~ubuntu-branches/debian/experimental/postgresql-11/experimental

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>E.26. Release 9.5.4</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets V1.79.1" /><link rel="prev" href="release-9-5-5.html" title="E.25. Release 9.5.5" /><link rel="next" href="release-9-5-3.html" title="E.27. Release 9.5.3" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">E.26. Release 9.5.4</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="release-9-5-5.html" title="E.25. Release 9.5.5">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="release.html" title="Appendix E. Release Notes">Up</a></td><th width="60%" align="center">Appendix E. Release Notes</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 11beta1 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="release-9-5-3.html" title="E.27. Release 9.5.3">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="RELEASE-9-5-4"><div class="titlepage"><div><div><h2 class="title" style="clear: both">E.26. Release 9.5.4</h2></div></div></div><div class="toc"><dl class="toc"><dt><a href="release-9-5-4.html#id-1.11.6.30.4">E.26.1. Migration to Version 9.5.4</a></dt><dt><a href="release-9-5-4.html#id-1.11.6.30.5">E.26.2. Changes</a></dt></dl></div>Release date: 2016-08-11

This release contains a variety of fixes from 9.5.3.

For information about new features in the 9.5 major release, see

<a class="xref" href="release-9-5.html" title="E.30. Release 9.5">Section E.30</a>.

<div class="sect2" id="id-1.11.6.30.4"><div class="titlepage"><div><div><h3 class="title">E.26.1. Migration to Version 9.5.4</h3></div></div></div>

A dump/restore is not required for those running 9.5.X.

However, if you are upgrading from a version earlier than 9.5.2,

see <a class="xref" href="release-9-5-2.html" title="E.28. Release 9.5.2">Section E.28</a>.

</div><div class="sect2" id="id-1.11.6.30.5"><div class="titlepage"><div><div><h3 class="title">E.26.2. Changes</h3></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem">

Fix possible mis-evaluation of

nested <code class="literal">CASE</code>-<code class="literal">WHEN</code> expressions (Heikki

Linnakangas, Michael Paquier, Tom Lane)

A <code class="literal">CASE</code> expression appearing within the test value

subexpression of another <code class="literal">CASE</code> could become confused about

whether its own test value was null or not. Also, inlining of a SQL

function implementing the equality operator used by

a <code class="literal">CASE</code> expression could result in passing the wrong test

value to functions called within a <code class="literal">CASE</code> expression in the

SQL function's body. If the test values were of different data

types, a crash might result; moreover such situations could be abused

to allow disclosure of portions of server memory. (CVE-2016-5423)

</li><li class="listitem">

Fix client programs' handling of special characters in database and

role names (Noah Misch, Nathan Bossart, Michael Paquier)

Numerous places in vacuumdb and other client programs

could become confused by database and role names containing double

quotes or backslashes. Tighten up quoting rules to make that safe.

Also, ensure that when a conninfo string is used as a database name

parameter to these programs, it is correctly treated as such throughout.

Fix handling of paired double quotes

in psql's <code class="command">\connect</code>

and <code class="command">\password</code> commands to match the documentation.

Introduce a new <code class="option">-reuse-previous</code> option

in psql's <code class="command">\connect</code> command to allow

explicit control of whether to re-use connection parameters from a

previous connection. (Without this, the choice is based on whether

the database name looks like a conninfo string, as before.) This

allows secure handling of database names containing special

characters in pg_dumpall scripts.

pg_dumpall now refuses to deal with database and role

names containing carriage returns or newlines, as it seems impractical

to quote those characters safely on Windows. In future we may reject

such names on the server side, but that step has not been taken yet.

These are considered security fixes because crafted object names

containing special characters could have been used to execute

commands with superuser privileges the next time a superuser

executes pg_dumpall or other routine maintenance

operations. (CVE-2016-5424)

</li><li class="listitem">

Fix corner-case misbehaviors for <code class="literal">IS NULL</code>/<code class="literal">IS NOT

NULL</code> applied to nested composite values (Andrew Gierth, Tom Lane)

The SQL standard specifies that <code class="literal">IS NULL</code> should return

TRUE for a row of all null values (thus <code class="literal">ROW(NULL,NULL) IS

NULL</code> yields TRUE), but this is not meant to apply recursively

(thus <code class="literal">ROW(NULL, ROW(NULL,NULL)) IS NULL</code> yields FALSE).

