[PATCH RFC 0/2] Make hvm_fep available to non-debug builds

[PATCH RFC 0/2] Make hvm_fep available to non-debug builds

[Wei Liu]

Wei Liu (2):
  xen/kernel: document 'C' in print_tainted
  xen: make available hvm_fep to non-debug build as well

 docs/misc/xen-command-line.markdown |  8 ++++++--
 xen/arch/x86/hvm/hvm.c              | 31 ++++++++++++++++++++++++++++---
 xen/common/kernel.c                 |  7 +++++--
 xen/include/asm-x86/hvm/hvm.h       |  4 ----
 xen/include/xen/lib.h               |  1 +
 5 files changed, 40 insertions(+), 11 deletions(-)

-- 
2.1.4


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH RFC 1/2] xen/kernel: document 'C' in print_tainted

[Wei Liu]

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
---
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: George Dunlap <George.Dunlap@eu.citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>
Cc: Jan Beulich <jbeulich@suse.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: Tim Deegan <tim@xen.org>
Cc: Wei Liu <wei.liu2@citrix.com>
---
 xen/common/kernel.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/xen/common/kernel.c b/xen/common/kernel.c
index 1a6823a..dae7e35 100644
--- a/xen/common/kernel.c
+++ b/xen/common/kernel.c
@@ -174,6 +174,7 @@ int __init parse_bool(const char *s)
  *  'S' - SMP with CPUs not designed for SMP.
  *  'M' - Machine had a machine check experience.
  *  'B' - System has hit bad_page.
+ *  'C' - Console output is synchronous.
  *
  *      The string is overwritten by the next call to print_taint().
  */
-- 
2.1.4


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 1/2] xen/kernel: document 'C' in print_tainted

[Jan Beulich]

>>> On 15.06.16 at 16:31, <wei.liu2@citrix.com> wrote:
> Signed-off-by: Wei Liu <wei.liu2@citrix.com>

Acked-by: Jan Beulich <jbeulich@suse.com>


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

[PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Wei Liu]

Originally hvm_fep was guarded by NDEBUG, which means it was only
available to debug builds.

However there is value to have it for non-debug builds as well. User can
use that to run tests in setup that replicates production setup.

Make it clear with a sync_console style warning that this option can't
be used in production setup. Update command line documentation
accordingly. Finally mark Xen as tainted when this option is enabled.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
---
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: Jan Beulich <jbeulich@suse.com>
---
 docs/misc/xen-command-line.markdown |  8 ++++++--
 xen/arch/x86/hvm/hvm.c              | 31 ++++++++++++++++++++++++++++---
 xen/common/kernel.c                 |  6 ++++--
 xen/include/asm-x86/hvm/hvm.h       |  4 ----
 xen/include/xen/lib.h               |  1 +
 5 files changed, 39 insertions(+), 11 deletions(-)

diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown
index fed732c..dc53e24 100644
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -878,8 +878,12 @@ Recognized in debug builds of the hypervisor only.
 Allow use of the Forced Emulation Prefix in HVM guests, to allow emulation of
 arbitrary instructions.
 
-This option is intended for development purposes, and is only available in
-debug builds of the hypervisor.
+This option is intended for development and testing purposes.
+
+*Warning*
+As this feature opens up the instruction emulator to HVM guest, don't
+use this in production system. No security support is provided when
+this flag is set.
 
 ### hvm\_port80
 > `= <boolean>`
diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
index 78db903..5bafaef 100644
--- a/xen/arch/x86/hvm/hvm.c
+++ b/xen/arch/x86/hvm/hvm.c
@@ -37,6 +37,7 @@
 #include <xen/mem_access.h>
 #include <xen/rangeset.h>
 #include <xen/vm_event.h>
+#include <xen/delay.h>
 #include <asm/shadow.h>
 #include <asm/hap.h>
 #include <asm/current.h>
@@ -95,11 +96,9 @@ unsigned long __section(".bss.page_aligned")
 static bool_t __initdata opt_hap_enabled = 1;
 boolean_param("hap", opt_hap_enabled);
 
