Skip to contentSkip to conversation

xen/xsm: Introduce new boot parameter xsm

PATCH In review Xin Li 4 participants 10 messages Sep 29, 2018 – Oct 8, 2018

lore
Share & tools
lore

Xin Li

author
#0 lore raw

PATCH 1/2 xen/xsm: Introduce new boot parameter xsm

Introduce new boot parameter xsm to choose which xsm module is enabled, and set default to dummy.

---
CC: Daniel De Graaf <[email protected]>
CC: George Dunlap <[email protected]>
CC: Jan Beulich <[email protected]>
CC: Konrad Rzeszutek Wilk <[email protected]>
CC: Stefano Stabellini <[email protected]>
CC: Tim Deegan <[email protected]>
CC: Wei Liu <[email protected]>
CC: Sergey Dyasli <[email protected]>
CC: Andrew Cooper <[email protected]>
CC: Ming Lu <[email protected]>

v4:
1. change the default XSM boot parameter name from "default" to "dummy".
2. Kconfig, remove XSM_FLASK from EXPERT.
3. Kconfig, add new choice to select the default XSM module.

---
 docs/misc/xen-command-line.markdown | 13 ++++++++
 xen/common/Kconfig                  | 13 +++++++-
 xen/xsm/xsm_core.c                  | 46 ++++++++++++++++++++++++++++-
 3 files changed, 70 insertions(+), 2 deletions(-)
docs/misc/xen-command-line.markdown +13 -0
index 1ffd586224..cf9924f53f 100644
@@ -899,6 +899,19 @@ hardware domain is architecture dependent.
 Note that specifying zero as domU value means zero, while for dom0 it means
 to use the default.
 
+### xsm
+> `= dummy | flask`
+
+> Default: `dummy`
+
+Specify which XSM module should be enabled.  This option is only available if
+the hypervisor was compiled with XSM support.
+
+* `dummy`: this is the default choice.  Basic restriction for common deployment
+  (the dummy module) will be applied.  it's also used when XSM is compiled out.
+* `flask`: this is the policy based access control.  To choose this, the
+  separated option in kconfig must also be enabled.
+
 ### flask
 > `= permissive | enforcing | late | disabled`
 
xen/common/Kconfig +12 -1
index 1a6d6281c1..f802efb625 100644
@@ -116,7 +116,7 @@ config XSM
 
 config XSM_FLASK
 	def_bool y
-	prompt "FLux Advanced Security Kernel support" if EXPERT = "y"
+	prompt "FLux Advanced Security Kernel support"
 	depends on XSM
 	---help---
 	  Enables FLASK (FLux Advanced Security Kernel) as the access control
@@ -154,6 +154,17 @@ config XSM_FLASK_POLICY
 
 	  If unsure, say Y.
 
+choice
+	prompt "Default XSM implementation"
+	depends on XSM
+	default XSM_FLASK_DEFAULT if XSM_FLASK
+	default XSM_DUMMY_DEFAULT
+	config XSM_DUMMY_DEFAULT
+		bool "Match non-XSM behavior"
+	config XSM_FLASK_DEFAULT
+		bool "FLux Advanced Security Kernel" if XSM_FLASK
+endchoice
+
 config LATE_HWDOM
 	bool "Dedicated hardware domain"
 	default n
xen/xsm/xsm_core.c +45 -1

58 lines not on this page · raw message

Signature · 7 lines
2.18.0


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
author
#1 lore raw

PATCH 2/2 xen/xsm: Add new SILO mode for XSM

When SILO is enabled, there would be no page-sharing or event notifications between unprivileged VMs (no grant tables or event channels).

---
CC: Daniel De Graaf <[email protected]>
CC: George Dunlap <[email protected]>
CC: Jan Beulich <[email protected]>
CC: Konrad Rzeszutek Wilk <[email protected]>
CC: Stefano Stabellini <[email protected]>
CC: Tim Deegan <[email protected]>
CC: Wei Liu <[email protected]>
CC: Sergey Dyasli <[email protected]>
CC: Andrew Cooper <[email protected]>
CC: Ming Lu <[email protected]>

v4:
1. include the dummy.h as a copy, and call the dummy module functions to avoid indirect all.
2. call register_xsm() to setup silo module.

