[PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Andrew Cooper]

When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:

  common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-Werror=unused-function]
   static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

due to an unguarded forward declaration.

It turns out there is no need to forward declare
sh_remove_write_access_from_sl1p() to begin with, so move it to just ahead of
its first users, which is within a larger #ifdef'd SHOPT_OUT_OF_SYNC block.

Fix up for style while moving it.  No functional change.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
---
CC: Jan Beulich <JBeulich@suse.com>
CC: Wei Liu <wl@xen.org>
CC: Roger Pau Monné <roger.pau@citrix.com>
CC: Tim Deegan <tim@xen.org>
---
 xen/arch/x86/mm/shadow/common.c | 56 ++++++++++++++++++-----------------------
 1 file changed, 25 insertions(+), 31 deletions(-)

diff --git a/xen/arch/x86/mm/shadow/common.c b/xen/arch/x86/mm/shadow/common.c
index 0ac3f880e1..6dff240e97 100644
--- a/xen/arch/x86/mm/shadow/common.c
+++ b/xen/arch/x86/mm/shadow/common.c
@@ -38,9 +38,6 @@
 #include <xen/numa.h>
 #include "private.h"
 
-static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
-                                            mfn_t smfn, unsigned long offset);
-
 DEFINE_PER_CPU(uint32_t,trace_shadow_path_flags);
 
 static int sh_enable_log_dirty(struct domain *, bool log_global);
@@ -252,6 +249,31 @@ static inline void _sh_resync_l1(struct vcpu *v, mfn_t gmfn, mfn_t snpmfn)
         SHADOW_INTERNAL_NAME(sh_resync_l1, 4)(v, gmfn, snpmfn);
 }
 
+static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
+                                            mfn_t smfn, unsigned long off)
+{
+    struct page_info *sp = mfn_to_page(smfn);
+
+    ASSERT(mfn_valid(smfn));
+    ASSERT(mfn_valid(gmfn));
+
+    if ( sp->u.sh.type == SH_type_l1_32_shadow ||
+         sp->u.sh.type == SH_type_fl1_32_shadow )
+        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 2)
+            (d, gmfn, smfn, off);
+
+    if ( sp->u.sh.type == SH_type_l1_pae_shadow ||
+         sp->u.sh.type == SH_type_fl1_pae_shadow )
+        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 3)
+            (d, gmfn, smfn, off);
+
+    if ( sp->u.sh.type == SH_type_l1_64_shadow ||
+         sp->u.sh.type == SH_type_fl1_64_shadow )
+        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 4)
+            (d, gmfn, smfn, off);
+
+    return 0;
+}
 
 /*
  * Fixup arrays: We limit the maximum number of writable mappings to
@@ -2001,34 +2023,6 @@ int sh_remove_write_access(struct domain *d, mfn_t gmfn,
 }
 #endif /* CONFIG_HVM */
 
-#if (SHADOW_OPTIMIZATIONS & SHOPT_OUT_OF_SYNC)
-static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
-                                            mfn_t smfn, unsigned long off)
-{
-    struct page_info *sp = mfn_to_page(smfn);
-
-    ASSERT(mfn_valid(smfn));
-    ASSERT(mfn_valid(gmfn));
-
-    if ( sp->u.sh.type == SH_type_l1_32_shadow
-         || sp->u.sh.type == SH_type_fl1_32_shadow )
-    {
-        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,2)
-            (d, gmfn, smfn, off);
-    }
-    else if ( sp->u.sh.type == SH_type_l1_pae_shadow
-              || sp->u.sh.type == SH_type_fl1_pae_shadow )
-        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,3)
-            (d, gmfn, smfn, off);
-    else if ( sp->u.sh.type == SH_type_l1_64_shadow
-              || sp->u.sh.type == SH_type_fl1_64_shadow )
-        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,4)
-            (d, gmfn, smfn, off);
-
-    return 0;
-}
-#endif
-
 /**************************************************************************/
 /* Remove all mappings of a guest frame from the shadow tables.
  * Returns non-zero if we need to flush TLBs. */
-- 
2.11.0

RE: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Paul Durrant]

> -----Original Message-----
> From: Xen-devel <xen-devel-bounces@lists.xenproject.org> On Behalf Of Andrew Cooper
> Sent: 21 May 2020 10:04
> To: Xen-devel <xen-devel@lists.xenproject.org>
> Cc: Andrew Cooper <andrew.cooper3@citrix.com>; Tim Deegan <tim@xen.org>; Wei Liu <wl@xen.org>; Jan
> Beulich <JBeulich@suse.com>; Roger Pau Monné <roger.pau@citrix.com>
> Subject: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()
> 
> When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:
> 
>   common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-
> Werror=unused-function]
>    static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> due to an unguarded forward declaration.

Is this, perhaps, an argument for making SHADOW_OPTIMIZATIONS tunable via Kconfig so that randconfig could catch things like this?

  Paul

> 
> It turns out there is no need to forward declare
> sh_remove_write_access_from_sl1p() to begin with, so move it to just ahead of
> its first users, which is within a larger #ifdef'd SHOPT_OUT_OF_SYNC block.
> 
> Fix up for style while moving it.  No functional change.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> ---
> CC: Jan Beulich <JBeulich@suse.com>
> CC: Wei Liu <wl@xen.org>
> CC: Roger Pau Monné <roger.pau@citrix.com>
> CC: Tim Deegan <tim@xen.org>
> ---
>  xen/arch/x86/mm/shadow/common.c | 56 ++++++++++++++++++-----------------------
>  1 file changed, 25 insertions(+), 31 deletions(-)
> 
> diff --git a/xen/arch/x86/mm/shadow/common.c b/xen/arch/x86/mm/shadow/common.c
> index 0ac3f880e1..6dff240e97 100644
> --- a/xen/arch/x86/mm/shadow/common.c
> +++ b/xen/arch/x86/mm/shadow/common.c
> @@ -38,9 +38,6 @@
>  #include <xen/numa.h>
>  #include "private.h"
> 
> -static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> -                                            mfn_t smfn, unsigned long offset);
> -
>  DEFINE_PER_CPU(uint32_t,trace_shadow_path_flags);
> 
>  static int sh_enable_log_dirty(struct domain *, bool log_global);
> @@ -252,6 +249,31 @@ static inline void _sh_resync_l1(struct vcpu *v, mfn_t gmfn, mfn_t snpmfn)
>          SHADOW_INTERNAL_NAME(sh_resync_l1, 4)(v, gmfn, snpmfn);
>  }
> 
> +static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> +                                            mfn_t smfn, unsigned long off)
> +{
> +    struct page_info *sp = mfn_to_page(smfn);
> +
> +    ASSERT(mfn_valid(smfn));
> +    ASSERT(mfn_valid(gmfn));
> +
> +    if ( sp->u.sh.type == SH_type_l1_32_shadow ||
> +         sp->u.sh.type == SH_type_fl1_32_shadow )
> +        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 2)
> +            (d, gmfn, smfn, off);
> +
> +    if ( sp->u.sh.type == SH_type_l1_pae_shadow ||
> +         sp->u.sh.type == SH_type_fl1_pae_shadow )
> +        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 3)
> +            (d, gmfn, smfn, off);
> +
> +    if ( sp->u.sh.type == SH_type_l1_64_shadow ||
> +         sp->u.sh.type == SH_type_fl1_64_shadow )
> +        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p, 4)
> +            (d, gmfn, smfn, off);
> +
> +    return 0;
> +}
> 
>  /*
>   * Fixup arrays: We limit the maximum number of writable mappings to
> @@ -2001,34 +2023,6 @@ int sh_remove_write_access(struct domain *d, mfn_t gmfn,
>  }
>  #endif /* CONFIG_HVM */
> 
> -#if (SHADOW_OPTIMIZATIONS & SHOPT_OUT_OF_SYNC)
> -static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> -                                            mfn_t smfn, unsigned long off)
> -{
> -    struct page_info *sp = mfn_to_page(smfn);
> -
> -    ASSERT(mfn_valid(smfn));
> -    ASSERT(mfn_valid(gmfn));
> -
> -    if ( sp->u.sh.type == SH_type_l1_32_shadow
> -         || sp->u.sh.type == SH_type_fl1_32_shadow )
> -    {
> -        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,2)
> -            (d, gmfn, smfn, off);
> -    }
> -    else if ( sp->u.sh.type == SH_type_l1_pae_shadow
> -              || sp->u.sh.type == SH_type_fl1_pae_shadow )
> -        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,3)
> -            (d, gmfn, smfn, off);
> -    else if ( sp->u.sh.type == SH_type_l1_64_shadow
> -              || sp->u.sh.type == SH_type_fl1_64_shadow )
> -        return SHADOW_INTERNAL_NAME(sh_rm_write_access_from_sl1p,4)
> -            (d, gmfn, smfn, off);
> -
> -    return 0;
> -}
> -#endif
> -
>  /**************************************************************************/
>  /* Remove all mappings of a guest frame from the shadow tables.
>   * Returns non-zero if we need to flush TLBs. */
> --
> 2.11.0
> 

Re: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Andrew Cooper]

On 21/05/2020 11:12, Paul Durrant wrote:
>> -----Original Message-----
>> From: Xen-devel <xen-devel-bounces@lists.xenproject.org> On Behalf Of Andrew Cooper
>> Sent: 21 May 2020 10:04
>> To: Xen-devel <xen-devel@lists.xenproject.org>
>> Cc: Andrew Cooper <andrew.cooper3@citrix.com>; Tim Deegan <tim@xen.org>; Wei Liu <wl@xen.org>; Jan
>> Beulich <JBeulich@suse.com>; Roger Pau Monné <roger.pau@citrix.com>
>> Subject: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()
>>
>> When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:
>>
>>   common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-
>> Werror=unused-function]
>>    static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>>               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>
>> due to an unguarded forward declaration.
> Is this, perhaps, an argument for making SHADOW_OPTIMIZATIONS tunable via Kconfig so that randconfig could catch things like this?

Given enough TUITS, yes.

~Andrew

Re: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Roger Pau Monné]

On Thu, May 21, 2020 at 10:04:28AM +0100, Andrew Cooper wrote:
> When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:
> 
>   common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-Werror=unused-function]
>    static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> due to an unguarded forward declaration.
> 
> It turns out there is no need to forward declare
> sh_remove_write_access_from_sl1p() to begin with, so move it to just ahead of
> its first users, which is within a larger #ifdef'd SHOPT_OUT_OF_SYNC block.
> 
> Fix up for style while moving it.  No functional change.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>

Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>

> ---
> CC: Jan Beulich <JBeulich@suse.com>
> CC: Wei Liu <wl@xen.org>
> CC: Roger Pau Monné <roger.pau@citrix.com>
> CC: Tim Deegan <tim@xen.org>
> ---
>  xen/arch/x86/mm/shadow/common.c | 56 ++++++++++++++++++-----------------------
>  1 file changed, 25 insertions(+), 31 deletions(-)
> 
> diff --git a/xen/arch/x86/mm/shadow/common.c b/xen/arch/x86/mm/shadow/common.c
> index 0ac3f880e1..6dff240e97 100644
> --- a/xen/arch/x86/mm/shadow/common.c
> +++ b/xen/arch/x86/mm/shadow/common.c
> @@ -38,9 +38,6 @@
>  #include <xen/numa.h>
>  #include "private.h"
>  
> -static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> -                                            mfn_t smfn, unsigned long offset);
> -
>  DEFINE_PER_CPU(uint32_t,trace_shadow_path_flags);
>  
>  static int sh_enable_log_dirty(struct domain *, bool log_global);
> @@ -252,6 +249,31 @@ static inline void _sh_resync_l1(struct vcpu *v, mfn_t gmfn, mfn_t snpmfn)
>          SHADOW_INTERNAL_NAME(sh_resync_l1, 4)(v, gmfn, snpmfn);
>  }
>  
> +static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> +                                            mfn_t smfn, unsigned long off)
> +{
> +    struct page_info *sp = mfn_to_page(smfn);
> +
> +    ASSERT(mfn_valid(smfn));
> +    ASSERT(mfn_valid(gmfn));
> +
> +    if ( sp->u.sh.type == SH_type_l1_32_shadow ||
> +         sp->u.sh.type == SH_type_fl1_32_shadow )

Using a switch would also be nice IMO and would avoid some of the code
churn.

Thanks, Roger.

