x86/vdso: Split virtual clock pages into dedicated mapping

The generic vdso data storage cannot handle the special pvclock and
hvclock pages. Split them into their own mapping, so the other vdso
storage can be migrated to the generic code.

Signed-off-by: Thomas Weißschuh <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]

Author: t-8ch

arch/x86/entry/vdso/vdso-layout.lds.S

@@ -17,14 +17,16 @@ SECTIONS
 	 * segment.
 	 */
 
-	vvar_start = . - 4 * PAGE_SIZE;
+	vvar_start = . - __VVAR_PAGES * PAGE_SIZE;
 	vvar_page  = vvar_start;
 
 	vdso_rng_data = vvar_page + __VDSO_RND_DATA_OFFSET;
 
-	pvclock_page = vvar_start + PAGE_SIZE;
-	hvclock_page = vvar_start + 2 * PAGE_SIZE;
-	timens_page  = vvar_start + 3 * PAGE_SIZE;
+	timens_page  = vvar_start + PAGE_SIZE;
+
+	vclock_pages = vvar_start + VDSO_NR_VCLOCK_PAGES * PAGE_SIZE;
+	pvclock_page = vclock_pages + VDSO_PAGE_PVCLOCK_OFFSET * PAGE_SIZE;
+	hvclock_page = vclock_pages + VDSO_PAGE_HVCLOCK_OFFSET * PAGE_SIZE;
 
 	. = SIZEOF_HEADERS;
 

arch/x86/entry/vdso/vma.c

@@ -24,6 +24,7 @@
 #include <asm/page.h>
 #include <asm/desc.h>
 #include <asm/cpufeature.h>
+#include <asm/vdso/vsyscall.h>
 #include <clocksource/hyperv_timer.h>
 
 struct vdso_data *arch_get_vdso_data(void *vvar_page)
@@ -175,19 +176,7 @@ static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
 		}
 
 		return vmf_insert_pfn(vma, vmf->address, pfn);
-	} else if (sym_offset == image->sym_pvclock_page) {
-		struct pvclock_vsyscall_time_info *pvti =
-			pvclock_get_pvti_cpu0_va();
-		if (pvti && vclock_was_used(VDSO_CLOCKMODE_PVCLOCK)) {
-			return vmf_insert_pfn_prot(vma, vmf->address,
-					__pa(pvti) >> PAGE_SHIFT,
-					pgprot_decrypted(vma->vm_page_prot));
-		}
-	} else if (sym_offset == image->sym_hvclock_page) {
-		pfn = hv_get_tsc_pfn();
 
-		if (pfn && vclock_was_used(VDSO_CLOCKMODE_HVCLOCK))
-			return vmf_insert_pfn(vma, vmf->address, pfn);
 	} else if (sym_offset == image->sym_timens_page) {
 		struct page *timens_page = find_timens_vvar_page(vma);
 
@@ -201,6 +190,33 @@ static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
 	return VM_FAULT_SIGBUS;
 }
 
+static vm_fault_t vvar_vclock_fault(const struct vm_special_mapping *sm,
+				    struct vm_area_struct *vma, struct vm_fault *vmf)
+{
+	switch (vmf->pgoff) {
+#ifdef CONFIG_PARAVIRT_CLOCK
+	case VDSO_PAGE_PVCLOCK_OFFSET:
+		struct pvclock_vsyscall_time_info *pvti =
+			pvclock_get_pvti_cpu0_va();
+		if (pvti && vclock_was_used(VDSO_CLOCKMODE_PVCLOCK))
+			return vmf_insert_pfn_prot(vma, vmf->address,
+					__pa(pvti) >> PAGE_SHIFT,
+					pgprot_decrypted(vma->vm_page_prot));
+		break;
+#endif /* CONFIG_PARAVIRT_CLOCK */
+#ifdef CONFIG_HYPERV_TIMER
+	case VDSO_PAGE_HVCLOCK_OFFSET:
+		unsigned long pfn = hv_get_tsc_pfn();
+
+		if (pfn && vclock_was_used(VDSO_CLOCKMODE_HVCLOCK))
+			return vmf_insert_pfn(vma, vmf->address, pfn);
+		break;
+#endif /* CONFIG_HYPERV_TIMER */
+	}
+
+	return VM_FAULT_SIGBUS;
+}
+
 static const struct vm_special_mapping vdso_mapping = {
 	.name = "[vdso]",
 	.fault = vdso_fault,
@@ -210,6 +226,10 @@ static const struct vm_special_mapping vvar_mapping = {
 	.name = "[vvar]",
 	.fault = vvar_fault,
 };
+static const struct vm_special_mapping vvar_vclock_mapping = {
+	.name = "[vvar_vclock]",
+	.fault = vvar_vclock_fault,
+};
 
 /*
  * Add vdso and vvar mappings to current process.
@@ -252,19 +272,34 @@ static int map_vdso(const struct vdso_image *image, unsigned long addr)
 
 	vma = _install_special_mapping(mm,
 				       addr,
-				       -image->sym_vvar_start,
+				       (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE,
 				       VM_READ|VM_MAYREAD|VM_IO|VM_DONTDUMP|
 				       VM_PFNMAP,
 				       &vvar_mapping);
 
 	if (IS_ERR(vma)) {
 		ret = PTR_ERR(vma);
 		do_munmap(mm, text_start, image->size, NULL);
-	} else {
-		current->mm->context.vdso = (void __user *)text_start;
-		current->mm->context.vdso_image = image;
+		goto up_fail;
 	}
 
+	vma = _install_special_mapping(mm,
+				       addr + (__VVAR_PAGES - VDSO_NR_VCLOCK_PAGES) * PAGE_SIZE,
+				       VDSO_NR_VCLOCK_PAGES * PAGE_SIZE,
+				       VM_READ|VM_MAYREAD|VM_IO|VM_DONTDUMP|
+				       VM_PFNMAP,
+				       &vvar_vclock_mapping);
+
+	if (IS_ERR(vma)) {
+		ret = PTR_ERR(vma);
+		do_munmap(mm, text_start, image->size, NULL);
+		do_munmap(mm, addr, image->size, NULL);
+		goto up_fail;
+	}
+
+	current->mm->context.vdso = (void __user *)text_start;
+	current->mm->context.vdso_image = image;
+
 up_fail:
 	mmap_write_unlock(mm);
 	return ret;
@@ -286,7 +321,8 @@ int map_vdso_once(const struct vdso_image *image, unsigned long addr)
 	 */
 	for_each_vma(vmi, vma) {
 		if (vma_is_special_mapping(vma, &vdso_mapping) ||
-				vma_is_special_mapping(vma, &vvar_mapping)) {
+				vma_is_special_mapping(vma, &vvar_mapping) ||
+				vma_is_special_mapping(vma, &vvar_vclock_mapping)) {
 			mmap_write_unlock(mm);
 			return -EEXIST;
 		}

arch/x86/include/asm/vdso/vsyscall.h

@@ -3,6 +3,11 @@
 #define __ASM_VDSO_VSYSCALL_H
 
 #define __VDSO_RND_DATA_OFFSET  640
+#define __VVAR_PAGES	4
+
+#define VDSO_NR_VCLOCK_PAGES	2
+#define VDSO_PAGE_PVCLOCK_OFFSET	0
+#define VDSO_PAGE_HVCLOCK_OFFSET	1
 
 #ifndef __ASSEMBLY__