Skip to contentSkip to conversation

Even more XSA-409 fixes

PATCH In review Andrew Cooper 5 participants 10 messages Nov 21, 2022

lore
Share & tools
lore
#0 lore raw

cover Even more XSA-409 fixes

Once more, with feeling...

Patch 1 has been posted previously, but has accumulated another bugfix.

Patch 2 has been discussed on list, but this is the first posting. I've confirmed that it fixes the issue reported by OSSTest.

Andrew Cooper (2):
tools/libxl: Fixes to libxl__domain_set_paging_mempool_size() xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

 tools/flask/policy/modules/dom0.te  |  3 ++-
 tools/flask/policy/modules/xen.if   |  2 +-
 tools/libs/light/libxl_dom.c        | 10 +++++++++-
 xen/xsm/flask/hooks.c               |  6 ++++++
 xen/xsm/flask/policy/access_vectors |  4 ++++
 5 files changed, 22 insertions(+), 3 deletions(-)
Signature · 1 line
2.11.0
#1 lore raw

PATCH 1/2 tools/libxl: Fixes to libxl__domain_set_paging_mempool_size()

The error message accidentally printed the bytes value as if it were kB.

Furthermore, both b_info.shadow_memkb and shadow_mem are uint64_t, meaning there is a risk of overflow if the user specified a stupidly large value in the vm.cfg file. Check and reject such a condition.

Fixes 7c3bbd940dd8 ("xen/arm, libxl: Revert XEN_DOMCTL_shadow_op; use p2m mempool hypercalls")

---
CC: Anthony PERARD <[email protected]>
CC: Henry Wang <[email protected]>

v2:
* Retain PRIu64
* Check for overflow

For 4.17. This is a low risk change, removes one overflow case, and makes an error message accurate.

---
 tools/libs/light/libxl_dom.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
tools/libs/light/libxl_dom.c +9 -1
index fa5c79e4f650..b454f988fbc5 100644
@@ -1458,10 +1458,18 @@ int libxl__domain_set_paging_mempool_size(
     shadow_mem = d_config->b_info.shadow_memkb;
     shadow_mem <<= 10;
 
+    if ((shadow_mem >> 10) != d_config->b_info.shadow_memkb) {
+        LOGED(ERROR, domid,
+              "shadow_memkb value %"PRIu64"kB too large",
+              d_config->b_info.shadow_memkb);
+        return ERROR_FAIL;
+    }
+
     int r = xc_set_paging_mempool_size(CTX->xch, domid, shadow_mem);
     if (r) {
         LOGED(ERROR, domid,
-              "Failed to set paging mempool size to %"PRIu64"kB", shadow_mem);
+              "Failed to set paging mempool size to %"PRIu64"kB",
+              d_config->b_info.shadow_memkb);
         return ERROR_FAIL;
     }
 
Signature · 1 line
2.11.0
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] tools/libxl: Fixes to libxl__domain_set_paging_mempool_size()

Quoted 8 lines · Andrew Cooper

The error message accidentally printed the bytes value as if it were kB.

Furthermore, both b_info.shadow_memkb and shadow_mem are uint64_t, meaning there is a risk of overflow if the user specified a stupidly large value in the vm.cfg file. Check and reject such a condition.

Fixes 7c3bbd940dd8 ("xen/arm, libxl: Revert XEN_DOMCTL_shadow_op; use p2m mempool hypercalls")
Quoted 8 lines

---

v2:
* Retain PRIu64
* Check for overflow

For 4.17. This is a low risk change, removes one overflow case, and makes an error message accurate.

Thanks,

Signature · 1 line
Anthony PERARD
#3 lore raw

RE: [PATCH 1/2] tools/libxl: Fixes to libxl__domain_set_paging_mempool_size()

Hi Andrew,

Quoted 21 lines

-----Original Message-----
Subject: Re: [PATCH 1/2] tools/libxl: Fixes to
libxl__domain_set_paging_mempool_size()

Quoted 3 lines · Andrew Cooper

The error message accidentally printed the bytes value as if it were kB.