Re: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Andrew Cooper]

On 21/05/2020 11:26, Roger Pau Monné wrote:
> On Thu, May 21, 2020 at 10:04:28AM +0100, Andrew Cooper wrote:
>> When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:
>>
>>   common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-Werror=unused-function]
>>    static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>>               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>
>> due to an unguarded forward declaration.
>>
>> It turns out there is no need to forward declare
>> sh_remove_write_access_from_sl1p() to begin with, so move it to just ahead of
>> its first users, which is within a larger #ifdef'd SHOPT_OUT_OF_SYNC block.
>>
>> Fix up for style while moving it.  No functional change.
>>
>> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
>
>> ---
>> CC: Jan Beulich <JBeulich@suse.com>
>> CC: Wei Liu <wl@xen.org>
>> CC: Roger Pau Monné <roger.pau@citrix.com>
>> CC: Tim Deegan <tim@xen.org>
>> ---
>>  xen/arch/x86/mm/shadow/common.c | 56 ++++++++++++++++++-----------------------
>>  1 file changed, 25 insertions(+), 31 deletions(-)
>>
>> diff --git a/xen/arch/x86/mm/shadow/common.c b/xen/arch/x86/mm/shadow/common.c
>> index 0ac3f880e1..6dff240e97 100644
>> --- a/xen/arch/x86/mm/shadow/common.c
>> +++ b/xen/arch/x86/mm/shadow/common.c
>> @@ -38,9 +38,6 @@
>>  #include <xen/numa.h>
>>  #include "private.h"
>>  
>> -static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>> -                                            mfn_t smfn, unsigned long offset);
>> -
>>  DEFINE_PER_CPU(uint32_t,trace_shadow_path_flags);
>>  
>>  static int sh_enable_log_dirty(struct domain *, bool log_global);
>> @@ -252,6 +249,31 @@ static inline void _sh_resync_l1(struct vcpu *v, mfn_t gmfn, mfn_t snpmfn)
>>          SHADOW_INTERNAL_NAME(sh_resync_l1, 4)(v, gmfn, snpmfn);
>>  }
>>  
>> +static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>> +                                            mfn_t smfn, unsigned long off)
>> +{
>> +    struct page_info *sp = mfn_to_page(smfn);
>> +
>> +    ASSERT(mfn_valid(smfn));
>> +    ASSERT(mfn_valid(gmfn));
>> +
>> +    if ( sp->u.sh.type == SH_type_l1_32_shadow ||
>> +         sp->u.sh.type == SH_type_fl1_32_shadow )
> Using a switch would also be nice IMO and would avoid some of the code
> churn.

Good point.  Happy to do that if Tim agrees (but I won't bother sending
a v2 just now).

~Andrew

Re: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Roger Pau Monné]

On Thu, May 21, 2020 at 11:32:04AM +0100, Andrew Cooper wrote:
> On 21/05/2020 11:26, Roger Pau Monné wrote:
> > On Thu, May 21, 2020 at 10:04:28AM +0100, Andrew Cooper wrote:
> >> +static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
> >> +                                            mfn_t smfn, unsigned long off)
> >> +{
> >> +    struct page_info *sp = mfn_to_page(smfn);
> >> +
> >> +    ASSERT(mfn_valid(smfn));
> >> +    ASSERT(mfn_valid(gmfn));
> >> +
> >> +    if ( sp->u.sh.type == SH_type_l1_32_shadow ||
> >> +         sp->u.sh.type == SH_type_fl1_32_shadow )
> > Using a switch would also be nice IMO and would avoid some of the code
> > churn.
> 
> Good point.  Happy to do that if Tim agrees (but I won't bother sending
> a v2 just now).

Sure, feel free to keep my RB after that.

Roger.

Re: [PATCH] x86/shadow: Reposition sh_remove_write_access_from_sl1p()

[Tim Deegan]

At 10:04 +0100 on 21 May (1590055468), Andrew Cooper wrote:
> When compiling with SHOPT_OUT_OF_SYNC disabled, the build fails with:
> 
>   common.c:41:12: error: ‘sh_remove_write_access_from_sl1p’ declared ‘static’ but never defined [-Werror=unused-function]
>    static int sh_remove_write_access_from_sl1p(struct domain *d, mfn_t gmfn,
>               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> due to an unguarded forward declaration.
> 
> It turns out there is no need to forward declare
> sh_remove_write_access_from_sl1p() to begin with, so move it to just ahead of
> its first users, which is within a larger #ifdef'd SHOPT_OUT_OF_SYNC block.
> 
> Fix up for style while moving it.  No functional change.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>

Thank you!  This is fine, either as-is or with the suggested change to
a switch.

Reviewed-by: Tim Deegan <tim@xen.org>