Re: [PATCH v2 2/3] x86/ldt: Make modify_ldt optional

[Ingo Molnar <mingo@kernel.org>]

* Andy Lutomirski <luto@amacapital.net> wrote:

> On Tue, Jul 21, 2015 at 11:06 PM, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > * Andy Lutomirski <luto@amacapital.net> wrote:
> >
> >> On Tue, Jul 21, 2015 at 1:28 PM, Brian Gerst <brgerst@gmail.com> wrote:
> >> > On Tue, Jul 21, 2015 at 3:59 PM, Andy Lutomirski <luto@kernel.org> wrote:
> >> >>
> >> >> The modify_ldt syscall exposes a large attack surface and is unnecessary for
> >> >> modern userspace.  Make it optional.
> >> >>
> >> >> Signed-off-by: Andy Lutomirski <luto@kernel.org>
> >> >> ---
> >> >>  arch/x86/Kconfig                   | 17 +++++++++++++++++
> >> >>  arch/x86/include/asm/mmu.h         |  2 ++
> >> >>  arch/x86/include/asm/mmu_context.h | 31 +++++++++++++++++++++++--------
> >> >>  arch/x86/kernel/Makefile           |  3 ++-
> >> >>  arch/x86/kernel/cpu/perf_event.c   |  4 ++++
> >> >>  arch/x86/kernel/process_64.c       |  2 ++
> >> >>  arch/x86/kernel/step.c             |  2 ++
> >> >>  kernel/sys_ni.c                    |  1 +
> >> >>  8 files changed, 53 insertions(+), 9 deletions(-)
> >> >>
> >> >> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> >> >> index 55bced17dc95..a7ff3980bd65 100644
> >> >> --- a/arch/x86/Kconfig
> >> >> +++ b/arch/x86/Kconfig
> >> >> @@ -1009,6 +1009,7 @@ config VM86
> >> >>  config X86_16BIT
> >> >>         bool "Enable support for 16-bit segments" if EXPERT
> >> >>         default y
> >> >> +       depends on MODIFY_LDT_SYSCALL
> >> >>         ---help---
> >> >>           This option is required by programs like Wine to run 16-bit
> >> >>           protected mode legacy code on x86 processors.  Disabling
> >> >> @@ -2047,6 +2048,22 @@ config CMDLINE_OVERRIDE
> >> >>           This is used to work around broken boot loaders.  This should
> >> >>           be set to 'N' under normal conditions.
> >> >>
> >> >> +config MODIFY_LDT_SYSCALL
> >> >> +       bool "Enable the LDT (local descriptor table)" if EXPERT
> >> >> +       default y
> >> >> +       ---help---
> >> >> +         Linux can allow user programs to install a per-process x86
> >> >> +        Local Descriptor Table (LDT) using the modify_ldt(2) system
> >> >> +        call.  This is required to run 16-bit or segmented code such as
> >> >> +        DOSEMU or some Wine programs.  It is also used by some very old
> >> >> +        threading libraries.
> >> >> +
> >> >> +        Enabling this feature adds a small amount of overhead to
> >> >> +        context switches and increases the low-level kernel attack
> >> >> +        surface.  Disabling it removes the modify_ldt(2) system call.
> >> >> +
> >> >> +        Saying 'N' here may make sense for embedded or server kernels.
> >> >> +
> >> >
> >> > I believe Wine still uses the LDT for thread-local data, even for 32
> >> > and 64-bit programs.  This is separate from the Linux runtime TLS.
> >> >
> >>
> >> Really?  I thought the whole reason we had three set_thread_area slots
> >> was for Wine.
> >
> > Too bad we have to guess, if only we had the Wine source code under a nicely
> > accessible Git archive or so to check?
> >
> >   git clone git://source.winehq.org/git/wine.git
> >
> > ;-)
> 
> You don't say?
> 
> It appears that Wine uses set_thread_area with a fallback to
> modify_ldt for 32-bit binaries and arch_prctl for 64-bit.

I also think it uses modify_ldt() unconditionally for 16-bit APIs:

triton:~/wine> git grep wine_ldt_set_entry
dlls/krnl386.exe16/dosvm.c:        wine_ldt_set_entry( sel, &entry );
dlls/krnl386.exe16/int31.c:        wine_ldt_set_entry( sel, &entry );
dlls/krnl386.exe16/int31.c:            wine_ldt_set_entry( BX_reg(context), entry );
dlls/krnl386.exe16/selector.c:            if (wine_ldt_set_entry( sel + (i << __AHSHIFT), &entry ) < 0)
dlls/krnl386.exe16/selector.c:        wine_ldt_set_entry( newsel + (i << __AHSHIFT), &entry );
dlls/krnl386.exe16/selector.c:        if (wine_ldt_set_entry( sel + (i << __AHSHIFT), &entry ) < 0) return FALSE;
dlls/krnl386.exe16/selector.c:    wine_ldt_set_entry( selDst, &entry );
dlls/krnl386.exe16/selector.c:    if (wine_ldt_set_entry( newsel, &entry ) >= 0) return newsel;
dlls/krnl386.exe16/selector.c:    if (wine_ldt_set_entry( newsel, &entry ) >= 0) return newsel;
dlls/krnl386.exe16/selector.c:    wine_ldt_set_entry( SELECTOROF(ptr), &entry );
dlls/krnl386.exe16/selector.c:    if (wine_ldt_set_entry( sel, &entry ) < 0) sel = 0;
dlls/krnl386.exe16/selector.c:    if (wine_ldt_set_entry( sel, &entry ) < 0) sel = 0;
dlls/krnl386.exe16/selector.c:        wine_ldt_set_entry( sel, &entry );
dlls/krnl386.exe16/wowthunk.c:    wine_ldt_set_entry( codesel, &entry );
dlls/user.exe16/message.c:        wine_ldt_set_entry( thunk_selector, &entry );
include/wine/library.h:extern int wine_ldt_set_entry( unsigned short sel, const LDT_ENTRY *entry );
libs/wine/ldt.c: *           wine_ldt_set_entry
libs/wine/ldt.c:int wine_ldt_set_entry( unsigned short sel, const LDT_ENTRY *entry )
libs/wine/wine.map:    wine_ldt_set_entry;

So the situation looks mostly encouraging, IMHO.

Plus an 'strace -fc' output of a modern, Wine driven Windows game would probably 
tell us pretty definitely whether there's anything particularly performance 
sensitive about modify_ldt().

Thanks,

	Ingo