Add huge_page_size setting for use on Linux.

This allows the huge page size to be set explicitly.  The default is 0,
meaning it will use the system default, as before.

Author: Odin Ugedal <odin@ugedal.com>
Discussion: https://postgr.es/m/20200608154639.20254-1-odin%40ugedal.com

6 files changed, 141 insertions, 38 deletions

diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index b353c616830..e0ea397ed40 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
@@ -1582,6 +1582,33 @@ include_dir 'conf.d'
       </listitem>
      </varlistentry>
 
+     <varlistentry id="guc-huge-page-size" xreflabel="huge_page_size">
+      <term><varname>huge_page_size</varname> (<type>integer</type>)
+      <indexterm>
+       <primary><varname>huge_page_size</varname> configuration parameter</primary>
+      </indexterm>
+      </term>
+      <listitem>
+       <para>
+        Controls the size of huge pages, when they are enabled with
+        <xref linkend="guc-huge-pages"/>.
+        The default is zero (<literal>0</literal>).
+        When set to <literal>0</literal>, the default huge page size on the
+        system will be used.
+       </para>
+       <para>
+        Some commonly available page sizes on modern 64 bit server architectures include:
+        <literal>2MB</literal> and <literal>1GB</literal> (Intel and AMD), <literal>16MB</literal> and
+        <literal>16GB</literal> (IBM POWER), and <literal>64kB</literal>, <literal>2MB</literal>,
+        <literal>32MB</literal> and <literal>1GB</literal> (ARM). For more information
+        about usage and support, see <xref linkend="linux-huge-pages"/>.
+       </para>
+       <para>
+        Non-default settings are currently supported only on Linux.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry id="guc-temp-buffers" xreflabel="temp_buffers">
       <term><varname>temp_buffers</varname> (<type>integer</type>)
       <indexterm>
diff --git a/doc/src/sgml/runtime.sgml b/doc/src/sgml/runtime.sgml
index 937bb2e8ac9..e09cb55efcd 100644
--- a/doc/src/sgml/runtime.sgml
+++ b/doc/src/sgml/runtime.sgml
@@ -1391,13 +1391,14 @@ export PG_OOM_ADJUST_VALUE=0
     using large values of <xref linkend="guc-shared-buffers"/>.  To use this
     feature in <productname>PostgreSQL</productname> you need a kernel
     with <varname>CONFIG_HUGETLBFS=y</varname> and
-    <varname>CONFIG_HUGETLB_PAGE=y</varname>. You will also have to adjust
-    the kernel setting <varname>vm.nr_hugepages</varname>. To estimate the
-    number of huge pages needed, start <productname>PostgreSQL</productname>
-    without huge pages enabled and check the
-    postmaster's anonymous shared memory segment size, as well as the system's
-    huge page size, using the <filename>/proc</filename> file system.  This might
-    look like:
+    <varname>CONFIG_HUGETLB_PAGE=y</varname>. You will also have to configure
+    the operating system to provide enough huge pages of the desired size.
+    To estimate the number of huge pages needed, start
+    <productname>PostgreSQL</productname> without huge pages enabled and check
+    the postmaster's anonymous shared memory segment size, as well as the
+    system's default and supported huge page sizes, using the
+    <filename>/proc</filename> and <filename>/sys</filename> file systems.
+    This might look like:
 <programlisting>
 $ <userinput>head -1 $PGDATA/postmaster.pid</userinput>
 4170
@@ -1405,27 +1406,40 @@ $ <userinput>pmap 4170 | awk '/rw-s/ &amp;&amp; /zero/ {print $2}'</userinput>
 6490428K
 $ <userinput>grep ^Hugepagesize /proc/meminfo</userinput>
 Hugepagesize:       2048 kB
+$ <userinput>ls /sys/kernel/mm/hugepages</userinput>
+hugepages-1048576kB  hugepages-2048kB
 </programlisting>
+
+     In this example the default is 2MB, but you can also explicitly request
+     either 2MB or 1GB with <xref linkend="guc-huge-page-size"/>.
+
+     Assuming <literal>2MB</literal> huge pages,
      <literal>6490428</literal> / <literal>2048</literal> gives approximately
      <literal>3169.154</literal>, so in this example we need at
-     least <literal>3170</literal> huge pages, which we can set with:
+     least <literal>3170</literal> huge pages.  A larger setting would be
+     appropriate if other programs on the machine also need huge pages.
+     We can set this with:
+<programlisting>
+# <userinput>sysctl -w vm.nr_hugepages=3170</userinput>
+</programlisting>
+     Don't forget to add this setting to <filename>/etc/sysctl.conf</filename>
+     so that it is reapplied after reboots.  For non-default huge page sizes,
+     we can instead use:
 <programlisting>
-$ <userinput>sysctl -w vm.nr_hugepages=3170</userinput>
+# <userinput>echo 3170 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages</userinput>
 </programlisting>
-    A larger setting would be appropriate if other programs on the machine
-    also need huge pages.  Don't forget to add this setting
-    to <filename>/etc/sysctl.conf</filename> so that it will be reapplied
-    after reboots.
+    It is also possible to provide these settings at boot time using
+    kernel parameters such as <literal>hugepagesz=2M hugepages=3170</literal>.
    </para>
 
