Re: [Patch] Optimize dropping of relation buffers using dlist - Mailing list pgsql-hackers

From Amit Kapila
Subject Re: [Patch] Optimize dropping of relation buffers using dlist
Date
Msg-id CAA4eK1Lagd-ZsjjD06bKdrPpD0-LMpOhMjZ3UhcScnFd23-aOQ@mail.gmail.com
Whole thread Raw
In response to RE: [Patch] Optimize dropping of relation buffers using dlist  ("tsunakawa.takay@fujitsu.com" <tsunakawa.takay@fujitsu.com>)
List pgsql-hackers
On Tue, Dec 22, 2020 at 8:18 AM tsunakawa.takay@fujitsu.com
<tsunakawa.takay@fujitsu.com> wrote:
>
> From: Amit Kapila <amit.kapila16@gmail.com>
> > Why would all client backends wait for AccessExclusive lock on this
> > relation? Say, a client needs a buffer for some other relation and
> > that might evict this buffer after we release the lock on the
> > partition. In StrategyGetBuffer, it is important to either have a pin
> > on the buffer or the buffer header itself must be locked to avoid
> > getting picked as victim buffer. Am I missing something?
>
> Ouch, right.  (The year-end business must be making me crazy...)
>
> So, there are two choices here:
>
> 1) The current patch.
> 2) Acquire the buffer header spinlock before releasing the buffer mapping lwlock, and eliminate the buffer tag
comparisonas follows:
 
>
>   BufTableLookup();
>   LockBufHdr();
>   LWLockRelease();
>   InvalidateBuffer();
>
> I think both are okay.  If I must choose either, I kind of prefer 1), because LWLockRelease() could take longer time
towake up other processes waiting on the lwlock, which is not very good to do while holding a spinlock.
 
>
>

I also prefer (1). I will add some comments about the locking protocol
in the next version of the patch.

-- 
With Regards,
Amit Kapila.



pgsql-hackers by date:

Previous
From: "tsunakawa.takay@fujitsu.com"
Date:
Subject: RE: [Patch] Optimize dropping of relation buffers using dlist
Next
From: Amit Kapila
Date:
Subject: Re: [Patch] Optimize dropping of relation buffers using dlist