* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 12:28 [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap Eduardo Habkost
@ 2018-11-30 12:35 ` Stefano Garzarella
2018-11-30 12:51 ` David Hildenbrand
` (3 subsequent siblings)
4 siblings, 0 replies; 10+ messages in thread
From: Stefano Garzarella @ 2018-11-30 12:35 UTC (permalink / raw)
To: ehabkost; +Cc: qemu-devel, david, Markus Armbruster, imammedo
On Fri, Nov 30, 2018 at 1:28 PM Eduardo Habkost <ehabkost@redhat.com> wrote:
>
> host_memory_backend_set_host_nodes() was not validating
> host-nodes before writing to backend->host_nodes, making QEMU
> write beyond the end of the bitmap.
>
> Fix the crash and add a simple regression test for the fix.
>
> While at it, fix memory leak of the list returned by
> visit_type_uint16List().
>
> Reported-by: Markus Armbruster <armbru@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> ---
> Changes v1 -> v2:
> * Remove extra `l = l->next` statement
> (reported by Stefano Garzarella)
> * Fix (existing) leak of `host_nodes`
> (reported by Markus Armbruster)
> ---
> backends/hostmem.c | 17 +++++++++----
> tests/acceptance/host-nodes-limit.py | 36 ++++++++++++++++++++++++++++
> 2 files changed, 49 insertions(+), 4 deletions(-)
> create mode 100644 tests/acceptance/host-nodes-limit.py
>
> diff --git a/backends/hostmem.c b/backends/hostmem.c
> index 1a89342039..af800284e0 100644
> --- a/backends/hostmem.c
> +++ b/backends/hostmem.c
> @@ -103,14 +103,23 @@ host_memory_backend_set_host_nodes(Object *obj, Visitor *v, const char *name,
> {
> #ifdef CONFIG_NUMA
> HostMemoryBackend *backend = MEMORY_BACKEND(obj);
> - uint16List *l = NULL;
> + uint16List *l, *host_nodes = NULL;
>
> - visit_type_uint16List(v, name, &l, errp);
> + visit_type_uint16List(v, name, &host_nodes, errp);
>
> - while (l) {
> + for (l = host_nodes; l; l = l->next) {
> + if (l->value >= MAX_NODES) {
> + error_setg(errp, "Invalid host-nodes value: %d", l->value);
> + goto out;
> + }
> + }
> +
> + for (l = host_nodes; l; l = l->next) {
> bitmap_set(backend->host_nodes, l->value, 1);
> - l = l->next;
> }
> +
> +out:
> + qapi_free_uint16List(host_nodes);
> #else
> error_setg(errp, "NUMA node binding are not supported by this QEMU");
> #endif
> diff --git a/tests/acceptance/host-nodes-limit.py b/tests/acceptance/host-nodes-limit.py
> new file mode 100644
> index 0000000000..e803e10104
> --- /dev/null
> +++ b/tests/acceptance/host-nodes-limit.py
> @@ -0,0 +1,36 @@
> +# Regression test for host-nodes limit validation
> +#
> +# Copyright (c) 2018 Red Hat, Inc.
> +#
> +# Author:
> +# Eduardo Habkost <ehabkost@redhat.com>
> +#
> +# This work is licensed under the terms of the GNU GPL, version 2 or
> +# later. See the COPYING file in the top-level directory.
> +
> +from avocado_qemu import Test
> +from subprocess import Popen, PIPE
> +
> +MAX_NODES = 128
> +
> +class HostNodesValidation(Test):
> + def test_large_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'Invalid host-nodes', stderr)
> + self.assertEquals(stdout, b'')
> + self.assertEquals(p.returncode, 1)
> +
> + def test_valid_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES - 1)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'host-nodes must be empty', stderr)
> + self.assertEquals(p.returncode, 1)
> --
> 2.18.0.rc1.1.g3f1ff2140
>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
--
Stefano Garzarella
Red Hat
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 12:28 [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap Eduardo Habkost
2018-11-30 12:35 ` Stefano Garzarella
@ 2018-11-30 12:51 ` David Hildenbrand
2018-11-30 13:22 ` Markus Armbruster
` (2 subsequent siblings)
4 siblings, 0 replies; 10+ messages in thread
From: David Hildenbrand @ 2018-11-30 12:51 UTC (permalink / raw)
To: Eduardo Habkost, qemu-devel
Cc: Markus Armbruster, Igor Mammedov, Stefano Garzarella
On 30.11.18 13:28, Eduardo Habkost wrote:
> host_memory_backend_set_host_nodes() was not validating
> host-nodes before writing to backend->host_nodes, making QEMU
> write beyond the end of the bitmap.
