On Fri, Aug 03, 2018 at 10:54:19AM +0100, Tim Deegan wrote:
> Hi,
>
> Apologies for the delay. Several of my other hats were on fire.
>
> > > I suspect the address, from which offset is derived, is bounded. But I
> > > haven't found the spec for KD.
> >
> > I don’t think there is one.
>
> Inde
Hi,
Apologies for the delay. Several of my other hats were on fire.
> > I suspect the address, from which offset is derived, is bounded. But I
> > haven't found the spec for KD.
>
> I don’t think there is one.
Indeed not. The official way to extend windbg &c is to write a plugin
that runs on
After some back and forth discussion on gcc-help, it is suggested this
is a bug in gcc and I'm asked to open a bug report.
Here is the bug report for reference:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86827
Wei.
___
Xen-devel mailing list
Xen-dev
One interesting observation is that if I revert 2de2b10b225 which turns
the type of offset from uint32_t back to uint64_t, kdd.c will build with
32 bit, but then of course 64 bit build is broken. :-/
Wei.
___
Xen-devel mailing list
Xen-devel@lists.xenpr
On Thu, Jul 26, 2018 at 01:37:45PM +0100, Ian Jackson wrote:
> Wei Liu writes ("[PATCH RFC] tools/kdd: avoid adversarial optimisation
> hazard"):
> > There have been two attempts to fix kdd build with gcc 8.1
> > (437e00fe and 2de2b10b), but building with gcc 8.1 32 bit non-debug
> > build still y
> -Original Message-
> From: Xen-devel [mailto:xen-devel-boun...@lists.xenproject.org] On Behalf
> Of Wei Liu
> Sent: 26 July 2018 13:54
> To: Ian Jackson
> Cc: Xen-devel ; Wei Liu
> ; Marek Marczykowski
> ; Jan Beulich ; Tim
> (Xen.org)
> Subject: Re: [Xe
On Thu, Jul 26, 2018 at 01:37:45PM +0100, Ian Jackson wrote:
> Wei Liu writes ("[PATCH RFC] tools/kdd: avoid adversarial optimisation
> hazard"):
> > There have been two attempts to fix kdd build with gcc 8.1
> > (437e00fe and 2de2b10b), but building with gcc 8.1 32 bit non-debug
> > build still y
Wei Liu writes ("[PATCH RFC] tools/kdd: avoid adversarial optimisation hazard"):
> There have been two attempts to fix kdd build with gcc 8.1
> (437e00fe and 2de2b10b), but building with gcc 8.1 32 bit non-debug
> build still yields the same error as in 437e00fe.
>
> Ian wrote about adversarial op
There have been two attempts to fix kdd build with gcc 8.1
(437e00fe and 2de2b10b), but building with gcc 8.1 32 bit non-debug
build still yields the same error as in 437e00fe.
Ian wrote about adversarial optimisation in [0], one of the key points
is that computing an out-of-bounds pointer is UB.