---
 docs/misc/xen-command-line.markdown |   5 +-
 xen/common/Kconfig                  |  15 ++++
 xen/include/xsm/dummy.h             |   5 ++
 xen/include/xsm/xsm.h               |   6 ++
 xen/xsm/Makefile                    |   1 +
 xen/xsm/dummy.c                     |   1 -
 xen/xsm/silo.c                      | 109 ++++++++++++++++++++++++++++
 xen/xsm/xsm_core.c                  |  11 +++
 8 files changed, 151 insertions(+), 2 deletions(-)
 create mode 100644 xen/xsm/silo.c
docs/misc/xen-command-line.markdown +4 -1
index cf9924f53f..1b49fda8fc 100644
@@ -900,7 +900,7 @@ Note that specifying zero as domU value means zero, while for dom0 it means
 to use the default.
 
 ### xsm
-> `= dummy | flask`
+> `= dummy | flask | silo`
 
 > Default: `dummy`
 
@@ -911,6 +911,9 @@ the hypervisor was compiled with XSM support.
   (the dummy module) will be applied.  it's also used when XSM is compiled out.
 * `flask`: this is the policy based access control.  To choose this, the
   separated option in kconfig must also be enabled.
+* `silo`: this will deny any unmediated communication channels between
+  unprivileged VMs.  To choose this, the separated option in kconfig must also
+  be enabled.
 
 ### flask
 > `= permissive | enforcing | late | disabled`
xen/common/Kconfig +15 -0
index f802efb625..ce965fbf17 100644
@@ -154,15 +154,30 @@ config XSM_FLASK_POLICY
 
 	  If unsure, say Y.
 
+config XSM_SILO
+	def_bool y
+	prompt "SILO support"
+	depends on XSM
+	---help---
+	  Enables SILO as the access control mechanism used by the XSM framework.
+	  This is not the default module, add boot parameter xsm=silo to choose
+	  it. This will deny any unmediated communication channels (grant tables
+	  and event channels) between unprivileged VMs.
+
+	  If unsure, say Y.
+
 choice
 	prompt "Default XSM implementation"
 	depends on XSM
 	default XSM_FLASK_DEFAULT if XSM_FLASK
+	default XSM_SILO_DEFAULT if XSM_SILO
 	default XSM_DUMMY_DEFAULT
 	config XSM_DUMMY_DEFAULT
 		bool "Match non-XSM behavior"
 	config XSM_FLASK_DEFAULT
 		bool "FLux Advanced Security Kernel" if XSM_FLASK
+	config XSM_SILO_DEFAULT
+		bool "SILO" if XSM_SILO
 endchoice
 
 config LATE_HWDOM
xen/include/xsm/dummy.h +5 -0
index b0ac1f66b3..b2bb16c55e 100644
@@ -48,7 +48,12 @@ void __xsm_action_mismatch_detected(void);
  * There is no xsm_default_t argument available, so the value from the assertion
  * is used to initialize the variable.
  */
+#ifdef CONFIG_XSM_SILO
+#define XSM_INLINE __attribute__ ((unused))
+#else
 #define XSM_INLINE /* */
+#endif
+
 #define XSM_DEFAULT_ARG /* */
 #define XSM_DEFAULT_VOID void
 #define XSM_ASSERT_ACTION(def) xsm_default_t action = def; (void)action
xen/include/xsm/xsm.h +6 -0
index 3d67962493..3b192b5c31 100644
@@ -733,6 +733,12 @@ extern const unsigned char xsm_flask_init_policy[];
 extern const unsigned int xsm_flask_init_policy_size;
 #endif
 
+#ifdef CONFIG_XSM_SILO
+extern void silo_init(void);
+#else
+static inline void silo_init(void) {}
+#endif
+
 #else /* CONFIG_XSM */
 
 #include <xsm/dummy.h>
xen/xsm/Makefile +1 -0
index 8bb4a24f09..e4d581e065 100644
@@ -1,5 +1,6 @@
 obj-y += xsm_core.o
 obj-$(CONFIG_XSM) += xsm_policy.o
 obj-$(CONFIG_XSM) += dummy.o
+obj-$(CONFIG_XSM_SILO) += silo.o
 
 subdir-$(CONFIG_XSM_FLASK) += flask
xen/xsm/dummy.c +0 -1
index 3290d04527..06a674fad0 100644
@@ -11,7 +11,6 @@
  */
 
