hppa: Define __NO_LONG_DOUBLE_MATH so headers are consistent with libm build
Commit Message
On 2015-11-22, at 5:15 PM, John David Anglin wrote:
> The attached patch fixes BZ #19270 and the Debian gmt package know builds successfully. Aside from the
> comment, the define of __NO_LONG_DOUBLE_MATH is similar to that in the generic version of glibc.
>
> Build tested on hppa-unknown-linux-gnu with no observed regressions.
It was found in further testing that the abi lists need updating. This is currently installed in Debian glibc-2.21.
Please review and install if okay.
Dave
--
John David Anglin dave.anglin@bell.net
2015-12-11 John David Anglin <danglin@gcc.gnu.org>
[BZ #19270]
* sysdeps/hppa/fpu/bits/mathdef.h (__NO_LONG_DOUBLE_MATH): Define.
2015-12-11 Aurelien Jarno <aurelien@aurel32.net>
* sysdeps/unix/sysv/linux/hppa/libc.abilist: Add __finitel, __isinfl,
and __isnanl.
* sysdeps/unix/sysv/linux/hppa/libm.abilist: Add _finitel.
Comments
John David Anglin <dave.anglin@bell.net> writes:
> diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
> index 1d30644..72435f8 100644
> --- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
> +++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
> @@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
> GLIBC_2.2 __ffs F
> GLIBC_2.2 __finite F
> GLIBC_2.2 __finitef F
> +GLIBC_2.2 __finitel F
> GLIBC_2.2 __flbf F
> GLIBC_2.2 __fork F
> GLIBC_2.2 __fpending F
> @@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
You can't do that. Please read the thread starting at
<http://sourceware.org/ml/libc-alpha/2015-11/msg00564.html>.
Andreas.
On 12/13/2015 4:13 AM, Andreas Schwab wrote:
> John David Anglin <dave.anglin@bell.net> writes:
>
>> diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>> index 1d30644..72435f8 100644
>> --- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
>> +++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>> @@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
>> GLIBC_2.2 __ffs F
>> GLIBC_2.2 __finite F
>> GLIBC_2.2 __finitef F
>> +GLIBC_2.2 __finitel F
>> GLIBC_2.2 __flbf F
>> GLIBC_2.2 __fork F
>> GLIBC_2.2 __fpending F
>> @@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
> You can't do that. Please read the thread starting at
> <http://sourceware.org/ml/libc-alpha/2015-11/msg00564.html>.
I committed my fix, so I expect if you simply merge up to the tip of master,
the hppa abilist problem will go away too.
On 2015-12-13, at 4:42 PM, Chris Metcalf wrote:
> On 12/13/2015 4:13 AM, Andreas Schwab wrote:
>> John David Anglin <dave.anglin@bell.net> writes:
>>
>>> diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>> index 1d30644..72435f8 100644
>>> --- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>> +++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>> @@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
>>> GLIBC_2.2 __ffs F
>>> GLIBC_2.2 __finite F
>>> GLIBC_2.2 __finitef F
>>> +GLIBC_2.2 __finitel F
>>> GLIBC_2.2 __flbf F
>>> GLIBC_2.2 __fork F
>>> GLIBC_2.2 __fpending F
>>> @@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
>> You can't do that. Please read the thread starting at
>> <http://sourceware.org/ml/libc-alpha/2015-11/msg00564.html>.
>
> I committed my fix, so I expect if you simply merge up to the tip of master,
> the hppa abilist problem will go away too.
I'm testing a new version of the change with the versions changed to 2.23 and an update
to the hppa/Versions file.
I'm not a glibc maintainer, so I shouldn't do the merge.
Dave
--
John David Anglin dave.anglin@bell.net
On 12/13/2015 5:02 PM, John David Anglin wrote:
> On 2015-12-13, at 4:42 PM, Chris Metcalf wrote:
>
>> On 12/13/2015 4:13 AM, Andreas Schwab wrote:
>>> John David Anglin <dave.anglin@bell.net> writes:
>>>
>>>> diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>> index 1d30644..72435f8 100644
>>>> --- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>> +++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>> @@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
>>>> GLIBC_2.2 __ffs F
>>>> GLIBC_2.2 __finite F
>>>> GLIBC_2.2 __finitef F
>>>> +GLIBC_2.2 __finitel F
>>>> GLIBC_2.2 __flbf F
>>>> GLIBC_2.2 __fork F
>>>> GLIBC_2.2 __fpending F
>>>> @@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
>>> You can't do that. Please read the thread starting at
>>> <http://sourceware.org/ml/libc-alpha/2015-11/msg00564.html>.
