Re: [PATCH 1/8] kasan,x86: Frob kasan_report() in an exception

[Dmitry Vyukov]

On Wed, Mar 6, 2019 at 3:40 PM Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:
>
> On Wed, Mar 6, 2019 at 3:34 PM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> >
> > On Wed, Mar 06, 2019 at 03:12:37PM +0100, Peter Zijlstra wrote:
> > > On Wed, Mar 06, 2019 at 03:01:33PM +0100, Dmitry Vyukov wrote:
> > > > On Wed, Mar 6, 2019 at 2:57 PM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> > >
> > > > > I've not found callers of __asan_report_load* with AC=1 in the kernel
> > > > > yet. Under what condtions does GCC emit calls to these functions?
> > > >
> > > > CONFIG_KASAN_INLINE=y
> > > > Then compiler inlines fast path into generated code and only calls
> > > > into runtime to report errors (also, faster, this should be a default
> > > > for anything other than tiny ROM controllers).
> > >
> > > *sigh*, clearly I've not build enough kernels yet... Lemme go try that.
> >
> > mm/kasan/generic_report.o: warning: objtool: __asan_report_load1_noabort()+0x0: call to __fentry__() with UACCESS enabled
> >
> > You want to do:
> >
> > CFLAGS_REMOVE_f= -pg
> >
> > like generic.o has?
>
> This should not matter for KASAN itself.
> KASAN will call into function tracer, and function tracer will call
> into KASAN, but unless function tracer is badly broken and causes a
> KASAN report on every invocation, the recursion will end (function
> tracer will get to the _report_ function). So we only disabled -pg for

tracer will _not_ get to the _report_ function

> fast paths.
> But if it makes things simpler for the objtool, then I think we can
> disable function tracer for generic_report.c too.