WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

[Xen-devel] Re: [PATCH 6/7 v2] xen-gntdev: Support mapping in HVM domain

To: jeremy@xxxxxxxx
Subject: [Xen-devel] Re: [PATCH 6/7 v2] xen-gntdev: Support mapping in HVM domains
From: Daniel De Graaf <dgdegra@xxxxxxxxxxxxx>
Date: Fri, 17 Dec 2010 10:22:55 -0500
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx, Ian.Campbell@xxxxxxxxxx
Delivery-date: Fri, 17 Dec 2010 07:25:33 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <1292545063-32107-7-git-send-email-dgdegra@xxxxxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Organization: National Security Agency
References: <1292545063-32107-1-git-send-email-dgdegra@xxxxxxxxxxxxx> <1292545063-32107-7-git-send-email-dgdegra@xxxxxxxxxxxxx>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.13) Gecko/20101209 Fedora/3.1.7-0.35.b3pre.fc13 Thunderbird/3.1.7
HVM does not allow direct PTE modification, so instead we request
that Xen change its internal p2m mappings on the allocated pages and
map the memory into userspace normally.
    
Signed-off-by: Daniel De Graaf <dgdegra@xxxxxxxxxxxxx>
---
 drivers/xen/gntdev.c      |  117 +++++++++++++++++++++++++++++++-------------
 drivers/xen/grant-table.c |    7 +++
 2 files changed, 89 insertions(+), 35 deletions(-)

diff --git a/drivers/xen/gntdev.c b/drivers/xen/gntdev.c
index 0c8e8a2..0985577 100644
--- a/drivers/xen/gntdev.c
+++ b/drivers/xen/gntdev.c
@@ -30,6 +30,7 @@
 #include <linux/sched.h>
 #include <linux/spinlock.h>
 #include <linux/slab.h>
+#include <linux/highmem.h>
 
 #include <xen/xen.h>
 #include <xen/grant_table.h>
@@ -49,6 +50,8 @@ static int limit = 1024*1024;
 module_param(limit, int, 0644);
 static atomic_t pages_mapped = ATOMIC_INIT(0);
 
+static int use_ptemod = 0;
+
 struct gntdev_priv {
        struct list_head maps;
        spinlock_t lock;
@@ -179,6 +182,9 @@ static void gntdev_put_map(struct grant_map *map)
 
        atomic_sub(map->count, &pages_mapped);
 
+       if (!use_ptemod)
+               unmap_grant_pages(map, 0, map->count);
+
        for (i = 0; i < map->count; i++) {
                if (map->pages[i])
                        __free_page(map->pages[i]);
@@ -206,9 +212,12 @@ static int find_grant_ptes(pte_t *pte, pgtable_t token, 
unsigned long addr, void
 static int map_grant_pages(struct grant_map *map)
 {
        int i, flags, err = 0;
+       phys_addr_t addr;
        struct gnttab_map_grant_ref* map_ops = NULL;
 
-       flags = GNTMAP_host_map | GNTMAP_application_map | GNTMAP_contains_pte;
+       flags = GNTMAP_host_map;
+       if (use_ptemod)
+               flags |= GNTMAP_application_map | GNTMAP_contains_pte;
        if (map->is_ro)
                flags |= GNTMAP_readonly;
 
@@ -218,7 +227,11 @@ static int map_grant_pages(struct grant_map *map)
                goto out;
 
        for(i=0; i < map->count; i++) {
-               gnttab_set_map_op(&map_ops[i], map->pginfo[i].pte_maddr, flags,
+               if (use_ptemod)
+                       addr = map->pginfo[i].pte_maddr;
+               else
+                       addr = 
(phys_addr_t)pfn_to_kaddr(page_to_pfn(map->pages[i]));
+               gnttab_set_map_op(&map_ops[i], addr, flags,
                                  map->pginfo[i].target.ref,
                                  map->pginfo[i].target.domid);
        }
@@ -250,6 +263,7 @@ static void unmap_grant_pages(struct grant_map *map, int 
offset, int pages)
        int i, flags, err = 0;
        struct gnttab_unmap_grant_ref *unmap_ops;
        struct gnttab_unmap_grant_ref unmap_single;
+       phys_addr_t addr;
 
        if (pages > 1) {
                unmap_ops = kzalloc(sizeof(unmap_ops[0]) * pages,
@@ -263,14 +277,23 @@ static void unmap_grant_pages(struct grant_map *map, int 
offset, int pages)
                unmap_ops = &unmap_single;
        }
 
