From patchwork Wed Nov 15 07:14:12 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Xiaoyao Li X-Patchwork-Id: 13456236 Received: from lindbergh.monkeyblade.net (lindbergh.monkeyblade.net [23.128.96.19]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id D53E2611A for ; Wed, 15 Nov 2023 07:15:55 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="AmJn1HXQ" Received: from mgamail.intel.com (mgamail.intel.com [192.55.52.115]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 0E69BE9 for ; Tue, 14 Nov 2023 23:15:55 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1700032555; x=1731568555; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=qY9pdRRi3gnZ4VLEVBqiSbNccgKMU+XVba+RGctQW98=; b=AmJn1HXQStnv9zXy1y0tfQUpFdY7pTck042UxmW9H8ZuuuGkufZDD8kV CCO9G73mzeUAJu9m1aN2qbAqj48Gly1J+EDRzvdoYhQuqp8X8462GHBZr ubmBKmYTTpUAC9TMVjaQlpUm5dr2g9hCR5eJBnocON4TjRTYCNeZPxqIw W+XqGxZfG+qaV6PmRw6NLBZQDJJWlnV3sGBUTp0hdeKBteRDimUIRekrx zBUbnL8wluh3GuPTGV/LQ44TukM6mUkPvU7C04iabSdkp/kDNHFlY4Whk arIrlQTGTL/iRiUn8uJauaOjtZVJfJNpC+30BnT3ddHnEJPRZR7NpLxkL Q==; X-IronPort-AV: E=McAfee;i="6600,9927,10894"; a="390622077" X-IronPort-AV: E=Sophos;i="6.03,304,1694761200"; d="scan'208";a="390622077" Received: from orsmga003.jf.intel.com ([10.7.209.27]) by fmsmga103.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 14 Nov 2023 23:15:54 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10894"; a="714796837" X-IronPort-AV: E=Sophos;i="6.03,304,1694761200"; d="scan'208";a="714796837" Received: from lxy-clx-4s.sh.intel.com ([10.239.48.52]) by orsmga003.jf.intel.com with ESMTP; 14 Nov 2023 23:15:46 -0800 From: Xiaoyao Li To: Paolo Bonzini , David Hildenbrand , Igor Mammedov , "Michael S . Tsirkin" , Marcel Apfelbaum , Richard Henderson , Peter Xu , =?utf-8?q?Philippe_Mathieu-Daud=C3=A9?= , Cornelia Huck , =?utf-8?q?Daniel_P_=2E_Berrang=C3=A9?= , Eric Blake , Markus Armbruster , Marcelo Tosatti Cc: qemu-devel@nongnu.org, kvm@vger.kernel.org, xiaoyao.li@intel.com, Michael Roth , Sean Christopherson , Claudio Fontana , Gerd Hoffmann , Isaku Yamahata , Chenyi Qiang Subject: [PATCH v3 03/70] RAMBlock/guest_memfd: Enable KVM_GUEST_MEMFD_ALLOW_HUGEPAGE Date: Wed, 15 Nov 2023 02:14:12 -0500 Message-Id: <20231115071519.2864957-4-xiaoyao.li@intel.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20231115071519.2864957-1-xiaoyao.li@intel.com> References: <20231115071519.2864957-1-xiaoyao.li@intel.com> Precedence: bulk X-Mailing-List: kvm@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 KVM allows KVM_GUEST_MEMFD_ALLOW_HUGEPAGE for guest memfd. When the flag is set, KVM tries to allocate memory with transparent hugeapge at first and falls back to non-hugepage on failure. However, KVM defines one restriction that size must be hugepage size aligned when KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is set. Signed-off-by: Xiaoyao Li --- v3: - New one in v3. --- system/physmem.c | 38 ++++++++++++++++++++++++++++++++++++-- 1 file changed, 36 insertions(+), 2 deletions(-) diff --git a/system/physmem.c b/system/physmem.c index 0af2213cbd9c..c56b17e44df6 100644 --- a/system/physmem.c +++ b/system/physmem.c @@ -1803,6 +1803,40 @@ static void dirty_memory_extend(ram_addr_t old_ram_size, } } +#ifdef CONFIG_KVM +#define HPAGE_PMD_SIZE_PATH "/sys/kernel/mm/transparent_hugepage/hpage_pmd_size" +#define DEFAULT_PMD_SIZE (1ul << 21) + +static uint32_t get_thp_size(void) +{ + gchar *content = NULL; + const char *endptr; + static uint64_t thp_size = 0; + uint64_t tmp; + + if (thp_size != 0) { + return thp_size; + } + + if (g_file_get_contents(HPAGE_PMD_SIZE_PATH, &content, NULL, NULL) && + !qemu_strtou64(content, &endptr, 0, &tmp) && + (!endptr || *endptr == '\n')) { + /* Sanity-check the value and fallback to something reasonable. */ + if (!tmp || !is_power_of_2(tmp)) { + warn_report("Read unsupported THP size: %" PRIx64, tmp); + } else { + thp_size = tmp; + } + } + + if (!thp_size) { + thp_size = DEFAULT_PMD_SIZE; + } + + return thp_size; +} +#endif + static void ram_block_add(RAMBlock *new_block, Error **errp) { const bool noreserve = qemu_ram_is_noreserve(new_block); @@ -1844,8 +1878,8 @@ static void ram_block_add(RAMBlock *new_block, Error **errp) #ifdef CONFIG_KVM if (kvm_enabled() && new_block->flags & RAM_GUEST_MEMFD && new_block->guest_memfd < 0) { - /* TODO: to decide if KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is supported */ - uint64_t flags = 0; + uint64_t flags = QEMU_IS_ALIGNED(new_block->max_length, get_thp_size()) ? + KVM_GUEST_MEMFD_ALLOW_HUGEPAGE : 0; new_block->guest_memfd = kvm_create_guest_memfd(new_block->max_length, flags, errp); if (new_block->guest_memfd < 0) {