>
> Fix the crash and add a simple regression test for the fix.
>
> While at it, fix memory leak of the list returned by
> visit_type_uint16List().
>
> Reported-by: Markus Armbruster <armbru@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> ---
> Changes v1 -> v2:
> * Remove extra `l = l->next` statement
> (reported by Stefano Garzarella)
> * Fix (existing) leak of `host_nodes`
> (reported by Markus Armbruster)
> ---
> backends/hostmem.c | 17 +++++++++----
> tests/acceptance/host-nodes-limit.py | 36 ++++++++++++++++++++++++++++
> 2 files changed, 49 insertions(+), 4 deletions(-)
> create mode 100644 tests/acceptance/host-nodes-limit.py
>
> diff --git a/backends/hostmem.c b/backends/hostmem.c
> index 1a89342039..af800284e0 100644
> --- a/backends/hostmem.c
> +++ b/backends/hostmem.c
> @@ -103,14 +103,23 @@ host_memory_backend_set_host_nodes(Object *obj, Visitor *v, const char *name,
> {
> #ifdef CONFIG_NUMA
> HostMemoryBackend *backend = MEMORY_BACKEND(obj);
> - uint16List *l = NULL;
> + uint16List *l, *host_nodes = NULL;
>
> - visit_type_uint16List(v, name, &l, errp);
> + visit_type_uint16List(v, name, &host_nodes, errp);
>
> - while (l) {
> + for (l = host_nodes; l; l = l->next) {
> + if (l->value >= MAX_NODES) {
> + error_setg(errp, "Invalid host-nodes value: %d", l->value);
> + goto out;
> + }
> + }
> +
> + for (l = host_nodes; l; l = l->next) {
> bitmap_set(backend->host_nodes, l->value, 1);
> - l = l->next;
> }
> +
> +out:
> + qapi_free_uint16List(host_nodes);
> #else
> error_setg(errp, "NUMA node binding are not supported by this QEMU");
> #endif
> diff --git a/tests/acceptance/host-nodes-limit.py b/tests/acceptance/host-nodes-limit.py
> new file mode 100644
> index 0000000000..e803e10104
> --- /dev/null
> +++ b/tests/acceptance/host-nodes-limit.py
> @@ -0,0 +1,36 @@
> +# Regression test for host-nodes limit validation
> +#
> +# Copyright (c) 2018 Red Hat, Inc.
> +#
> +# Author:
> +# Eduardo Habkost <ehabkost@redhat.com>
> +#
> +# This work is licensed under the terms of the GNU GPL, version 2 or
> +# later. See the COPYING file in the top-level directory.
> +
> +from avocado_qemu import Test
> +from subprocess import Popen, PIPE
> +
> +MAX_NODES = 128
> +
> +class HostNodesValidation(Test):
> + def test_large_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'Invalid host-nodes', stderr)
> + self.assertEquals(stdout, b'')
> + self.assertEquals(p.returncode, 1)
> +
> + def test_valid_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES - 1)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'host-nodes must be empty', stderr)
> + self.assertEquals(p.returncode, 1)
>
Reviewed-by: David Hildenbrand <david@redhat.com>
--
Thanks,
David / dhildenb
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 12:28 [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap Eduardo Habkost
2018-11-30 12:35 ` Stefano Garzarella
2018-11-30 12:51 ` David Hildenbrand
@ 2018-11-30 13:22 ` Markus Armbruster
2018-11-30 14:27 ` Eduardo Habkost
2018-11-30 14:53 ` [Qemu-devel] [PATCH for-3.1? " Eric Blake
2018-12-04 13:29 ` [Qemu-devel] [PATCH " Igor Mammedov
4 siblings, 1 reply; 10+ messages in thread
From: Markus Armbruster @ 2018-11-30 13:22 UTC (permalink / raw)
To: Eduardo Habkost
Cc: qemu-devel, Igor Mammedov, Stefano Garzarella, David Hildenbrand
Eduardo Habkost <ehabkost@redhat.com> writes:
> host_memory_backend_set_host_nodes() was not validating
> host-nodes before writing to backend->host_nodes, making QEMU
> write beyond the end of the bitmap.
