[PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain

Oleksandr Tyshchenko posted 24 patches 1 month, 2 weeks ago

[PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain

Posted by Oleksandr Tyshchenko 1 month, 2 weeks ago
From: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>

The IOREQ is a common feature now and this struct will be used
on Arm as is. Move it to common struct domain. This also
significantly reduces the layering violation in the common code
(*arch.hvm* usage).

We don't move ioreq_gfn since it is not used in the common code
(the "legacy" mechanism is x86 specific).

Signed-off-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
CC: Julien Grall <julien.grall@arm.com>
[On Arm only]
Tested-by: Wei Chen <Wei.Chen@arm.com>

---
Please note, this is a split/cleanup/hardening of Julien's PoC:
"Add support for Guest IO forwarding to a device emulator"

Changes V1 -> V2:
   - new patch

Changes V2 -> V3:
   - remove the mention of "ioreq_gfn" from patch subject/description
   - update patch according the "legacy interface" is x86 specific
   - drop hvm_params related changes in arch/x86/hvm/hvm.c
   - leave ioreq_gfn in hvm_domain

Changes V3 -> V4:
   - rebase
   - drop the stale part of the comment above struct ioreq_server
   - add Jan's A-b
---
 xen/common/ioreq.c               | 60 ++++++++++++++++++++--------------------
 xen/include/asm-x86/hvm/domain.h |  8 ------
 xen/include/xen/sched.h          | 10 +++++++
 3 files changed, 40 insertions(+), 38 deletions(-)

diff --git a/xen/common/ioreq.c b/xen/common/ioreq.c
index 3f631ec..a319c88 100644
--- a/xen/common/ioreq.c
+++ b/xen/common/ioreq.c
@@ -38,13 +38,13 @@ static void set_ioreq_server(struct domain *d, unsigned int id,
                              struct ioreq_server *s)
 {
     ASSERT(id < MAX_NR_IOREQ_SERVERS);
-    ASSERT(!s || !d->arch.hvm.ioreq_server.server[id]);
+    ASSERT(!s || !d->ioreq_server.server[id]);
 
-    d->arch.hvm.ioreq_server.server[id] = s;
+    d->ioreq_server.server[id] = s;
 }
 
 #define GET_IOREQ_SERVER(d, id) \
-    (d)->arch.hvm.ioreq_server.server[id]
+    (d)->ioreq_server.server[id]
 
 static struct ioreq_server *get_ioreq_server(const struct domain *d,
                                              unsigned int id)
@@ -285,7 +285,7 @@ bool is_ioreq_server_page(struct domain *d, const struct page_info *page)
     unsigned int id;
     bool found = false;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     FOR_EACH_IOREQ_SERVER(d, id, s)
     {
@@ -296,7 +296,7 @@ bool is_ioreq_server_page(struct domain *d, const struct page_info *page)
         }
     }
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return found;
 }
@@ -606,7 +606,7 @@ int hvm_create_ioreq_server(struct domain *d, int bufioreq_handling,
         return -ENOMEM;
 
     domain_pause(d);
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     for ( i = 0; i < MAX_NR_IOREQ_SERVERS; i++ )
     {
@@ -634,13 +634,13 @@ int hvm_create_ioreq_server(struct domain *d, int bufioreq_handling,
     if ( id )
         *id = i;
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
     domain_unpause(d);
 
     return 0;
 
  fail:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
     domain_unpause(d);
 
     xfree(s);
@@ -652,7 +652,7 @@ int hvm_destroy_ioreq_server(struct domain *d, ioservid_t id)
     struct ioreq_server *s;
     int rc;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -684,7 +684,7 @@ int hvm_destroy_ioreq_server(struct domain *d, ioservid_t id)
     rc = 0;
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -697,7 +697,7 @@ int hvm_get_ioreq_server_info(struct domain *d, ioservid_t id,
     struct ioreq_server *s;
     int rc;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -731,7 +731,7 @@ int hvm_get_ioreq_server_info(struct domain *d, ioservid_t id,
     rc = 0;
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -744,7 +744,7 @@ int hvm_get_ioreq_server_frame(struct domain *d, ioservid_t id,
 
