[PATCH v1 13/13] xen: make static shared memory supported in SUPPORT.md

Penny Zheng posted 13 patches 2 months, 2 weeks ago
[PATCH v1 13/13] xen: make static shared memory supported in SUPPORT.md
Posted by Penny Zheng 2 months, 2 weeks ago
After patching previous commits, we could make feature of "static shared memory"
supported in SUPPORT.md.

Signed-off-by: Penny Zheng <penny.zheng@arm.com>
---
 SUPPORT.md           | 2 +-
 xen/arch/arm/Kconfig | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/SUPPORT.md b/SUPPORT.md
index ab71464cf6..c9fe7daf56 100644
--- a/SUPPORT.md
+++ b/SUPPORT.md
@@ -316,7 +316,7 @@ Allow sharing of identical pages between guests
 Allow to statically set up shared memory on dom0less system,
 enabling domains to do shm-based communication
 
-    Status, ARM: Tech Preview
+    Status, ARM: Supported
 
 ### Memory Paging
 
diff --git a/xen/arch/arm/Kconfig b/xen/arch/arm/Kconfig
index 52a05f704d..e0ea6e1762 100644
--- a/xen/arch/arm/Kconfig
+++ b/xen/arch/arm/Kconfig
@@ -141,7 +141,7 @@ config TEE
 source "arch/arm/tee/Kconfig"
 
 config STATIC_SHM
-	bool "Statically shared memory on a dom0less system" if UNSUPPORTED
+	bool "Statically shared memory on a dom0less system"
 	depends on STATIC_MEMORY
 	help
 	  This option enables statically shared memory on a dom0less system.
-- 
2.25.1
Re: [PATCH v1 13/13] xen: make static shared memory supported in SUPPORT.md
Posted by Julien Grall 3 weeks, 2 days ago
Hi Penny,

On 15/11/2022 02:52, Penny Zheng wrote:
> After patching previous commits, we could make feature of "static shared memory"

Are you referring to the patch in this series? If so, they seem to had 
new features which I don't think are necessary to mark the "static 
shared memory".

Instead, "static shared memory" could be marked as supported if we 
believe that the new code has no security hole.

Looking below, the STATIC_SHM depends on STATIC_MEMORY which is 
currently unsupported. So it seems a bit strange to mark one supported 
but not the other.

Now, in order to support them, we need to make sure that the XENMEM_* 
operations are working as intended. I know there are some works that 
were done in the past, but I can't exactly remember if we fixed 
everything. So what happen if the domain (consider the case where the 
domain is directmapped or not):
   1) Remove the page?
   2) Remove the page twice? (Only in the directmap case)
   3) Request to map the page?

Cheers,

-- 
Julien Grall