2015-09-24 19:41:17 +08:00
|
|
|
/*
|
|
|
|
* Support for RAM backed by mmaped host memory.
|
|
|
|
*
|
|
|
|
* Copyright (c) 2015 Red Hat, Inc.
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Michael S. Tsirkin <mst@redhat.com>
|
|
|
|
*
|
|
|
|
* This work is licensed under the terms of the GNU GPL, version 2 or
|
|
|
|
* later. See the COPYING file in the top-level directory.
|
|
|
|
*/
|
2016-06-23 01:11:19 +08:00
|
|
|
|
2016-01-30 01:49:55 +08:00
|
|
|
#include "qemu/osdep.h"
|
2016-06-23 01:11:19 +08:00
|
|
|
#include "qemu/mmap-alloc.h"
|
2016-11-02 21:44:46 +08:00
|
|
|
#include "qemu/host-utils.h"
|
2015-09-24 19:41:17 +08:00
|
|
|
|
util/mmap-alloc: fix hugetlb support on ppc64
Since commit 8561c9244ddf1122d "exec: allocate PROT_NONE pages on top of
RAM", it is no longer possible to back guest RAM with hugepages on ppc64
hosts:
mmap(NULL, 285212672, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x3fff57000000
mmap(0x3fff57000000, 268435456, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 19, 0) = -1 EBUSY (Device or resource busy)
This is because on ppc64, Linux fixes a page size for a virtual address
at mmap time, so we can't switch a range of memory from anonymous
small pages to hugetlbs with MAP_FIXED.
See commit d0f13e3c20b6fb73ccb467bdca97fa7cf5a574cd
("[POWERPC] Introduce address space "slices"") in Linux
history for the details.
Detect this and create the PROT_NONE mapping using the same fd.
Naturally, this makes the guard page bigger with hugetlbfs.
Based on patch by Greg Kurz.
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2015-12-03 03:14:12 +08:00
|
|
|
#define HUGETLBFS_MAGIC 0x958458f6
|
|
|
|
|
|
|
|
#ifdef CONFIG_LINUX
|
|
|
|
#include <sys/vfs.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
size_t qemu_fd_getpagesize(int fd)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_LINUX
|
|
|
|
struct statfs fs;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (fd != -1) {
|
|
|
|
do {
|
|
|
|
ret = fstatfs(fd, &fs);
|
|
|
|
} while (ret != 0 && errno == EINTR);
|
|
|
|
|
|
|
|
if (ret == 0 && fs.f_type == HUGETLBFS_MAGIC) {
|
|
|
|
return fs.f_bsize;
|
|
|
|
}
|
|
|
|
}
|
2017-12-09 00:57:28 +08:00
|
|
|
#ifdef __sparc__
|
|
|
|
/* SPARC Linux needs greater alignment than the pagesize */
|
|
|
|
return QEMU_VMALLOC_ALIGN;
|
|
|
|
#endif
|
util/mmap-alloc: fix hugetlb support on ppc64
Since commit 8561c9244ddf1122d "exec: allocate PROT_NONE pages on top of
RAM", it is no longer possible to back guest RAM with hugepages on ppc64
hosts:
mmap(NULL, 285212672, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x3fff57000000
mmap(0x3fff57000000, 268435456, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 19, 0) = -1 EBUSY (Device or resource busy)
This is because on ppc64, Linux fixes a page size for a virtual address
at mmap time, so we can't switch a range of memory from anonymous
small pages to hugetlbs with MAP_FIXED.
See commit d0f13e3c20b6fb73ccb467bdca97fa7cf5a574cd
("[POWERPC] Introduce address space "slices"") in Linux
history for the details.
Detect this and create the PROT_NONE mapping using the same fd.
Naturally, this makes the guard page bigger with hugetlbfs.
Based on patch by Greg Kurz.