The core executor got this right, but certain planner optimizations

treated the test as recursive (thus producing TRUE in both cases),

and <code class="filename">contrib/postgres_fdw</code> could produce remote queries

that misbehaved similarly.

</li><li class="listitem">

Fix “unrecognized node type” error for <code class="command">INSERT ... ON

CONFLICT</code> within a recursive CTE (a <code class="literal">WITH</code> item) (Peter

Geoghegan)

</li><li class="listitem">

Fix <code class="command">INSERT ... ON CONFLICT</code> to successfully match index

expressions or index predicates that are simplified during the

planner's expression preprocessing phase (Tom Lane)

</li><li class="listitem">

Correctly handle violations of exclusion constraints that apply to

the target table of an <code class="literal">INSERT ... ON CONFLICT</code> command,

but are not one of the selected arbiter indexes (Tom Lane)

Such a case should raise a normal constraint-violation error, but it

got into an infinite loop instead.

</li><li class="listitem">

Fix <code class="command">INSERT ... ON CONFLICT</code> to not fail if the target

table has a unique index on OID (Tom Lane)

</li><li class="listitem">

Make the <code class="type">inet</code> and <code class="type">cidr</code> data types properly reject

IPv6 addresses with too many colon-separated fields (Tom Lane)

</li><li class="listitem">

Prevent crash in <code class="function">close_ps()</code>

(the <code class="type">point</code> <code class="literal">##</code> <code class="type">lseg</code> operator)

for NaN input coordinates (Tom Lane)

Make it return NULL instead of crashing.

</li><li class="listitem">

Avoid possible crash in <code class="function">pg_get_expr()</code> when inconsistent

values are passed to it (Michael Paquier, Thomas Munro)

</li><li class="listitem">

100

Fix several one-byte buffer over-reads in <code class="function">to_number()</code>

101

(Peter Eisentraut)

102

103

In several cases the <code class="function">to_number()</code> function would read one

104

more character than it should from the input string. There is a

105

small chance of a crash, if the input happens to be adjacent to the

106

end of memory.

107

</li><li class="listitem">

108

Do not run the planner on the query contained in <code class="literal">CREATE

109

MATERIALIZED VIEW</code> or <code class="literal">CREATE TABLE AS</code>

110

when <code class="literal">WITH NO DATA</code> is specified (Michael Paquier,

111

Tom Lane)

112

113

This avoids some unnecessary failure conditions, for example if a

114

stable function invoked by the materialized view depends on a table

115

that doesn't exist yet.

116

</li><li class="listitem">

117

Avoid unsafe intermediate state during expensive paths

118

through <code class="function">heap_update()</code> (Masahiko Sawada, Andres Freund)

119

120

Previously, these cases locked the target tuple (by setting its XMAX)

121

but did not WAL-log that action, thus risking data integrity problems

122

if the page were spilled to disk and then a database crash occurred

123

before the tuple update could be completed.

124

</li><li class="listitem">

125

Fix hint bit update during WAL replay of row locking operations

126

(Andres Freund)

127

128

The only known consequence of this problem is that row locks held by

129

a prepared, but uncommitted, transaction might fail to be enforced

130

after a crash and restart.

131

</li><li class="listitem">

132

Avoid unnecessary “could not serialize access” errors when

133

acquiring <code class="literal">FOR KEY SHARE</code> row locks in serializable mode

134

(Álvaro Herrera)

135

</li><li class="listitem">

136

Make sure “expanded” datums returned by a plan node are

137

read-only (Tom Lane)

138

139

This avoids failures in some cases where the result of a lower plan

140

node is referenced in multiple places in upper nodes. So far as

141

core PostgreSQL is concerned, only array values

142

returned by PL/pgSQL functions are at risk; but extensions might

143

use expanded datums for other things.