 #define XSM_NO_WRAPPERS
-#define XSM_INLINE /* */
 #include <xsm/dummy.h>
 
 struct xsm_operations dummy_xsm_ops;
xen/xsm/silo.c new file +109 -0

109 lines not on this page · raw message

xen/xsm/xsm_core.c +11 -0
index df284ec463..262af5ec9b 100644
@@ -34,11 +34,14 @@ struct xsm_operations *xsm_ops;
 enum xsm_bootparam {
     XSM_BOOTPARAM_DUMMY,
     XSM_BOOTPARAM_FLASK,
+    XSM_BOOTPARAM_SILO,
 };
 
 static enum xsm_bootparam __initdata xsm_bootparam =
 #ifdef CONFIG_XSM_FLASK_DEFAULT
     XSM_BOOTPARAM_FLASK;
+#elif CONFIG_XSM_SILO_DEFAULT
+    XSM_BOOTPARAM_SILO;
 #else
     XSM_BOOTPARAM_DUMMY;
 #endif
@@ -52,6 +55,10 @@ static int __init parse_xsm_param(const char *s)
 #ifdef CONFIG_XSM_FLASK
     else if ( !strcmp(s, "flask") )
         xsm_bootparam = XSM_BOOTPARAM_FLASK;
+#endif
+#ifdef CONFIG_XSM_SILO
+    else if ( !strcmp(s, "silo") )
+        xsm_bootparam = XSM_BOOTPARAM_SILO;
 #endif
     else {
         printk("XSM: can't parse boot parameter xsm=%s\n", s);
@@ -98,6 +105,10 @@ static int __init xsm_core_init(const void *policy_buffer, size_t policy_size)
         flask_init(policy_buffer, policy_size);
         break;
 
+    case XSM_BOOTPARAM_SILO:
+        silo_init();
+        break;
+
     default:
         printk("XSM: Invalid value for xsm= boot parameter\n");
         break;
Signature · 7 lines
2.18.0


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Acked-by Daniel De Graaf on v1 2/2
#2 lore raw

Re: [Non-DoD Source] [PATCH 2/2] xen/xsm: Add new SILO mode for XSM

Quoted 6 lines

When SILO is enabled, there would be no page-sharing or event notifications between unprivileged VMs (no grant tables or event channels).

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#3 lore raw

Re: [PATCH 2/2] xen/xsm: Add new SILO mode for XSM

In reply to Xin Li

xen/include/xsm/dummy.h void __xsm_action_mismatch_detected(void); · line 52

Quoted 9 lines
Quoted 1 line
Quoted 1 line · On 29.09.18 at 11:22, <[email protected]> wrote:
xen/include/xsm/dummy.h +2 -0
@@ -48,7 +48,12 @@ void __xsm_action_mismatch_detected(void);
  * There is no xsm_default_t argument available, so the value from the assertion
  * is used to initialize the variable.
  */
+#ifdef CONFIG_XSM_SILO
+#define XSM_INLINE __attribute__ ((unused))

Please don't open-code __maybe_unused. Furthermore I question the dependency on CONFIG_XSM_SILO here: Afaict you want this for just the single new inclusion site, without affecting any others.

xen/include/xsm/xsm.h extern const unsigned char xsm_flask_init_policy[]; · line 740

Quoted 11 lines
xen/include/xsm/xsm.h +5 -0
@@ -733,6 +733,12 @@ extern const unsigned char xsm_flask_init_policy[];
 extern const unsigned int xsm_flask_init_policy_size;
 #endif
 
+#ifdef CONFIG_XSM_SILO
+extern void silo_init(void);
+#else
+static inline void silo_init(void) {}
+#endif

Along the lines of one of my remarks on patch 1, I think you would better get away without the inline variant, by adding suitable #ifdef-s to eliminate the risk of wrong use of the new enumerator.

xen/xsm/dummy.c line 14

Quoted 8 lines
xen/xsm/dummy.c +0 -1
@@ -11,7 +11,6 @@
  */
 
 #define XSM_NO_WRAPPERS
-#define XSM_INLINE /* */
 #include <xsm/dummy.h>

The change looks unmotivated here, perhaps because of the questionable CONFIG_XSM_SILO dependency further up. That said, it looks as if the #define could go away currently, as being redundant with what dummy.h already has. That would then better be a small separate prereq patch imo.

xen/xsm/silo.c line 4

Quoted 7 lines
xen/xsm/silo.c new file +4 -0
@@ -0,0 +1,109 @@
+/******************************************************************************
+ * xsm/silo.c
+ *
+ * SILO module for XSM(Xen Security Modules)

Would you mind adding the missing blank here?

