Re: + i386-fix-get_segment_eip-with-vm86.patch added to -mm tree

[Zachary Amsden]

akpm@xxxxxxxx wrote:
> The patch titled
>
>      i386: fix get_segment_eip() with vm86 segments
>
> has been added to the -mm tree.  Its filename is
>
>      i386-fix-get_segment_eip-with-vm86.patch
>
> See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
> out what to do about this
>
> ------------------------------------------------------
> Subject: i386: fix get_segment_eip() with vm86 segments
> From: Chuck Ebbert <76306.1226@xxxxxxxxxxxxxx>
>
>
> We need to check for vm86 mode first before looking at selector privilege
> bits.
>
> Segment limit is always base + 64k and only the low 16 bits of EIP are
> significant in vm86 mode.
>
> Signed-off-by: Chuck Ebbert <76306.1226@xxxxxxxxxxxxxx>
> Cc: Andi Kleen <ak@xxxxxx>
> Cc: Zachary Amsden <zach@xxxxxxxxxx>
> Cc: Rohit Seth <rohitseth@xxxxxxxxxx>
> Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
> ---
>
>  arch/i386/mm/fault.c |   11 +++++++----
>  1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff -puN arch/i386/mm/fault.c~i386-fix-get_segment_eip-with-vm86 arch/i386/mm/fault.c
> --- devel/arch/i386/mm/fault.c~i386-fix-get_segment_eip-with-vm86	2006-05-29 20:06:19.000000000 -0700
> +++ devel-akpm/arch/i386/mm/fault.c	2006-05-29 20:06:19.000000000 -0700
> @@ -77,12 +77,15 @@ static inline unsigned long get_segment_
>  	unsigned seg = regs->xcs & 0xffff;
>  	u32 seg_ar, seg_limit, base, *desc;
>  
> +	/* Unlikely, but must come before segment checks. */
> +	if (unlikely(regs->eflags & VM_MASK)) {
> +		base = seg << 4;
> +		*eip_limit = base + 0xffff;
> +		return base + (eip & 0xffff);
> +	}
> +
>  	/* The standard kernel/user address space limit. */
>  	*eip_limit = (seg & 3) ? USER_DS.seg : KERNEL_DS.seg;
> -
> -	/* Unlikely, but must come before segment checks. */
> -	if (unlikely((regs->eflags & VM_MASK) != 0))
> -		return eip + (seg << 4);
>  	
>  	/* By far the most common cases. */
>  	if (likely(seg == __USER_CS || seg == __KERNEL_CS))
> _
>
> Patches currently in -mm which might be from 76306.1226@xxxxxxxxxxxxxx are
>
> i386-let-usermode-execute-the-enter.patch
> i386-fix-get_segment_eip-with-vm86.patch
>
>   

This looks great.  While we're in the spirit let's fix kprobes v8086 
handling as well by filtering out int3s from v8086 mode.
Never allow int3 traps from V8086 mode to enter the kprobes handler.

Signed-off-by: Zachary Amsden <zach@xxxxxxxxxx>

Index: linux-2.6.17-rc/arch/i386/kernel/kprobes.c
===================================================================
--- linux-2.6.17-rc.orig/arch/i386/kernel/kprobes.c	2006-05-18 13:31:50.000000000 -0700
+++ linux-2.6.17-rc/arch/i386/kernel/kprobes.c	2006-05-31 13:09:26.000000000 -0700
@@ -607,7 +607,7 @@ int __kprobes kprobe_exceptions_notify(s
 	struct die_args *args = (struct die_args *)data;
 	int ret = NOTIFY_DONE;
 
-	if (args->regs && user_mode(args->regs))
+	if (args->regs && user_mode_vm(args->regs))
 		return ret;
 
 	switch (val) {