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2015-12-03 03:14:12 +08:00
|
|
|
#endif
|
|
|
|
|
|
|
|
return getpagesize();
|
|
|
|
}
|
|
|
|
|
2017-03-02 10:36:11 +08:00
|
|
|
size_t qemu_mempath_getpagesize(const char *mem_path)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_LINUX
|
|
|
|
struct statfs fs;
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
do {
|
|
|
|
ret = statfs(mem_path, &fs);
|
|
|
|
} while (ret != 0 && errno == EINTR);
|
|
|
|
|
|
|
|
if (ret != 0) {
|
|
|
|
fprintf(stderr, "Couldn't statfs() memory path: %s\n",
|
|
|
|
strerror(errno));
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (fs.f_type == HUGETLBFS_MAGIC) {
|
|
|
|
/* It's hugepage, return the huge page size */
|
|
|
|
return fs.f_bsize;
|
|
|
|
}
|
2017-12-09 00:57:28 +08:00
|
|
|
#ifdef __sparc__
|
|
|
|
/* SPARC Linux needs greater alignment than the pagesize */
|
|
|
|
return QEMU_VMALLOC_ALIGN;
|
|
|
|
#endif
|
2017-03-02 10:36:11 +08:00
|
|
|
#endif
|
|
|
|
|
|
|
|
return getpagesize();
|
|
|
|
}
|
|
|
|
|
2015-09-24 19:41:17 +08:00
|
|
|
void *qemu_ram_mmap(int fd, size_t size, size_t align, bool shared)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Note: this always allocates at least one extra page of virtual address
|
|
|
|
* space, even if size is already aligned.
|
|
|
|
*/
|
|
|
|
size_t total = size + align;
|
util/mmap-alloc: fix hugetlb support on ppc64
Since commit 8561c9244ddf1122d "exec: allocate PROT_NONE pages on top of
RAM", it is no longer possible to back guest RAM with hugepages on ppc64
hosts:
mmap(NULL, 285212672, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x3fff57000000
mmap(0x3fff57000000, 268435456, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 19, 0) = -1 EBUSY (Device or resource busy)
This is because on ppc64, Linux fixes a page size for a virtual address
at mmap time, so we can't switch a range of memory from anonymous
small pages to hugetlbs with MAP_FIXED.
See commit d0f13e3c20b6fb73ccb467bdca97fa7cf5a574cd
("[POWERPC] Introduce address space "slices"") in Linux
history for the details.
Detect this and create the PROT_NONE mapping using the same fd.
Naturally, this makes the guard page bigger with hugetlbfs.
Based on patch by Greg Kurz.
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2015-12-03 03:14:12 +08:00
|
|
|
#if defined(__powerpc64__) && defined(__linux__)
|
|
|
|
/* On ppc64 mappings in the same segment (aka slice) must share the same
|
|
|
|
* page size. Since we will be re-allocating part of this segment
|
2015-12-03 16:35:31 +08:00
|
|
|
* from the supplied fd, we should make sure to use the same page size, to
|
|
|
|
* this end we mmap the supplied fd. In this case, set MAP_NORESERVE to
|
|
|
|
* avoid allocating backing store memory.
|
|
|
|
* We do this unless we are using the system page size, in which case
|
|
|
|
* anonymous memory is OK.
|
util/mmap-alloc: fix hugetlb support on ppc64
Since commit 8561c9244ddf1122d "exec: allocate PROT_NONE pages on top of
RAM", it is no longer possible to back guest RAM with hugepages on ppc64
hosts:
mmap(NULL, 285212672, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x3fff57000000
mmap(0x3fff57000000, 268435456, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 19, 0) = -1 EBUSY (Device or resource busy)
This is because on ppc64, Linux fixes a page size for a virtual address
at mmap time, so we can't switch a range of memory from anonymous
small pages to hugetlbs with MAP_FIXED.
See commit d0f13e3c20b6fb73ccb467bdca97fa7cf5a574cd
("[POWERPC] Introduce address space "slices"") in Linux
history for the details.
Detect this and create the PROT_NONE mapping using the same fd.
Naturally, this makes the guard page bigger with hugetlbfs.
Based on patch by Greg Kurz.