Quoted 7 lines
xen/xsm/silo.c new file +7 -0
+ * Copyright (c) 2018 Citrix Systems Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT

xen/xsm/silo.c line 22

Quoted 10 lines
xen/xsm/silo.c new file +10 -0
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; If not, see <http://www.gnu.org/licenses/>.
+ */
+#define XSM_NO_WRAPPERS
+
+#include <xsm/dummy.h>

Please switch around the blank and the #define lines, matching how dummy.c is written. This helps readers understand that the #define is specifically in preparation of the #include.

xen/xsm/silo.c line 31

Quoted 8 lines
xen/xsm/silo.c new file +8 -0
+/*
+ * Check if inter-domain communication is allowed.
+ * Return true when pass check.
+ */
+static bool silo_mode_dom_check(const struct domain *ldom,
+                                const struct domain *rdom)
+{
+    const struct domain *cur_dom = current->domain;

We commonly call such variables currd.

xen/xsm/silo.c line 43

Quoted 11 lines
xen/xsm/silo.c new file +11 -0
+    return (is_control_domain(cur_dom) || is_control_domain(ldom) ||
+            is_control_domain(rdom) || ldom == rdom);
+}
+
+static int silo_evtchn_unbound(struct domain *d1, struct evtchn *chn,
+                               domid_t id2)
+{
+    int rc = -EPERM;
+    struct domain *d2 = rcu_lock_domain_by_any_id(id2);
+
+    if ( d2 == NULL )

We generally try to use the shorter !d2 in new code. But it's a matter of personal preference, I agree.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#4 lore raw

Re: [PATCH 2/2] xen/xsm: Add new SILO mode for XSM

Thanks Jan.

Quoted 18 lines

-----Original Message-----
From: Jan Beulich [mailto:[email protected]]
Sent: Tuesday, October 2, 2018 5:33 PM
To: Xin Li <[email protected]>
Cc: Andrew Cooper <[email protected]>; Ming Lu <[email protected]>; Sergey Dyasli <[email protected]>; Wei Liu <[email protected]>; Xin Li (Talons) <[email protected]>; George Dunlap <[email protected]>; Stefano Stabellini <[email protected]>; xen- [email protected]; Konrad Rzeszutek Wilk <[email protected]>; Daniel de Graaf <[email protected]>; Tim (Xen.org) <[email protected]>
Subject: Re: [PATCH 2/2] xen/xsm: Add new SILO mode for XSM

xen/include/xsm/dummy.h void __xsm_action_mismatch_detected(void); · line 48

Quoted 5 lines
Quoted 1 line
Quoted 1 line · On 29.09.18 at 11:22, <[email protected]> wrote:
xen/include/xsm/dummy.h +0 -0
@@ -48,7 +48,12 @@ void __xsm_action_mismatch_detected(void);
  * There is no xsm_default_t argument available, so the value from the

assertion

Quoted 8 lines

xen/include/xsm/dummy.h void __xsm_action_mismatch_detected(void); · line 52

Quoted 4 lines
xen/include/xsm/dummy.h +2 -0
  * is used to initialize the variable.
  */
+#ifdef CONFIG_XSM_SILO
+#define XSM_INLINE __attribute__ ((unused))

Please don't open-code __maybe_unused. Furthermore I question the dependency on CONFIG_XSM_SILO here: Afaict you want this for just the single new inclusion site, without affecting any others.

OK, use __maybe_unused directly and remove this #ifdef.

Quoted 7 lines
Quoted 7 lines
xen/include/xsm/xsm.h +0 -0
@@ -733,6 +733,12 @@ extern const unsigned char

xsm_flask_init_policy[]; extern const unsigned int xsm_flask_init_policy_size; #endif

Quoted 8 lines

xen/include/xsm/xsm.h extern const unsigned char xsm_flask_init_policy[];

Quoted 4 lines
xen/include/xsm/xsm.h +4 -0
+#ifdef CONFIG_XSM_SILO
+extern void silo_init(void);
+#else
+static inline void silo_init(void) {} #endif

Along the lines of one of my remarks on patch 1, I think you would better get away without the inline variant, by adding suitable #ifdef-s to eliminate the risk of wrong use of the new enumerator.

