Blame SOURCES/cve-2017-13693.patch

88c41d
From 987a3b5cf7175916e2a4b6ea5b8e70f830dfe732 Mon Sep 17 00:00:00 2001
88c41d
From: Seunghun Han <kkamagui@gmail.com>
88c41d
Date: Wed, 19 Jul 2017 16:47:53 +0900
88c41d
Subject: [PATCH] acpi: acpica: fix acpi operand cache leak in dswstate.c
88c41d
88c41d
I found an ACPI cache leak in ACPI early termination and boot continuing case.
88c41d
88c41d
When early termination occurs due to malicious ACPI table, Linux kernel
88c41d
terminates ACPI function and continues to boot process. While kernel terminates
88c41d
ACPI function, kmem_cache_destroy() reports Acpi-Operand cache leak.
88c41d
88c41d
Boot log of ACPI operand cache leak is as follows:
88c41d
>[    0.585957] ACPI: Added _OSI(Module Device)
88c41d
>[    0.587218] ACPI: Added _OSI(Processor Device)
88c41d
>[    0.588530] ACPI: Added _OSI(3.0 _SCP Extensions)
88c41d
>[    0.589790] ACPI: Added _OSI(Processor Aggregator Device)
88c41d
>[    0.591534] ACPI Error: Illegal I/O port address/length above 64K: C806E00000004002/0x2 (20170303/hwvalid-155)
88c41d
>[    0.594351] ACPI Exception: AE_LIMIT, Unable to initialize fixed events (20170303/evevent-88)
88c41d
>[    0.597858] ACPI: Unable to start the ACPI Interpreter
88c41d
>[    0.599162] ACPI Error: Could not remove SCI handler (20170303/evmisc-281)
88c41d
>[    0.601836] kmem_cache_destroy Acpi-Operand: Slab cache still has objects
88c41d
>[    0.603556] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.12.0-rc5 #26
88c41d
>[    0.605159] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
88c41d
>[    0.609177] Call Trace:
88c41d
>[    0.610063]  ? dump_stack+0x5c/0x81
88c41d
>[    0.611118]  ? kmem_cache_destroy+0x1aa/0x1c0
88c41d
>[    0.612632]  ? acpi_sleep_proc_init+0x27/0x27
88c41d
>[    0.613906]  ? acpi_os_delete_cache+0xa/0x10
88c41d
>[    0.617986]  ? acpi_ut_delete_caches+0x3f/0x7b
88c41d
>[    0.619293]  ? acpi_terminate+0xa/0x14
88c41d
>[    0.620394]  ? acpi_init+0x2af/0x34f
88c41d
>[    0.621616]  ? __class_create+0x4c/0x80
88c41d
>[    0.623412]  ? video_setup+0x7f/0x7f
88c41d
>[    0.624585]  ? acpi_sleep_proc_init+0x27/0x27
88c41d
>[    0.625861]  ? do_one_initcall+0x4e/0x1a0
88c41d
>[    0.627513]  ? kernel_init_freeable+0x19e/0x21f
88c41d
>[    0.628972]  ? rest_init+0x80/0x80
88c41d
>[    0.630043]  ? kernel_init+0xa/0x100
88c41d
>[    0.631084]  ? ret_from_fork+0x25/0x30
88c41d
>[    0.633343] vgaarb: loaded
88c41d
>[    0.635036] EDAC MC: Ver: 3.0.0
88c41d
>[    0.638601] PCI: Probing PCI hardware
88c41d
>[    0.639833] PCI host bridge to bus 0000:00
88c41d
>[    0.641031] pci_bus 0000:00: root bus resource [io  0x0000-0xffff]
88c41d
> ... Continue to boot and log is omitted ...
88c41d
88c41d
I analyzed this memory leak in detail and found acpi_ds_obj_stack_pop_and_
88c41d
delete() function miscalculated the top of the stack. acpi_ds_obj_stack_push()
88c41d
function uses walk_state->operand_index for start position of the top, but
88c41d
acpi_ds_obj_stack_pop_and_delete() function considers index 0 for it.
88c41d
Therefore, this causes acpi operand memory leak.
88c41d
88c41d
This cache leak causes a security threat because an old kernel (<= 4.9) shows
88c41d
memory locations of kernel functions in stack dump. Some malicious users
88c41d
could use this information to neutralize kernel ASLR.
88c41d
88c41d
I made a patch to fix ACPI operand cache leak.
88c41d
88c41d
Signed-off-by: Seunghun Han <kkamagui@gmail.com>
88c41d
88c41d
Github-Location: https://github.com/acpica/acpica/pull/295/commits/987a3b5cf7175916e2a4b6ea5b8e70f830dfe732
88c41d
---
88c41d
 source/components/dispatcher/dsutils.c | 9 ++++++++-
88c41d
 1 file changed, 8 insertions(+), 1 deletion(-)
88c41d
88c41d
Index: acpica-unix2-20180209/source/components/dispatcher/dsutils.c
88c41d
===================================================================
88c41d
--- acpica-unix2-20180209.orig/source/components/dispatcher/dsutils.c
88c41d
+++ acpica-unix2-20180209/source/components/dispatcher/dsutils.c
88c41d
@@ -761,6 +761,8 @@ AcpiDsCreateOperands (
88c41d
     ACPI_PARSE_OBJECT       *Arguments[ACPI_OBJ_NUM_OPERANDS];
88c41d
     UINT32                  ArgCount = 0;
88c41d
     UINT32                  Index = WalkState->NumOperands;
88c41d
+    UINT32                  PrevNumOperands = WalkState->NumOperands;
88c41d
+    UINT32                  NewNumOperands;
88c41d
     UINT32                  i;
88c41d
 
88c41d
 
88c41d
@@ -793,6 +795,7 @@ AcpiDsCreateOperands (
88c41d
 
88c41d
     /* Create the interpreter arguments, in reverse order */
88c41d
 
88c41d
+    NewNumOperands = Index;
88c41d
     Index--;
88c41d
     for (i = 0; i < ArgCount; i++)
88c41d
     {
88c41d
@@ -820,7 +823,11 @@ Cleanup:
88c41d
      * pop everything off of the operand stack and delete those
88c41d
      * objects
88c41d
      */
88c41d
-    AcpiDsObjStackPopAndDelete (ArgCount, WalkState);
88c41d
+    WalkState->NumOperands = i;
88c41d
+    AcpiDsObjStackPopAndDelete (NewNumOperands, WalkState);
88c41d
+
88c41d
+    /* Restore operand count */
88c41d
+    WalkState->NumOperands = PrevNumOperands;
88c41d
 
88c41d
     ACPI_EXCEPTION ((AE_INFO, Status, "While creating Arg %u", Index));
88c41d
     return_ACPI_STATUS (Status);