x86/mm: Do not reference phys addr beyond kernel
authorThomas Garnier <thgarnie@google.com>
Wed, 15 Jun 2016 19:05:45 +0000 (12:05 -0700)
committerIngo Molnar <mingo@kernel.org>
Sun, 10 Jul 2016 15:21:37 +0000 (17:21 +0200)
The new physical address randomized KASLR implementation can cause the
kernel to be aligned close to the end of physical memory. In this case,
_brk_end aligned to PMD will go beyond what is expected safe and hit
the assert in __phys_addr_symbol():

VIRTUAL_BUG_ON(y >= KERNEL_IMAGE_SIZE);

Instead, perform an inclusive range check to avoid incorrectly triggering
the assert:

kernel BUG at arch/x86/mm/physaddr.c:38!
invalid opcode: 0000 [#1] SMP
...
RIP: 0010:[<ffffffffbe055721>] __phys_addr_symbol+0x41/0x50
...
Call Trace:
[<ffffffffbe052eb9>] cpa_process_alias+0xa9/0x210
[<ffffffffbe109011>] ? do_raw_spin_unlock+0xc1/0x100
[<ffffffffbe051eef>] __change_page_attr_set_clr+0x8cf/0xbd0
[<ffffffffbe201a4d>] ? vm_unmap_aliases+0x7d/0x210
[<ffffffffbe05237c>] change_page_attr_set_clr+0x18c/0x4e0
[<ffffffffbe0534ec>] set_memory_4k+0x2c/0x40
[<ffffffffbefb08b3>] check_bugs+0x28/0x2a
[<ffffffffbefa4f52>] start_kernel+0x49d/0x4b9
[<ffffffffbefa4120>] ? early_idt_handler_array+0x120/0x120
[<ffffffffbefa4423>] x86_64_start_reservations+0x29/0x2b
[<ffffffffbefa4568>] x86_64_start_kernel+0x143/0x152

Signed-off-by: Thomas Garnier <thgarnie@google.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Dexuan Cui <decui@microsoft.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sai Praneeth <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Toshi Kani <toshi.kani@hpe.com>
Link: http://lkml.kernel.org/r/20160615190545.GA26071@www.outflux.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/mm/pageattr.c

index 7a1f7bbf4105b6ec570c9c15497a00237c5185e9..379b5111ac6b904915086e735b9502855cb04f3d 100644 (file)
@@ -101,7 +101,8 @@ static inline unsigned long highmap_start_pfn(void)
 
 static inline unsigned long highmap_end_pfn(void)
 {
-       return __pa_symbol(roundup(_brk_end, PMD_SIZE)) >> PAGE_SHIFT;
+       /* Do not reference physical address outside the kernel. */
+       return __pa_symbol(roundup(_brk_end, PMD_SIZE) - 1) >> PAGE_SHIFT;
 }
 
 #endif
@@ -112,6 +113,12 @@ within(unsigned long addr, unsigned long start, unsigned long end)
        return addr >= start && addr < end;
 }
 
+static inline int
+within_inclusive(unsigned long addr, unsigned long start, unsigned long end)
+{
+       return addr >= start && addr <= end;
+}
+
 /*
  * Flushing functions
  */
@@ -1316,7 +1323,8 @@ static int cpa_process_alias(struct cpa_data *cpa)
         * to touch the high mapped kernel as well:
         */
        if (!within(vaddr, (unsigned long)_text, _brk_end) &&
-           within(cpa->pfn, highmap_start_pfn(), highmap_end_pfn())) {
+           within_inclusive(cpa->pfn, highmap_start_pfn(),
+                            highmap_end_pfn())) {
                unsigned long temp_cpa_vaddr = (cpa->pfn << PAGE_SHIFT) +
                                               __START_KERNEL_map - phys_base;
                alias_cpa = *cpa;