>
> Fix the crash and add a simple regression test for the fix.
>
> While at it, fix memory leak of the list returned by
> visit_type_uint16List().
>
> Reported-by: Markus Armbruster <armbru@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> ---
> Changes v1 -> v2:
> * Remove extra `l = l->next` statement
> (reported by Stefano Garzarella)
> * Fix (existing) leak of `host_nodes`
> (reported by Markus Armbruster)
> ---
> backends/hostmem.c | 17 +++++++++----
> tests/acceptance/host-nodes-limit.py | 36 ++++++++++++++++++++++++++++
> 2 files changed, 49 insertions(+), 4 deletions(-)
> create mode 100644 tests/acceptance/host-nodes-limit.py
>
> diff --git a/backends/hostmem.c b/backends/hostmem.c
> index 1a89342039..af800284e0 100644
> --- a/backends/hostmem.c
> +++ b/backends/hostmem.c
> @@ -103,14 +103,23 @@ host_memory_backend_set_host_nodes(Object *obj, Visitor *v, const char *name,
> {
> #ifdef CONFIG_NUMA
> HostMemoryBackend *backend = MEMORY_BACKEND(obj);
> - uint16List *l = NULL;
> + uint16List *l, *host_nodes = NULL;
>
> - visit_type_uint16List(v, name, &l, errp);
> + visit_type_uint16List(v, name, &host_nodes, errp);
>
> - while (l) {
> + for (l = host_nodes; l; l = l->next) {
> + if (l->value >= MAX_NODES) {
> + error_setg(errp, "Invalid host-nodes value: %d", l->value);
> + goto out;
> + }
> + }
> +
> + for (l = host_nodes; l; l = l->next) {
> bitmap_set(backend->host_nodes, l->value, 1);
> - l = l->next;
> }
> +
> +out:
> + qapi_free_uint16List(host_nodes);
> #else
> error_setg(errp, "NUMA node binding are not supported by this QEMU");
> #endif
Care to fix the leak in host_memory_backend_get_host_nodes(), too?
[...]
Reviewed-by: Markus Armbruster <armbru@redhat.com>
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 13:22 ` Markus Armbruster
@ 2018-11-30 14:27 ` Eduardo Habkost
2018-11-30 15:47 ` Markus Armbruster
0 siblings, 1 reply; 10+ messages in thread
From: Eduardo Habkost @ 2018-11-30 14:27 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel, Igor Mammedov, Stefano Garzarella, David Hildenbrand
On Fri, Nov 30, 2018 at 02:22:21PM +0100, Markus Armbruster wrote:
> Eduardo Habkost <ehabkost@redhat.com> writes:
>
> > host_memory_backend_set_host_nodes() was not validating
> > host-nodes before writing to backend->host_nodes, making QEMU
> > write beyond the end of the bitmap.
> >
> > Fix the crash and add a simple regression test for the fix.
> >
> > While at it, fix memory leak of the list returned by
> > visit_type_uint16List().
