overflow _main SUBQ $0x90,SP
33489: breakpoint main+0x8 RET
acid: *L2
0x8000000000000001
acid: -9223372036854775808
0x8000000000000000
And the patch:
diff -r a01d0802d023 sys/src/cmd/cc/lexbody
--- a/sys/src/cmd/cc/lexbody Mon Mar 20 19:15:40 2017 +0100
+++ b/sys/src/cmd/cc/lexbody Tue Mar 21 21:45:44 2017 -0700
@@ -343,10 +343,9 @@
goto casee;
*cp = 0;
if(sizeof(yylval.lval) == sizeof(vlong))
- yylval.lval = strtoll(symb, nil, 10);
+ yylval.lval = strtoull(symb, nil, 10);
else
- yylval.lval = strtol(symb, nil, 10);
-
+ yylval.lval = strtoul(symb, nil, 10);
ncu:
while(c == 'U' || c == 'u' || c == 'l' || c == 'L')
c = GETC();
--
Ori Bernstein
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [9front] Patch: Assemblers do not handle minimum vlong value correctly.
2017-03-22 4:48 Patch: Assemblers do not handle minimum vlong value correctly Ori Bernstein
@ 2017-03-22 5:05 ` Benjamin Purcell
0 siblings, 0 replies; 2+ messages in thread
From: Benjamin Purcell @ 2017-03-22 5:05 UTC (permalink / raw)
To: 9front
Thank you, committed.
-spew
On Tue, Mar 21, 2017 at 11:48 PM, Ori Bernstein <ori@eigenstate.org> wrote:
> The Plan 9 assemblers use strtoll to parse the integer literals
> in their input. It turns out that this is almost correct, but
> VLONG_MIN is clamped. This patch changes to use strtoull
> in order to allow the full range of integers.
>
> Below is a program that demosntrates the problem:
>
> term% cat test.s
> GLOBL L2<>+0(SB),$8
> DATA L2<>+0(SB)/8,$-9223372036854775808
>
> TEXT main+0(SB),2,$0
> MOVQ L2<>+0(SB),AX
> RET
>
> term% 6a test.s
> term% 6l test.6 /amd64/lib/libc.a
> term% acid 6.out
> 6.out:amd64 plan 9 executable
> /sys/lib/acid/port
> /sys/lib/acid/amd64
> acid: new()
> 33489: overflow _main SUBQ $0x90,SP
> 33489: breakpoint main+0x8 RET
> acid: *L2
> 0x8000000000000001
> acid: -9223372036854775808
> 0x8000000000000000
>
> And the patch:
>
>
> diff -r a01d0802d023 sys/src/cmd/cc/lexbody
> --- a/sys/src/cmd/cc/lexbody Mon Mar 20 19:15:40 2017 +0100
> +++ b/sys/src/cmd/cc/lexbody Tue Mar 21 21:45:44 2017 -0700
> @@ -343,10 +343,9 @@
> goto casee;
> *cp = 0;
> if(sizeof(yylval.lval) == sizeof(vlong))
> - yylval.lval = strtoll(symb, nil, 10);
> + yylval.lval = strtoull(symb, nil, 10);
> else
> - yylval.lval = strtol(symb, nil, 10);
> -
> + yylval.lval = strtoul(symb, nil, 10);
> ncu:
> while(c == 'U' || c == 'u' || c == 'l' || c == 'L')
> c = GETC();
>
> --
> Ori Bernstein
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2017-03-22 5:05 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-03-22 4:48 Patch: Assemblers do not handle minimum vlong value correctly Ori Bernstein
2017-03-22 5:05 ` [9front] " Benjamin Purcell
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).