Re: [Xen-devel] [PATCH v3 09/16] xen/arm: Introduce alternative runtime patching

[Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>]

On Wed, Jun 08, 2016 at 07:22:45PM +0100, Julien Grall wrote:
> Hi Konrad,
> 
> On 08/06/16 19:17, Konrad Rzeszutek Wilk wrote:
> >>>>diff --git a/xen/arch/arm/xen.lds.S b/xen/arch/arm/xen.lds.S
> >>>>index 1f010bd..495f9d8 100644
> >>>>--- a/xen/arch/arm/xen.lds.S
> >>>>+++ b/xen/arch/arm/xen.lds.S
> >>>>@@ -129,6 +129,9 @@ SECTIONS
> >>>>        _sinittext = .;
> >>>>        *(.init.text)
> >>>>        _einittext = .;
> >>>>+#ifdef CONFIG_ALTERNATIVE
> >>>>+       *(.altinstr_replacement)
> >>>>+#endif
> >>>
> >>>This is outside the _einittext? x86 looks to have .altinstr_replacement
> >>>inside the _einittext.
> >>
> >>Yes, I looked at the x86 code when I did the implement and I did not find
> >>any good reason to keep .altinstr_replace inside the inittext.
> >>
> >>altinstr_replacement contains replacement instructions. Anything inside the
> >>inittext region will be mark executable, which is not what we want here.
> >
> >Right, but we don't this code after the bootup (as in we patch the
> >.text and we can eject the .altinstr_replacement).
> 
> I don't see any problem, .altinstr_replacement is within __init_begin and
> __init_end. So it will be free when free_init_memory will be called.
> 
> _sinittex and _einittext are used to know which page will be executable.

Aaah, right! In that case, pls disregard my question.
> 
> Regards,
> 
> -- 
> Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel