Обсуждение: PostgreSQL 8.0.0beta2 compile error with Sun CC.

Поиск
Список
Период
Сортировка

PostgreSQL 8.0.0beta2 compile error with Sun CC.

От
Mark Round
Дата:
============================================================================
                        POSTGRESQL BUG REPORT TEMPLATE
============================================================================


Your name               : Mark Round
Your email address      : mark@markround.com


System Configuration
---------------------
  Architecture (example: Intel Pentium)         : Sparc (UltraSparc IIi)

  Operating System (example: Linux 2.4.18)      : Solaris 8

  PostgreSQL version (example: PostgreSQL-8.0):   PostgreSQL-8.0.0beta2

  Compiler used (example:  gcc 2.95.2)          : Sun C 5.5 2003/03/12


Please enter a FULL description of your problem:
-----------------------------------------------
Compilation fails with the error that ".seg" is no a recognised opcode.
The error is in src/backend/storage/lmgr/s_lock.c , lines 228 and 229.


Please describe a way to repeat the problem.   Please try to provide a
concise reproducible example, if at all possible:
----------------------------------------------------------------------
Compile using Sun's C compiler.


If you know how this problem might be fixed, list the solution below:
---------------------------------------------------------------------
.seg should be changed to .section.
Lines 229 and 229 in s_lock.c should be changed to the following :

  asm(".section \"data\"");
  asm(".section \"text\"");

 From docs.sun.com SPARC assembly manual : "This pseudo-op (.seg) is
currently supported for compatibility with existing SunOS 4.1 SPARC
assembly language programs. This pseudo-op has been replaced by the
.section pseudo-op."

Re: PostgreSQL 8.0.0beta2 compile error with Sun CC.

От
Tom Lane
Дата:
Mark Round <mark@markround.com> writes:
> Compilation fails with the error that ".seg" is no a recognised opcode.
> The error is in src/backend/storage/lmgr/s_lock.c , lines 228 and 229.

We heard about this before, but the previous complainant didn't answer
the obvious followup question: which versions of Solaris will fail if we
change it to ".section"?  Or more to the point, how can we #ifdef the
code so that this works on all Solaris versions?

            regards, tom lane

Re: PostgreSQL 8.0.0beta2 compile error with Sun CC.

От
Mark Round
Дата:
>We heard about this before, but the previous complainant didn't answer
>the obvious followup question: which versions of Solaris will fail if we
>change it to ".section"?  Or more to the point, how can we #ifdef the
>code so that this works on all Solaris versions?
>
>
Hi,

If it is changed to ".section", any SunOS 4.1.x -based (pre-1992) system
will fail to compile it.

I am guessing here (not a programmer), but as the ".seg" pseudo-op was a
SunOS 4.1.x construct, you could probably check for a 5.x version of
SunOS, and in which case, set the lines to ".section". I am not sure
what the appropriate #ifdef lines should look like, but will try to find
out for you.

-Mark

Re: PostgreSQL 8.0.0beta2 compile error with Sun CC.

От
Tom Lane
Дата:
Mark Round <mark@markround.com> writes:
> If it is changed to ".section", any SunOS 4.1.x -based (pre-1992) system
> will fail to compile it.

Hmm.  AFAIK no one has cared about SunOS for a couple of years now.
I was more concerned about whether there were any Solaris versions
that would fail to recognize .section.

> I am guessing here (not a programmer), but as the ".seg" pseudo-op was a
> SunOS 4.1.x construct, you could probably check for a 5.x version of
> SunOS, and in which case, set the lines to ".section". I am not sure
> what the appropriate #ifdef lines should look like, but will try to find
> out for you.

When I went looking a few days ago, it wasn't clear that there were any
predefined symbols that would let one tell the difference --- "__sun"
seems to be the only OS-specific predefine on Solaris, and I imagine it
was defined in SunOS too.  But if you find out anything please do let me
know.

            regards, tom lane