Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
After more test and looking better Test result on ppc64 max, >0, >=1 9223372036854775808., 9223372036854775807, 0x7fff 9223372036854775808., -9223372036854775808, 0x8000 max+1, >0, >=1 9223372036854775808., 9223372036854775807, 0x7fff 9223372036854775808., -9223372036854775808, 0x8000 max+1, >1, >=1 9223372036854786048., 9223372036854775807, 0x7fff 9223372036854786048., -9223372036854765568, 0x80002800 Test result on x86_64 max, >0, >=1 9223372036854775808., -9223372036854775808, 0x8000 9223372036854775808., -9223372036854775808, 0x8000 max+1, >0, >=1 9223372036854775808., -9223372036854775808, 0x8000 9223372036854775808., -9223372036854775808, 0x8000 max+1, >1, >=1 9223372036854786048., -9223372036854775808, 0x8000 9223372036854786048., -9223372036854765568, 0x80002800 Test result on x86_64, build with -O2 (so, same as ppc64) 9223372036854775808., 9223372036854775807, 0x7fff 9223372036854775808., -9223372036854775808, 0x8000 max+1, >0, >=1 9223372036854775808., 9223372036854775807, 0x7fff 9223372036854775808., -9223372036854775808, 0x8000 max+1, >1, >=1 9223372036854786048., 9223372036854775807, 0x7fff 9223372036854786048., -9223372036854765568, 0x80002800 So, this show that > is the good test, so I have reverted the fix. So, the problem is elsewhere, I will search, probably using the others failed tests output. Thanks for catching this. Remi. -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
On 02/05/2013 12:33 AM, Remi Collet wrote:
> Le 05/02/2013 08:54, Stas Malyshev a écrit :
>> Hi!
>>
>>> - if (d > LONG_MAX) {
>>> + if (d >= LONG_MAX) {
>>> return (long)(unsigned long) d;
>>> }
>>> return (long) d;
>>
>> Could you please explain this one? I thought LONG_MAX is a largest
>> permissible long value, why when d==LONG_MAX we need to convert?
>
> LONG_MAX is the largest "unsigned" value.
> We have different int/float conversion result on some arch (detected, at
> least on ppc64).
>
>long l = 0x7fff;
>double d;
>d = l;
>d++;
>printf("max+1=%lf, %ld, 0x%lx\n", d, (long)d, (long)d);
>printf("max+1=%lf, %ld, 0x%lx\n", d, (long)(unsigned long)d,
> (long)(unsigned long)d);
>
>
> On x86_64
> max+1=9223372036854786048.00, -9223372036854775808, 0x8000
> max+1=9223372036854786048.00, -9223372036854765568, 0x80002800
>
> ON ppc64
>
> max+1=9223372036854775808.00, 9223372036854775807, 0x7fff
> max+1=9223372036854775808.00, -9223372036854775808, 0x8000
>
> So we are in a really border case...
>
>> Also, why this patch - which seems to be a BC break since it exhibits
>> different behavior on 64-bit systems - needs to be in 5.4?
>
> This patch doesn't change behavior on x86 arch.
> It fixes 9 failed test on ppc64.
I guess the confusion here is that without an #ifdef PPC around this,
how is this not triggering the conversion earlier on x86?
Also, I tried your test program on my x86_64 Linux box and I don't get
the same result:
I get:
max+1=9223372036854775808.00, -9223372036854775808, 0x8000
max+1=9223372036854775808.00, -9223372036854775808, 0x8000
gcc-4.7, glibc-2.15
-Rasmus
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
Le 05/02/2013 08:54, Stas Malyshev a écrit :
> Hi!
>
>> -if (d > LONG_MAX) {
>> +if (d >= LONG_MAX) {
>> return (long)(unsigned long) d;
>> }
>> return (long) d;
>
> Could you please explain this one? I thought LONG_MAX is a largest
> permissible long value, why when d==LONG_MAX we need to convert?
LONG_MAX is the largest "unsigned" value.
We have different int/float conversion result on some arch (detected, at
least on ppc64).
long l = 0x7fff;
double d;
d = l;
d++;
printf("max+1=%lf, %ld, 0x%lx\n", d, (long)d, (long)d);
printf("max+1=%lf, %ld, 0x%lx\n", d, (long)(unsigned long)d,
(long)(unsigned long)d);
On x86_64
max+1=9223372036854786048.00, -9223372036854775808, 0x8000
max+1=9223372036854786048.00, -9223372036854765568, 0x80002800
ON ppc64
max+1=9223372036854775808.00, 9223372036854775807, 0x7fff
max+1=9223372036854775808.00, -9223372036854775808, 0x8000
So we are in a really border case...
> Also, why this patch - which seems to be a BC break since it exhibits
> different behavior on 64-bit systems - needs to be in 5.4?
This patch doesn't change behavior on x86 arch.
It fixes 9 failed test on ppc64.
This is a very very old bug
http://marc.info/?l=php-internals&m=109360955403511
(this patch wasn't correct, but the issue was the same)
Probably most people don't care in previous version (<5.4), where test
suite was in a quite awfull state.
Now than we have a good test suite (0 failed test in x86), I think it is
interesting to do some work on other arch (and hopefully, I have a ppc64
test machine available).
Regards,
Remi.
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
Le 05/02/2013 08:25, Rasmus Lerdorf a écrit : > On 02/04/2013 06:19 AM, Remi Collet wrote: >> Commit:6533094d9e4d0c85044a170d9fad9a3986a68675 >> Author:Remi Collet Mon, 4 Feb 2013 15:19:32 +0100 >> Parents: 740b823be80a29e4c6537d4053eb7bad0fae6c41 >> Branches: PHP-5.4 PHP-5.5 master >> >> Link: >> http://git.php.net/?p=php-src.git;a=commitdiff;h=6533094d9e4d0c85044a170d9fad9a3986a68675 > > Maybe add a test case for this one? We already have: tests/lang/operators/bitwiseAnd_basiclong_64bit.phpt tests/lang/operators/bitwiseNot_basiclong_64bit.phpt tests/lang/operators/bitwiseOr_basiclong_64bit.phpt tests/lang/operators/bitwiseXor_basiclong_64bit.phpt tests/lang/operators/modulus_basiclong_64bit.phpt ext/standard/tests/math/decbin_basiclong_64bit.phpt ext/standard/tests/math/dechex_basiclong_64bit.phpt ext/standard/tests/math/decoct_basiclong_64bit.phpt ext/standard/tests/strings/chunk_split_variation2.php All this tests were failing on ppc64. For your information Before built-in server fix : 46 failed test Before this fix : 25 Still to fix : 16 Remi. -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
Hi!
> - if (d > LONG_MAX) {
> + if (d >= LONG_MAX) {
> return (long)(unsigned long) d;
> }
> return (long) d;
Could you please explain this one? I thought LONG_MAX is a largest
permissible long value, why when d==LONG_MAX we need to convert?
Also, why this patch - which seems to be a BC break since it exhibits
different behavior on 64-bit systems - needs to be in 5.4?
--
Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-CVS] com php-src: Fixed bug #64142 (dval to lval different behavior on ppc64): NEWS Zend/zend_operators.h
On 02/04/2013 06:19 AM, Remi Collet wrote: > Commit:6533094d9e4d0c85044a170d9fad9a3986a68675 > Author:Remi Collet Mon, 4 Feb 2013 15:19:32 +0100 > Parents: 740b823be80a29e4c6537d4053eb7bad0fae6c41 > Branches: PHP-5.4 PHP-5.5 master > > Link: > http://git.php.net/?p=php-src.git;a=commitdiff;h=6533094d9e4d0c85044a170d9fad9a3986a68675 Maybe add a test case for this one? -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