    <para>
     Sometimes the kernel is not able to allocate the desired number of huge
-    pages immediately, so it might be necessary to repeat the command or to
-    reboot.  (Immediately after a reboot, most of the machine's memory
-    should be available to convert into huge pages.)  To verify the huge
-    page allocation situation, use:
+    pages immediately due to fragmentation, so it might be necessary
+    to repeat the command or to reboot.  (Immediately after a reboot, most of
+    the machine's memory should be available to convert into huge pages.)
+    To verify the huge page allocation situation for a given size, use:
 <programlisting>
-$ <userinput>grep Huge /proc/meminfo</userinput>
+$ <userinput>cat /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages</userinput>
 </programlisting>
    </para>
 
@@ -1438,8 +1452,9 @@ $ <userinput>grep Huge /proc/meminfo</userinput>
 
    <para>
     The default behavior for huge pages in
-    <productname>PostgreSQL</productname> is to use them when possible and
-    to fall back to normal pages when failing. To enforce the use of huge
+    <productname>PostgreSQL</productname> is to use them when possible, with
+    the system's default huge page size, and
+    to fall back to normal pages on failure. To enforce the use of huge
     pages, you can set <xref linkend="guc-huge-pages"/>
     to <literal>on</literal> in <filename>postgresql.conf</filename>.
     Note that with this setting <productname>PostgreSQL</productname> will fail to
diff --git a/src/backend/port/sysv_shmem.c b/src/backend/port/sysv_shmem.c
index 198a6985bf3..203555822d9 100644
--- a/src/backend/port/sysv_shmem.c
+++ b/src/backend/port/sysv_shmem.c
@@ -32,6 +32,7 @@
 #endif
 
 #include "miscadmin.h"
+#include "port/pg_bitutils.h"
 #include "portability/mem.h"
 #include "storage/dsm.h"
 #include "storage/fd.h"
@@ -448,7 +449,7 @@ PGSharedMemoryAttach(IpcMemoryId shmId,
 #ifdef MAP_HUGETLB
 
