[v2,03/13] uprobes: allow arch access to xol slot

Commit Message

Oleg Nesterov Oct. 29, 2013, 3:40 p.m. UTC

On 10/22, David Long wrote:
>
> On 10/19/13 12:36, Oleg Nesterov wrote:
>>
>> How about
>>
>> 	void __weak arch_uprobe_xol_copy(...)
>> 	{
>> 		copy_to_page(...);
>> 	}
>>
>> then just
>>
>> 	- copy_to_page(...);
>> 	+ arch_uprobe_xol_copy(...);
>>
>> ?
>>
>
> I was trying to avoid duplicating the VM calls in the
> architecture-specific implementations,

Yes, I understand... But this uglifies the generic code.

> but maybe that is the cleaner way
> to do it after all.  I've made changes as suggested above.

Thanks. and this uglifies arm code I guess ;)


David. Perhaps we can avoid the new hook altogether? What if we do
the simple change below (it ignores powerpc) ?

Then arm can add "unsigned long ixol[2]" into its arch_uprobe, and
arch_uprobe_analyze_insn() can initialize this member correctly.

What do you think?

Oleg.

Patch

--- x/arch/x86/include/asm/uprobes.h
+++ x/arch/x86/include/asm/uprobes.h
@@ -35,7 +35,10 @@  typedef u8 uprobe_opcode_t;
 
 struct arch_uprobe {
 	u16				fixups;
-	u8				insn[MAX_UINSN_BYTES];
+	union {
+		u8			insn[MAX_UINSN_BYTES];
+		u8			ixol[MAX_UINSN_BYTES];
+	}
 #ifdef CONFIG_X86_64
 	unsigned long			rip_rela_target_address;
 #endif
--- x/kernel/events/uprobes.c
+++ x/kernel/events/uprobes.c
@@ -1264,7 +1264,8 @@  static unsigned long xol_get_insn_slot(s
 		return 0;
 
 	/* Initialize the slot */
-	copy_to_page(area->page, xol_vaddr, uprobe->arch.insn, MAX_UINSN_BYTES);
+	copy_to_page(area->page, xol_vaddr,
+			uprobe->arch.ixol, sizeof(uprobe->arch.ixol));
 	/*
 	 * We probably need flush_icache_user_range() but it needs vma.
 	 * This should work on supported architectures too.