> >
> > Reported-by: Markus Armbruster <armbru@redhat.com>
> > Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> > ---
> > Changes v1 -> v2:
> > * Remove extra `l = l->next` statement
> > (reported by Stefano Garzarella)
> > * Fix (existing) leak of `host_nodes`
> > (reported by Markus Armbruster)
> > ---
> > backends/hostmem.c | 17 +++++++++----
> > tests/acceptance/host-nodes-limit.py | 36 ++++++++++++++++++++++++++++
> > 2 files changed, 49 insertions(+), 4 deletions(-)
> > create mode 100644 tests/acceptance/host-nodes-limit.py
> >
> > diff --git a/backends/hostmem.c b/backends/hostmem.c
> > index 1a89342039..af800284e0 100644
> > --- a/backends/hostmem.c
> > +++ b/backends/hostmem.c
> > @@ -103,14 +103,23 @@ host_memory_backend_set_host_nodes(Object *obj, Visitor *v, const char *name,
> > {
> > #ifdef CONFIG_NUMA
> > HostMemoryBackend *backend = MEMORY_BACKEND(obj);
> > - uint16List *l = NULL;
> > + uint16List *l, *host_nodes = NULL;
> >
> > - visit_type_uint16List(v, name, &l, errp);
> > + visit_type_uint16List(v, name, &host_nodes, errp);
> >
> > - while (l) {
> > + for (l = host_nodes; l; l = l->next) {
> > + if (l->value >= MAX_NODES) {
> > + error_setg(errp, "Invalid host-nodes value: %d", l->value);
> > + goto out;
> > + }
> > + }
> > +
> > + for (l = host_nodes; l; l = l->next) {
> > bitmap_set(backend->host_nodes, l->value, 1);
> > - l = l->next;
> > }
> > +
> > +out:
> > + qapi_free_uint16List(host_nodes);
> > #else
> > error_setg(errp, "NUMA node binding are not supported by this QEMU");
> > #endif
>
> Care to fix the leak in host_memory_backend_get_host_nodes(), too?
I don't understand yet if there's a leak at
host_memory_backend_get_host_nodes(). Won't
visit_type_uint16List() take ownership of the list on that case?
>
> [...]
>
> Reviewed-by: Markus Armbruster <armbru@redhat.com>
--
Eduardo
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 14:27 ` Eduardo Habkost
@ 2018-11-30 15:47 ` Markus Armbruster
0 siblings, 0 replies; 10+ messages in thread
From: Markus Armbruster @ 2018-11-30 15:47 UTC (permalink / raw)
To: Eduardo Habkost
Cc: Markus Armbruster, Igor Mammedov, David Hildenbrand, qemu-devel,
Stefano Garzarella
Eduardo Habkost <ehabkost@redhat.com> writes:
> I don't understand yet if there's a leak at
> host_memory_backend_get_host_nodes(). Won't
> visit_type_uint16List() take ownership of the list on that case?
Nope. I checked with valgrind:
$ valgrind --leak-check=full upstream-qemu -nodefaults -S -display none -qmp stdio -object memory-backend-file,id=mem0,mem-path=x,size=4096,host-nodes=1,policy=bind
[...]
{"QMP": {"version": {"qemu": {"micro": 92, "minor": 0, "major": 3}, "package": "v3.1.0-rc2-48-g039d4e3df0-dirty"}, "capabilities": []}}
{"execute": "qmp_capabilities"}
{"return": {}}
{ "execute": "qom-get", "arguments": { "path": "mem0", "property": "host-nodes" {"execute": "qom-get", "arguments": {"path": "mem0", "property": "host-nodes"}}
{"return": [1]}
{"execute": "quit"}
{"return": {}}
{"timestamp": {"seconds": 1543592652, "microseconds": 950994}, "event": "SHUTDOWN", "data": {"guest": false}}
==4954==
==4954== HEAP SUMMARY:
==4954== in use at exit: 3,631,673 bytes in 14,706 blocks
==4954== total heap usage: 51,347 allocs, 36,641 frees, 24,195,921 bytes allocated
[...]
