Hi,
PG95/pg_test_fsync utility (MSVC built) is crashing on Windows and It is caused by gettimeofday (src\port\gettimeofday.c) function. The following check-in raised this issue i.e.
commit 8001fe67a3d66c95861ce1f7075ef03953670d13
Author: Simon Riggs <simon@2ndQuadrant.com>
Date: Mon Dec 8 23:36:06 2014 +0900
Windows: use GetSystemTimePreciseAsFileTime if available
PostgreSQL on Windows 8 or Windows Server 2012 will now
get high-resolution timestamps by dynamically loading the
GetSystemTimePreciseAsFileTime function. It'll fall back to
to GetSystemTimeAsFileTime if the higher precision variant
isn't found, so the same binaries without problems on older
Windows releases.
No attempt is made to detect the Windows version. Only the
presence or absence of the desired function is considered.
Craig Ringer
It make PG gettimeofday() implementation depend on init_win32_gettimeofday() function call, otherwise gettimeofday() function is going to crash the related process because of NULL pg_get_system_time function pointer. PFA patch gettimeofday_win32_v1.patch, it enables necessary initialization in related utilities.
PFA patch gettimeofday_win32_suggestion_v1.patch, it suggest minor change to make it more versatile version of PG gettimeofday() function. It makes it more crash safe and by default it uses much faster or lesser CPU intensive GetSystemTimeAsFileTime function. If a pg utility do require high precision, it can call init_win32_precise_gettimeofday() to try to elevate the precision when available. Please do let me know if I missed something or more information is required. Thanks.
Regards,
Muhammad Asif Naeem