-       flags = GNTMAP_host_map | GNTMAP_application_map | GNTMAP_contains_pte;
+       flags = GNTMAP_host_map;
+       if (use_ptemod)
+               flags |= GNTMAP_application_map | GNTMAP_contains_pte;
        if (map->is_ro)
                flags |= GNTMAP_readonly;
 
-       for(i=0; i < pages; i++)
-               gnttab_set_unmap_op(&unmap_ops[i],
-                                   map->pginfo[offset+i].pte_maddr, flags,
+       for(i=0; i < pages; i++) {
+               if (WARN_ON(!map->pages[i]))
+                       continue;
+               if (use_ptemod)
+                       addr = map->pginfo[i].pte_maddr;
+               else
+                       addr = 
(phys_addr_t)pfn_to_kaddr(page_to_pfn(map->pages[i]));
+               gnttab_set_unmap_op(&unmap_ops[i], addr, flags,
                                    map->pginfo[offset+i].handle);
+       }
+
        if (debug)
                printk("%s: map %d+%d [%d+%d]\n", __FUNCTION__,
                       map->index, map->count, offset, pages);
@@ -281,8 +304,25 @@ static void unmap_grant_pages(struct grant_map *map, int 
offset, int pages)
                goto out;
 
        for (i = 0; i < pages; i++) {
+               uint32_t check, *tmp;
                WARN_ON(unmap_ops[i].status);
-               __free_page(map->pages[offset+i]);
+               if (!map->pages[i])
+                       continue;
+               /* XXX When unmapping, Xen will sometimes end up mapping the GFN
+                * to an invalid MFN. In this case, writes will be discarded and
+                * reads will return all 0xFF bytes. Leak these unusable GFNs
+                * until a way to restore them is found.
+                */
+               tmp = kmap(map->pages[i]);
+               tmp[0] = 0xdeaddead;
+               mb();
+               check = tmp[0];
+               kunmap(map->pages[i]);
+               if (check == 0xdeaddead)
+                       __free_page(map->pages[i]);
+               else if (debug)
+                       printk("%s: Discard page %d=%ld\n", __func__,
+                               i, page_to_pfn(map->pages[i]));
                map->pages[offset+i] = NULL;
                map->pginfo[offset+i].handle = 0;
        }
@@ -305,18 +345,8 @@ static void gntdev_vma_close(struct vm_area_struct *vma)
        gntdev_put_map(map);
 }
 
-static int gntdev_vma_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
-{
-       if (debug)
-               printk("%s: vaddr %p, pgoff %ld (shouldn't happen)\n",
-                      __FUNCTION__, vmf->virtual_address, vmf->pgoff);
-       vmf->flags = VM_FAULT_ERROR;
-       return 0;
-}
-
 static struct vm_operations_struct gntdev_vmops = {
        .close = gntdev_vma_close,
-       .fault = gntdev_vma_fault,
 };
 
 /* ------------------------------------------------------------------ */
@@ -398,14 +428,16 @@ static int gntdev_open(struct inode *inode, struct file 
*flip)
        INIT_LIST_HEAD(&priv->maps);
        spin_lock_init(&priv->lock);
 
-       priv->mm = get_task_mm(current);
-       if (!priv->mm) {
-               kfree(priv);
-               return -ENOMEM;
+       if (use_ptemod) {
+               priv->mm = get_task_mm(current);
+               if (!priv->mm) {
+                       kfree(priv);
+                       return -ENOMEM;
+               }
+               priv->mn.ops = &gntdev_mmu_ops;
+               mmu_notifier_register(&priv->mn, priv->mm);
+               mmput(priv->mm);
        }
-       priv->mn.ops = &gntdev_mmu_ops;
-       mmu_notifier_register(&priv->mn, priv->mm);
-       mmput(priv->mm);
 
        flip->private_data = priv;
        if (debug)
@@ -430,7 +462,8 @@ static int gntdev_release(struct inode *inode, struct file 
*flip)
        }
        spin_unlock(&priv->lock);
 
