Thread: BUG #2546: PostgreSQL does not have native spinlock support on this platform

BUG #2546: PostgreSQL does not have native spinlock support on this platform

From
"John Weekley"
Date:
The following bug has been logged online:

Bug reference:      2546
Logged by:          John Weekley
Email address:      weekleyj@foobarlabs.net
PostgreSQL version: 8.1.4
Operating system:   Solaris unknown 5.11 snv_43 i86pc i386 i86pc
Description:        PostgreSQL does not have native spinlock support on this
platform
Details:

Platform is 64 bit Solaris x86.
unknown:/export/home/weekleyj/postgresql-8.1.4:>psrinfo -vp
The physical processor has 1 virtual processor (0)
  x86 (AuthenticAMD family 15 model 28 step 0 clock 2000 MHz)
        Mobile AMD Athlon 64 Processor 3000+

Relevant build output:

gmake[5]: Entering directory
`/export/home/weekleyj/postgresql-8.1.4/src/backend/access/transam'
/opt/SUNWspro/bin/cc -Xa -xO4 -xarch=amd64 -I../../../../src/include
-I/opt/X64/include   -c -o xlog.o xlog.c
"../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
does not have native spinlock support on this platform.  To continue the
compilation, rerun configure using --disable-spinlocks.  However,
performance will be poor.  Please report this to pgsql-bugs@postgresql.org.
"John Weekley" <weekleyj@foobarlabs.net> writes:
> Platform is 64 bit Solaris x86.
> ...
> "../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
> does not have native spinlock support on this platform.  To continue the
> compilation, rerun configure using --disable-spinlocks.  However,
> performance will be poor.  Please report this to pgsql-bugs@postgresql.org.

It looks like s_lock.h currently has coverage for amd64 using gcc, or
sparc64 using either compiler, but not for amd64 using Sun's cc.
Please feel free to send a patch, or you could pester Sun to send us
one (Josh?)

            regards, tom lane

Re: BUG #2546: PostgreSQL does not have native spinlock support

From
Zdenek Kotala
Date:
Tom Lane wrote:
> "John Weekley" <weekleyj@foobarlabs.net> writes:
>> Platform is 64 bit Solaris x86.
>> ...
>> "../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
>> does not have native spinlock support on this platform.  To continue the
>> compilation, rerun configure using --disable-spinlocks.  However,
>> performance will be poor.  Please report this to pgsql-bugs@postgresql.org.
>
> It looks like s_lock.h currently has coverage for amd64 using gcc, or
> sparc64 using either compiler, but not for amd64 using Sun's cc.
> Please feel free to send a patch, or you could pester Sun to send us
> one (Josh?)

SunStudio is supported on the HEAD (8.2). Spin lock problem is know
issue in version 8.1.4. I don't know if latest 8.1 branch contains
backport.


        Zdenek

Re: BUG #2546: PostgreSQL does not have native spinlock

From
Bruce Momjian
Date:
Tom Lane wrote:
> "John Weekley" <weekleyj@foobarlabs.net> writes:
> > Platform is 64 bit Solaris x86.
> > ...
> > "../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
> > does not have native spinlock support on this platform.  To continue the
> > compilation, rerun configure using --disable-spinlocks.  However,
> > performance will be poor.  Please report this to pgsql-bugs@postgresql.org.
>
> It looks like s_lock.h currently has coverage for amd64 using gcc, or
> sparc64 using either compiler, but not for amd64 using Sun's cc.
> Please feel free to send a patch, or you could pester Sun to send us
> one (Josh?)

amd64 support for Sun compilers will be in 8.2.  It is already in CVS.

--
  Bruce Momjian   bruce@momjian.us
  EnterpriseDB    http://www.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +

Re: BUG #2546: PostgreSQL does not have native spinlock

From
Bruce Momjian
Date:
Zdenek Kotala wrote:
> Tom Lane wrote:
> > "John Weekley" <weekleyj@foobarlabs.net> writes:
> >> Platform is 64 bit Solaris x86.
> >> ...
> >> "../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
> >> does not have native spinlock support on this platform.  To continue the
> >> compilation, rerun configure using --disable-spinlocks.  However,
> >> performance will be poor.  Please report this to pgsql-bugs@postgresql.org.
> >
> > It looks like s_lock.h currently has coverage for amd64 using gcc, or
> > sparc64 using either compiler, but not for amd64 using Sun's cc.
> > Please feel free to send a patch, or you could pester Sun to send us
> > one (Josh?)
>
> SunStudio is supported on the HEAD (8.2). Spin lock problem is know
> issue in version 8.1.4. I don't know if latest 8.1 branch contains
> backport.

No, we did not backport because of concern that not enough people
reviewed the changes.

--
  Bruce Momjian   bruce@momjian.us
  EnterpriseDB    http://www.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +

Re: BUG #2546: PostgreSQL does not have native spinlock

From
John Weekley
Date:
On Mon, 2006-07-24 at 21:47 -0400, Bruce Momjian wrote:
> Zdenek Kotala wrote:
> > Tom Lane wrote:
> > > "John Weekley" <weekleyj@foobarlabs.net> writes:
> > >> Platform is 64 bit Solaris x86.
> > >> ...
> > >> "../../../../src/include/storage/s_lock.h", line 814: #error: PostgreSQL
> > >> does not have native spinlock support on this platform.  To continue the
> > >> compilation, rerun configure using --disable-spinlocks.  However,
> > >> performance will be poor.  Please report this to pgsql-bugs@postgresql.org.
> > >
> > > It looks like s_lock.h currently has coverage for amd64 using gcc, or
> > > sparc64 using either compiler, but not for amd64 using Sun's cc.
> > > Please feel free to send a patch, or you could pester Sun to send us
> > > one (Josh?)
> >
> > SunStudio is supported on the HEAD (8.2). Spin lock problem is know
> > issue in version 8.1.4. I don't know if latest 8.1 branch contains
> > backport.
>
> No, we did not backport because of concern that not enough people
> reviewed the changes.

Wow,  you guys are really,really good.
Many thanks!

John


P.S.  May I ask when we can expect 8.2?  Or is is possible to beat on an
alpha/beta?