-#ifndef opt_hvm_fep
 /* Permit use of the Forced Emulation Prefix in HVM guests */
-bool_t opt_hvm_fep;
+bool_t __read_mostly opt_hvm_fep;
 boolean_param("hvm_fep", opt_hvm_fep);
-#endif
 
 /* Xen command-line option to enable altp2m */
 static bool_t __initdata opt_altp2m_enabled = 0;
@@ -182,6 +181,32 @@ static int __init hvm_enable(void)
     if ( !opt_altp2m_enabled )
         hvm_funcs.altp2m_supported = 0;
 
+    if ( opt_hvm_fep )
+    {
+        unsigned i, j;
+
+        printk("**********************************************\n");
+        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
+        printk("******* This option is *ONLY* intended to aid debugging "
+               "and testing of Xen\n");
+        printk("******* that HVM guest can enter instruction emulator "
+               "with UD instruction.\n");
+        printk("******* It has implication on the security of the system.\n");
+        printk("******* Please *DO NOT* use this in production.\n");
+        printk("**********************************************\n");
+        add_taint(TAINT_HVM_FEP);
+        for ( i = 0; i < 3; i++ )
+        {
+            printk("%d... ", 3-i);
+            for ( j = 0; j < 100; j++ )
+            {
+                process_pending_softirqs();
+                mdelay(10);
+            }
+        }
+        printk("\n");
+    }
+
     /*
      * Allow direct access to the PC debug ports 0x80 and 0xed (they are
      * often used for I/O delays, but the vmexits simply slow things down).
diff --git a/xen/common/kernel.c b/xen/common/kernel.c
index dae7e35..5bf77aa 100644
--- a/xen/common/kernel.c
+++ b/xen/common/kernel.c
@@ -175,6 +175,7 @@ int __init parse_bool(const char *s)
  *  'M' - Machine had a machine check experience.
  *  'B' - System has hit bad_page.
  *  'C' - Console output is synchronous.
+ *  'H' - HVM forced emulation prefix is permitted.
  *
  *      The string is overwritten by the next call to print_taint().
  */
@@ -182,11 +183,12 @@ char *print_tainted(char *str)
 {
     if ( tainted )
     {
-        snprintf(str, TAINT_STRING_MAX_LEN, "Tainted: %c%c%c%c",
+        snprintf(str, TAINT_STRING_MAX_LEN, "Tainted: %c%c%c%c%c",
                  tainted & TAINT_UNSAFE_SMP ? 'S' : ' ',
                  tainted & TAINT_MACHINE_CHECK ? 'M' : ' ',
                  tainted & TAINT_BAD_PAGE ? 'B' : ' ',
-                 tainted & TAINT_SYNC_CONSOLE ? 'C' : ' ');
+                 tainted & TAINT_SYNC_CONSOLE ? 'C' : ' ',
+                 tainted & TAINT_HVM_FEP ? 'H' : ' ');
     }
     else
     {
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index f486ee9..217112d 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -27,12 +27,8 @@
 #include <public/hvm/save.h>
 #include <xen/mm.h>
 
-#ifndef NDEBUG
 /* Permit use of the Forced Emulation Prefix in HVM guests */
 extern bool_t opt_hvm_fep;
-#else
-#define opt_hvm_fep 0
-#endif
 
 /* Interrupt acknowledgement sources. */
 enum hvm_intsrc {
diff --git a/xen/include/xen/lib.h b/xen/include/xen/lib.h
index 1c652bb..b1b0fb2 100644
--- a/xen/include/xen/lib.h
+++ b/xen/include/xen/lib.h
@@ -142,6 +142,7 @@ uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c);
 #define TAINT_BAD_PAGE                  (1<<2)
 #define TAINT_SYNC_CONSOLE              (1<<3)
 #define TAINT_ERROR_INJECT              (1<<4)
+#define TAINT_HVM_FEP                   (1<<5)
 extern int tainted;
 #define TAINT_STRING_MAX_LEN            20
 extern char *print_tainted(char *str);
-- 
2.1.4


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Doug Goldstein]

[-- Attachment #1.1.1: Type: text/plain, Size: 6329 bytes --]