     ASSERT(is_hvm_domain(d));
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -782,7 +782,7 @@ int hvm_get_ioreq_server_frame(struct domain *d, ioservid_t id,
     }
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -798,7 +798,7 @@ int hvm_map_io_range_to_ioreq_server(struct domain *d, ioservid_t id,
     if ( start > end )
         return -EINVAL;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -834,7 +834,7 @@ int hvm_map_io_range_to_ioreq_server(struct domain *d, ioservid_t id,
     rc = rangeset_add_range(r, start, end);
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -850,7 +850,7 @@ int hvm_unmap_io_range_from_ioreq_server(struct domain *d, ioservid_t id,
     if ( start > end )
         return -EINVAL;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -886,7 +886,7 @@ int hvm_unmap_io_range_from_ioreq_server(struct domain *d, ioservid_t id,
     rc = rangeset_remove_range(r, start, end);
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -911,7 +911,7 @@ int hvm_map_mem_type_to_ioreq_server(struct domain *d, ioservid_t id,
     if ( flags & ~XEN_DMOP_IOREQ_MEM_ACCESS_WRITE )
         return -EINVAL;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -926,7 +926,7 @@ int hvm_map_mem_type_to_ioreq_server(struct domain *d, ioservid_t id,
     rc = arch_ioreq_server_map_mem_type(d, s, flags);
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     if ( rc == 0 )
         arch_ioreq_server_map_mem_type_completed(d, s, flags);
@@ -940,7 +940,7 @@ int hvm_set_ioreq_server_state(struct domain *d, ioservid_t id,
     struct ioreq_server *s;
     int rc;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     s = get_ioreq_server(d, id);
 
@@ -964,7 +964,7 @@ int hvm_set_ioreq_server_state(struct domain *d, ioservid_t id,
     rc = 0;
 
  out:
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
     return rc;
 }
 
@@ -974,7 +974,7 @@ int hvm_all_ioreq_servers_add_vcpu(struct domain *d, struct vcpu *v)
     unsigned int id;
     int rc;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     FOR_EACH_IOREQ_SERVER(d, id, s)
     {
@@ -983,7 +983,7 @@ int hvm_all_ioreq_servers_add_vcpu(struct domain *d, struct vcpu *v)
             goto fail;
     }
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return 0;
 
@@ -998,7 +998,7 @@ int hvm_all_ioreq_servers_add_vcpu(struct domain *d, struct vcpu *v)
         hvm_ioreq_server_remove_vcpu(s, v);
     }
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 
     return rc;
 }
@@ -1008,12 +1008,12 @@ void hvm_all_ioreq_servers_remove_vcpu(struct domain *d, struct vcpu *v)
     struct ioreq_server *s;
     unsigned int id;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     FOR_EACH_IOREQ_SERVER(d, id, s)
         hvm_ioreq_server_remove_vcpu(s, v);
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 }
 
 void hvm_destroy_all_ioreq_servers(struct domain *d)
@@ -1024,7 +1024,7 @@ void hvm_destroy_all_ioreq_servers(struct domain *d)
     if ( !arch_ioreq_server_destroy_all(d) )
         return;
 
-    spin_lock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_recursive(&d->ioreq_server.lock);
 
     /* No need to domain_pause() as the domain is being torn down */
 
@@ -1042,7 +1042,7 @@ void hvm_destroy_all_ioreq_servers(struct domain *d)
         xfree(s);
     }
 
-    spin_unlock_recursive(&d->arch.hvm.ioreq_server.lock);
+    spin_unlock_recursive(&d->ioreq_server.lock);
 }
 
 struct ioreq_server *hvm_select_ioreq_server(struct domain *d,
@@ -1274,7 +1274,7 @@ unsigned int hvm_broadcast_ioreq(ioreq_t *p, bool buffered)
 
 void hvm_ioreq_init(struct domain *d)
 {
-    spin_lock_init(&d->arch.hvm.ioreq_server.lock);
+    spin_lock_init(&d->ioreq_server.lock);
 
     arch_ioreq_domain_init(d);
 }
diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm-x86/hvm/domain.h
index 1c4ca47..b8be1ad 100644
--- a/xen/include/asm-x86/hvm/domain.h
+++ b/xen/include/asm-x86/hvm/domain.h
@@ -63,8 +63,6 @@ struct hvm_pi_ops {
     void (*vcpu_block)(struct vcpu *);
 };
 
