Re: Backend memory dump analysis

On 3/27/18 12:47, Tom Lane wrote:
> Here's an updated patch that adjusts the output format per discussion:
>
> - context identifier at the end of the line, so it's easier to see the
> numbers
>
> - identifiers truncated at 100 bytes, control characters replaced by
> spaces
>
> Also, I hacked things so that dynahash hash tables continue to print
> the way they did before, since the hash table name is really what
> you want to see there.
>
> Sample output is same test case as last time (dump at end of plpgsql.sql
> regression test script).
>
> Barring objection I plan to push this shortly.

Cool.

How about this one as well:

diff --git a/src/backend/utils/mmgr/portalmem.c
b/src/backend/utils/mmgr/portalmem.c
index 75a6dde32b..c08dc260e2 100644
--- a/src/backend/utils/mmgr/portalmem.c
+++ b/src/backend/utils/mmgr/portalmem.c
@@ -200,6 +200,7 @@ CreatePortal(const char *name, bool allowDup, bool
dupSilent)
portal->portalContext = AllocSetContextCreate(TopPortalContext,
"PortalContext",
ALLOCSET_SMALL_SIZES);
+ MemoryContextCopySetIdentifier(portal->portalContext, name);

/* create a resource owner for the portal */
portal->resowner = ResourceOwnerCreate(CurTransactionResourceOwner,

The term CopySetIdentifier has confused me a bit. (What's a "set
identifier"?) Maybe use CopyAndSetIdentifier? (We similarly have
MemoryContextResetAndDeleteChildren.)

I'm also not clear why this doesn't undo the previous optimization that
preferred making the identifier a compile time-constant. Aren't we now
just going back to doing a pstrdup() every time?

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services