[PATCH v6 22/27] vfio: selftests: Add helpers for Live Update DMA memory and sysfs state

[Vipin Sharma]

Add shared libvfio helpers used by the Live Update kexec selftest:

 - vfio_thp_memfd_alloc() and vfio_memfd_mmap_aligned() to allocate a
   shmem memfd, map it at a 2MB-aligned virtual address, and back it
   with 2MB THP folios via MADV_HUGEPAGE, write-fault population, and
   MADV_COLLAPSE.
 - __sysfs_iommu_group_get(), a variant of sysfs_iommu_group_get() that
   returns an error instead of asserting when the device has no IOMMU
   group.

Signed-off-by: Vipin Sharma <vipinsh@xxxxxxxxxx>
---
 .../selftests/vfio/lib/include/libvfio.h      |  3 ++
 .../vfio/lib/include/libvfio/sysfs.h          |  1 +
 tools/testing/selftests/vfio/lib/libvfio.c    | 54 +++++++++++++++++++
 tools/testing/selftests/vfio/lib/sysfs.c      | 22 ++++++--
 4 files changed, 77 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/vfio/lib/include/libvfio.h b/tools/testing/selftests/vfio/lib/include/libvfio.h
index 07862b470777..1c245962525b 100644
--- a/tools/testing/selftests/vfio/lib/include/libvfio.h
+++ b/tools/testing/selftests/vfio/lib/include/libvfio.h
@@ -33,4 +33,7 @@ char **vfio_selftests_get_bdfs(int *argc, char *argv[], int *nr_bdfs);
  */
 void *mmap_reserve(size_t size, size_t align, size_t offset);
 
+int vfio_thp_memfd_alloc(size_t size, void **vaddr_out);
+void *vfio_memfd_mmap_aligned(int fd, size_t size, size_t align);
+
 #endif /* SELFTESTS_VFIO_LIB_INCLUDE_LIBVFIO_H */
diff --git a/tools/testing/selftests/vfio/lib/include/libvfio/sysfs.h b/tools/testing/selftests/vfio/lib/include/libvfio/sysfs.h
index c9ab1ea8f5a9..b9d6f6f69f88 100644
--- a/tools/testing/selftests/vfio/lib/include/libvfio/sysfs.h
+++ b/tools/testing/selftests/vfio/lib/include/libvfio/sysfs.h
@@ -6,6 +6,7 @@ int sysfs_sriov_totalvfs_get(const char *bdf);
 int sysfs_sriov_numvfs_get(const char *bdf);
 void sysfs_sriov_numvfs_set(const char *bdf, int numvfs);
 char *sysfs_sriov_vf_bdf_get(const char *pf_bdf, int i);
+int __sysfs_iommu_group_get(const char *bdf, int *group);
 int sysfs_iommu_group_get(const char *bdf);
 char *sysfs_driver_get(const char *bdf);
 
diff --git a/tools/testing/selftests/vfio/lib/libvfio.c b/tools/testing/selftests/vfio/lib/libvfio.c
index 3a3d1ed635c1..00369fe7d074 100644
--- a/tools/testing/selftests/vfio/lib/libvfio.c
+++ b/tools/testing/selftests/vfio/lib/libvfio.c
@@ -1,14 +1,23 @@
 // SPDX-License-Identifier: GPL-2.0-only
 
+#include <errno.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
 #include <sys/mman.h>
 
 #include <linux/align.h>
+#include <linux/mman.h>
+#include <linux/sizes.h>
 
 #include "../../../kselftest.h"
 #include <libvfio.h>
 
+#ifndef MADV_COLLAPSE
+#define MADV_COLLAPSE 25
+#endif
+
 static bool is_bdf(const char *str)
 {
 	unsigned int s, b, d, f;
@@ -101,3 +110,48 @@ void *mmap_reserve(size_t size, size_t align, size_t offset)
 
 	return map_align;
 }
+
+void *vfio_memfd_mmap_aligned(int fd, size_t size, size_t align)
+{
+	void *vaddr;
+
+	vaddr = mmap_reserve(size, align, 0);
+	vaddr = mmap(vaddr, size, PROT_READ | PROT_WRITE,
+		     MAP_SHARED | MAP_FIXED, fd, 0);
+	VFIO_ASSERT_NE(vaddr, MAP_FAILED);
+
+	return vaddr;
+}
+
+int vfio_thp_memfd_alloc(size_t size, void **vaddr_out)
+{
+	void *vaddr;
+	int fd;
+
+	VFIO_ASSERT_TRUE(IS_ALIGNED(size, SZ_2M));
+
+	fd = memfd_create("vfio-lu-dma", 0);
+	VFIO_ASSERT_GE(fd, 0);
+
+	VFIO_ASSERT_EQ(ftruncate(fd, size), 0);
+
+	vaddr = vfio_memfd_mmap_aligned(fd, size, SZ_2M);
+
+	if (madvise(vaddr, size, MADV_HUGEPAGE))
+		ksft_exit_skip("madvise(MADV_HUGEPAGE) failed: %s\n",
+			       strerror(errno));
+
+	/*
+	 * Write-fault the mapping rather than using fallocate() so that shmem
+	 * allocates PMD-sized THP folios when shmem_enabled is always,
+	 * within_size, or advise, then collapse in case shmem_enabled is never.
+	 */
+	memset(vaddr, 0, size);
+
+	if (madvise(vaddr, size, MADV_COLLAPSE))
+		ksft_exit_skip("madvise(MADV_COLLAPSE) failed: %s\n",
+			       strerror(errno));
+
+	*vaddr_out = vaddr;
+	return fd;
+}
diff --git a/tools/testing/selftests/vfio/lib/sysfs.c b/tools/testing/selftests/vfio/lib/sysfs.c
index 98a46a2543cd..219db6aaa101 100644
--- a/tools/testing/selftests/vfio/lib/sysfs.c
+++ b/tools/testing/selftests/vfio/lib/sysfs.c
@@ -115,14 +115,30 @@ char *sysfs_sriov_vf_bdf_get(const char *pf_bdf, int i)
 	return out_vf_bdf;
 }
 
-int sysfs_iommu_group_get(const char *bdf)
+int __sysfs_iommu_group_get(const char *bdf, int *group)
 {
+	char rl_path[PATH_MAX];
 	char path[PATH_MAX];
-	int group;
+	char *end;
+	int ret;
 
 	snprintf_assert(path, PATH_MAX, "/sys/bus/pci/devices/%s/iommu_group", bdf);
-	readlink_base(path, "%d", &group);
+	ret = readlink_safe(path, rl_path);
+	if (ret == -1)
+		return -errno;
+
+	errno = 0;
+	*group = strtol(basename(rl_path), &end, 10);
+	VFIO_ASSERT_EQ(errno, 0);
+	VFIO_ASSERT_EQ(*end, '\0');
+	return 0;
+}
+
+int sysfs_iommu_group_get(const char *bdf)
+{
+	int group;
 
+	VFIO_ASSERT_EQ(__sysfs_iommu_group_get(bdf, &group), 0);
 	return group;
 }
 
-- 
2.56.0.317.g454f8302e1-goog