Re: Parallel bitmap heap scan

On Tue, Feb 14, 2017 at 12:48 AM, Dilip Kumar <dilipbalaut@gmail.com> wrote:

> On Mon, Feb 13, 2017 at 6:24 PM, Robert Haas <robertmhaas@gmail.com>
> wrote:
> > I don't think it's acceptable (or necessary) to move the DSA
> > definitions into postgres.h.  Why do you think you need to do that,
> > vs. just including dsa.h in a few more places?
>
> I need to access dsa_pointer in tidbitmap.h, which is included from
> FRONTEND as well. Now, problem is that dsa.h is including #include
> "port/atomics.h", but atomic.h can not be included if FRONTEND is
> defined.
>
> #ifndef ATOMICS_H
> #define ATOMICS_H
> #ifdef FRONTEND
> #error "atomics.h may not be included from frontend code"
> #endif
>
> Is there any other solution to this ?


How about creating another header file with the parallel changes
and include it only in necessary places?

Following are my observations, while going through the patch.

+#if SIZEOF_DSA_POINTER == 4
+typedef uint32 dsa_pointer;
+#else
+typedef uint64 dsa_pointer;
+#endif

I feel the declaration of the above typdef can be moved into the
section above if we going with the current move into postgres.h
file.

+/*
+ * tbm_alloc_shared
+ *
+ * Callback function for allocating the memory for hashtable elements.
+ * It allocates memory from DSA if tbm holds a reference to a dsa.
+ */
+static inline void *
+pagetable_allocate(pagetable_hash *pagetable, Size size)

Function name and comments mismatch?


Regards,
Hari Babu
Fujitsu Australia