I can remove the inline function, and add #idef in xsm_core.c But, this way does not match the current code style(like flask_init()). So I would like to keep this way.

Quoted 6 lines
Quoted 6 lines
xen/xsm/dummy.c +0 -0
@@ -11,7 +11,6 @@
  */
 
Quoted 8 lines

xen/xsm/dummy.c line 14

Quoted 3 lines
xen/xsm/dummy.c +0 -1
 #define XSM_NO_WRAPPERS
-#define XSM_INLINE /* */
 #include <xsm/dummy.h>

The change looks unmotivated here, perhaps because of the questionable CONFIG_XSM_SILO dependency further up. That said, it looks as if the #define could go away currently, as being redundant with what dummy.h already has. That would then better be a small separate prereq patch imo.

This is intended, to avoid compile error, can't refine XSM_INLINE. Do you mean a separate patch to just remove this line?

Quoted 7 lines

xen/xsm/silo.c

Quoted 5 lines
xen/xsm/silo.c new file +0 -0
@@ -0,0 +1,109 @@

+/***************************************************************
*****

Quoted 6 lines

xen/xsm/silo.c line 4

Quoted 4 lines
xen/xsm/silo.c new file +4 -0
+**********
+ * xsm/silo.c
+ *
+ * SILO module for XSM(Xen Security Modules)

Would you mind adding the missing blank here?

Do you mean add one space after XSM?
* SILO module for XSM (Xen Security Modules)

Quoted 12 lines

xen/xsm/silo.c

Quoted 11 lines
xen/xsm/silo.c new file +10 -0
+ * Copyright (c) 2018 Citrix Systems Ltd.
+ *
+ * This program is free software; you can redistribute it and/or
+modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but
+WITHOUT
+ * ANY WARRANTY; without even the implied warranty of

MERCHANTABILITY

Quoted 16 lines

xen/xsm/silo.c line 22

Quoted 12 lines
xen/xsm/silo.c new file +12 -0
+or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
+License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+along with
+ * this program; If not, see <http://www.gnu.org/licenses/>.
+ */
+#define XSM_NO_WRAPPERS
+
+#include <xsm/dummy.h>

Please switch around the blank and the #define lines, matching how dummy.c is written. This helps readers understand that the #define is specifically in preparation of the #include.

OK. Remove the line before #include.

Quoted 9 lines

xen/xsm/silo.c line 31

Quoted 7 lines
xen/xsm/silo.c new file +7 -0
+/*
+ * Check if inter-domain communication is allowed.
+ * Return true when pass check.
+ */
+static bool silo_mode_dom_check(const struct domain *ldom,
+                                const struct domain *rdom) {
+    const struct domain *cur_dom = current->domain;

We commonly call such variables currd.

OK. Rename this var.

Quoted 13 lines

xen/xsm/silo.c line 43

Quoted 10 lines
xen/xsm/silo.c new file +9 -0
+    return (is_control_domain(cur_dom) || is_control_domain(ldom) ||
+            is_control_domain(rdom) || ldom == rdom); }
+
+static int silo_evtchn_unbound(struct domain *d1, struct evtchn *chn,
+                               domid_t id2) {
+    int rc = -EPERM;
+    struct domain *d2 = rcu_lock_domain_by_any_id(id2);
+
+    if ( d2 == NULL )

We generally try to use the shorter !d2 in new code. But it's a matter of personal preference, I agree.

Thanks. I will just keep it.

Quoted 3 lines

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#5 lore raw

Re: [PATCH 2/2] xen/xsm: Add new SILO mode for XSM

Quoted 19 lines
Quoted 10 lines
Quoted 1 line · On 08.10.18 at 09:49, <[email protected]> wrote:

From: Jan Beulich [mailto:[email protected]]
Sent: Tuesday, October 2, 2018 5:33 PM

Quoted 6 lines
Quoted 1 line
Quoted 1 line · On 29.09.18 at 11:22, <[email protected]> wrote:
xen/xsm/dummy.c +0 -0
@@ -11,7 +11,6 @@
  */
 
Quoted 8 lines

xen/xsm/dummy.c line 14

Quoted 3 lines
xen/xsm/dummy.c +0 -1
 #define XSM_NO_WRAPPERS
-#define XSM_INLINE /* */
 #include <xsm/dummy.h>

The change looks unmotivated here, perhaps because of the questionable CONFIG_XSM_SILO dependency further up. That said, it looks as if the #define could go away currently, as being redundant with what dummy.h already has. That would then better be a small separate prereq patch imo.

Quoted 2 lines

This is intended, to avoid compile error, can't refine XSM_INLINE. Do you mean a separate patch to just remove this line?

Yes, and then with an explanation of why this line is not needed.

Quoted 11 lines
Quoted 11 lines

xen/xsm/silo.c

Quoted 4 lines
xen/xsm/silo.c new file +0 -0
@@ -0,0 +1,109 @@

+/***************************************************************
*****

Quoted 4 lines
xen/xsm/silo.c new file +4 -0
+**********
+ * xsm/silo.c
+ *
+ * SILO module for XSM(Xen Security Modules)
Quoted 4 lines
Quoted 1 line

Would you mind adding the missing blank here?

Do you mean add one space after XSM?
* SILO module for XSM (Xen Security Modules)

Yes.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

Acked-by Daniel De Graaf on v1 1/2
#6 lore raw
In reply to Xin Li
Quoted 5 lines

Introduce new boot parameter xsm to choose which xsm module is enabled, and set default to dummy.

It might be useful for the commit message to also reference the new Kconfig option; thanks for adding it.

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#7 lore raw

Thanks Daniel.

Quoted 18 lines

-----Original Message-----
From: DeGraaf, Daniel G [mailto:[email protected]]
Sent: Monday, October 1, 2018 11:18 PM
To: 'Xin Li' <[email protected]>; [email protected]
Cc: Xin Li (Talons) <[email protected]>; Daniel De Graaf <[email protected]>; George Dunlap <[email protected]>; Jan Beulich <[email protected]>; Konrad Rzeszutek Wilk <[email protected]>; Stefano Stabellini <[email protected]>; Tim (Xen.org) <[email protected]>; Wei Liu <[email protected]>; Sergey Dyasli <[email protected]>; Andrew Cooper <[email protected]>; Ming Lu <[email protected]>
Subject: RE: [Non-DoD Source] [PATCH 1/2] xen/xsm: Introduce new boot parameter xsm

Quoted 4 lines

Introduce new boot parameter xsm to choose which xsm module is enabled, and set default to dummy.

Quoted 6 lines
Quoted 1 line

It might be useful for the commit message to also reference the new Kconfig option; thanks for adding it.

OK. Sure.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#8 lore raw
In reply to Xin Li
Quoted 7 lines
Quoted 1 line
Quoted 1 line · On 29.09.18 at 11:22, <[email protected]> wrote:
docs/misc/xen-command-line.markdown +0 -0
@@ -899,6 +899,19 @@ hardware domain is architecture dependent.
 Note that specifying zero as domU value means zero, while for dom0 it means
 to use the default.
 

docs/misc/xen-command-line.markdown hardware domain is architecture dependent. · line 911

Quoted 10 lines
docs/misc/xen-command-line.markdown +10 -0
+### xsm
+> `= dummy | flask`
+
+> Default: `dummy`
+
+Specify which XSM module should be enabled.  This option is only available if
+the hypervisor was compiled with XSM support.
+
+* `dummy`: this is the default choice.  Basic restriction for common deployment
+  (the dummy module) will be applied.  it's also used when XSM is compiled out.

"It's" (upper case I).

docs/misc/xen-command-line.markdown hardware domain is architecture dependent. · line 913

Quoted 2 lines
docs/misc/xen-command-line.markdown +2 -0
+* `flask`: this is the policy based access control.  To choose this, the
+  separated option in kconfig must also be enabled.

Perhaps better "separate" (but I'm not a native speaker)?

Quoted 4 lines
docs/misc/xen-command-line.markdown +0 -0
@@ -154,6 +154,17 @@ config XSM_FLASK_POLICY
 
 	  If unsure, say Y.
 

docs/misc/xen-command-line.markdown config XSM_FLASK_POLICY · line 166

Quoted 10 lines
docs/misc/xen-command-line.markdown +10 -0
+choice
+	prompt "Default XSM implementation"
+	depends on XSM
+	default XSM_FLASK_DEFAULT if XSM_FLASK
+	default XSM_DUMMY_DEFAULT
+	config XSM_DUMMY_DEFAULT
+		bool "Match non-XSM behavior"
+	config XSM_FLASK_DEFAULT
+		bool "FLux Advanced Security Kernel" if XSM_FLASK
+endchoice

Personally I'd prefer XSM_DEFAULT_*; not sure what others think.

xen/xsm/xsm_core.c line 36

Quoted 9 lines
xen/xsm/xsm_core.c +3 -0
@@ -31,6 +31,37 @@
 
 struct xsm_operations *xsm_ops;
 
+enum xsm_bootparam {
+    XSM_BOOTPARAM_DUMMY,
+    XSM_BOOTPARAM_FLASK,

Considering the #ifdef-s further down, should this perhaps also be put in "#ifdef CONFIG_XSM_FLASK", to avoid it mistakenly getting used when the code was compiled out?

