From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Geoghegan <pg@bowt.ie>
Cc: Michael Paquier <michael@paquier.xyz>
Cc: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: Adrien Nayrat <adrien.nayrat@anayrat.info>
Cc: Thomas Munro <thomas.munro@enterprisedb.com>
Cc: Dmitry Dolgov <9erthalion6@gmail.com>
Cc: Tomas Vondra <tomas.vondra@2ndquadrant.com>
Cc: Vik Fearing <vik.fearing@2ndquadrant.com>
Cc: Robert Haas <robertmhaas@gmail.com>
Cc: David Rowley <david.rowley@2ndquadrant.com>
Cc: Pg Hackers <pgsql-hackers@postgresql.org>
Subject: Re: random() (was Re: New GUC to sample log queries)
Date: Fri, 28 Dec 2018 16:42:20 -0500
Message-ID: <22401.1546033340@sss.pgh.pa.us> (raw)
In-Reply-To: <10518.1545940836@sss.pgh.pa.us>
References: <20181129214850.kiq6mvofsgplqjdr@alvherre.pgsql>
<3859.1545849900@sss.pgh.pa.us>
<20181227013458.GF2106@paquier.xyz>
<32696.1545875185@sss.pgh.pa.us>
<CAH2-Wz=xJiy+BgU7M49L6EHSkMbJWbzbprge=AqyOC8aDXRj8A@mail.gmail.com>
<2820.1545878340@sss.pgh.pa.us>
<CAH2-WznuXsWmjnvbt=odKMVxpj1zutgzRb7s6Vu3VPXCjejS5g@mail.gmail.com>
<9585.1545939214@sss.pgh.pa.us>
<10518.1545940836@sss.pgh.pa.us>
I wrote:
> On further reflection, it seems likely that in most installations a lot
> of processes never invoke drandom()/setseed() at all, making such work
> in InitProcessGlobals a waste of cycles. Probably a better idea is to
> have drandom() initialize the seed on first use, if it wasn't already
> set by setseed().
Here's a proposed patch for that.
It occurs to me that there's still another good reason to decouple
drandom from everything else: the point of setseed(), if it has any
at all, is to allow a repeatable sequence of drandom values to be
generated. Without this patch, no such guarantee exists because of
the possibility of the backend making additional internal calls of
libc random().
regards, tom lane
Attachments:
[text/x-diff] decouple-drandom-from-other-random-calls-1.patch (4.2K, ../22401.1546033340@sss.pgh.pa.us/2-decouple-drandom-from-other-random-calls-1.patch)
download | inline diff:diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgmlindex 3786099..1df4356 100644
*** a/doc/src/sgml/func.sgml
--- b/doc/src/sgml/func.sgml
***************
*** 1136,1146 ****
</table>
<para>
! The characteristics of the values returned by
! <literal><function>random()</function></literal> depend
! on the system implementation. It is not suitable for cryptographic
! applications; see <xref linkend="pgcrypto"/> module for an alternative.
! </para>
<para>
Finally, <xref linkend="functions-math-trig-table"/> shows the
--- 1136,1150 ----
</table>
<para>
! The <function>random()</function> function uses a simple linear
! congruential algorithm. It is fast but not suitable for cryptographic
! applications; see the <xref linkend="pgcrypto"/> module for an
! alternative.
! If <function>setseed()</function> is called, the results of
! subsequent <function>random()</function> calls in the current session are
! repeatable by re-issuing <function>setseed()</function> with the same
! argument.
! </para>
<para>
Finally, <xref linkend="functions-math-trig-table"/> shows the
diff --git a/src/backend/utils/adt/float.c b/src/backend/utils/adt/float.cindex cf9327f..add099e 100644
*** a/src/backend/utils/adt/float.c
--- b/src/backend/utils/adt/float.c
***************
*** 22,31 ****
--- 22,34 ----
#include "catalog/pg_type.h"
#include "common/int.h"
#include "libpq/pqformat.h"
+ #include "miscadmin.h"
#include "utils/array.h"
+ #include "utils/backend_random.h"
#include "utils/float.h"
#include "utils/fmgrprotos.h"
#include "utils/sortsupport.h"
+ #include "utils/timestamp.h"
/* Configurable GUC parameter */
*************** float8 degree_c_sixty = 60.0;
*** 53,58 ****
--- 56,65 ----
float8 degree_c_one_half = 0.5;
float8 degree_c_one = 1.0;
+ /* State for drandom() and setseed() */+ static bool drandom_seed_set = false;+ static unsigned short drandom_seed[3] = {0, 0, 0};+
/* Local function prototypes */
static double sind_q1(double x);
static double cosd_q1(double x);
*************** drandom(PG_FUNCTION_ARGS)
*** 2378,2385 ****
{
float8 result;
! /* result [0.0 - 1.0) */
! result = (double) random() / ((double) MAX_RANDOM_VALUE + 1);
PG_RETURN_FLOAT8(result);
}
--- 2385,2414 ----
{
float8 result;
! /* Initialize random seed, if not done yet in this process */
! if (unlikely(!drandom_seed_set))
! {
! /*
! * If possible, initialize the seed using high-quality random bits.
! * Should that fail for some reason, we fall back on a lower-quality
! * seed based on current time and PID.
! */
! if (!pg_backend_random((char *) drandom_seed, sizeof(drandom_seed)))
! {
! TimestampTz now = GetCurrentTimestamp();
! uint64 iseed;
!
! /* Mix the PID with the most predictable bits of the timestamp */
! iseed = (uint64) now ^ ((uint64) MyProcPid << 32);
! drandom_seed[0] = (unsigned short) iseed;
! drandom_seed[1] = (unsigned short) (iseed >> 16);
! drandom_seed[2] = (unsigned short) (iseed >> 32);
! }
! drandom_seed_set = true;
! }
!
! /* pg_erand48 produces desired result range [0.0 - 1.0) */
! result = pg_erand48(drandom_seed);
PG_RETURN_FLOAT8(result);
}
*************** Datum
*** 2392,2404 ****
setseed(PG_FUNCTION_ARGS)
{
float8 seed = PG_GETARG_FLOAT8(0);
! int iseed;
! if (seed < -1 || seed > 1)
! elog(ERROR, "setseed parameter %f out of range [-1,1]", seed);
! iseed = (int) (seed * MAX_RANDOM_VALUE);
! srandom((unsigned int) iseed);
PG_RETURN_VOID();
}
--- 2421,2440 ----
setseed(PG_FUNCTION_ARGS)
{
float8 seed = PG_GETARG_FLOAT8(0);
! uint64 iseed;
! if (seed < -1 || seed > 1 || isnan(seed))
! ereport(ERROR,
! (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
! errmsg("setseed parameter %g is out of allowed range [-1,1]",
! seed)));
! /* Use sign bit + 47 fractional bits to fill drandom_seed[] */
! iseed = (int64) (seed * (float8) UINT64CONST(0x7FFFFFFFFFFF));
! drandom_seed[0] = (unsigned short) iseed;
! drandom_seed[1] = (unsigned short) (iseed >> 16);
! drandom_seed[2] = (unsigned short) (iseed >> 32);
! drandom_seed_set = true;
PG_RETURN_VOID();
}
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, pg@bowt.ie, michael@paquier.xyz, alvherre@2ndquadrant.com, adrien.nayrat@anayrat.info, thomas.munro@enterprisedb.com, 9erthalion6@gmail.com, tomas.vondra@2ndquadrant.com, vik.fearing@2ndquadrant.com, robertmhaas@gmail.com, david.rowley@2ndquadrant.com
Subject: Re: random() (was Re: New GUC to sample log queries)
In-Reply-To: <22401.1546033340@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