From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@postgreSQL.org
Subject: Minor codegen silliness in ExecInterpExpr()
Date: Thu, 28 Sep 2017 16:21:34 -0400
Message-ID: <2508.1506630094@sss.pgh.pa.us> (raw)
List-Unsubscribe: <mailto:majordomo@postgresql.org?body=unsub%20pgsql-hackers>
I noticed the following while poking around with perf:
| fcinfo->isnull = false;
|b5b: movb $0x0,0x1c(%rdx)
| *op->resvalue = op->d.func.fn_addr(fcinfo);
0.02 | mov 0x8(%rbx),%rcx
1.19 | mov %rdx,%rdi
0.93 | mov %rdx,(%rsp)
| mov %rcx,0x8(%rsp)
0.01 | callq *0x28(%rbx)
2.17 | mov 0x8(%rsp),%rcx
| mov %rax,(%rcx)
| *op->resnull = fcinfo->isnull;
1.18 | mov (%rsp),%rdx
4.32 | mov 0x10(%rbx),%rax
0.06 | movzbl 0x1c(%rdx),%edx
9.14 | mov %dl,(%rax)
It looks to me like gcc believes it is required to evaluate "op->resvalue"
before invoking the called function, just in case the function somehow has
access to *op and modifies that. We could save a pointless register spill
and reload if there were a temporary variable in there, ie
EEO_CASE(EEOP_FUNCEXPR)
{
FunctionCallInfo fcinfo = op->d.func.fcinfo_data;
+ Datum fvalue;
fcinfo->isnull = false;
- *op->resvalue = op->d.func.fn_addr(fcinfo);
+ fvalue = op->d.func.fn_addr(fcinfo);
+ *op->resvalue = fvalue;
*op->resnull = fcinfo->isnull;
EEO_NEXT();
}
and likewise in the other FUNCEXPR cases.
This is on a rather old gcc, haven't checked on bleeding-edge versions.
regards, tom lane
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-hackers@postgresql.org
Cc: tgl@sss.pgh.pa.us, andres@anarazel.de
Subject: Re: Minor codegen silliness in ExecInterpExpr()
In-Reply-To: <2508.1506630094@sss.pgh.pa.us>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox