|
23 | 23 | </para>
|
24 | 24 |
|
25 | 25 | <para>
|
26 |
| - However, if you use BRIN indexes, see the first changelog entry below. |
| 26 | + However, if you use BRIN indexes, see the fourth changelog entry below. |
27 | 27 | </para>
|
28 | 28 |
|
29 | 29 | <para>
|
|
37 | 37 |
|
38 | 38 | <itemizedlist>
|
39 | 39 |
|
| 40 | + <listitem> |
| 41 | + <para> |
| 42 | + Ensure that <literal>INSERT ... ON CONFLICT DO UPDATE</literal> checks |
| 43 | + table permissions and RLS policies in all cases (Dean Rasheed) |
| 44 | + </para> |
| 45 | + |
| 46 | + <para> |
| 47 | + The update path of <literal>INSERT ... ON CONFLICT DO UPDATE</literal> |
| 48 | + requires <literal>SELECT</literal> permission on the columns of the |
| 49 | + arbiter index, but it failed to check for that in the case of an |
| 50 | + arbiter specified by constraint name. |
| 51 | + In addition, for a table with row level security enabled, it failed to |
| 52 | + check updated rows against the table's <literal>SELECT</literal> |
| 53 | + policies (regardless of how the arbiter index was specified). |
| 54 | + (CVE-2017-15099) |
| 55 | + </para> |
| 56 | + </listitem> |
| 57 | + |
| 58 | + <listitem> |
| 59 | + <para> |
| 60 | + Fix crash due to rowtype mismatch |
| 61 | + in <function>json{b}_populate_recordset()</function> |
| 62 | + (Michael Paquier, Tom Lane) |
| 63 | + </para> |
| 64 | + |
| 65 | + <para> |
| 66 | + These functions used the result rowtype specified in the <literal>FROM |
| 67 | + ... AS</literal> clause without checking that it matched the actual |
| 68 | + rowtype of the supplied tuple value. If it didn't, that would usually |
| 69 | + result in a crash, though disclosure of server memory contents seems |
| 70 | + possible as well. |
| 71 | + (CVE-2017-15098) |
| 72 | + </para> |
| 73 | + </listitem> |
| 74 | + |
| 75 | + <listitem> |
| 76 | + <para> |
| 77 | + Fix sample server-start scripts to become <literal>$PGUSER</literal> |
| 78 | + before opening <literal>$PGLOG</literal> (Noah Misch) |
| 79 | + </para> |
| 80 | + |
| 81 | + <para> |
| 82 | + Previously, the postmaster log file was opened while still running as |
| 83 | + root. The database owner could therefore mount an attack against |
| 84 | + another system user by making <literal>$PGLOG</literal> be a symbolic |
| 85 | + link to some other file, which would then become corrupted by appending |
| 86 | + log messages. |
| 87 | + </para> |
| 88 | + |
| 89 | + <para> |
| 90 | + By default, these scripts are not installed anywhere. Users who have |
| 91 | + made use of them will need to manually recopy them, or apply the same |
| 92 | + changes to their modified versions. If the |
| 93 | + existing <literal>$PGLOG</literal> file is root-owned, it will need to |
| 94 | + be removed or renamed out of the way before restarting the server with |
| 95 | + the corrected script. |
| 96 | + (CVE-2017-12172) |
| 97 | + </para> |
| 98 | + </listitem> |
| 99 | + |
40 | 100 | <listitem>
|
41 | 101 | <para>
|
42 | 102 | Fix BRIN index summarization to handle concurrent table extension
|
@@ -459,6 +519,19 @@ Branch: REL9_6_STABLE [407e66078] 2017-09-14 01:17:15 +0200
|
459 | 519 | </para>
|
460 | 520 | </listitem>
|
461 | 521 |
|
| 522 | + <listitem> |
| 523 | + <para> |
| 524 | + Fix missing temp-install prerequisites |
| 525 | + for <literal>check</literal>-like Make targets (Noah Misch) |
| 526 | + </para> |
| 527 | + |
| 528 | + <para> |
| 529 | + Some non-default test procedures that are meant to work |
| 530 | + like <literal>make check</literal> failed to ensure that the temporary |
| 531 | + installation was up to date. |
| 532 | + </para> |
| 533 | + </listitem> |
| 534 | + |
462 | 535 | <listitem>
|
463 | 536 | <!--
|
464 | 537 | Author: Tom Lane < [email protected]>
|
|
0 commit comments