[RFC XEN PATCH v2 13/15] tools/libxl: add support to map host pmem device to guests

From: Haozhong Zhang <haozhong.zhang@intel.com>
To: xen-devel@lists.xen.org
Cc: Konrad Rzeszutek Wilk <konrad@darnok.org>,
	Dan Williams <dan.j.williams@intel.com>,
	Ian Jackson <ian.jackson@eu.citrix.com>,
	Wei Liu <wei.liu2@citrix.com>,
	Haozhong Zhang <haozhong.zhang@intel.com>
Subject: [RFC XEN PATCH v2 13/15] tools/libxl: add support to map host pmem device to guests
Date: Mon, 20 Mar 2017 08:09:47 +0800	[thread overview]
Message-ID: <20170320000949.24675-14-haozhong.zhang@intel.com> (raw)
In-Reply-To: <20170320000949.24675-1-haozhong.zhang@intel.com>

We can map host pmem devices or files on pmem devices to guests. This
patch adds support to map pmem devices. The implementation relies on
the Linux pmem driver (CONFIG_ACPI_NFIT, CONFIG_LIBNVDIMM, CONFIG_BLK_DEV_PMEM),
so it functions only when libxl is compiled for Linux right now.

Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
---
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Wei Liu <wei.liu2@citrix.com>
---
 tools/libxl/Makefile       |   4 +
 tools/libxl/libxl_nvdimm.c | 182 +++++++++++++++++++++++++++++++++++++++++++++
 tools/libxl/libxl_nvdimm.h |  42 +++++++++++
 3 files changed, 228 insertions(+)
 create mode 100644 tools/libxl/libxl_nvdimm.c
 create mode 100644 tools/libxl/libxl_nvdimm.h

diff --git a/tools/libxl/Makefile b/tools/libxl/Makefile
index a3e5e9909f..6bfc78972f 100644
--- a/tools/libxl/Makefile
+++ b/tools/libxl/Makefile
@@ -115,6 +115,10 @@ endif
 endif
 endif
 