-       mmu_notifier_unregister(&priv->mn, priv->mm);
+       if (use_ptemod)
+               mmu_notifier_unregister(&priv->mn, priv->mm);
        kfree(priv);
        return 0;
 }
@@ -574,7 +607,7 @@ static int gntdev_mmap(struct file *flip, struct 
vm_area_struct *vma)
        int index = vma->vm_pgoff;
        int count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
        struct grant_map *map;
-       int err = -EINVAL;
+       int i, err = -EINVAL;
 
        if ((vma->vm_flags & VM_WRITE) && !(vma->vm_flags & VM_SHARED))
                return -EINVAL;
@@ -587,9 +620,9 @@ static int gntdev_mmap(struct file *flip, struct 
vm_area_struct *vma)
        map = gntdev_find_map_index(priv, index, count);
        if (!map)
                goto unlock_out;
-       if (map->vma)
+       if (use_ptemod && map->vma)
                goto unlock_out;
-       if (priv->mm != vma->vm_mm) {
+       if (use_ptemod && priv->mm != vma->vm_mm) {
                printk("%s: Huh? Other mm?\n", __FUNCTION__);
                goto unlock_out;
        }
@@ -605,24 +638,36 @@ static int gntdev_mmap(struct file *flip, struct 
vm_area_struct *vma)
 
        vma->vm_private_data = map;
 
-       map->vma = vma;
+       if (use_ptemod)
+               map->vma = vma;
 
        map->is_ro = !(vma->vm_flags & VM_WRITE);
 
        spin_unlock(&priv->lock);
 
-       err = apply_to_page_range(vma->vm_mm, vma->vm_start,
-                                 vma->vm_end - vma->vm_start,
-                                 find_grant_ptes, map);
-       if (err)
-               return err;
+       if (use_ptemod) {
+               err = apply_to_page_range(vma->vm_mm, vma->vm_start,
+                                         vma->vm_end - vma->vm_start,
+                                         find_grant_ptes, map);
+               if (err)
+                       return err;
+       }
 
        err = map_grant_pages(map);
        if (err)
                return err;
-
+       
        map->is_mapped = 1;
 
+       if (!use_ptemod) {
+               for(i = 0; i < count; i++) {
+                       err = vm_insert_page(vma, vma->vm_start + i*PAGE_SIZE,
+                                            map->pages[i]);
+                       if (err)
+                               return err;
+               }
+       }
+
        return 0;
 
 unlock_out:
@@ -653,6 +698,8 @@ static int __init gntdev_init(void)
        if (!xen_domain())
                return -ENODEV;
 
+       use_ptemod = xen_pv_domain();
+
        err = misc_register(&gntdev_miscdev);
        if (err != 0) {
                printk(KERN_ERR "Could not register gntdev device\n");
diff --git a/drivers/xen/grant-table.c b/drivers/xen/grant-table.c
index a5cf820..c8ab76e 100644
--- a/drivers/xen/grant-table.c
+++ b/drivers/xen/grant-table.c
@@ -457,6 +457,9 @@ int gnttab_map_refs(struct gnttab_map_grant_ref *map_ops,
 
        ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map_ops, count);
 
+       if (xen_feature(XENFEAT_auto_translated_physmap))
+               return ret;
+
        for (i = 0; i < count; i++) {
                pfn = mfn_to_pfn(map_ops[i].host_addr >> PAGE_SHIFT);
                pte = (pte_t *) __va((pfn << PAGE_SHIFT) +
@@ -476,6 +479,10 @@ int gnttab_unmap_refs(struct gnttab_unmap_grant_ref 
*unmap_ops,
        int i, ret;
 
        ret = HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, unmap_ops, 
count);
+
+       if (xen_feature(XENFEAT_auto_translated_physmap))
+               return ret;
+
        for (i = 0; i < count; i++)
                m2p_remove_override(pages[i]);
 


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel