From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:34671) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eU5eJ-0000Pe-OD for qemu-devel@nongnu.org; Wed, 27 Dec 2017 01:57:12 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eU5eF-0002QG-3P for qemu-devel@nongnu.org; Wed, 27 Dec 2017 01:57:11 -0500 Received: from mga11.intel.com ([192.55.52.93]:40921) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1eU5eE-0002Ng-Rb for qemu-devel@nongnu.org; Wed, 27 Dec 2017 01:57:07 -0500 From: Haozhong Zhang Date: Wed, 27 Dec 2017 14:56:20 +0800 Message-Id: <20171227065620.20889-1-haozhong.zhang@intel.com> Subject: [Qemu-devel] [PATCH] util/mmap-alloc: support MAP_SYNC in qemu_ram_mmap() List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Xiao Guangrong , mst@redhat.com, Eduardo Habkost , Igor Mammedov , Stefan Hajnoczi , Dan Williams , Haozhong Zhang When a file supporting DAX is used as vNVDIMM backend, mmap it with MAP_SYNC flag in addition can guarantee the persistence of guest write to the backend file without other QEMU actions (e.g., periodic fsync() by QEMU). By using MAP_SHARED_VALIDATE flag with MAP_SYNC, we can ensure mmap with MAP_SYNC fails if MAP_SYNC is not supported by the kernel or the backend file. On such failures, QEMU retries mmap without MAP_SYNC and MAP_SHARED_VALIDATE. Signed-off-by: Haozhong Zhang --- util/mmap-alloc.c | 36 +++++++++++++++++++++++++++++++++++- 1 file changed, 35 insertions(+), 1 deletion(-) diff --git a/util/mmap-alloc.c b/util/mmap-alloc.c index 2fd8cbcc6f..37b302f057 100644 --- a/util/mmap-alloc.c +++ b/util/mmap-alloc.c @@ -18,7 +18,18 @@ #ifdef CONFIG_LINUX #include + +/* + * MAP_SHARED_VALIDATE and MAP_SYNC were introduced in 4.15 kernel, so + * they may not be defined when compiling on older kernels. + */ +#ifndef MAP_SHARED_VALIDATE +#define MAP_SHARED_VALIDATE 0x3 #endif +#ifndef MAP_SYNC +#define MAP_SYNC 0x80000 +#endif +#endif /* CONFIG_LINUX */ size_t qemu_fd_getpagesize(int fd) { @@ -97,6 +108,7 @@ void *qemu_ram_mmap(int fd, size_t size, size_t align, bool shared) #endif size_t offset; void *ptr1; + int xflags = 0; if (ptr == MAP_FAILED) { return MAP_FAILED; @@ -107,12 +119,34 @@ void *qemu_ram_mmap(int fd, size_t size, size_t align, bool shared) assert(align >= getpagesize()); offset = QEMU_ALIGN_UP((uintptr_t)ptr, align) - (uintptr_t)ptr; + +#if defined(__linux__) + /* + * If 'fd' refers to a file supporting DAX, mmap it with MAP_SYNC + * will guarantee the guest write persistence without other + * actions in QEMU (e.g., fsync() in QEMU). + * + * MAP_SHARED_VALIDATE ensures mmap with MAP_SYNC fails if + * MAP_SYNC is not supported by the kernel or the file. + * + * On failures of mmap with xflags, QEMU will retry mmap without + * xflags. + */ + xflags = shared ? (MAP_SHARED_VALIDATE | MAP_SYNC) : 0; +#endif + + retry_mmap_fd: ptr1 = mmap(ptr + offset, size, PROT_READ | PROT_WRITE, MAP_FIXED | (fd == -1 ? MAP_ANONYMOUS : 0) | - (shared ? MAP_SHARED : MAP_PRIVATE), + (shared ? MAP_SHARED : MAP_PRIVATE) | xflags, fd, 0); if (ptr1 == MAP_FAILED) { + if (xflags) { + xflags = 0; + goto retry_mmap_fd; + } + munmap(ptr, total); return MAP_FAILED; } -- 2.14.1