Re: [PATCH 2/6] kexec jump: check code size in control page

Previous thread: [PATCH 6/6] kexec jump: fix for ftrace by Huang Ying on Thursday, August 7, 2008 - 2:05 am. (4 messages)

Next thread: [PATCH] lib/vsprintf.c: wrong conversion function used by Swen Schillig on Thursday, August 7, 2008 - 2:24 am. (2 messages)
From: Huang Ying
Date: Thursday, August 7, 2008 - 2:05 am

Kexec/Kexec-jump requires code size in control page is less than
PAGE_SIZE/2. This patch adds runtime checking for this.

Signed-off-by: Huang Ying <ying.huang@intel.com>

---
 arch/x86/kernel/machine_kexec_32.c   |    4 ++++
 arch/x86/kernel/relocate_kernel_32.S |    3 +++
 include/asm-x86/kexec.h              |    1 +
 3 files changed, 8 insertions(+)

--- a/arch/x86/kernel/machine_kexec_32.c
+++ b/arch/x86/kernel/machine_kexec_32.c
@@ -92,6 +92,10 @@ int machine_kexec_prepare(struct kimage 
 {
 	if (nx_enabled)
 		set_pages_x(image->control_code_page, 1);
+
+	BUG_ON((unsigned long)kexec_control_page_code_end - \
+	       (unsigned long)relocate_kernel >= PAGE_SIZE/2);
+
 	return 0;
 }
 
--- a/arch/x86/kernel/relocate_kernel_32.S
+++ b/arch/x86/kernel/relocate_kernel_32.S
@@ -376,3 +376,6 @@ swap_pages:
 	popl	%ebx
 	popl	%ebp
 	ret
+
+	.globl kexec_control_page_code_end
+kexec_control_page_code_end:
--- a/include/asm-x86/kexec.h
+++ b/include/asm-x86/kexec.h
@@ -159,6 +159,7 @@ relocate_kernel(unsigned long indirectio
 		unsigned long start_address,
 		unsigned int has_pae,
 		unsigned int preserve_context);
+void kexec_control_page_code_end(void);
 #else
 NORET_TYPE void
 relocate_kernel(unsigned long indirection_page,


--

From: Vivek Goyal
Date: Thursday, August 7, 2008 - 6:15 am

Hi Huang,

Run time check is better than nothing but I think in this case it would
be better if we can catch it at compile time. 

One of the methods will be to write a small program of your own and
put in script/ and at build time check for the size and flag error. May
be there are other better ways to do this.

Thanks
Vivek
--

From: Huang Ying
Date: Thursday, August 7, 2008 - 7:44 pm

I tried with BUILD_BUG_ON(), and compiling is OK for both of following
statement:

BUILD_BUG_ON((unsigned long)kexec_control_page_code_end - \
	     (unsigned long)relocate_kernel >= PAGE_SIZE/2);

BUILD_BUG_ON((unsigned long)kexec_control_page_code_end - \
	     (unsigned long)relocate_kernel < PAGE_SIZE/2);

In general, I think value of kexec_control_page_code_end and
relocate_kernel is not determined during compiling time. So
BUILD_BUG_ON() doesn't work.

Another idea, use ASSERT() command of ld link script as in the following
patch:

--- a/arch/x86/kernel/vmlinux_32.lds.S
+++ b/arch/x86/kernel/vmlinux_32.lds.S
@@ -209,3 +209,5 @@ SECTIONS
 
   DWARF_DEBUG
 }
+
+#include "vmlinux_check_32.lds.S"
--- /dev/null
+++ b/arch/x86/kernel/vmlinux_check_32.lds.S
@@ -0,0 +1,3 @@
+#include <asm/kexec.h>
+
+ASSERT(kexec_control_page_code_end - relocate_kernel >= 2048, "kexec control page code size is too big")


It works for me. What do you think about that?

Best Regards,
Huang Ying


--

From: Vivek Goyal
Date: Friday, August 8, 2008 - 6:52 am

Use of ASSERT() looks good to me. I think creati