==4954== 16 bytes in 1 blocks are definitely lost in loss record 1,964 of 5,297
==4954== at 0x4C3111A: calloc (vg_replace_malloc.c:752)
==4954== by 0x574948D: g_malloc0 (in /usr/lib64/libglib-2.0.so.0.5600.3)
==4954== by 0x9E1CE0: opts_start_list (opts-visitor.c:228)
==4954== by 0x9DAB35: visit_start_list (qapi-visit-core.c:78)
==4954== by 0x99BA3A: visit_type_uint16List (qapi-builtin-visit.c:272)
==4954== by 0x5F911B: host_memory_backend_set_host_nodes (hostmem.c:108)
==4954== by 0x8AC7D4: object_property_set (object.c:1183)
==4954== by 0x8AFC82: user_creatable_add_type (object_interfaces.c:73)
==4954== by 0x8AFED2: user_creatable_add_opts (object_interfaces.c:131)
==4954== by 0x8AFFCD: user_creatable_add_opts_foreach (object_interfaces.c:154)
==4954== by 0xA0B9B9: qemu_opts_foreach (qemu-option.c:1171)
==4954== by 0x5C6C44: main (vl.c:4415)
==4954==
==4954== 16 bytes in 1 blocks are definitely lost in loss record 1,965 of 5,297
==4954== at 0x4C3111A: calloc (vg_replace_malloc.c:752)
==4954== by 0x574948D: g_malloc0 (in /usr/lib64/libglib-2.0.so.0.5600.3)
==4954== by 0x5F8FF5: host_memory_backend_get_host_nodes (hostmem.c:82)
==4954== by 0x8AC739: object_property_get (object.c:1168)
==4954== by 0x8AF910: object_property_get_qobject (qom-qobject.c:39)
==4954== by 0x5E1736: qmp_qom_get (qmp.c:249)
==4954== by 0x5D872F: qmp_marshal_qom_get (qapi-commands-misc.c:1284)
==4954== by 0x9DF5C1: do_qmp_dispatch (qmp-dispatch.c:129)
==4954== by 0x9DF788: qmp_dispatch (qmp-dispatch.c:171)
==4954== by 0x42C0C1: monitor_qmp_dispatch (monitor.c:4085)
==4954== by 0x42C3E1: monitor_qmp_bh_dispatcher (monitor.c:4157)
==4954== by 0x9EEDB1: aio_bh_call (async.c:90)
[...]
==4954== LEAK SUMMARY:
==4954== definitely lost: 32 bytes in 2 blocks
==4954== indirectly lost: 0 bytes in 0 blocks
==4954== possibly lost: 2,504 bytes in 20 blocks
==4954== still reachable: 3,629,137 bytes in 14,684 blocks
==4954== of which reachable via heuristic:
==4954== newarray : 1,536 bytes in 16 blocks
==4954== suppressed: 0 bytes in 0 blocks
==4954== Reachable blocks (those to which a pointer was found) are not shown.
==4954== To see them, rerun with: --leak-check=full --show-leak-kinds=all
==4954==
==4954== For counts of detected and suppressed errors, rerun with: -v
==4954== Use --track-origins=yes to see where uninitialised values come from
==4954== ERROR SUMMARY: 24 errors from 24 contexts (suppressed: 0 from 0)
The first block shown is leaked in host_memory_backend_set_host_nodes()
on behalf of -object, the second block in
host_memory_backend_get_host_nodes() on behalf of qom-get.
Full disclosure: I hacked host_memory_backend_complete() to skip
mbind():
diff --git a/backends/hostmem.c b/backends/hostmem.c
index 1a89342039..0e40bb1ad4 100644
--- a/backends/hostmem.c
+++ b/backends/hostmem.c
@@ -333,7 +333,7 @@ host_memory_backend_memory_complete(UserCreatable *uc, Error **errp)
assert(sizeof(backend->host_nodes) >=
BITS_TO_LONGS(MAX_NODES + 1) * sizeof(unsigned long));
assert(maxnode <= MAX_NODES);
- if (mbind(ptr, sz, backend->policy,
+ if (0 && mbind(ptr, sz, backend->policy,
maxnode ? backend->host_nodes : NULL, maxnode + 1, flags)) {
if (backend->policy != MPOL_DEFAULT || errno != ENOSYS) {
error_setg_errno(errp, errno,
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [Qemu-devel] [PATCH for-3.1? v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 12:28 [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap Eduardo Habkost
` (2 preceding siblings ...)