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2015-12-03 03:14:12 +08:00
|
|
|
*/
|
|
|
|
int anonfd = fd == -1 || qemu_fd_getpagesize(fd) == getpagesize() ? -1 : fd;
|
|
|
|
int flags = anonfd == -1 ? MAP_ANONYMOUS : MAP_NORESERVE;
|
|
|
|
void *ptr = mmap(0, total, PROT_NONE, flags | MAP_PRIVATE, anonfd, 0);
|
|
|
|
#else
|
2015-09-24 19:41:17 +08:00
|
|
|
void *ptr = mmap(0, total, PROT_NONE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
|
util/mmap-alloc: fix hugetlb support on ppc64
Since commit 8561c9244ddf1122d "exec: allocate PROT_NONE pages on top of
RAM", it is no longer possible to back guest RAM with hugepages on ppc64
hosts:
mmap(NULL, 285212672, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x3fff57000000
mmap(0x3fff57000000, 268435456, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, 19, 0) = -1 EBUSY (Device or resource busy)
This is because on ppc64, Linux fixes a page size for a virtual address
at mmap time, so we can't switch a range of memory from anonymous
small pages to hugetlbs with MAP_FIXED.
See commit d0f13e3c20b6fb73ccb467bdca97fa7cf5a574cd
("[POWERPC] Introduce address space "slices"") in Linux
history for the details.
Detect this and create the PROT_NONE mapping using the same fd.
Naturally, this makes the guard page bigger with hugetlbfs.
Based on patch by Greg Kurz.
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Tested-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2015-12-03 03:14:12 +08:00
|
|
|
#endif
|
2016-11-02 21:44:46 +08:00
|
|
|
size_t offset;
|
2015-09-24 19:41:17 +08:00
|
|
|
void *ptr1;
|
|
|
|
|
|
|
|
if (ptr == MAP_FAILED) {
|
2015-10-25 23:07:45 +08:00
|
|
|
return MAP_FAILED;
|
2015-09-24 19:41:17 +08:00
|
|
|
}
|
|
|
|
|
2016-11-02 21:44:46 +08:00
|
|
|
assert(is_power_of_2(align));
|
2015-09-24 19:41:17 +08:00
|
|
|
/* Always align to host page size */
|
|
|
|
assert(align >= getpagesize());
|
|
|
|
|
2016-11-02 21:44:46 +08:00
|
|
|
offset = QEMU_ALIGN_UP((uintptr_t)ptr, align) - (uintptr_t)ptr;
|
2015-09-24 19:41:17 +08:00
|
|
|
ptr1 = mmap(ptr + offset, size, PROT_READ | PROT_WRITE,
|
|
|
|
MAP_FIXED |
|
|
|
|
(fd == -1 ? MAP_ANONYMOUS : 0) |
|
|
|
|
(shared ? MAP_SHARED : MAP_PRIVATE),
|
|
|
|
fd, 0);
|
|
|
|
if (ptr1 == MAP_FAILED) {
|
|
|
|
munmap(ptr, total);
|
2015-10-25 23:07:45 +08:00
|
|
|
return MAP_FAILED;
|
2015-09-24 19:41:17 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
if (offset > 0) {
|
2016-11-02 21:44:47 +08:00
|
|
|
munmap(ptr, offset);
|
2015-09-24 19:41:17 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Leave a single PROT_NONE page allocated after the RAM block, to serve as
|
|
|
|
* a guard page guarding against potential buffer overflows.
|
|
|
|
*/
|
2016-11-02 21:44:47 +08:00
|
|
|
total -= offset;
|
2015-09-24 19:41:17 +08:00
|
|
|
if (total > size + getpagesize()) {
|
2016-11-02 21:44:47 +08:00
|
|
|
munmap(ptr1 + size + getpagesize(), total - size - getpagesize());
|
2015-09-24 19:41:17 +08:00
|
|
|
}
|
|
|
|
|
2016-11-02 21:44:47 +08:00
|
|
|
return ptr1;
|
2015-09-24 19:41:17 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void qemu_ram_munmap(void *ptr, size_t size)
|
|
|
|
{
|
|
|
|
if (ptr) {
|
|
|
|
/* Unmap both the RAM block and the guard page */
|
|
|
|
munmap(ptr, size + getpagesize());
|
|
|
|
}
|
|
|
|
}
|