RE: AIX support - Mailing list pgsql-hackers

    > The PPC asm code was originally written in 2002, and the first use of
    > _ sync_lock_test_and_set(), for ARM, appeared in 2012. The commit that
    > made __sync_lock_test_and_set() be chosen automatically for platforms
    > that don't specify anything else was added in 2022.

Thanks for the info.

------------------
    > Ok, if we don't need the assembler code at all, that's good. A patch to
    > introduce AIX support should not change it for non-AIX powerpc systems
    > though. That might be a good change, but would need to be justified
    > separately, e.g.  by some performance testing, and should be a separate
    > patch.

    > If you make no changes to s_lock.h at all, will it work? Why not?

With the existing asm code I see there are some syntax errors, being hit.

But after reverting the old changes the issues resolved. Below are diffs.

     static __inline__ int

     tas(volatile slock_t *lock)

     {

    @@ -424,17 +413,15 @@ tas(volatile slock_t *lock)

            __asm__ __volatile__(

     "      lwarx   %0,0,%3,1       \n"

     "      cmpwi   %0,0            \n"

    -"      bne     1f                      \n"

    +"      bne     $+16            \n"             /* branch to li %1,1 */

     "      addi    %0,%0,1         \n"

     "      stwcx.  %0,0,%3         \n"

    -"      beq     2f                      \n"

    -"1: \n"

    +"      beq     $+12            \n"             /* branch to lwsync */

     "      li      %1,1            \n"

    -"      b       3f                      \n"

    -"2: \n"

    +"      b       $+12            \n"             /* branch to end of asm sequence */

     "      lwsync                          \n"

     "      li      %1,0            \n"

    -"3: \n"

    +

     :      "=&b"(_t), "=r"(_res), "+m"(*lock)

     :      "r"(lock)

     :      "memory", "cc");

Let me know if I need to run any perf tools to check the performance of

the __sync_lock_test_and_set change.

---------------

    > > I mean this part of the code is needed as this is specific to AIX kernel memory
    > > operation which is different from __sync_lock_test_and_set().
    > >
    > > I would like to mention that the changes made in src/include/storage/s_lock.h
    > > are pretty much required and need to be operated in assemble specific to IBM
    > > Power architecture.

    > Was that earlier statement incorrect? Is the manual wrong or outdated or
    > not applicable to us?

Here this change is specific to AIX, but since we are compiling with gcc, this

is not applicable. But I will try with __sync* routines and check.

---------------

    > Do you still need mkldexport.sh? Surely there's a better way to do that

    > in year 2024. Some quick googling says there's a '-bexpall' option to

    > 'ld', which kind of sounds like what we want. Will that work? How do

    > other programs do this?

Thanks for looking into this, I’m working on this, I will let you know.

Thanks,

Sriram.