Date
1 - 4 of 4
[dunfell 07/16] QEMU: CVE-2022-4144 QXL: qxl_phys2virt unsafe address translation can lead to out-of-bounds read
|
Steve Sakoman
From: Hitendra Prajapati <hprajapati@...>
Upstream-Status: Backport from https://gitlab.com/qemu-project/qemu/-/commit/6dbbf055148c6f1b7d8a3251a65bd6f3d1e1f622 Signed-off-by: Hitendra Prajapati <hprajapati@...> Signed-off-by: Steve Sakoman <steve@...> --- meta/recipes-devtools/qemu/qemu.inc | 1 + .../qemu/qemu/CVE-2022-4144.patch | 103 ++++++++++++++++++ 2 files changed, 104 insertions(+) create mode 100644 meta/recipes-devtools/qemu/qemu/CVE-2022-4144.patch diff --git a/meta/recipes-devtools/qemu/qemu.inc b/meta/recipes-devtools/qemu/qemu.inc index fff2c87780..898fa1a8d8 100644 --- a/meta/recipes-devtools/qemu/qemu.inc +++ b/meta/recipes-devtools/qemu/qemu.inc @@ -115,6 +115,7 @@ SRC_URI = "https://download.qemu.org/${BPN}-${PV}.tar.xz \ file://CVE-2021-3638.patch \ file://CVE-2021-20196.patch \ file://CVE-2021-3507.patch \ + file://CVE-2022-4144.patch \ " UPSTREAM_CHECK_REGEX = "qemu-(?P<pver>\d+(\.\d+)+)\.tar" diff --git a/meta/recipes-devtools/qemu/qemu/CVE-2022-4144.patch b/meta/recipes-devtools/qemu/qemu/CVE-2022-4144.patch new file mode 100644 index 0000000000..3f0d5fbd5c --- /dev/null +++ b/meta/recipes-devtools/qemu/qemu/CVE-2022-4144.patch @@ -0,0 +1,103 @@ +From 6dbbf055148c6f1b7d8a3251a65bd6f3d1e1f622 Mon Sep 17 00:00:00 2001 +From: =?UTF-8?q?Philippe=20Mathieu-Daud=C3=A9?= <philmd@...> +Date: Mon, 28 Nov 2022 21:27:40 +0100 +Subject: [PATCH] hw/display/qxl: Avoid buffer overrun in qxl_phys2virt + (CVE-2022-4144) +MIME-Version: 1.0 +Content-Type: text/plain; charset=UTF-8 +Content-Transfer-Encoding: 8bit + +Have qxl_get_check_slot_offset() return false if the requested +buffer size does not fit within the slot memory region. + +Similarly qxl_phys2virt() now returns NULL in such case, and +qxl_dirty_one_surface() aborts. + +This avoids buffer overrun in the host pointer returned by +memory_region_get_ram_ptr(). + +Fixes: CVE-2022-4144 (out-of-bounds read) +Reported-by: Wenxu Yin (@awxylitol) +Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1336 + +Signed-off-by: Philippe Mathieu-Daudé <philmd@...> +Signed-off-by: Stefan Hajnoczi <stefanha@...> +Message-Id: <20221128202741.4945-5-philmd@...> + +Upstream-Status: Backport [https://gitlab.com/qemu-project/qemu/-/commit/6dbbf055148c6f1b7d8a3251a65bd6f3d1e1f622] +CVE: CVE-2022-4144 +Comments: Deleted patch hunk in qxl.h,as it contains change +in comments which is not present in current version of qemu. + +Signed-off-by: Hitendra Prajapati <hprajapati@...> +--- + hw/display/qxl.c | 27 +++++++++++++++++++++++---- + 1 file changed, 23 insertions(+), 4 deletions(-) + +diff --git a/hw/display/qxl.c b/hw/display/qxl.c +index cd7eb39d..6bc8385b 100644 +--- a/hw/display/qxl.c ++++ b/hw/display/qxl.c +@@ -1440,11 +1440,13 @@ static void qxl_reset_surfaces(PCIQXLDevice *d) + + /* can be also called from spice server thread context */ + static bool qxl_get_check_slot_offset(PCIQXLDevice *qxl, QXLPHYSICAL pqxl, +- uint32_t *s, uint64_t *o) ++ uint32_t *s, uint64_t *o, ++ size_t size_requested) + { + uint64_t phys = le64_to_cpu(pqxl); + uint32_t slot = (phys >> (64 - 8)) & 0xff; + uint64_t offset = phys & 0xffffffffffff; ++ uint64_t size_available; + + if (slot >= NUM_MEMSLOTS) { + qxl_set_guest_bug(qxl, "slot too large %d >= %d", slot, +@@ -1468,6 +1470,23 @@ static bool qxl_get_check_slot_offset(PCIQXLDevice *qxl, QXLPHYSICAL pqxl, + slot, offset, qxl->guest_slots[slot].size); + return false; + } ++ size_available = memory_region_size(qxl->guest_slots[slot].mr); ++ if (qxl->guest_slots[slot].offset + offset >= size_available) { ++ qxl_set_guest_bug(qxl, ++ "slot %d offset %"PRIu64" > region size %"PRIu64"\n", ++ slot, qxl->guest_slots[slot].offset + offset, ++ size_available); ++ return false; ++ } ++ size_available -= qxl->guest_slots[slot].offset + offset; ++ if (size_requested > size_available) { ++ qxl_set_guest_bug(qxl, ++ "slot %d offset %"PRIu64" size %zu: " ++ "overrun by %"PRIu64" bytes\n", ++ slot, offset, size_requested, ++ size_requested - size_available); ++ return false; ++ } + + *s = slot; + *o = offset; +@@ -1486,7 +1505,7 @@ void *qxl_phys2virt(PCIQXLDevice *qxl, QXLPHYSICAL pqxl, int group_id) + offset = le64_to_cpu(pqxl) & 0xffffffffffff; + return (void *)(intptr_t)offset; + case MEMSLOT_GROUP_GUEST: +- if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset)) { ++ if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset, size)) { + return NULL; + } + ptr = memory_region_get_ram_ptr(qxl->guest_slots[slot].mr); +@@ -1944,9 +1963,9 @@ static void qxl_dirty_one_surface(PCIQXLDevice *qxl, QXLPHYSICAL pqxl, + uint32_t slot; + bool rc; + +- rc = qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset); +- assert(rc == true); + size = (uint64_t)height * abs(stride); ++ rc = qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset, size); ++ assert(rc == true); + trace_qxl_surfaces_dirty(qxl->id, offset, size); + qxl_set_dirty(qxl->guest_slots[slot].mr, + qxl->guest_slots[slot].offset + offset, +-- +2.25.1 + -- 2.25.1 |
|
|
|
Martin Jansa
I haven't checked yet in dunfell, but in kirkstone this commit merged last week breaks nativesdk-qemu builds. This chunk: +@@ -1486,7 +1505,7 @@ void *qxl_phys2virt(PCIQXLDevice *qxl, QXLPHYSICAL pqxl, int group_id) + offset = le64_to_cpu(pqxl) & 0xffffffffffff; + return (void *)(intptr_t)offset; + case MEMSLOT_GROUP_GUEST: +- if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset)) { ++ if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset, size)) { + return NULL; + } + ptr = memory_region_get_ram_ptr(qxl->guest_slots[slot].mr); Uses size parameter which was added in previous commit (which wasn't backported): So either both commits need to be backported or this one reworked not to use undeclared size as it leads to: | ../qemu-6.2.0/hw/display/qxl.c: In function 'qxl_phys2virt': | ../qemu-6.2.0/hw/display/qxl.c:1477:67: error: 'size' undeclared (first use in this function); did you mean 'gsize'? | 1477 | if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset, size)) { | | ^~~~ | | gsize I'm surprised this wasn't caught on autobuilder already, maybe it's triggered only with extra PACKAGECONFIG options we have enabled: PACKAGECONFIG:append:class-nativesdk = " virglrenderer epoxy spice libusb usb-redir" Regards, On Wed, Jan 25, 2023 at 3:42 PM Steve Sakoman <steve@...> wrote: From: Hitendra Prajapati <hprajapati@...> |
|
|
|
Martin Jansa
On Tue, Jan 31, 2023 at 8:18 AM Martin Jansa via lists.openembedded.org <Martin.Jansa=gmail.com@...> wrote:
I can confirm it fails the same with this commit from Hitendra in dunfell: nativesdk-qemu/4.2.0-r0/qemu-4.2.0/hw/display/qxl.c:1508:67: error: 'size' undeclared (first use in this function); did you mean 'gsize'? | 1508 | if (!qxl_get_check_slot_offset(qxl, pqxl, &slot, &offset, size)) { | | ^~~~ | | gsize as well as the same backport from Bhabu in kirkstone (already merged): To reproduce this just enable "spice" PACKAGECONFIG in nativesdk-qemu. Hitendra, Bhabu: please have a look. Regards,
|
|
|
|
Steve Sakoman
On Mon, Jan 30, 2023 at 9:39 PM Martin Jansa <Martin.Jansa@...> wrote:
Yes, please do! Unless I get a fix for the spice regression sometime soon I will revert the patch in kirkstone and of course won't take the dunfell version. Steve This chunk: |
|
|