On 6/15/16 9:31 AM, Wei Liu wrote:
> Originally hvm_fep was guarded by NDEBUG, which means it was only
> available to debug builds.
> 
> However there is value to have it for non-debug builds as well. User can
> use that to run tests in setup that replicates production setup.
> 
> Make it clear with a sync_console style warning that this option can't
> be used in production setup. Update command line documentation
> accordingly. Finally mark Xen as tainted when this option is enabled.
> 
> Signed-off-by: Wei Liu <wei.liu2@citrix.com>
> ---
> Cc: Andrew Cooper <andrew.cooper3@citrix.com>
> Cc: Jan Beulich <jbeulich@suse.com>
> ---
>  docs/misc/xen-command-line.markdown |  8 ++++++--
>  xen/arch/x86/hvm/hvm.c              | 31 ++++++++++++++++++++++++++++---
>  xen/common/kernel.c                 |  6 ++++--
>  xen/include/asm-x86/hvm/hvm.h       |  4 ----
>  xen/include/xen/lib.h               |  1 +
>  5 files changed, 39 insertions(+), 11 deletions(-)
> 
> diff --git a/docs/misc/xen-command-line.markdown b/docs/misc/xen-command-line.markdown
> index fed732c..dc53e24 100644
> --- a/docs/misc/xen-command-line.markdown
> +++ b/docs/misc/xen-command-line.markdown
> @@ -878,8 +878,12 @@ Recognized in debug builds of the hypervisor only.
>  Allow use of the Forced Emulation Prefix in HVM guests, to allow emulation of
>  arbitrary instructions.
>  
> -This option is intended for development purposes, and is only available in
> -debug builds of the hypervisor.
> +This option is intended for development and testing purposes.
> +
> +*Warning*
> +As this feature opens up the instruction emulator to HVM guest, don't
> +use this in production system. No security support is provided when
> +this flag is set.
>  
>  ### hvm\_port80
>  > `= <boolean>`
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index 78db903..5bafaef 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -37,6 +37,7 @@
>  #include <xen/mem_access.h>
>  #include <xen/rangeset.h>
>  #include <xen/vm_event.h>
> +#include <xen/delay.h>
>  #include <asm/shadow.h>
>  #include <asm/hap.h>
>  #include <asm/current.h>
> @@ -95,11 +96,9 @@ unsigned long __section(".bss.page_aligned")
>  static bool_t __initdata opt_hap_enabled = 1;
>  boolean_param("hap", opt_hap_enabled);
>  
> -#ifndef opt_hvm_fep
>  /* Permit use of the Forced Emulation Prefix in HVM guests */
> -bool_t opt_hvm_fep;
> +bool_t __read_mostly opt_hvm_fep;
>  boolean_param("hvm_fep", opt_hvm_fep);
> -#endif
>  
>  /* Xen command-line option to enable altp2m */
>  static bool_t __initdata opt_altp2m_enabled = 0;
> @@ -182,6 +181,32 @@ static int __init hvm_enable(void)
>      if ( !opt_altp2m_enabled )
>          hvm_funcs.altp2m_supported = 0;
>  
> +    if ( opt_hvm_fep )
> +    {
> +        unsigned i, j;
> +
> +        printk("**********************************************\n");
> +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
> +        printk("******* This option is *ONLY* intended to aid debugging "
> +               "and testing of Xen\n");
> +        printk("******* that HVM guest can enter instruction emulator "
> +               "with UD instruction.\n");
> +        printk("******* It has implication on the security of the system.\n");
> +        printk("******* Please *DO NOT* use this in production.\n");
> +        printk("**********************************************\n");
> +        add_taint(TAINT_HVM_FEP);
> +        for ( i = 0; i < 3; i++ )
> +        {
> +            printk("%d... ", 3-i);
> +            for ( j = 0; j < 100; j++ )
> +            {
> +                process_pending_softirqs();
> +                mdelay(10);
> +            }
> +        }
> +        printk("\n");
> +    }
> +
>      /*
>       * Allow direct access to the PC debug ports 0x80 and 0xed (they are
>       * often used for I/O delays, but the vmexits simply slow things down).
> diff --git a/xen/common/kernel.c b/xen/common/kernel.c
> index dae7e35..5bf77aa 100644
> --- a/xen/common/kernel.c
> +++ b/xen/common/kernel.c
> @@ -175,6 +175,7 @@ int __init parse_bool(const char *s)
>   *  'M' - Machine had a machine check experience.
>   *  'B' - System has hit bad_page.
>   *  'C' - Console output is synchronous.
> + *  'H' - HVM forced emulation prefix is permitted.
>   *
>   *      The string is overwritten by the next call to print_taint().
>   */
> @@ -182,11 +183,12 @@ char *print_tainted(char *str)
>  {
>      if ( tainted )
>      {
> -        snprintf(str, TAINT_STRING_MAX_LEN, "Tainted: %c%c%c%c",
> +        snprintf(str, TAINT_STRING_MAX_LEN, "Tainted: %c%c%c%c%c",
>                   tainted & TAINT_UNSAFE_SMP ? 'S' : ' ',
>                   tainted & TAINT_MACHINE_CHECK ? 'M' : ' ',
>                   tainted & TAINT_BAD_PAGE ? 'B' : ' ',
> -                 tainted & TAINT_SYNC_CONSOLE ? 'C' : ' ');
> +                 tainted & TAINT_SYNC_CONSOLE ? 'C' : ' ',
> +                 tainted & TAINT_HVM_FEP ? 'H' : ' ');
>      }
>      else
>      {
> diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
> index f486ee9..217112d 100644
> --- a/xen/include/asm-x86/hvm/hvm.h
> +++ b/xen/include/asm-x86/hvm/hvm.h
> @@ -27,12 +27,8 @@
>  #include <public/hvm/save.h>
>  #include <xen/mm.h>
>  
> -#ifndef NDEBUG
>  /* Permit use of the Forced Emulation Prefix in HVM guests */
>  extern bool_t opt_hvm_fep;
> -#else
> -#define opt_hvm_fep 0
> -#endif

