Re: Broken stuff in new dtrace probes

Greg Stark <stark(at)enterprisedb(dot)com> writes:
> On Wed, Mar 11, 2009 at 11:50 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Furthermore, an isExtend call doesn't actually do a read(), so lumping
>> them together with regular reads doesn't seem like quite the right thing
>> for performance measurement purposes anyway. Maybe we actually ought to
>> have different probes for isExtend and regular cases.

> i like the idea of just have a separate pair of probes for table
> extension. I bet there are people who would actually like to see that
> alone sometimes too.

After further thought I concluded that the best solution for this is to
add the isExtend flag to the buffer_read_start/read_done probe parameter
lists. This allows the dtrace script writer to make the distinction if
he chooses, without adding any extra overhead for normal non-traced
operation. AFAICS using a separate probe type would add at least a
couple of if-tests even with tracing turned off.

regards, tom lane