Furthermore, both b_info.shadow_memkb and shadow_mem are uint64_t,

meaning

Quoted 4 lines

there is a risk of overflow if the user specified a stupidly large value in the vm.cfg file. Check and reject such a condition.

Fixes 7c3bbd940dd8 ("xen/arm, libxl: Revert XEN_DOMCTL_shadow_op;

use p2m mempool hypercalls")

Quoted 7 lines

---

v2:
* Retain PRIu64
* Check for overflow

Quoted 5 lines
Quoted 1 line

For 4.17. This is a low risk change, removes one overflow case, and makes

an

Quoted 1 line

error message accurate.

Kind regards,
Henry

#4 lore raw

PATCH 2/2 xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

In reply to Andrew Cooper

These were overlooked in the original patch, and noticed by OSSTest which does run some Flask tests.

Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging mempool size")

---
CC: Daniel De Graaf <[email protected]>
CC: Daniel Smith <[email protected]>
CC: Jason Andryuk <[email protected]>
CC: Henry Wang <[email protected]>

It should be noted that the original XSA-409 fix broke Flask on ARM but no testing noticed. c/s 7c3bbd940dd8 ("xen/arm, libxl: Revert XEN_DOMCTL_shadow_op; use p2m mempool hypercalls") "fixes" the original breakage and introduced this breakage instead.

For 4.17. It's a fix for an issue that OSSTest is currently blocking as a regression.

---
 tools/flask/policy/modules/dom0.te  | 3 ++-
 tools/flask/policy/modules/xen.if   | 2 +-
 xen/xsm/flask/hooks.c               | 6 ++++++
 xen/xsm/flask/policy/access_vectors | 4 ++++
 4 files changed, 13 insertions(+), 2 deletions(-)
tools/flask/policy/modules/dom0.te +2 -1
index f710ff9941c0..f1dcff48e227 100644
@@ -35,7 +35,8 @@ allow dom0_t dom0_t:domain {
 	setvcpucontext max_vcpus setaffinity getaffinity getscheduler
 	getdomaininfo getvcpuinfo getvcpucontext setdomainmaxmem setdomainhandle
 	setdebugging hypercall settime setaddrsize getaddrsize trigger
-	getpodtarget setpodtarget set_misc_info set_virq_handler
+	getpodtarget setpodtarget getpagingmempool setpagingmempool set_misc_info
+	set_virq_handler
 };
 allow dom0_t dom0_t:domain2 {
 	set_cpu_policy gettsc settsc setscheduler set_vnumainfo
tools/flask/policy/modules/xen.if +1 -1
index 424daab6a022..6b7b7d403ab4 100644
@@ -92,7 +92,7 @@ define(`manage_domain', `
 	allow $1 $2:domain { getdomaininfo getvcpuinfo getaffinity
 			getaddrsize pause unpause trigger shutdown destroy
 			setaffinity setdomainmaxmem getscheduler resume
-			setpodtarget getpodtarget };
+			setpodtarget getpodtarget getpagingmempool setpagingmempool };
     allow $1 $2:domain2 set_vnumainfo;
 ')
 
xen/xsm/flask/hooks.c +6 -0
index 391aec4dc221..78225f68c15c 100644
@@ -822,6 +822,12 @@ static int cf_check flask_domctl(struct domain *d, int cmd)
     case XEN_DOMCTL_get_cpu_policy:
         return current_has_perm(d, SECCLASS_DOMAIN2, DOMAIN2__GET_CPU_POLICY);
 
+    case XEN_DOMCTL_get_paging_mempool_size:
+        return current_has_perm(d, SECCLASS_DOMAIN, DOMAIN__GETPAGINGMEMPOOL);
+
+    case XEN_DOMCTL_set_paging_mempool_size:
+        return current_has_perm(d, SECCLASS_DOMAIN, DOMAIN__SETPAGINGMEMPOOL);
+
     default:
         return avc_unknown_permission("domctl", cmd);
     }
xen/xsm/flask/policy/access_vectors +4 -0
index 6359c7fc8757..4e6710a63e1b 100644
@@ -180,6 +180,10 @@ class domain
     set_misc_info
 # XEN_DOMCTL_set_virq_handler
     set_virq_handler
+# XEN_DOMCTL_get_paging_mempool_size
+    getpagingmempool
+# XEN_DOMCTL_set_paging_mempool_size
+    setpagingmempool
 }
 
