[PATCH] Avoid overwriting the current pgd (V4, x86_64)

kexec: Avoid overwriting the current pgd (V4, x86_64)

This patch upgrades the x86_64-specific kexec code to avoid overwriting the
current pgd. Overwriting the current pgd is bad when CONFIG_CRASH_DUMP is used
to start a secondary kernel that dumps the memory of the previous kernel.

The code introduces a new set of page tables. These tables are used to provide
an executable identity mapping without overwriting the current pgd.

Signed-off-by: Magnus Damm <magnus@valinux.co.jp>
Signed-off-by: Andi Kleen <ak@suse.de>

1 files changed, 152 insertions, 19 deletions

diff --git a/arch/x86_64/kernel/relocate_kernel.S b/arch/x86_64/kernel/relocate_kernel.S
index d24fa9b72a2b..14e95872c6a3 100644
--- a/arch/x86_64/kernel/relocate_kernel.S
+++ b/arch/x86_64/kernel/relocate_kernel.S
@@ -7,31 +7,169 @@
  */
 
 #include <linux/linkage.h>
+#include <asm/page.h>
+#include <asm/kexec.h>
 
-        /*
-         * Must be relocatable PIC code callable as a C function, that once
-         * it starts can not use the previous processes stack.
-         */
-        .globl relocate_new_kernel
+/*
+ * Must be relocatable PIC code callable as a C function
+ */
+
+#define PTR(x) (x << 3)
+#define PAGE_ALIGNED (1 << PAGE_SHIFT)
+#define PAGE_ATTR 0x63 /* _PAGE_PRESENT|_PAGE_RW|_PAGE_ACCESSED|_PAGE_DIRTY */
+
+        .text
+        .align PAGE_ALIGNED
         .code64
+        .globl relocate_kernel
+relocate_kernel:
+        /* %rdi indirection_page
+         * %rsi page_list
+         * %rdx start address
+         */
+
+        /* map the control page at its virtual address */
+
+        movq    $0x0000ff8000000000, %r10        /* mask */
+        mov     $(39 - 3), %cl                   /* bits to shift */
+        movq    PTR(VA_CONTROL_PAGE)(%rsi), %r11 /* address to map */
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PGD)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PUD_0)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PUD_0)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PMD_0)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PMD_0)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PTE_0)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PTE_0)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_CONTROL_PAGE)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        /* identity map the control page at its physical address */
+
+        movq    $0x0000ff8000000000, %r10        /* mask */
+        mov     $(39 - 3), %cl                   /* bits to shift */
+        movq    PTR(PA_CONTROL_PAGE)(%rsi), %r11 /* address to map */
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PGD)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PUD_1)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PUD_1)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PMD_1)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PMD_1)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_PTE_1)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
+        shrq    $9, %r10
+        sub     $9, %cl
+
+        movq    %r11, %r9
+        andq    %r10, %r9
+        shrq    %cl, %r9
+
+        movq    PTR(VA_PTE_1)(%rsi), %r8
+        addq    %r8, %r9
+        movq    PTR(PA_CONTROL_PAGE)(%rsi), %r8
+        orq     $PAGE_ATTR, %r8
+        movq    %r8, (%r9)
+
 relocate_new_kernel:
-        /* %rdi page_list
-         * %rsi reboot_code_buffer
+        /* %rdi indirection_page
+         * %rsi page_list
          * %rdx start address
-         * %rcx page_table
-         * %r8  arg5
-         * %r9  arg6
          */
 
         /* zero out flags, and disable interrupts */
         pushq $0
         popfq
 
-        /* set a new stack at the bottom of our page... */
-        lea   4096(%rsi), %rsp
+        /* get physical address of control page now */
+        /* this is impossible after page table switch */
+        movq    PTR(PA_CONTROL_PAGE)(%rsi), %r8
+
+        /* get physical address of page table now too */
+        movq    PTR(PA_TABLE_PAGE)(%rsi), %rcx
 
-        /* store the parameters back on the stack */
-        pushq   %rdx /* store the start address */
+        /* switch to new set of page tables */
+        movq    PTR(PA_PGD)(%rsi), %r9
+        movq    %r9, %cr3
+
+        /* setup a new stack at the end of the physical control page */
+        lea     4096(%r8), %rsp
+
+        /* jump to identity mapped page */
+        addq    $(identity_mapped - relocate_kernel), %r8
+        pushq   %r8
+        ret
+
+identity_mapped:
+        /* store the start address on the stack */
+        pushq   %rdx
 
         /* Set cr0 to a known state:
          * 31 1 == Paging enabled
@@ -136,8 +274,3 @@ relocate_new_kernel:
         xorq    %r15, %r15
 
         ret
-relocate_new_kernel_end:
-
-        .globl relocate_new_kernel_size
-relocate_new_kernel_size:
-        .quad relocate_new_kernel_end - relocate_new_kernel