Re: Injection points: some tools to wait and wake - Mailing list pgsql-hackers

From Bertrand Drouvot
Subject Re: Injection points: some tools to wait and wake
Date
Msg-id ZdWvU6zQNE6D7yRz@ip-10-97-1-34.eu-west-3.compute.internal
Whole thread Raw
In response to Re: Injection points: some tools to wait and wake  (Michael Paquier <michael@paquier.xyz>)
List pgsql-hackers
Hi,

On Wed, Feb 21, 2024 at 07:08:03AM +0900, Michael Paquier wrote:
> On Tue, Feb 20, 2024 at 03:55:08PM +0000, Bertrand Drouvot wrote:
> > +PG_FUNCTION_INFO_V1(injection_points_wake);
> > +Datum
> > +injection_points_wake(PG_FUNCTION_ARGS)
> > +{
> > 
> > I think that This function will wake up all the "wait" injection points.
> > Would that make sense to implement some filtering based on the name? That could
> > be useful for tests that would need multiple wait injection points and that want
> > to wake them up "sequentially".
> > 
> > We could think about it if there is such a need in the future though.
> 
> Well, both you and Andrey are asking for it now, so let's do it.

Thanks!

> The implementation is simple:
> - Store in InjectionPointSharedState an array of wait_counts and an
> array of names.  There is only one condition variable.
> - When a point wants to wait, it takes the spinlock and looks within
> the array of names until it finds a free slot, adds its name into the
> array to reserve a wait counter at the same position, releases the
> spinlock.  Then it loops on the condition variable for an update of
> the counter it has reserved.  It is possible to make something more
> efficient, but at a small size it would not really matter.
> - The wakeup takes a point name in argument, acquires the spinlock,
> and checks if it can find the point into the array, pinpoints the
> location of the counter to update and updates it.  Then it broadcasts
> the change.
> - The wait loop checks its counter, leaves its loop, cancels the
> sleep, takes the spinlock to unregister from the array, and leaves.
>

I think that makes sense and now the "counter" makes more sense to me (thanks to
it we don't need multiple CV).
 
> I would just hardcode the number of points that can wait, say 5 of
> them tracked in shmem?  Does that look like what you are looking at?

I think so yes and more than 5 points would look like a complicated test IMHO.

Regards,

-- 
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com



pgsql-hackers by date:

Previous
From: jian he
Date:
Subject: Re: Catalog domain not-null constraints
Next
From: Richard Guo
Date:
Subject: Re: POC: GROUP BY optimization