 /*
- * Identify the huge page size to use.
+ * Identify the huge page size to use, and compute the related mmap flags.
  *
  * Some Linux kernel versions have a bug causing mmap() to fail on requests
  * that are not a multiple of the hugepage size.  Versions without that bug
@@ -464,25 +465,13 @@ PGSharedMemoryAttach(IpcMemoryId shmId,
  * hugepage sizes, we might want to think about more invasive strategies,
  * such as increasing shared_buffers to absorb the extra space.
  *
- * Returns the (real or assumed) page size into *hugepagesize,
+ * Returns the (real, assumed or config provided) page size into *hugepagesize,
  * and the hugepage-related mmap flags to use into *mmap_flags.
- *
- * Currently *mmap_flags is always just MAP_HUGETLB.  Someday, on systems
- * that support it, we might OR in additional bits to specify a particular
- * non-default huge page size.
  */
 static void
 GetHugePageSize(Size *hugepagesize, int *mmap_flags)
 {
-	/*
-	 * If we fail to find out the system's default huge page size, assume it
-	 * is 2MB.  This will work fine when the actual size is less.  If it's
-	 * more, we might get mmap() or munmap() failures due to unaligned
-	 * requests; but at this writing, there are no reports of any non-Linux
-	 * systems being picky about that.
-	 */
-	*hugepagesize = 2 * 1024 * 1024;
-	*mmap_flags = MAP_HUGETLB;
+	Size		default_hugepagesize = 0;
 
 	/*
 	 * System-dependent code to find out the default huge page size.
@@ -491,6 +480,7 @@ GetHugePageSize(Size *hugepagesize, int *mmap_flags)
 	 * nnnn kB".  Ignore any failures, falling back to the preset default.
 	 */
 #ifdef __linux__
+
 	{
 		FILE	   *fp = AllocateFile("/proc/meminfo", "r");
 		char		buf[128];
@@ -505,7 +495,7 @@ GetHugePageSize(Size *hugepagesize, int *mmap_flags)
 				{
 					if (ch == 'k')
 					{
-						*hugepagesize = sz * (Size) 1024;
+						default_hugepagesize = sz * (Size) 1024;
 						break;
 					}
 					/* We could accept other units besides kB, if needed */
@@ -515,6 +505,44 @@ GetHugePageSize(Size *hugepagesize, int *mmap_flags)
 		}
 	}
 #endif							/* __linux__ */
+
+	if (huge_page_size != 0)
+	{
+		/* If huge page size is requested explicitly, use that. */
+		*hugepagesize = (Size) huge_page_size * 1024;
+	}
+	else if (default_hugepagesize != 0)
+	{
+		/* Otherwise use the system default, if we have it. */
+		*hugepagesize = default_hugepagesize;
+	}
+	else
+	{
+		/*
+		 * If we fail to find out the system's default huge page size, or no
+		 * huge page size is requested explicitly, assume it is 2MB. This will
+		 * work fine when the actual size is less.  If it's more, we might get
+		 * mmap() or munmap() failures due to unaligned requests; but at this
+		 * writing, there are no reports of any non-Linux systems being picky
+		 * about that.
+		 */
+		*hugepagesize = 2 * 1024 * 1024;
+	}
+
+	*mmap_flags = MAP_HUGETLB;
+
+	/*
+	 * On recent enough Linux, also include the explicit page size, if
+	 * necessary.
+	 */
+#if defined(MAP_HUGE_MASK) && defined(MAP_HUGE_SHIFT)
+	if (*hugepagesize != default_hugepagesize)
+	{
+		int			shift = pg_ceil_log2_64(*hugepagesize);
+
+		*mmap_flags |= (shift & MAP_HUGE_MASK) << MAP_HUGE_SHIFT;
+	}
+#endif
 }
 
 #endif							/* MAP_HUGETLB */
@@ -583,7 +611,7 @@ CreateAnonymousSegment(Size *size)
 						 "(currently %zu bytes), reduce PostgreSQL's shared "
 						 "memory usage, perhaps by reducing shared_buffers or "
 						 "max_connections.",
-						 *size) : 0));
+						 allocsize) : 0));
 	}
 