-#define MAX_NR_IOREQ_SERVERS 8
-
 struct hvm_domain {
     /* Guest page range used for non-default ioreq servers */
     struct {
@@ -73,12 +71,6 @@ struct hvm_domain {
         unsigned long legacy_mask; /* indexed by HVM param number */
     } ioreq_gfn;
 
-    /* Lock protects all other values in the sub-struct and the default */
-    struct {
-        spinlock_t              lock;
-        struct ioreq_server *server[MAX_NR_IOREQ_SERVERS];
-    } ioreq_server;
-
     /* Cached CF8 for guest PCI config cycles */
     uint32_t                pci_cf8;
 
diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index 3e46384..ad0d761 100644
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -318,6 +318,8 @@ struct sched_unit {
 
 struct evtchn_port_ops;
 
+#define MAX_NR_IOREQ_SERVERS 8
+
 struct domain
 {
     domid_t          domain_id;
@@ -533,6 +535,14 @@ struct domain
     struct {
         unsigned int val;
     } teardown;
+
+#ifdef CONFIG_IOREQ_SERVER
+    /* Lock protects all other values in the sub-struct */
+    struct {
+        spinlock_t              lock;
+        struct ioreq_server     *server[MAX_NR_IOREQ_SERVERS];
+    } ioreq_server;
+#endif
 };
 
 static inline struct page_list_head *page_to_list(
-- 
2.7.4


Re: [PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain

Posted by Alex Bennée 1 month, 1 week ago
Oleksandr Tyshchenko <olekstysh@gmail.com> writes:

> From: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
>
> The IOREQ is a common feature now and this struct will be used
> on Arm as is. Move it to common struct domain. This also
> significantly reduces the layering violation in the common code
> (*arch.hvm* usage).
>
> We don't move ioreq_gfn since it is not used in the common code
> (the "legacy" mechanism is x86 specific).
>
> Signed-off-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
> Acked-by: Jan Beulich <jbeulich@suse.com>
> CC: Julien Grall <julien.grall@arm.com>
> [On Arm only]
> Tested-by: Wei Chen <Wei.Chen@arm.com>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée

RE: [PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain

Posted by Paul Durrant 1 month, 1 week ago
> -----Original Message-----
> From: Xen-devel <xen-devel-bounces@lists.xenproject.org> On Behalf Of Oleksandr Tyshchenko
> Sent: 12 January 2021 21:52
> To: xen-devel@lists.xenproject.org
> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>; Paul Durrant <paul@xen.org>; Andrew Cooper
> <andrew.cooper3@citrix.com>; George Dunlap <george.dunlap@citrix.com>; Ian Jackson
> <iwj@xenproject.org>; Jan Beulich <jbeulich@suse.com>; Julien Grall <julien@xen.org>; Stefano
> Stabellini <sstabellini@kernel.org>; Wei Liu <wl@xen.org>; Roger Pau Monné <roger.pau@citrix.com>;
> Julien Grall <julien.grall@arm.com>
> Subject: [PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain
> 
> From: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
> 
> The IOREQ is a common feature now and this struct will be used
> on Arm as is. Move it to common struct domain. This also
> significantly reduces the layering violation in the common code
> (*arch.hvm* usage).
> 
> We don't move ioreq_gfn since it is not used in the common code
> (the "legacy" mechanism is x86 specific).
> 
> Signed-off-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>

Reviewed-by: Paul Durrant <paul@xen.org>

...and I see you fix the alignment issue as part of the code movement in this patch :-)



Re: [PATCH V4 08/24] xen/ioreq: Move x86's ioreq_server to struct domain

Posted by Julien Grall 1 month, 1 week ago
Hi Oleksandr,

On 12/01/2021 21:52, Oleksandr Tyshchenko wrote:
> From: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
> 
> The IOREQ is a common feature now and this struct will be used
> on Arm as is. Move it to common struct domain. This also
> significantly reduces the layering violation in the common code
> (*arch.hvm* usage).
> 
> We don't move ioreq_gfn since it is not used in the common code
> (the "legacy" mechanism is x86 specific).
> 
> Signed-off-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
> Acked-by: Jan Beulich <jbeulich@suse.com>
> CC: Julien Grall <julien.grall@arm.com>
> [On Arm only]
> Tested-by: Wei Chen <Wei.Chen@arm.com>

Reviewed-by: Julien Grall <jgrall@amazon.com>

Cheers,

-- 
Julien Grall