Quoted 10 lines
xen/xsm/xsm_core.c +10 -0
+};
+
+static enum xsm_bootparam __initdata xsm_bootparam =
+#ifdef CONFIG_XSM_FLASK_DEFAULT
+    XSM_BOOTPARAM_FLASK;
+#else
+    XSM_BOOTPARAM_DUMMY;
+#endif
+
+static int __init parse_xsm_param(const char *s)

xen/xsm/xsm_core.c line 56

Quoted 10 lines
xen/xsm/xsm_core.c +10 -0
+{
+    int rc = 0;
+
+    if ( !strcmp(s, "dummy") )
+        xsm_bootparam = XSM_BOOTPARAM_DUMMY;
+#ifdef CONFIG_XSM_FLASK
+    else if ( !strcmp(s, "flask") )
+        xsm_bootparam = XSM_BOOTPARAM_FLASK;
+#endif
+    else {

Style (brace goes on its own line).

xen/xsm/xsm_core.c line 57

Quoted 1 line
xen/xsm/xsm_core.c +1 -0
+        printk("XSM: can't parse boot parameter xsm=%s\n", s);

Again, not being a native speaker, "can't parse" sounds odd. Why not just "unknown" or "unknown or unsupported"?

Quoted 8 lines
xen/xsm/xsm_core.c +3 -1
@@ -57,7 +88,20 @@ static int __init xsm_core_init(const void *policy_buffer, size_t policy_size)
     }
 
     xsm_ops = &dummy_xsm_ops;
-    flask_init(policy_buffer, policy_size);
+
+    switch ( xsm_bootparam )
+    {

xen/xsm/xsm_core.c static int __init xsm_core_init(const void *policy_buffer, size_t policy_size) · line 103

Quoted 10 lines
xen/xsm/xsm_core.c +10 -0
+    case XSM_BOOTPARAM_DUMMY:
+        break;
+
+    case XSM_BOOTPARAM_FLASK:
+        flask_init(policy_buffer, policy_size);
+        break;
+
+    default:
+        printk("XSM: Invalid value for xsm= boot parameter\n");
+        break;

What situation is this covering, when all enumerators already have their case label? Simply ASSERT_UNREACHABLE()?

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#9 lore raw

Thanks Jan.

Quoted 20 lines
Quoted 8 lines
Quoted 1 line
Quoted 1 line · On 29.09.18 at 11:22, <[email protected]> wrote:
docs/misc/xen-command-line.markdown +0 -0
@@ -899,6 +899,19 @@ hardware domain is architecture dependent.
 Note that specifying zero as domU value means zero, while for dom0 it

means to use the default.

docs/misc/xen-command-line.markdown +1 -0
+### xsm

docs/misc/xen-command-line.markdown hardware domain is architecture dependent.

Quoted 10 lines
docs/misc/xen-command-line.markdown +10 -0
+> `= dummy | flask`
+
+> Default: `dummy`
+
+Specify which XSM module should be enabled.  This option is only
+available if the hypervisor was compiled with XSM support.
+
+* `dummy`: this is the default choice.  Basic restriction for common
+deployment
+  (the dummy module) will be applied.  it's also used when XSM is compiled

out.

Quoted 2 lines

"It's" (upper case I).

OK. Done.

Quoted 6 lines

docs/misc/xen-command-line.markdown

Quoted 3 lines
docs/misc/xen-command-line.markdown +3 -0
+* `flask`: this is the policy based access control.  To choose this,
+the
+  separated option in kconfig must also be enabled.

Perhaps better "separate" (but I'm not a native speaker)?

I prefer separated, as I checked here: https://english.stackexchange.com/questions/13144/separated-versus-separate
But generally both are OK I think.