2018-11-30 13:22 ` Markus Armbruster
@ 2018-11-30 14:53 ` Eric Blake
2018-11-30 17:55 ` Markus Armbruster
2018-12-04 13:29 ` [Qemu-devel] [PATCH " Igor Mammedov
4 siblings, 1 reply; 10+ messages in thread
From: Eric Blake @ 2018-11-30 14:53 UTC (permalink / raw)
To: Eduardo Habkost, qemu-devel
Cc: Igor Mammedov, Stefano Garzarella, Markus Armbruster,
David Hildenbrand, Peter Maydell
On 11/30/18 6:28 AM, Eduardo Habkost wrote:
> host_memory_backend_set_host_nodes() was not validating
> host-nodes before writing to backend->host_nodes, making QEMU
> write beyond the end of the bitmap.
>
> Fix the crash and add a simple regression test for the fix.
>
> While at it, fix memory leak of the list returned by
> visit_type_uint16List().
>
> Reported-by: Markus Armbruster <armbru@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> ---
Looks like we may have 3.1-rc4 due to some CVE fixes; is this worth
including in 3.1 as well?
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [Qemu-devel] [PATCH for-3.1? v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 14:53 ` [Qemu-devel] [PATCH for-3.1? " Eric Blake
@ 2018-11-30 17:55 ` Markus Armbruster
2018-11-30 18:19 ` Eduardo Habkost
0 siblings, 1 reply; 10+ messages in thread
From: Markus Armbruster @ 2018-11-30 17:55 UTC (permalink / raw)
To: Eric Blake
Cc: Eduardo Habkost, qemu-devel, Peter Maydell, Igor Mammedov,
David Hildenbrand, Stefano Garzarella
Eric Blake <eblake@redhat.com> writes:
> On 11/30/18 6:28 AM, Eduardo Habkost wrote:
>> host_memory_backend_set_host_nodes() was not validating
>> host-nodes before writing to backend->host_nodes, making QEMU
>> write beyond the end of the bitmap.
>>
>> Fix the crash and add a simple regression test for the fix.
>>
>> While at it, fix memory leak of the list returned by
>> visit_type_uint16List().
>>
>> Reported-by: Markus Armbruster <armbru@redhat.com>
>> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
>> ---
>
> Looks like we may have 3.1-rc4 due to some CVE fixes; is this worth
> including in 3.1 as well?
You could conceivably crash a running VM with object-add. On the other
hand, the bug has been around for a while, and was only found by code
inspection.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [Qemu-devel] [PATCH for-3.1? v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 17:55 ` Markus Armbruster
@ 2018-11-30 18:19 ` Eduardo Habkost
0 siblings, 0 replies; 10+ messages in thread
From: Eduardo Habkost @ 2018-11-30 18:19 UTC (permalink / raw)
To: Markus Armbruster
Cc: Eric Blake, qemu-devel, Peter Maydell, Igor Mammedov,
David Hildenbrand, Stefano Garzarella
On Fri, Nov 30, 2018 at 06:55:39PM +0100, Markus Armbruster wrote:
> Eric Blake <eblake@redhat.com> writes:
>
> > On 11/30/18 6:28 AM, Eduardo Habkost wrote:
> >> host_memory_backend_set_host_nodes() was not validating
> >> host-nodes before writing to backend->host_nodes, making QEMU
> >> write beyond the end of the bitmap.
> >>
> >> Fix the crash and add a simple regression test for the fix.
> >>
> >> While at it, fix memory leak of the list returned by
> >> visit_type_uint16List().
> >>
> >> Reported-by: Markus Armbruster <armbru@redhat.com>
> >> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
> >> ---
> >
> > Looks like we may have 3.1-rc4 due to some CVE fixes; is this worth
> > including in 3.1 as well?
>
> You could conceivably crash a running VM with object-add. On the other
> hand, the bug has been around for a while, and was only found by code
> inspection.
I think the fix would be appropriate for -rc2 or -rc3, but I
don't think it's critical enough for -rc4.
--
Eduardo
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap
2018-11-30 12:28 [Qemu-devel] [PATCH v2] hostmem: Validate host-nodes before setting bitmap Eduardo Habkost
` (3 preceding siblings ...)
