Re: [PATCH] rpcrt4: Handle absense of RPC flags in call_stubless_func() thunk.

[Zebediah Figura]

On 04/10/18 07:51, Huw Davies wrote:
> On Wed, Oct 03, 2018 at 12:08:21PM -0500, Zebediah Figura wrote:
> > If Oi_HAS_RPCFLAGS is not set, then the stack size is at offset 4, not 8.
> > Signed-off-by: Zebediah Figura
> >
> > dlls/rpcrt4/cproxy.c | 6 ++++++
> >  1 file changed, 6 insertions(+)
> >
> > diff --git a/dlls/rpcrt4/cproxy.c b/dlls/rpcrt4/cproxy.c
> > index 2f49bb5..39dd34e 100644
> > --- a/dlls/rpcrt4/cproxy.c
> > +++ b/dlls/rpcrt4/cproxy.c
> > @@ -64,7 +64,13 @@ __ASM_GLOBAL_FUNC(call_stubless_func,
> >                    "movl 8(%ecx),%edx\n\t"         /* info->FormatStringOffset */
> >                    "movzwl (%edx,%eax,2),%edx\n\t" /* FormatStringOffset[index] */
> >                    "addl 4(%ecx),%edx\n\t"         /* info->ProcFormatString + offset */
> > -
> >              "testb $0x08,1(%edx)\n\t"       /* Oi_HAS_RPCFLAGS */
> > -
> >              "jz 1f\n\t"
> >              "movzwl 8(%edx),%eax\n\t"       /* arguments size */
> > -
> >              "jmp 2f\n\t"
> > -
> >              "1:\n\t"
> > -
> >              "movzwl 4(%edx),%eax\n\t"
> > -
> >              "2:\n\t"
> >              "pushl %eax\n\t"
> >              __ASM_CFI(".cfi_adjust_cfa_offset 4\n\t")
> >              "leal 8(%esp),%eax\n\t"         /* &This */
>
> Actually, thinking about this a bit more, it would be nice to avoid
> the jumps.  It should be easy enough to mask off the flag bit and
> then use that to calculate the correct offset from %edx.
>
> Huw.

Sure, can do.