Quoted 4 lines
Quoted 4 lines
docs/misc/xen-command-line.markdown +0 -0
@@ -154,6 +154,17 @@ config XSM_FLASK_POLICY
 
 	  If unsure, say Y.
 
Quoted 12 lines

docs/misc/xen-command-line.markdown config XSM_FLASK_POLICY · line 165

Quoted 9 lines
docs/misc/xen-command-line.markdown +9 -0
+choice
+	prompt "Default XSM implementation"
+	depends on XSM
+	default XSM_FLASK_DEFAULT if XSM_FLASK
+	default XSM_DUMMY_DEFAULT
+	config XSM_DUMMY_DEFAULT
+		bool "Match non-XSM behavior"
+	config XSM_FLASK_DEFAULT
+		bool "FLux Advanced Security Kernel" if XSM_FLASK endchoice

Personally I'd prefer XSM_DEFAULT_*; not sure what others think.

I would like to keep this style, because I see "
default "credit" if SCHED_CREDIT_DEFAULT default "credit2" if SCHED_CREDIT2_DEFAULT
" in config file.

Quoted 6 lines
Quoted 6 lines
xen/xsm/xsm_core.c +0 -0
@@ -31,6 +31,37 @@
 
 struct xsm_operations *xsm_ops;
 
Quoted 8 lines

xen/xsm/xsm_core.c line 36

Quoted 3 lines
xen/xsm/xsm_core.c +3 -0
+enum xsm_bootparam {
+    XSM_BOOTPARAM_DUMMY,
+    XSM_BOOTPARAM_FLASK,

Considering the #ifdef-s further down, should this perhaps also be put in "#ifdef CONFIG_XSM_FLASK", to avoid it mistakenly getting used when the code was compiled out?

I would like to avoid #ifdef here, because I prefer the same enum value in different config.

Quoted 15 lines
Quoted 4 lines
xen/xsm/xsm_core.c +4 -0
+};
+
+static enum xsm_bootparam __initdata xsm_bootparam = #ifdef
+CONFIG_XSM_FLASK_DEFAULT

xen/xsm/xsm_core.c

Quoted 10 lines
xen/xsm/xsm_core.c +10 -0
+    XSM_BOOTPARAM_FLASK;
+#else
+    XSM_BOOTPARAM_DUMMY;
+#endif
+
+static int __init parse_xsm_param(const char *s) {
+    int rc = 0;
+
+    if ( !strcmp(s, "dummy") )
+        xsm_bootparam = XSM_BOOTPARAM_DUMMY; #ifdef

CONFIG_XSM_FLASK

Quoted 5 lines

xen/xsm/xsm_core.c line 56

Quoted 3 lines
xen/xsm/xsm_core.c +3 -0
+    else if ( !strcmp(s, "flask") )
+        xsm_bootparam = XSM_BOOTPARAM_FLASK; #endif
+    else {

Style (brace goes on its own line).

OK. Done.

Quoted 5 lines

xen/xsm/xsm_core.c line 57

Quoted 2 lines
xen/xsm/xsm_core.c +1 -0
+        printk("XSM: can't parse boot parameter xsm=%s\n", s);

Again, not being a native speaker, "can't parse" sounds odd. Why not just "unknown" or "unknown or unsupported"?

OK. Done.

Quoted 15 lines

xen/xsm/xsm_core.c static int __init xsm_core_init(const void

Quoted 2 lines
xen/xsm/xsm_core.c +0 -0
@@ -57,7 +88,20 @@ static int __init xsm_core_init(const void

*policy_buffer, size_t policy_size)

Quoted 12 lines
xen/xsm/xsm_core.c +8 -1
     }
 
     xsm_ops = &dummy_xsm_ops;
-    flask_init(policy_buffer, policy_size);
+
+    switch ( xsm_bootparam )
+    {
+    case XSM_BOOTPARAM_DUMMY:
+        break;
+
+    case XSM_BOOTPARAM_FLASK:
+        flask_init(policy_buffer, policy_size);
Quoted 9 lines

xen/xsm/xsm_core.c static int __init xsm_core_init(const void *policy_buffer, size_t policy_size) · line 103

Quoted 5 lines
xen/xsm/xsm_core.c +5 -0
+        break;
+
+    default:
+        printk("XSM: Invalid value for xsm= boot parameter\n");
+        break;

What situation is this covering, when all enumerators already have their case label? Simply ASSERT_UNREACHABLE()?

OK. Done.

Quoted 2 lines

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel