Re: [Xen-devel] [PATCH v3 6/9] livepatch: Add parsing for the symbol+0x<offset>/<len>

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

On Mon, Aug 15, 2016 at 09:12:21AM -0600, Jan Beulich wrote:
> >>> On 15.08.16 at 16:35, <konrad.wilk@xxxxxxxxxx> wrote:
> > On Mon, Aug 15, 2016 at 04:53:48AM -0600, Jan Beulich wrote:
> >> >>> On 14.08.16 at 23:52, <konrad.wilk@xxxxxxxxxx> wrote:
> >> > in case we want to patch at specific offsets inside
> >> > a function. (for example if we want to do NOP patching).
> >> > 
> >> > We also assume that the 'len' is only the size of
> >> > an isns that would be for a call opcode (so 5 bytes
> >> > on x86, and 4 on ARM 32/64).
> >> 
> >> Which makes the notation using a slash quite confusing: Normally
> >> if we see <symbol>+<offset>/<length> the length part is assumed
> >> to be the overall size of the object/function the name refers to.
> > 
> > Oh, I somehow had the length of instruction on my mind.
> > 
> >> Could I talk you into using a different separator, like colon or
> >> semicolon?
> > 
> > Of course. Whatever is easier to folks.
> > 
> > I can also ditch the <length> part altogether?
> 
> The question is why you added it in the first place (see also the
> comments on the other patch, where I'm puzzled by the restriction
> to 5 bytes length).

Oh, I was looking at the dump stacks and got in my mind that /<len>
MUST be part of it. But it most certainly does not and I can remove it.

> 
> Jan
> 

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