 # This is a continuation of class domain, since only 32 permissions can be
Signature · 1 line
2.11.0
#5 lore raw

RE: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

Hi Andrew,

Quoted 11 lines

-----Original Message-----
From: Andrew Cooper <[email protected]>
Sent: Monday, November 21, 2022 10:38 PM
To: Xen-devel <[email protected]>
Cc: Andrew Cooper <[email protected]>; Daniel De Graaf <[email protected]>; Daniel Smith <[email protected]>; Jason Andryuk <[email protected]>; Henry Wang <[email protected]>
Subject: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

Quoted 8 lines

These were overlooked in the original patch, and noticed by OSSTest which does
run some Flask tests.

Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging

mempool size")

Kind regards,
Henry

Reviewed-by on v1 2/2
#6 lore raw

Re: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

In reply to Andrew Cooper
Quoted 4 lines · Andrew Cooper

These were overlooked in the original patch, and noticed by OSSTest which does run some Flask tests.

Quoted 8 lines
Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging mempool size")

---
CC: Daniel De Graaf <[email protected]>
CC: Daniel Smith <[email protected]>
CC: Jason Andryuk <[email protected]>
CC: Henry Wang <[email protected]>

Thanks, Andrew. Though we might want a small tweak - possibly as a follow up?

tools/flask/policy/modules/xen.if define(`manage_domain', ` · line 95

Quoted 10 lines
tools/flask/policy/modules/xen.if +1 -1
index 424daab6a022..6b7b7d403ab4 100644
@@ -92,7 +92,7 @@ define(`manage_domain', `
        allow $1 $2:domain { getdomaininfo getvcpuinfo getaffinity
                        getaddrsize pause unpause trigger shutdown destroy
                        setaffinity setdomainmaxmem getscheduler resume
-                       setpodtarget getpodtarget };
+                       setpodtarget getpodtarget getpagingmempool setpagingmempool };

There is also create_domain_common which is for a dedicated "domain builder" that creates but does not manage domains. I think that should gain setpagingmempool permission?

Regards,
Jason

#7 lore raw

Re: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

Quoted 16 lines · Jason Andryuk
Quoted 11 lines · Andrew Cooper

These were overlooked in the original patch, and noticed by OSSTest which does run some Flask tests.

Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging mempool size")

---
CC: Daniel De Graaf <[email protected]>
CC: Daniel Smith <[email protected]>
CC: Jason Andryuk <[email protected]>
CC: Henry Wang <[email protected]>

Thanks, Andrew. Though we might want a small tweak - possibly as a follow up?

Quoted 13 lines

tools/flask/policy/modules/xen.if define(`manage_domain', ` · line 95

Quoted 10 lines
tools/flask/policy/modules/xen.if +1 -1
index 424daab6a022..6b7b7d403ab4 100644
@@ -92,7 +92,7 @@ define(`manage_domain', `
        allow $1 $2:domain { getdomaininfo getvcpuinfo getaffinity
                        getaddrsize pause unpause trigger shutdown destroy
                        setaffinity setdomainmaxmem getscheduler resume
-                       setpodtarget getpodtarget };
+                       setpodtarget getpodtarget getpagingmempool setpagingmempool };

There is also create_domain_common which is for a dedicated "domain builder" that creates but does not manage domains. I think that should gain setpagingmempool permission?

Sounds like it should.  Something like this?

