From 9430fb407bb64cc842e56fb5844265a9343dafba Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Wed, 20 Jul 2022 13:50:57 +1200 Subject: [PATCH] Add wal_sync_method=fdatasync for Windows. Windows 10 gained support for flushing NTFS files with fdatasync() semantics. The main advantage over open_datasync (in Windows API terms FILE_FLAG_WRITE_THROUGH) is that the latter does not flush SATA drive caches. The default setting is not changed, so users have to opt in to this. Discussion: https://postgr.es/m/CA%2BhUKGJZJVO%3DiX%2Beb-PXi2_XS9ZRqnn_4URh0NUQOwt6-_51xQ%40mail.gmail.com --- configure | 6 ++++ configure.ac | 1 + doc/src/sgml/wal.sgml | 3 +- src/include/c.h | 2 +- src/include/port/win32_port.h | 6 ++++ src/include/port/win32ntdll.h | 10 ++++++- src/port/fdatasync.c | 53 +++++++++++++++++++++++++++++++++++ src/port/win32ntdll.c | 6 +++- src/tools/msvc/Mkvcbuild.pm | 3 +- src/tools/msvc/Solution.pm | 2 +- 10 files changed, 86 insertions(+), 6 deletions(-) create mode 100644 src/port/fdatasync.c diff --git a/configure b/configure index a4f4d321fb..59fa82b8d7 100755 --- a/configure +++ b/configure @@ -17132,6 +17132,12 @@ fi ;; esac + case " $LIBOBJS " in + *" fdatasync.$ac_objext "* ) ;; + *) LIBOBJS="$LIBOBJS fdatasync.$ac_objext" + ;; +esac + case " $LIBOBJS " in *" kill.$ac_objext "* ) ;; *) LIBOBJS="$LIBOBJS kill.$ac_objext" diff --git a/configure.ac b/configure.ac index 5bd29a4d2f..612dabf698 100644 --- a/configure.ac +++ b/configure.ac @@ -1988,6 +1988,7 @@ if test "$PORTNAME" = "win32"; then AC_CHECK_FUNCS(_configthreadlocale) AC_REPLACE_FUNCS(gettimeofday) AC_LIBOBJ(dirmod) + AC_LIBOBJ(fdatasync) AC_LIBOBJ(kill) AC_LIBOBJ(open) AC_LIBOBJ(system) diff --git a/doc/src/sgml/wal.sgml b/doc/src/sgml/wal.sgml index 4b6ef283c1..01f7379ebb 100644 --- a/doc/src/sgml/wal.sgml +++ b/doc/src/sgml/wal.sgml @@ -108,7 +108,8 @@ open_datasync (the default), write caching can be disabled by unchecking My Computer\Open\disk drive\Properties\Hardware\Properties\Policies\Enable write caching on the disk. Alternatively, set wal_sync_method to - fsync or fsync_writethrough, which prevent + fdatasync (NTFS only), fsync or + fsync_writethrough, which prevent write caching. diff --git a/src/include/c.h b/src/include/c.h index 2cc2784750..d35405f191 100644 --- a/src/include/c.h +++ b/src/include/c.h @@ -1290,7 +1290,7 @@ typedef union PGAlignedXLogBlock * standard C library. */ -#if defined(HAVE_FDATASYNC) && !HAVE_DECL_FDATASYNC +#if !HAVE_DECL_FDATASYNC extern int fdatasync(int fildes); #endif diff --git a/src/include/port/win32_port.h b/src/include/port/win32_port.h index 5121c0c626..5ea66528fa 100644 --- a/src/include/port/win32_port.h +++ b/src/include/port/win32_port.h @@ -83,6 +83,12 @@ #define HAVE_FSYNC_WRITETHROUGH #define FSYNC_WRITETHROUGH_IS_FSYNC +/* + * We have a replacement for fdatasync() in src/port/fdatasync.c, which is + * unconditionally used by MSVC and Mingw builds. + */ +#define HAVE_FDATASYNC + #define USES_WINSOCK /* diff --git a/src/include/port/win32ntdll.h b/src/include/port/win32ntdll.h index 291b067ea4..34cebddd54 100644 --- a/src/include/port/win32ntdll.h +++ b/src/include/port/win32ntdll.h @@ -23,9 +23,17 @@ #include #include -typedef NTSTATUS (__stdcall * RtlGetLastNtStatus_t) (void); +#ifndef FLUSH_FLAGS_FILE_DATA_SYNC_ONLY +#define FLUSH_FLAGS_FILE_DATA_SYNC_ONLY 0x4 +#endif + +typedef NTSTATUS (__stdcall *RtlGetLastNtStatus_t) (void); +typedef ULONG (__stdcall *RtlNtStatusToDosError_t) (NTSTATUS); +typedef NTSTATUS (__stdcall *NtFlushBuffersFileEx_t) (HANDLE, ULONG, PVOID, ULONG, PIO_STATUS_BLOCK); extern PGDLLIMPORT RtlGetLastNtStatus_t pg_RtlGetLastNtStatus; +extern PGDLLIMPORT RtlNtStatusToDosError_t pg_RtlNtStatusToDosError; +extern PGDLLIMPORT NtFlushBuffersFileEx_t pg_NtFlushBuffersFileEx; extern int initialize_ntdll(void); diff --git a/src/port/fdatasync.c b/src/port/fdatasync.c new file mode 100644 index 0000000000..afef853aa3 --- /dev/null +++ b/src/port/fdatasync.c @@ -0,0 +1,53 @@ +/*------------------------------------------------------------------------- + * + * fdatasync.c + * Win32 fdatasync() replacement + * + * + * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group + * + * src/port/fdatasync.c + * + *------------------------------------------------------------------------- + */ + +#define UMDF_USING_NTSTATUS + +#ifdef FRONTEND +#include "postgres_fe.h" +#else +#include "postgres.h" +#endif + +#include "port/win32ntdll.h" + +int +fdatasync(int fd) +{ + IO_STATUS_BLOCK iosb; + NTSTATUS status; + HANDLE handle; + + handle = (HANDLE) _get_osfhandle(fd); + if (handle == INVALID_HANDLE_VALUE) + { + errno = EBADF; + return -1; + } + + if (initialize_ntdll() < 0) + return -1; + + memset(&iosb, 0, sizeof(iosb)); + status = pg_NtFlushBuffersFileEx(handle, + FLUSH_FLAGS_FILE_DATA_SYNC_ONLY, + NULL, + 0, + &iosb); + + if (NT_SUCCESS(status)) + return 0; + + _dosmaperr(pg_RtlNtStatusToDosError(status)); + return -1; +} diff --git a/src/port/win32ntdll.c b/src/port/win32ntdll.c index 10c33c6a01..eb61407754 100644 --- a/src/port/win32ntdll.c +++ b/src/port/win32ntdll.c @@ -20,6 +20,8 @@ #include "port/win32ntdll.h" RtlGetLastNtStatus_t pg_RtlGetLastNtStatus; +RtlNtStatusToDosError_t pg_RtlNtStatusToDosError; +NtFlushBuffersFileEx_t pg_NtFlushBuffersFileEx; typedef struct NtDllRoutine { @@ -28,7 +30,9 @@ typedef struct NtDllRoutine } NtDllRoutine; static const NtDllRoutine routines[] = { - {"RtlGetLastNtStatus", (pg_funcptr_t *) &pg_RtlGetLastNtStatus} + {"RtlGetLastNtStatus", (pg_funcptr_t *) &pg_RtlGetLastNtStatus}, + {"RtlNtStatusToDosError", (pg_funcptr_t *) &pg_RtlNtStatusToDosError}, + {"NtFlushBuffersFileEx", (pg_funcptr_t *) &pg_NtFlushBuffersFileEx} }; static bool initialized; diff --git a/src/tools/msvc/Mkvcbuild.pm b/src/tools/msvc/Mkvcbuild.pm index e4feda10fd..cc7a908d10 100644 --- a/src/tools/msvc/Mkvcbuild.pm +++ b/src/tools/msvc/Mkvcbuild.pm @@ -99,7 +99,8 @@ sub mkvcbuild $solution = CreateSolution($vsVersion, $config); our @pgportfiles = qw( - chklocale.c explicit_bzero.c fls.c getpeereid.c getrusage.c inet_aton.c + chklocale.c explicit_bzero.c fls.c fdatasync.c + getpeereid.c getrusage.c inet_aton.c getaddrinfo.c gettimeofday.c inet_net_ntop.c kill.c open.c snprintf.c strlcat.c strlcpy.c dirmod.c noblock.c path.c dirent.c dlopen.c getopt.c getopt_long.c link.c diff --git a/src/tools/msvc/Solution.pm b/src/tools/msvc/Solution.pm index 1e125aef94..3ddcd024a7 100644 --- a/src/tools/msvc/Solution.pm +++ b/src/tools/msvc/Solution.pm @@ -257,7 +257,7 @@ sub GenerateFiles HAVE_EDITLINE_READLINE_H => undef, HAVE_EXECINFO_H => undef, HAVE_EXPLICIT_BZERO => undef, - HAVE_FDATASYNC => undef, + HAVE_FDATASYNC => 1, HAVE_FLS => undef, HAVE_FSEEKO => 1, HAVE_FUNCNAME__FUNC => undef,