2018-11-30 14:53 ` [Qemu-devel] [PATCH for-3.1? " Eric Blake
@ 2018-12-04 13:29 ` Igor Mammedov
4 siblings, 0 replies; 10+ messages in thread
From: Igor Mammedov @ 2018-12-04 13:29 UTC (permalink / raw)
To: Eduardo Habkost
Cc: qemu-devel, David Hildenbrand, Markus Armbruster,
Stefano Garzarella
On Fri, 30 Nov 2018 10:28:44 -0200
Eduardo Habkost <ehabkost@redhat.com> wrote:
> host_memory_backend_set_host_nodes() was not validating
> host-nodes before writing to backend->host_nodes, making QEMU
> write beyond the end of the bitmap.
>
> Fix the crash and add a simple regression test for the fix.
>
> While at it, fix memory leak of the list returned by
> visit_type_uint16List().
>
> Reported-by: Markus Armbruster <armbru@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
> ---
> Changes v1 -> v2:
> * Remove extra `l = l->next` statement
> (reported by Stefano Garzarella)
> * Fix (existing) leak of `host_nodes`
> (reported by Markus Armbruster)
> ---
> backends/hostmem.c | 17 +++++++++----
> tests/acceptance/host-nodes-limit.py | 36 ++++++++++++++++++++++++++++
> 2 files changed, 49 insertions(+), 4 deletions(-)
> create mode 100644 tests/acceptance/host-nodes-limit.py
>
> diff --git a/backends/hostmem.c b/backends/hostmem.c
> index 1a89342039..af800284e0 100644
> --- a/backends/hostmem.c
> +++ b/backends/hostmem.c
> @@ -103,14 +103,23 @@ host_memory_backend_set_host_nodes(Object *obj, Visitor *v, const char *name,
> {
> #ifdef CONFIG_NUMA
> HostMemoryBackend *backend = MEMORY_BACKEND(obj);
> - uint16List *l = NULL;
> + uint16List *l, *host_nodes = NULL;
>
> - visit_type_uint16List(v, name, &l, errp);
> + visit_type_uint16List(v, name, &host_nodes, errp);
>
> - while (l) {
> + for (l = host_nodes; l; l = l->next) {
> + if (l->value >= MAX_NODES) {
> + error_setg(errp, "Invalid host-nodes value: %d", l->value);
> + goto out;
> + }
> + }
> +
> + for (l = host_nodes; l; l = l->next) {
> bitmap_set(backend->host_nodes, l->value, 1);
> - l = l->next;
> }
> +
> +out:
> + qapi_free_uint16List(host_nodes);
> #else
> error_setg(errp, "NUMA node binding are not supported by this QEMU");
> #endif
> diff --git a/tests/acceptance/host-nodes-limit.py b/tests/acceptance/host-nodes-limit.py
> new file mode 100644
> index 0000000000..e803e10104
> --- /dev/null
> +++ b/tests/acceptance/host-nodes-limit.py
> @@ -0,0 +1,36 @@
> +# Regression test for host-nodes limit validation
> +#
> +# Copyright (c) 2018 Red Hat, Inc.
> +#
> +# Author:
> +# Eduardo Habkost <ehabkost@redhat.com>
> +#
> +# This work is licensed under the terms of the GNU GPL, version 2 or
> +# later. See the COPYING file in the top-level directory.
> +
> +from avocado_qemu import Test
> +from subprocess import Popen, PIPE
> +
> +MAX_NODES = 128
> +
> +class HostNodesValidation(Test):
> + def test_large_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'Invalid host-nodes', stderr)
> + self.assertEquals(stdout, b'')
> + self.assertEquals(p.returncode, 1)
> +
> + def test_valid_host_nodes(self):
> + p = Popen([self.qemu_bin, '-display', 'none', '-nodefaults',
> + '-object', 'memory-backend-ram,id=m0,'
> + 'size=4096,host-nodes=%d' % (MAX_NODES - 1)],
> + stderr=PIPE, stdout=PIPE)
> + stdout,stderr = p.communicate()
> +
> + self.assertIn(b'host-nodes must be empty', stderr)
> + self.assertEquals(p.returncode, 1)
^ permalink raw reply [flat|nested] 10+ messages in thread