Please instead add this as a Kconfig option and you can default it to
enabled.

>  
>  /* Interrupt acknowledgement sources. */
>  enum hvm_intsrc {
> diff --git a/xen/include/xen/lib.h b/xen/include/xen/lib.h
> index 1c652bb..b1b0fb2 100644
> --- a/xen/include/xen/lib.h
> +++ b/xen/include/xen/lib.h
> @@ -142,6 +142,7 @@ uint64_t muldiv64(uint64_t a, uint32_t b, uint32_t c);
>  #define TAINT_BAD_PAGE                  (1<<2)
>  #define TAINT_SYNC_CONSOLE              (1<<3)
>  #define TAINT_ERROR_INJECT              (1<<4)
> +#define TAINT_HVM_FEP                   (1<<5)
>  extern int tainted;
>  #define TAINT_STRING_MAX_LEN            20
>  extern char *print_tainted(char *str);
> 


-- 
Doug Goldstein


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 959 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Wei Liu]

On Wed, Jun 15, 2016 at 09:39:24AM -0500, Doug Goldstein wrote:
> On 6/15/16 9:31 AM, Wei Liu wrote:
[...]
> > -#ifndef NDEBUG
> >  /* Permit use of the Forced Emulation Prefix in HVM guests */
> >  extern bool_t opt_hvm_fep;
> > -#else
> > -#define opt_hvm_fep 0
> > -#endif
> 
> Please instead add this as a Kconfig option and you can default it to
> enabled.
> 

Sure, it is reasonable that you want to compile this out.

But which section does it belong to? Architecture Features I guess?