>> I committed my fix, so I expect if you simply merge up to the tip of master,
>> the hppa abilist problem will go away too.
>
> I'm testing a new version of the change with the versions changed to 2.23 and an update
> to the hppa/Versions file.
>
> I'm not a glibc maintainer, so I shouldn't do the merge.
OK, but, if you are working on the glibc tip, you should not see these extra exports
for __finitel and friends at all any more, so you should not have to change your abilist files.
On 2015-12-13, at 6:11 PM, John David Anglin wrote:
> On 2015-12-13, at 5:12 PM, Chris Metcalf wrote:
>
>> On 12/13/2015 5:02 PM, John David Anglin wrote:
>>> On 2015-12-13, at 4:42 PM, Chris Metcalf wrote:
>>>
>>>> On 12/13/2015 4:13 AM, Andreas Schwab wrote:
>>>>> John David Anglin <dave.anglin@bell.net> writes:
>>>>>
>>>>>> diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>>>> index 1d30644..72435f8 100644
>>>>>> --- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>>>> +++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
>>>>>> @@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
>>>>>> GLIBC_2.2 __ffs F
>>>>>> GLIBC_2.2 __finite F
>>>>>> GLIBC_2.2 __finitef F
>>>>>> +GLIBC_2.2 __finitel F
>>>>>> GLIBC_2.2 __flbf F
>>>>>> GLIBC_2.2 __fork F
>>>>>> GLIBC_2.2 __fpending F
>>>>>> @@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
>>>>> You can't do that. Please read the thread starting at
>>>>> <http://sourceware.org/ml/libc-alpha/2015-11/msg00564.html>.
>>>> I committed my fix, so I expect if you simply merge up to the tip of master,
>>>> the hppa abilist problem will go away too.
>>>
>>> I'm testing a new version of the change with the versions changed to 2.23 and an update
>>> to the hppa/Versions file.
>>>
>>> I'm not a glibc maintainer, so I shouldn't do the merge.
>>
>> OK, but, if you are working on the glibc tip, you should not see these extra exports
>> for __finitel and friends at all any more, so you should not have to change your abilist files.
>
>
> Thanks, I'll retest with no changes to the hppa abi files.
There were no abi failures, so all that is needed is the original patch. The v2 patch is withdrawn,
Dave
--
John David Anglin dave.anglin@bell.net
@@ -34,5 +34,10 @@ typedef double double_t; /* `double' expressions are evaluated as
#endif /* ISO C99 */
-/* On hppa `long double' is 64-bits. */
-#undef __NO_LONG_DOUBLE_MATH
+#ifndef __NO_LONG_DOUBLE_MATH
+/* On hppa `long double' and `double' are 64-bits. So, libm is built
+ with NO_LONG_DOUBLE defined. The following define ensures the library
+ and headers are consistent. This disables the declaration of all the
+ `long double' function variants. */
+# define __NO_LONG_DOUBLE_MATH 1
+#endif
@@ -263,6 +263,7 @@ GLIBC_2.2 __fcntl F
GLIBC_2.2 __ffs F
GLIBC_2.2 __finite F
GLIBC_2.2 __finitef F
+GLIBC_2.2 __finitel F
GLIBC_2.2 __flbf F
GLIBC_2.2 __fork F
GLIBC_2.2 __fpending F
@@ -294,9 +295,11 @@ GLIBC_2.2 __isdigit_l F
GLIBC_2.2 __isgraph_l F
GLIBC_2.2 __isinf F
GLIBC_2.2 __isinff F
+GLIBC_2.2 __isinfl F
GLIBC_2.2 __islower_l F
GLIBC_2.2 __isnan F
GLIBC_2.2 __isnanf F
+GLIBC_2.2 __isnanl F
GLIBC_2.2 __isprint_l F
GLIBC_2.2 __ispunct_l F
GLIBC_2.2 __isspace_l F
@@ -63,6 +63,7 @@ GLIBC_2.2 __clog10f F
GLIBC_2.2 __clog10l F
GLIBC_2.2 __finite F
GLIBC_2.2 __finitef F
+GLIBC_2.2 __finitel F
GLIBC_2.2 __fpclassify F
GLIBC_2.2 __fpclassifyf F
GLIBC_2.2 __signbit F