Blame SOURCES/cve-2017-13693.patch

4496ab
From 987a3b5cf7175916e2a4b6ea5b8e70f830dfe732 Mon Sep 17 00:00:00 2001
4496ab
From: Seunghun Han <kkamagui@gmail.com>
4496ab
Date: Wed, 19 Jul 2017 16:47:53 +0900
4496ab
Subject: [PATCH] acpi: acpica: fix acpi operand cache leak in dswstate.c
4496ab
4496ab
I found an ACPI cache leak in ACPI early termination and boot continuing case.
4496ab
4496ab
When early termination occurs due to malicious ACPI table, Linux kernel
4496ab
terminates ACPI function and continues to boot process. While kernel terminates
4496ab
ACPI function, kmem_cache_destroy() reports Acpi-Operand cache leak.
4496ab
4496ab
Boot log of ACPI operand cache leak is as follows:
4496ab
>[    0.585957] ACPI: Added _OSI(Module Device)
4496ab
>[    0.587218] ACPI: Added _OSI(Processor Device)
4496ab
>[    0.588530] ACPI: Added _OSI(3.0 _SCP Extensions)
4496ab
>[    0.589790] ACPI: Added _OSI(Processor Aggregator Device)
4496ab
>[    0.591534] ACPI Error: Illegal I/O port address/length above 64K: C806E00000004002/0x2 (20170303/hwvalid-155)
4496ab
>[    0.594351] ACPI Exception: AE_LIMIT, Unable to initialize fixed events (20170303/evevent-88)
4496ab
>[    0.597858] ACPI: Unable to start the ACPI Interpreter
4496ab
>[    0.599162] ACPI Error: Could not remove SCI handler (20170303/evmisc-281)
4496ab
>[    0.601836] kmem_cache_destroy Acpi-Operand: Slab cache still has objects
4496ab
>[    0.603556] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.12.0-rc5 #26
4496ab
>[    0.605159] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
4496ab
>[    0.609177] Call Trace:
4496ab
>[    0.610063]  ? dump_stack+0x5c/0x81
4496ab
>[    0.611118]  ? kmem_cache_destroy+0x1aa/0x1c0
4496ab
>[    0.612632]  ? acpi_sleep_proc_init+0x27/0x27
4496ab
>[    0.613906]  ? acpi_os_delete_cache+0xa/0x10
4496ab
>[    0.617986]  ? acpi_ut_delete_caches+0x3f/0x7b
4496ab
>[    0.619293]  ? acpi_terminate+0xa/0x14
4496ab
>[    0.620394]  ? acpi_init+0x2af/0x34f
4496ab
>[    0.621616]  ? __class_create+0x4c/0x80
4496ab
>[    0.623412]  ? video_setup+0x7f/0x7f
4496ab
>[    0.624585]  ? acpi_sleep_proc_init+0x27/0x27
4496ab
>[    0.625861]  ? do_one_initcall+0x4e/0x1a0
4496ab
>[    0.627513]  ? kernel_init_freeable+0x19e/0x21f
4496ab
>[    0.628972]  ? rest_init+0x80/0x80
4496ab
>[    0.630043]  ? kernel_init+0xa/0x100
4496ab
>[    0.631084]  ? ret_from_fork+0x25/0x30
4496ab
>[    0.633343] vgaarb: loaded
4496ab
>[    0.635036] EDAC MC: Ver: 3.0.0
4496ab
>[    0.638601] PCI: Probing PCI hardware
4496ab
>[    0.639833] PCI host bridge to bus 0000:00
4496ab
>[    0.641031] pci_bus 0000:00: root bus resource [io  0x0000-0xffff]
4496ab
> ... Continue to boot and log is omitted ...
4496ab
4496ab
I analyzed this memory leak in detail and found acpi_ds_obj_stack_pop_and_
4496ab
delete() function miscalculated the top of the stack. acpi_ds_obj_stack_push()
4496ab
function uses walk_state->operand_index for start position of the top, but
4496ab
acpi_ds_obj_stack_pop_and_delete() function considers index 0 for it.
4496ab
Therefore, this causes acpi operand memory leak.
4496ab
4496ab
This cache leak causes a security threat because an old kernel (<= 4.9) shows
4496ab
memory locations of kernel functions in stack dump. Some malicious users
4496ab
could use this information to neutralize kernel ASLR.
4496ab
4496ab
I made a patch to fix ACPI operand cache leak.
4496ab
4496ab
Signed-off-by: Seunghun Han <kkamagui@gmail.com>
4496ab
4496ab
Github-Location: https://github.com/acpica/acpica/pull/295/commits/987a3b5cf7175916e2a4b6ea5b8e70f830dfe732
4496ab
---
4496ab
 source/components/dispatcher/dsutils.c | 9 ++++++++-
4496ab
 1 file changed, 8 insertions(+), 1 deletion(-)
4496ab
4496ab
Index: acpica-unix2-20180209/source/components/dispatcher/dsutils.c
4496ab
===================================================================
4496ab
--- acpica-unix2-20180209.orig/source/components/dispatcher/dsutils.c
4496ab
+++ acpica-unix2-20180209/source/components/dispatcher/dsutils.c
4496ab
@@ -761,6 +761,8 @@ AcpiDsCreateOperands (
4496ab
     ACPI_PARSE_OBJECT       *Arguments[ACPI_OBJ_NUM_OPERANDS];
4496ab
     UINT32                  ArgCount = 0;
4496ab
     UINT32                  Index = WalkState->NumOperands;
4496ab
+    UINT32                  PrevNumOperands = WalkState->NumOperands;
4496ab
+    UINT32                  NewNumOperands;
4496ab
     UINT32                  i;
4496ab
 
4496ab
 
4496ab
@@ -793,6 +795,7 @@ AcpiDsCreateOperands (
4496ab
 
4496ab
     /* Create the interpreter arguments, in reverse order */
4496ab
 
4496ab
+    NewNumOperands = Index;
4496ab
     Index--;
4496ab
     for (i = 0; i < ArgCount; i++)
4496ab
     {
4496ab
@@ -820,7 +823,11 @@ Cleanup:
4496ab
      * pop everything off of the operand stack and delete those
4496ab
      * objects
4496ab
      */
4496ab
-    AcpiDsObjStackPopAndDelete (ArgCount, WalkState);
4496ab
+    WalkState->NumOperands = i;
4496ab
+    AcpiDsObjStackPopAndDelete (NewNumOperands, WalkState);
4496ab
+
4496ab
+    /* Restore operand count */
4496ab
+    WalkState->NumOperands = PrevNumOperands;
4496ab
 
4496ab
     ACPI_EXCEPTION ((AE_INFO, Status, "While creating Arg %u", Index));
4496ab
     return_ACPI_STATUS (Status);