[hurd,commited]: Fix static initialization with -fstack-protector-strong

[hurd,commited]: Fix static initialization with -fstack-protector-strong

[Samuel Thibault]

    When linked statically, TLS initialization is not achieved before
    mach_init and alike, so ssp accesses to tcbhead's stack_guard would
    crash.  We can just avoid using ssp in the few functions needed before
    TLS is set up.
    
            * mach/Makefile (CFLAGS-mach_init.o, CFLAGS-RPC_vm_statistics.o,
            CFLAGS-RPC_vm_map.o, CFLAGS-RPC_vm_protect.o,
            CFLAGS-RPC_i386_set_gdt.o, CFLAGS-RPC_i386_set_ldt.o,
            CFLAGS-RPC_task_get_special_port.o): Add $(no-stack-protector).
            * hurd/Makefile (CFLAGS-hurdstartup.o,
            CFLAGS-RPC_exec_startup_get_info.o): Add $(no-stack-protector).

diff --git a/hurd/Makefile b/hurd/Makefile
index 9205822b24..b44b9b80c6 100644
--- a/hurd/Makefile
+++ b/hurd/Makefile
@@ -81,6 +81,10 @@ $(inlines:%=$(objpfx)%.c): $(objpfx)%-inlines.c: %.h
 	 echo '#include "$<"') > $@-new
 	mv -f $@-new $@
 generated += $(inlines:=.c)
+
+# Avoid ssp before TLS is initialized.
+CFLAGS-hurdstartup.o = $(no-stack-protector)
+CFLAGS-RPC_exec_startup_get_info.o = $(no-stack-protector)
 \f
 include ../mach/Machrules
 include ../Rules
diff --git a/mach/Makefile b/mach/Makefile
index 127cad00ca..b4f18b1525 100644
--- a/mach/Makefile
+++ b/mach/Makefile
@@ -53,6 +53,15 @@ server-interfaces := mach/exc
 
 # Clear any environment	value.
 generated =
+
+# Avoid ssp before TLS is initialized.
+CFLAGS-mach_init.o = $(no-stack-protector)
+CFLAGS-RPC_vm_statistics.o = $(no-stack-protector)
+CFLAGS-RPC_vm_map.o = $(no-stack-protector)
+CFLAGS-RPC_vm_protect.o = $(no-stack-protector)
+CFLAGS-RPC_i386_set_gdt.o = $(no-stack-protector)
+CFLAGS-RPC_i386_set_ldt.o = $(no-stack-protector)
+CFLAGS-RPC_task_get_special_port.o = $(no-stack-protector)
 \f
 # Translate GNU names for CPUs into the names used in Mach header files.
 mach-machine = $(patsubst powerpc,ppc,$(base-machine))

Re: [hurd,commited]: Fix static initialization with -fstack-protector-strong

[Nick Alcock]

On 27 Aug 2017, Samuel Thibault outgrape:

>     When linked statically, TLS initialization is not achieved before
>     mach_init and alike, so ssp accesses to tcbhead's stack_guard would
>     crash.  We can just avoid using ssp in the few functions needed before
>     TLS is set up.

Oh blast I forgot all about Hurd didn't I.

Sorry.

> diff --git a/hurd/Makefile b/hurd/Makefile
> index 9205822b24..b44b9b80c6 100644
> --- a/hurd/Makefile
> +++ b/hurd/Makefile
> @@ -81,6 +81,10 @@ $(inlines:%=$(objpfx)%.c): $(objpfx)%-inlines.c: %.h
>  	 echo '#include "$<"') > $@-new
>  	mv -f $@-new $@
>  generated += $(inlines:=.c)
> +
> +# Avoid ssp before TLS is initialized.
> +CFLAGS-hurdstartup.o = $(no-stack-protector)
> +CFLAGS-RPC_exec_startup_get_info.o = $(no-stack-protector)

This certainly looks like the right sort of thing to be doing, though I
don't have anything I can test it on right now.

-fstack-protector-all might well comb out some more cases.

-- 
NULL && (void)

Re: [hurd,commited]: Fix static initialization with -fstack-protector-strong

[Samuel Thibault]

Nick Alcock, on ven. 15 sept. 2017 17:38:38 +0100, wrote:
> On 27 Aug 2017, Samuel Thibault outgrape:
> >     When linked statically, TLS initialization is not achieved before
> >     mach_init and alike, so ssp accesses to tcbhead's stack_guard would
> >     crash.  We can just avoid using ssp in the few functions needed before
> >     TLS is set up.
> 
> Oh blast I forgot all about Hurd didn't I.
> 
> Sorry.

No problem. Determining which functions need it is not trivial anyway :)

Samuel