(quoted diff) +0 -0

b/tools/flask/policy/modules/xen.if
index 6b7b7d403ab4..11c1562aa5da 100644

tools/flask/policy/modules/xen.if +0 -0
@@ -49,7 +49,8 @@ define(`create_domain_common', `

        allow $1 $2:domain { create max_vcpus setdomainmaxmem setaddrsize getdomaininfo hypercall setvcpucontext getscheduler getvcpuinfo getaddrsize getaffinity setaffinity -                       settime setdomainhandle getvcpucontext set_misc_info };
+                       settime setdomainhandle getvcpucontext set_misc_info +                       getpagingmempool setpagingmempool }; allow $1 $2:domain2 { set_cpu_policy settsc setscheduler setclaim set_vnumainfo get_vnumainfo cacheflush psr_cmt_op psr_alloc soft_reset

I can fold this in on commit.

~Andrew

#8 lore raw

Re: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

Quoted 51 lines · Andrew Cooper
Quoted 16 lines · Jason Andryuk
Quoted 11 lines · Andrew Cooper

These were overlooked in the original patch, and noticed by OSSTest which does run some Flask tests.

Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging mempool size")

---
CC: Daniel De Graaf <[email protected]>
CC: Daniel Smith <[email protected]>
CC: Jason Andryuk <[email protected]>
CC: Henry Wang <[email protected]>

Thanks, Andrew. Though we might want a small tweak - possibly as a follow up?

Quoted 13 lines

tools/flask/policy/modules/xen.if define(`manage_domain', ` · line 95

Quoted 10 lines
tools/flask/policy/modules/xen.if +1 -1
index 424daab6a022..6b7b7d403ab4 100644
@@ -92,7 +92,7 @@ define(`manage_domain', `
        allow $1 $2:domain { getdomaininfo getvcpuinfo getaffinity
                        getaddrsize pause unpause trigger shutdown destroy
                        setaffinity setdomainmaxmem getscheduler resume
-                       setpodtarget getpodtarget };
+                       setpodtarget getpodtarget getpagingmempool setpagingmempool };

There is also create_domain_common which is for a dedicated "domain builder" that creates but does not manage domains. I think that should gain setpagingmempool permission?

Sounds like it should. Something like this?

tools/flask/policy/modules/xen.if +2 -1
index 6b7b7d403ab4..11c1562aa5da 100644
@@ -49,7 +49,8 @@ define(`create_domain_common', `
        allow $1 $2:domain { create max_vcpus setdomainmaxmem setaddrsize
                        getdomaininfo hypercall setvcpucontext getscheduler
                        getvcpuinfo getaddrsize getaffinity setaffinity
-                       settime setdomainhandle getvcpucontext
 set_misc_info };
+                       settime setdomainhandle getvcpucontext set_misc_info
+                       getpagingmempool setpagingmempool };
        allow $1 $2:domain2 { set_cpu_policy settsc setscheduler setclaim
                        set_vnumainfo get_vnumainfo cacheflush

psr_cmt_op psr_alloc soft_reset

Quoted 1 line

I can fold this in on commit.

Yes, though strictly speaking it may only need setpagingmempool and not getpagingmempool. These are all calls that would execute before the domain is run. But adding both is probably fine since the builder is setting these itself. i.e. allowing it to read what it set is fine.

So, yes, I'd just fold this in.

Thanks,
Jason

Acked-by on v1 2/2
#9 lore raw

Re: [PATCH 2/2] xen/flask: Wire up XEN_DOMCTL_{get,set}_paging_mempool_size

In reply to Andrew Cooper
Quoted 12 lines · Andrew Cooper

These were overlooked in the original patch, and noticed by OSSTest which does run some Flask tests.

Fixes 22b20bd98c02 ("xen: Introduce non-broken hypercalls for the paging mempool size")

---
CC: Daniel De Graaf <[email protected]>
CC: Daniel Smith <[email protected]>
CC: Jason Andryuk <[email protected]>
CC: Henry Wang <[email protected]>

Quoted 7 lines

It should be noted that the original XSA-409 fix broke Flask on ARM but no testing noticed. c/s 7c3bbd940dd8 ("xen/arm, libxl: Revert XEN_DOMCTL_shadow_op; use p2m mempool hypercalls") "fixes" the original breakage and introduced this breakage instead.

For 4.17. It's a fix for an issue that OSSTest is currently blocking as a regression.