/* $Id: RTSemEventGetResolution-nt.cpp 93115 2022-01-01 11:31:46Z vboxsync $ */ /** @file * IPRT - Single Release Event Semaphores, RTSemEventGetResolution. */ /* * Copyright (C) 2006-2022 Oracle Corporation * * This file is part of VirtualBox Open Source Edition (OSE), as * available from http://www.virtualbox.org. This file is free software; * you can redistribute it and/or modify it under the terms of the GNU * General Public License (GPL) as published by the Free Software * Foundation, in version 2 as it comes in the "COPYING" file of the * VirtualBox OSE distribution. VirtualBox OSE is distributed in the * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind. * * The contents of this file may alternatively be used under the terms * of the Common Development and Distribution License Version 1.0 * (CDDL) only, as it comes in the "COPYING.CDDL" file of the * VirtualBox OSE distribution, in which case the provisions of the * CDDL are applicable instead of those of the GPL. * * You may elect to license modified versions of this file under the * terms and conditions of either the GPL or the CDDL or both. */ /********************************************************************************************************************************* * Header Files * *********************************************************************************************************************************/ #define RTSEMEVENT_WITHOUT_REMAPPING #ifdef IN_RING0 # include "../r0drv/nt/the-nt-kernel.h" #else # include #endif #include #include #include #ifdef IN_RING3 # include #endif RTDECL(uint32_t) RTSemEventGetResolution(void) { /* * We need to figure the KeWaitForSingleObject / NtWaitForSingleObject timeout * resolution, i.e. if we wish to wait for 1000ns how long are we likely to * actually wait before woken up. * * In older versions of NT, these timeout were implemented using KTIMERs and * have the same resolution as what them. This should be found using * ExSetTimerResolution or NtQueryTimerResolution. * * Probably since windows 8.1 the value returned by NtQueryTimerResolution (and * set NtSetTimerResolution) have been virtualized and no longer reflects the * timer wheel resolution, at least from what I can tell. ExSetTimerResolution * still works as before, but it accesses variable that I cannot find out how * to access from user land. So, kernel will get (and be able to set) the right * granularity, while in user land we'll be forced to reporting the max value. * * (The reason why I suspect it's since 8.1 is because the high resolution * ExSetTimer APIs were introduced back then.) */ #ifdef IN_RING0 return RTTimerGetSystemGranularity(); #else ULONG cNtTicksMin = 0; ULONG cNtTicksMax = 0; ULONG cNtTicksCur = 0; NTSTATUS rcNt = NtQueryTimerResolution(&cNtTicksMin, &cNtTicksMax, &cNtTicksCur); if (NT_SUCCESS(rcNt)) { Assert(cNtTicksMin >= cNtTicksMax); if (RTSystemGetNtVersion() >= RTSYSTEM_MAKE_NT_VERSION(6,3,9600)) /** @todo check when the switch happened, might be much later... */ return cNtTicksMin * 100; return cNtTicksCur * 100; } AssertFailed(); return 16 * RT_NS_1MS; /* the default on 64-bit windows 10 */ #endif }