144

</li><li class="listitem">

145

Avoid crash in <code class="literal">postgres -C</code> when the specified variable

146

has a null string value (Michael Paquier)

147

</li><li class="listitem">

148

Prevent unintended waits for the receiver in WAL sender processes

149

(Kyotaro Horiguchi)

150

</li><li class="listitem">

151

Fix possible loss of large subtransactions in logical decoding

152

(Petru-Florin Mihancea)

153

</li><li class="listitem">

154

Fix failure of logical decoding when a subtransaction contains no

155

actual changes (Marko Tiikkaja, Andrew Gierth)

156

</li><li class="listitem">

157

Ensure that backends see up-to-date statistics for shared catalogs

158

(Tom Lane)

159

160

The statistics collector failed to update the statistics file for

161

shared catalogs after a request from a regular backend. This problem

162

was partially masked because the autovacuum launcher regularly makes

163

requests that did cause such updates; however, it became obvious with

164

autovacuum disabled.

165

</li><li class="listitem">

166

Avoid redundant writes of the statistics files when multiple

167

backends request updates close together (Tom Lane, Tomas Vondra)

168

</li><li class="listitem">

169

Avoid consuming a transaction ID during <code class="command">VACUUM</code>

170

(Alexander Korotkov)

171

172

Some cases in <code class="command">VACUUM</code> unnecessarily caused an XID to be

173

assigned to the current transaction. Normally this is negligible,

174

but if one is up against the XID wraparound limit, consuming more

175

XIDs during anti-wraparound vacuums is a very bad thing.

176

</li><li class="listitem">

177

Prevent possible failure when vacuuming multixact IDs in an

178

installation that has been pg_upgrade'd from pre-9.3 (Andrew Gierth,

179

Álvaro Herrera)

180

181

The usual symptom of this bug is errors

182

like “MultiXactId <code>NNN</code> has not been created

183

yet -- apparent wraparound”.

184

</li><li class="listitem">

185

When a manual <code class="command">ANALYZE</code> specifies a column list, don't

186

reset the table's <code class="literal">changes_since_analyze</code> counter

187

(Tom Lane)

188

189

If we're only analyzing some columns, we should not prevent routine

190

auto-analyze from happening for the other columns.

191

</li><li class="listitem">

192

Fix <code class="command">ANALYZE</code>'s overestimation of <code class="literal">n_distinct</code>

193

for a unique or nearly-unique column with many null entries (Tom

194

Lane)

195

196

The nulls could get counted as though they were themselves distinct

197

values, leading to serious planner misestimates in some types of

198

queries.

199

</li><li class="listitem">

200

Prevent autovacuum from starting multiple workers for the same shared

201

catalog (Álvaro Herrera)

202

203

Normally this isn't much of a problem because the vacuum doesn't take

204

long anyway; but in the case of a severely bloated catalog, it could

205

result in all but one worker uselessly waiting instead of doing

206

useful work on other tables.

207

</li><li class="listitem">

208

Fix bug in b-tree mark/restore processing (Kevin Grittner)

209

210

This error could lead to incorrect join results or assertion failures

211

in a merge join whose inner source node is a b-tree indexscan.

212

</li><li class="listitem">

213

Avoid duplicate buffer lock release when abandoning a b-tree index

214

page deletion attempt (Tom Lane)

215

216

This mistake prevented <code class="command">VACUUM</code> from completing in some

217

cases involving corrupt b-tree indexes.

218

</li><li class="listitem">

219

Fix building of large (bigger than <code class="varname">shared_buffers</code>)

220

hash indexes (Tom Lane)

221

222

The code path used for large indexes contained a bug causing

223

incorrect hash values to be inserted into the index, so that

224

subsequent index searches always failed, except for tuples inserted

225

into the index after the initial build.

226

</li><li class="listitem">

227

Prevent infinite loop in GiST index build for geometric columns

228

containing NaN component values (Tom Lane)

229

</li><li class="listitem">

230

Fix possible crash during a nearest-neighbor (<code class="literal">ORDER BY</code>

231

distance) indexscan on a <code class="filename">contrib/btree_gist</code> index on

232

an <code class="type">interval</code> column (Peter Geoghegan)

233

</li><li class="listitem">

234

Fix “PANIC: failed to add BRIN tuple” error when attempting

235

to update a BRIN index entry (Álvaro Herrera)

236

</li><li class="listitem">

237

Fix possible crash during background worker shutdown (Dmitry Ivanov)

238

</li><li class="listitem">

239

Fix PL/pgSQL's handling of the <code class="literal">INTO</code> clause

240

within <code class="command">IMPORT FOREIGN SCHEMA</code> commands (Tom Lane)

241

</li><li class="listitem">

242

Fix <code class="filename">contrib/btree_gin</code> to handle the smallest

243

possible <code class="type">bigint</code> value correctly (Peter Eisentraut)

244

</li><li class="listitem">

245

Teach libpq to correctly decode server version from future servers

246

(Peter Eisentraut)

247

248

It's planned to switch to two-part instead of three-part server

249

version numbers for releases after 9.6. Make sure

250

that <code class="function">PQserverVersion()</code> returns the correct value for

251

such cases.