+ifeq ($(CONFIG_Linux), y)
+LIBXL_OBJS-y += libxl_nvdimm.o
+endif
+
 ifeq ($(FLEX),)
 %.c %.h:: %.l
 	$(warning Flex is needed to rebuild some libxl parsers and \
diff --git a/tools/libxl/libxl_nvdimm.c b/tools/libxl/libxl_nvdimm.c
new file mode 100644
index 0000000000..1b3c83f2ca
--- /dev/null
+++ b/tools/libxl/libxl_nvdimm.c
@@ -0,0 +1,182 @@
+/*
+ * tools/libxl/libxl_nvdimm.c
+ *
+ * Copyright (C) 2017,  Intel Corporation
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License, version 2.1, as published by the Free Software Foundation.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <errno.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+
+#include "libxl_internal.h"
+#include "libxl_arch.h"
+#include "libxl_nvdimm.h"
+
+#include <xenctrl.h>
+
+#define BLK_DEVICE_ROOT "/sys/dev/block"
+
+static int nvdimm_sysfs_read(libxl__gc *gc,
+                             unsigned int major, unsigned int minor,
+                             const char *name, void **data_r)
+{
+    char *path = libxl__sprintf(gc, BLK_DEVICE_ROOT"/%u:%u/device/%s",
+                                major, minor, name);
+    return libxl__read_sysfs_file_contents(gc, path, data_r, NULL);
+}
+
+static int nvdimm_get_spa(libxl__gc *gc, unsigned int major, unsigned int minor,
+                          uint64_t *spa_r)
+{
+    void *data;
+    int ret = nvdimm_sysfs_read(gc, major, minor, "resource", &data);
+
+    if ( ret )
+        return ret;
+
+    *spa_r = strtoll(data, NULL, 0);
+    return 0;
+}
+
+static int nvdimm_get_size(libxl__gc *gc, unsigned int major, unsigned int minor,
+                           uint64_t *size_r)
+{
+    void *data;
+    int ret = nvdimm_sysfs_read(gc, major, minor, "size", &data);
+
+    if ( ret )
+        return ret;
+
+    *size_r = strtoll(data, NULL, 0);
+
+    return 0;
+}
+
+static int add_pages(libxl__gc *gc, uint32_t domid,
+                     xen_pfn_t mfn, xen_pfn_t gpfn, unsigned long nr_mfns)
+{
+    unsigned int nr;
+    int ret = 0;
+
+    while ( nr_mfns )
+    {
+        nr = min(nr_mfns, (unsigned long) UINT_MAX);
+
+        ret = xc_domain_populate_pmemmap(CTX->xch, domid, mfn, gpfn, nr);
+        if ( ret )
+        {
+            LOG(ERROR, "failed to map pmem pages, "
+                "mfn 0x%" PRIx64", gpfn 0x%" PRIx64 ", nr_mfns %u, err %d",
+                mfn, gpfn, nr, ret);
+            break;
+        }
+
+        nr_mfns -= nr;
+        mfn += nr;
+        gpfn += nr;
+    }
+
+    return ret;
+}
+
+int libxl_nvdimm_add_device(libxl__gc *gc,
+                            uint32_t domid, const char *path,
+                            uint64_t guest_spa, uint64_t guest_size)
+{
+    int fd;
+    struct stat st;
+    unsigned int major, minor;
+    uint64_t host_spa, host_size;
+    xen_pfn_t mfn, gpfn;
+    unsigned long nr_gpfns;
+    int ret;
+
+    if ( (guest_spa & ~XC_PAGE_MASK) || (guest_size & ~XC_PAGE_MASK) )
+        return -EINVAL;
+
+    fd = open(path, O_RDONLY);
+    if ( fd < 0 )
+    {
+        LOG(ERROR, "failed to open file %s (err: %d)", path, errno);
+        return -EIO;
+    }
+
+    ret = fstat(fd, &st);
+    if ( ret )
+    {
+        LOG(ERROR, "failed to get status of file %s (err: %d)",
+            path, errno);
+        goto out;
+    }
+
+    switch ( st.st_mode & S_IFMT )
+    {
+    case S_IFBLK:
+        major = major(st.st_rdev);
+        minor = minor(st.st_rdev);
+        break;
+
+    default:
+        LOG(ERROR, "only support block device now");
+        ret = -EINVAL;
+        goto out;
+    }
+
+    ret = nvdimm_get_spa(gc, major, minor, &host_spa);
+    if ( ret )
+    {
+        LOG(ERROR, "failed to get SPA of device %u:%u", major, minor);
+        goto out;
+    }
+    else if ( host_spa & ~XC_PAGE_MASK )
+    {
+        ret = -EINVAL;
+        goto out;
+    }
+
+    ret = nvdimm_get_size(gc, major, minor, &host_size);
+    if ( ret )
+    {
+        LOG(ERROR, "failed to get size of device %u:%u", major, minor);
+        goto out;
+    }
+    else if ( guest_size > host_size )
+    {
+        LOG(ERROR, "vNVDIMM size %" PRIu64 " expires NVDIMM size %" PRIu64,
+            guest_size, host_size);
+        ret = -EINVAL;
+        goto out;
+    }
+
+    mfn = host_spa >> XC_PAGE_SHIFT;
+    gpfn = guest_spa >> XC_PAGE_SHIFT;
+    nr_gpfns = guest_size >> XC_PAGE_SHIFT;
+    ret = add_pages(gc, domid, mfn, gpfn, nr_gpfns);
+
+ out:
+    close(fd);
+    return ret;
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/tools/libxl/libxl_nvdimm.h b/tools/libxl/libxl_nvdimm.h
new file mode 100644
index 0000000000..e95be99c67
--- /dev/null
+++ b/tools/libxl/libxl_nvdimm.h
@@ -0,0 +1,42 @@
+/*
+ * tools/libxl/libxl_nvdimm.h
+ *
+ * Copyright (C) 2017,  Intel Corporation
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License, version 2.1, as published by the Free Software Foundation.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef LIBXL_NVDIMM_H
+#define LIBXL_NVDIMM_H
+
+#include <stdint.h>
+#include "libxl_internal.h"
+
+#if defined(__linux__)
+
+int libxl_nvdimm_add_device(libxl__gc *gc,
+                            uint32_t domid, const char *path,
+                            uint64_t spa, uint64_t length);
+
+#else
+
+static inline int libxl_nvdimm_add_device(libxl__gc *gc,
+                                          uint32_t domid, const char *path,
+                                          uint64_t spa, uint64_t length)
+{
+    return -ENOSYS;
+}
+
+#endif /* __linux__ */
+
+#endif /* !LIBXL_NVDIMM_H */
-- 
2.12.0


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