 	*size = allocsize;
diff --git a/src/backend/utils/misc/guc.c b/src/backend/utils/misc/guc.c
index 031ca0327f0..99a3e4f6f65 100644
--- a/src/backend/utils/misc/guc.c
+++ b/src/backend/utils/misc/guc.c
@@ -20,11 +20,14 @@
 #include <float.h>
 #include <math.h>
 #include <limits.h>
-#include <unistd.h>
+#ifndef WIN32
+#include <sys/mman.h>
+#endif
 #include <sys/stat.h>
 #ifdef HAVE_SYSLOG
 #include <syslog.h>
 #endif
+#include <unistd.h>
 
 #include "access/commit_ts.h"
 #include "access/gin.h"
@@ -198,6 +201,7 @@ static bool check_max_wal_senders(int *newval, void **extra, GucSource source);
 static bool check_autovacuum_work_mem(int *newval, void **extra, GucSource source);
 static bool check_effective_io_concurrency(int *newval, void **extra, GucSource source);
 static bool check_maintenance_io_concurrency(int *newval, void **extra, GucSource source);
+static bool check_huge_page_size(int *newval, void **extra, GucSource source);
 static void assign_pgstat_temp_directory(const char *newval, void *extra);
 static bool check_application_name(char **newval, void **extra, GucSource source);
 static void assign_application_name(const char *newval, void *extra);
@@ -576,6 +580,7 @@ int			ssl_renegotiation_limit;
  * need to be duplicated in all the different implementations of pg_shmem.c.
  */
 int			huge_pages;
+int			huge_page_size;
 
 /*
  * These variables are all dummies that don't do anything, except in some
@@ -3381,6 +3386,17 @@ static struct config_int ConfigureNamesInt[] =
 		NULL, assign_tcp_user_timeout, show_tcp_user_timeout
 	},
 
+	{
+		{"huge_page_size", PGC_POSTMASTER, RESOURCES_MEM,
+			gettext_noop("The size of huge page that should be requested."),
+			NULL,
+			GUC_UNIT_KB
+		},
+		&huge_page_size,
+		0, 0, INT_MAX,
+		check_huge_page_size, NULL, NULL
+	},
+
 	/* End-of-list marker */
 	{
 		{NULL, 0, 0, NULL, NULL}, NULL, 0, 0, 0, NULL, NULL, NULL
@@ -11565,6 +11581,20 @@ check_maintenance_io_concurrency(int *newval, void **extra, GucSource source)
 	return true;
 }
 
+static bool
+check_huge_page_size(int *newval, void **extra, GucSource source)
+{
+#if !(defined(MAP_HUGE_MASK) && defined(MAP_HUGE_SHIFT))
+	/* Recent enough Linux only, for now.  See GetHugePageSize(). */
+	if (*newval != 0)
+	{
+		GUC_check_errdetail("huge_page_size must be 0 on this platform.");
+		return false;
+	}
+#endif
+	return true;
+}
+
 static void
 assign_pgstat_temp_directory(const char *newval, void *extra)
 {
diff --git a/src/backend/utils/misc/postgresql.conf.sample b/src/backend/utils/misc/postgresql.conf.sample
index e430e33c7b4..29e01521966 100644
--- a/src/backend/utils/misc/postgresql.conf.sample
+++ b/src/backend/utils/misc/postgresql.conf.sample
@@ -122,6 +122,8 @@
 					# (change requires restart)
 #huge_pages = try			# on, off, or try
 					# (change requires restart)
+#huge_page_size = 0			# zero for system default
+					# (change requires restart)
 #temp_buffers = 8MB			# min 800kB
 #max_prepared_transactions = 0		# zero disables the feature
 					# (change requires restart)
diff --git a/src/include/storage/pg_shmem.h b/src/include/storage/pg_shmem.h
index 0de26b34276..9992932a005 100644
--- a/src/include/storage/pg_shmem.h
+++ b/src/include/storage/pg_shmem.h
@@ -44,6 +44,7 @@ typedef struct PGShmemHeader	/* standard header for all Postgres shmem */
 /* GUC variables */
 extern int	shared_memory_type;
 extern int	huge_pages;
+extern int	huge_page_size;
 
 /* Possible values for huge_pages */
 typedef enum