Wei.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Doug Goldstein]

[-- Attachment #1.1.1: Type: text/plain, Size: 707 bytes --]

On 6/15/16 9:47 AM, Wei Liu wrote:
> On Wed, Jun 15, 2016 at 09:39:24AM -0500, Doug Goldstein wrote:
>> On 6/15/16 9:31 AM, Wei Liu wrote:
> [...]
>>> -#ifndef NDEBUG
>>>  /* Permit use of the Forced Emulation Prefix in HVM guests */
>>>  extern bool_t opt_hvm_fep;
>>> -#else
>>> -#define opt_hvm_fep 0
>>> -#endif
>>
>> Please instead add this as a Kconfig option and you can default it to
>> enabled.
>>
> 
> Sure, it is reasonable that you want to compile this out.
> 
> But which section does it belong to? Architecture Features I guess?
> 
> Wei.
> 

That sounds reasonable to me. You can add it to arch/Kconfig if it makes
sense for both ARM and x86.

-- 
Doug Goldstein


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 959 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Wei Liu]

On Wed, Jun 15, 2016 at 11:12:08AM -0500, Doug Goldstein wrote:
> On 6/15/16 9:47 AM, Wei Liu wrote:
> > On Wed, Jun 15, 2016 at 09:39:24AM -0500, Doug Goldstein wrote:
> >> On 6/15/16 9:31 AM, Wei Liu wrote:
> > [...]
> >>> -#ifndef NDEBUG
> >>>  /* Permit use of the Forced Emulation Prefix in HVM guests */
> >>>  extern bool_t opt_hvm_fep;
> >>> -#else
> >>> -#define opt_hvm_fep 0
> >>> -#endif
> >>
> >> Please instead add this as a Kconfig option and you can default it to
> >> enabled.
> >>
> > 
> > Sure, it is reasonable that you want to compile this out.
> > 
> > But which section does it belong to? Architecture Features I guess?
> > 
> > Wei.
> > 
> 
> That sounds reasonable to me. You can add it to arch/Kconfig if it makes
> sense for both ARM and x86.
> 

I think it should be x86 only for now. ARM doesn't have instruction
emulator.

Wei.

> -- 
> Doug Goldstein
> 




_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Jan Beulich]

>>> On 15.06.16 at 16:31, <wei.liu2@citrix.com> wrote:
> @@ -182,6 +181,32 @@ static int __init hvm_enable(void)
>      if ( !opt_altp2m_enabled )
>          hvm_funcs.altp2m_supported = 0;
>  
> +    if ( opt_hvm_fep )
> +    {
> +        unsigned i, j;

unsigned int

> +        printk("**********************************************\n");
> +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
> +        printk("******* This option is *ONLY* intended to aid debugging "
> +               "and testing of Xen\n");
> +        printk("******* that HVM guest can enter instruction emulator "
> +               "with UD instruction.\n");
> +        printk("******* It has implication on the security of the system.\n");
> +        printk("******* Please *DO NOT* use this in production.\n");
> +        printk("**********************************************\n");
> +        add_taint(TAINT_HVM_FEP);

Should we perhaps taint the system only the first time a guest
makes use of this?

> +        for ( i = 0; i < 3; i++ )
> +        {
> +            printk("%d... ", 3-i);

%u and spaces around - please.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Wei Liu]

On Thu, Jun 16, 2016 at 05:37:03AM -0600, Jan Beulich wrote:
> >>> On 15.06.16 at 16:31, <wei.liu2@citrix.com> wrote:
> > @@ -182,6 +181,32 @@ static int __init hvm_enable(void)
> >      if ( !opt_altp2m_enabled )
> >          hvm_funcs.altp2m_supported = 0;
> >  
> > +    if ( opt_hvm_fep )
> > +    {
> > +        unsigned i, j;
> 
> unsigned int
> 

Ack.

> > +        printk("**********************************************\n");
> > +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
> > +        printk("******* This option is *ONLY* intended to aid debugging "
> > +               "and testing of Xen\n");
> > +        printk("******* that HVM guest can enter instruction emulator "
> > +               "with UD instruction.\n");
> > +        printk("******* It has implication on the security of the system.\n");
> > +        printk("******* Please *DO NOT* use this in production.\n");
> > +        printk("**********************************************\n");
> > +        add_taint(TAINT_HVM_FEP);
> 
> Should we perhaps taint the system only the first time a guest
> makes use of this?
> 

Doesn't that add overhead to a potential hot path? Arguably it is only
setting a bit in a flag, but still...

