Re: ntdll: Fix stack alignment in RtlRaiseException assembly code.

[Alexandre Julliard]

Sebastian Lackner  writes:
> Signed-off-by: Sebastian Lackner
>
> Please review carefully, don't want to break anything shortly before the
> release. ;)
>
> dlls/ntdll/signal_i386.c |   41 ++++++++++++++++++++++++-----------------
>  1 file changed, 24 insertions(+), 17 deletions(-)

All it does is call other API functions that will align the stack
anyway, so I'm not sure it's worth the trouble.

ntdll: Fix stack alignment in RtlRaiseException assembly code.

[Sebastian Lackner]

On 21.07.2017 14:59, Alexandre Julliard wrote:
> Sebastian Lackner  writes:
> > Signed-off-by: Sebastian Lackner
> >
> > Please review carefully, don't want to break anything shortly before the
> > release. ;)
> >
> > dlls/ntdll/signal_i386.c |   41 ++++++++++++++++++++++++-----------------
> >  1 file changed, 24 insertions(+), 17 deletions(-)
>
> All it does is call other API functions that will align the stack
> anyway, so I'm not sure it's worth the trouble.

The idea was to avoid misaligning the stack when __force_align_arg_pointer__
is not present, but not sure if we still care about such ancient compilers.
And yes, in such a case the places where we intentionally misalign the stack
(in kernel32 for example) are probably more critical. I'll leave the decision
up to you. ;)

ntdll: Fix stack alignment in RtlRaiseException assembly code.

[Alexandre Julliard]

Sebastian Lackner  writes:
> On 21.07.2017 14:59, Alexandre Julliard wrote:
> > Sebastian Lackner  writes:
> > > Signed-off-by: Sebastian Lackner
> > >
> > > Please review carefully, don't want to break anything shortly before the
> > > release. ;)
> > >
> > > dlls/ntdll/signal_i386.c |   41 ++++++++++++++++++++++++-----------------
> > >  1 file changed, 24 insertions(+), 17 deletions(-)
> >
> > All it does is call other API functions that will align the stack
> > anyway, so I'm not sure it's worth the trouble.
>
> The idea was to avoid misaligning the stack when __force_align_arg_pointer__
> is not present, but not sure if we still care about such ancient compilers.
> And yes, in such a case the places where we intentionally misalign the stack
> (in kernel32 for example) are probably more critical. I'll leave the decision
> up to you. ;)

Yes, the stack is already misaligned on purpose, so on a system that
requires alignment, using an old compiler would break anyway. We still
need to be careful in places that call non-WINAPI functions of course.