252

</li><li class="listitem">

253

Fix ecpg's code for <code class="literal">unsigned long long</code>

254

array elements (Michael Meskes)

255

</li><li class="listitem">

256

In pg_dump with both <code class="option">-c</code> and <code class="option">-C</code>

257

options, avoid emitting an unwanted <code class="literal">CREATE SCHEMA public</code>

258

command (David Johnston, Tom Lane)

259

</li><li class="listitem">

260

Improve handling of SIGTERM/control-C in

261

parallel pg_dump and pg_restore (Tom

262

Lane)

263

264

Make sure that the worker processes will exit promptly, and also arrange

265

to send query-cancel requests to the connected backends, in case they

266

are doing something long-running such as a <code class="command">CREATE INDEX</code>.

267

</li><li class="listitem">

268

Fix error reporting in parallel pg_dump

269

and pg_restore (Tom Lane)

270

271

Previously, errors reported by pg_dump

272

or pg_restore worker processes might never make it to

273

the user's console, because the messages went through the master

274

process, and there were various deadlock scenarios that would prevent

275

the master process from passing on the messages. Instead, just print

276

everything to <code class="literal">stderr</code>. In some cases this will result in

277

duplicate messages (for instance, if all the workers report a server

278

shutdown), but that seems better than no message.

279

</li><li class="listitem">

280

Ensure that parallel pg_dump

281

or pg_restore on Windows will shut down properly

282

after an error (Kyotaro Horiguchi)

283

284

Previously, it would report the error, but then just sit until

285

manually stopped by the user.

286

</li><li class="listitem">

287

Make parallel pg_dump fail cleanly when run against a

288

standby server (Magnus Hagander)

289

290

This usage is not supported

291

unless <code class="option">--no-synchronized-snapshots</code> is specified, but the

292

error was not handled very well.

293

</li><li class="listitem">

294

Make pg_dump behave better when built without zlib

295

support (Kyotaro Horiguchi)

296

297

It didn't work right for parallel dumps, and emitted some rather

298

pointless warnings in other cases.

299

</li><li class="listitem">

300

Make pg_basebackup accept <code class="literal">-Z 0</code> as

301

specifying no compression (Fujii Masao)

302

</li><li class="listitem">

303

Fix makefiles' rule for building AIX shared libraries to be safe for

304

parallel make (Noah Misch)

305

</li><li class="listitem">

306

Fix TAP tests and MSVC scripts to work when build directory's path

307

name contains spaces (Michael Paquier, Kyotaro Horiguchi)

308

</li><li class="listitem">

309

Be more predictable about reporting “statement timeout”

310

versus “lock timeout” (Tom Lane)

311

312

On heavily loaded machines, the regression tests sometimes failed due

313

to reporting “lock timeout” even though the statement timeout

314

should have occurred first.

315

</li><li class="listitem">

316

Make regression tests safe for Danish and Welsh locales (Jeff Janes,

317

Tom Lane)

318

319

Change some test data that triggered the unusual sorting rules of

320

these locales.

321

</li><li class="listitem">

322

Update our copy of the timezone code to match

323

IANA's tzcode release 2016c (Tom Lane)

324

325

This is needed to cope with anticipated future changes in the time

326

zone data files. It also fixes some corner-case bugs in coping with

327

unusual time zones.

328

</li><li class="listitem">

329

Update time zone data files to tzdata release 2016f

330

for DST law changes in Kemerovo and Novosibirsk, plus historical

331

corrections for Azerbaijan, Belarus, and Morocco.

332

</li></ul></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="release-9-5-5.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="release.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="release-9-5-3.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">E.25. Release 9.5.5 </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> E.27. Release 9.5.3</td></tr></table></div></body></html>

b'\\ No newline at end of file'

Older »