> > +        for ( i = 0; i < 3; i++ )
> > +        {
> > +            printk("%d... ", 3-i);
> 
> %u and spaces around - please.
> 

Ack.

Wei.

> Jan
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Andrew Cooper]

On 16/06/16 12:52, Wei Liu wrote:
>
>>> +        printk("**********************************************\n");
>>> +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");

I would say "available" rather than permitted in this case.

>>> +        printk("******* This option is *ONLY* intended to aid debugging "
>>> +               "and testing of Xen\n");

Despite the line length, I would keep this string on a single line.  If
you want it a little shorter, you can drop "debugging and", leaving just
testing.

>>> +        printk("******* that HVM guest can enter instruction emulator "
>>> +               "with UD instruction.\n");

I think this like isn't necessary.  Anyone who is unclear what FEP is
can look it up.

>>> +        printk("******* It has implication on the security of the system.\n");

implications.

>>> +        printk("******* Please *DO NOT* use this in production.\n");
>>> +        printk("**********************************************\n");
>>> +        add_taint(TAINT_HVM_FEP);
>> Should we perhaps taint the system only the first time a guest
>> makes use of this?
>>
> Doesn't that add overhead to a potential hot path? Arguably it is only
> setting a bit in a flag, but still...

FEP is not a fastpath at all.  It would be fine to defer to
hvm_ud_intercept().

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Wei Liu]

On Thu, Jun 16, 2016 at 01:12:34PM +0100, Andrew Cooper wrote:
> On 16/06/16 12:52, Wei Liu wrote:
> >
> >>> +        printk("**********************************************\n");
> >>> +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
> 
> I would say "available" rather than permitted in this case.
> 
> >>> +        printk("******* This option is *ONLY* intended to aid debugging "
> >>> +               "and testing of Xen\n");
> 
> Despite the line length, I would keep this string on a single line.  If
> you want it a little shorter, you can drop "debugging and", leaving just
> testing.
> 
> >>> +        printk("******* that HVM guest can enter instruction emulator "
> >>> +               "with UD instruction.\n");
> 
> I think this like isn't necessary.  Anyone who is unclear what FEP is
> can look it up.
> 
> >>> +        printk("******* It has implication on the security of the system.\n");
> 
> implications.
> 

All fixed.

> >>> +        printk("******* Please *DO NOT* use this in production.\n");
> >>> +        printk("**********************************************\n");
> >>> +        add_taint(TAINT_HVM_FEP);
> >> Should we perhaps taint the system only the first time a guest
> >> makes use of this?
> >>
> > Doesn't that add overhead to a potential hot path? Arguably it is only
> > setting a bit in a flag, but still...
> 
> FEP is not a fastpath at all.  It would be fine to defer to
> hvm_ud_intercept().
> 

NP.

Wei.

> ~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [PATCH RFC 2/2] xen: make available hvm_fep to non-debug build as well

[Jan Beulich]

>>> On 16.06.16 at 13:52, <wei.liu2@citrix.com> wrote:
> On Thu, Jun 16, 2016 at 05:37:03AM -0600, Jan Beulich wrote:
>> >>> On 15.06.16 at 16:31, <wei.liu2@citrix.com> wrote:
>> > +        printk("**********************************************\n");
>> > +        printk("******* WARNING: HVM FORCED EMULATION PREFIX IS PERMITTED\n");
>> > +        printk("******* This option is *ONLY* intended to aid debugging "
>> > +               "and testing of Xen\n");
>> > +        printk("******* that HVM guest can enter instruction emulator "
>> > +               "with UD instruction.\n");
>> > +        printk("******* It has implication on the security of the system.\n");
>> > +        printk("******* Please *DO NOT* use this in production.\n");
>> > +        printk("**********************************************\n");
>> > +        add_taint(TAINT_HVM_FEP);
>> 
>> Should we perhaps taint the system only the first time a guest
>> makes use of this?
>> 
> 
> Doesn't that add overhead to a potential hot path? Arguably it is only
> setting a bit in a flag, but still...

How can that be a hot path, if it's not even usable without the
option set?

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel