| Message ID | 1506972007-80614-3-git-send-email-keescook@chromium.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Kees, 2017-10-03 4:20 GMT+09:00 Kees Cook <keescook@chromium.org>: > Various portions of the kernel, especially per-architecture pieces, > need to know if the compiler is building it with the stack protector. > This was done in the arch/Kconfig with 'select', but this doesn't > allow a way to do auto-detected compiler support. In preparation for > creating an on-if-available default, move the logic for the definition of > CONFIG_CC_STACKPROTECTOR into the Makefile. > > Cc: Masahiro Yamada <yamada.masahiro@socionext.com> > Cc: Michal Marek <mmarek@suse.com> > Cc: Andrew Morton <akpm@linux-foundation.org> > Cc: Ingo Molnar <mingo@kernel.org> > Cc: Laura Abbott <labbott@redhat.com> > Cc: Nicholas Piggin <npiggin@gmail.com> > Cc: Al Viro <viro@zeniv.linux.org.uk> > Cc: linux-kbuild@vger.kernel.org > Signed-off-by: Kees Cook <keescook@chromium.org> > --- > Makefile | 7 +++++-- > arch/Kconfig | 8 -------- > 2 files changed, 5 insertions(+), 10 deletions(-) > > diff --git a/Makefile b/Makefile > index d1119941261c..e122a9cf0399 100644 > --- a/Makefile > +++ b/Makefile > @@ -688,8 +688,11 @@ else > stackp-flag := $(call cc-option, -fno-stack-protector) > endif > endif > -# Find arch-specific stack protector compiler sanity-checking script. > -ifdef CONFIG_CC_STACKPROTECTOR > +ifdef stackp-name > + # If the stack protector has been selected, inform the rest of the build. > + KBUILD_CFLAGS += -DCONFIG_CC_STACKPROTECTOR > + KBUILD_AFLAGS += -DCONFIG_CC_STACKPROTECTOR > + # Find arch-specific stack protector compiler sanity-checking script. > stackp-path := $(srctree)/scripts/gcc-$(SRCARCH)_$(BITS)-has-stack-protector.sh > stackp-check := $(wildcard $(stackp-path)) > endif I have not tested this series, but I think this commit is bad (with the follow-up patch applied). I thought of this scenario: [1] Kernel is configured with CONFIG_CC_STACKPROTECTOR_AUTO [2] Kernel is built with a compiler without stack protector support. [3] CONFIG_CC_STACKPROTECTOR is not defined, so __stack_chk_fail() is not compiled. [4] Out-of-tree modules are compiled with a compiler with stack protector support. __stack_chk_fail() is inserted to functions of the modules. [5] insmod fails because reference to __stack_chk_fail() can not be resolved. I think "select CC_STACKPROTECTOR" should be kept in Kconfig.
On Wed, Oct 04, 2017 at 11:33:38PM +0900, Masahiro Yamada wrote: > Hi Kees, > > > 2017-10-03 4:20 GMT+09:00 Kees Cook <keescook@chromium.org>: > > Various portions of the kernel, especially per-architecture pieces, > > need to know if the compiler is building it with the stack protector. > > This was done in the arch/Kconfig with 'select', but this doesn't > > allow a way to do auto-detected compiler support. In preparation for > > creating an on-if-available default, move the logic for the definition of > > CONFIG_CC_STACKPROTECTOR into the Makefile. > > > > Cc: Masahiro Yamada <yamada.masahiro@socionext.com> > > Cc: Michal Marek <mmarek@suse.com> > > Cc: Andrew Morton <akpm@linux-foundation.org> > > Cc: Ingo Molnar <mingo@kernel.org> > > Cc: Laura Abbott <labbott@redhat.com> > > Cc: Nicholas Piggin <npiggin@gmail.com> > > Cc: Al Viro <viro@zeniv.linux.org.uk> > > Cc: linux-kbuild@vger.kernel.org > > Signed-off-by: Kees Cook <keescook@chromium.org> > > --- > > Makefile | 7 +++++-- > > arch/Kconfig | 8 -------- > > 2 files changed, 5 insertions(+), 10 deletions(-) > > > > diff --git a/Makefile b/Makefile > > index d1119941261c..e122a9cf0399 100644 > > --- a/Makefile > > +++ b/Makefile > > @@ -688,8 +688,11 @@ else > > stackp-flag := $(call cc-option, -fno-stack-protector) > > endif > > endif > > -# Find arch-specific stack protector compiler sanity-checking script. > > -ifdef CONFIG_CC_STACKPROTECTOR > > +ifdef stackp-name > > + # If the stack protector has been selected, inform the rest of the build. > > + KBUILD_CFLAGS += -DCONFIG_CC_STACKPROTECTOR > > + KBUILD_AFLAGS += -DCONFIG_CC_STACKPROTECTOR > > + # Find arch-specific stack protector compiler sanity-checking script. > > stackp-path := $(srctree)/scripts/gcc-$(SRCARCH)_$(BITS)-has-stack-protector.sh > > stackp-check := $(wildcard $(stackp-path)) > > endif > > > I have not tested this series, > but I think this commit is bad (with the follow-up patch applied). > > > I thought of this scenario: > > [1] Kernel is configured with CONFIG_CC_STACKPROTECTOR_AUTO > > [2] Kernel is built with a compiler without stack protector support. > > [3] CONFIG_CC_STACKPROTECTOR is not defined, > so __stack_chk_fail() is not compiled. > > [4] Out-of-tree modules are compiled with a compiler with > stack protector support. > __stack_chk_fail() is inserted to functions of the modules. We don't ever support the system of loading a module built with anything other than the _exact_ same compiler than the kernel was. So this will not happen (well, if someone tries it, they get to keep the pieces their kernel image is now in...) > [5] insmod fails because reference to __stack_chk_fail() > can not be resolved. Even nicer, we failed "cleanly" :) This isn't a real-world issue, sorry. thanks, greg k-h
diff --git a/Makefile b/Makefile index d1119941261c..e122a9cf0399 100644 --- a/Makefile +++ b/Makefile @@ -688,8 +688,11 @@ else stackp-flag := $(call cc-option, -fno-stack-protector) endif endif -# Find arch-specific stack protector compiler sanity-checking script. -ifdef CONFIG_CC_STACKPROTECTOR +ifdef stackp-name + # If the stack protector has been selected, inform the rest of the build. + KBUILD_CFLAGS += -DCONFIG_CC_STACKPROTECTOR + KBUILD_AFLAGS += -DCONFIG_CC_STACKPROTECTOR + # Find arch-specific stack protector compiler sanity-checking script. stackp-path := $(srctree)/scripts/gcc-$(SRCARCH)_$(BITS)-has-stack-protector.sh stackp-check := $(wildcard $(stackp-path)) endif diff --git a/arch/Kconfig b/arch/Kconfig index 1aafb4efbb51..7007c1bfa79c 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -523,12 +523,6 @@ config HAVE_CC_STACKPROTECTOR - its compiler supports the -fstack-protector option - it has implemented a stack canary (e.g. __stack_chk_guard) -config CC_STACKPROTECTOR - def_bool n - help - Set when a stack-protector mode is enabled, so that the build - can enable kernel-side support for the GCC feature. - choice prompt "Stack Protector buffer overflow detection" depends on HAVE_CC_STACKPROTECTOR @@ -549,7 +543,6 @@ config CC_STACKPROTECTOR_NONE config CC_STACKPROTECTOR_REGULAR bool "Regular" - select CC_STACKPROTECTOR help Functions will have the stack-protector canary logic added if they have an 8-byte or larger character array on the stack. @@ -563,7 +556,6 @@ config CC_STACKPROTECTOR_REGULAR config CC_STACKPROTECTOR_STRONG bool "Strong" - select CC_STACKPROTECTOR help Functions will have the stack-protector canary logic added in any of the following conditions:
Various portions of the kernel, especially per-architecture pieces, need to know if the compiler is building it with the stack protector. This was done in the arch/Kconfig with 'select', but this doesn't allow a way to do auto-detected compiler support. In preparation for creating an on-if-available default, move the logic for the definition of CONFIG_CC_STACKPROTECTOR into the Makefile. Cc: Masahiro Yamada <yamada.masahiro@socionext.com> Cc: Michal Marek <mmarek@suse.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Ingo Molnar <mingo@kernel.org> Cc: Laura Abbott <labbott@redhat.com> Cc: Nicholas Piggin <npiggin@gmail.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: linux-kbuild@vger.kernel.org Signed-off-by: Kees Cook <keescook@chromium.org> --- Makefile | 7 +++++-- arch/Kconfig | 8 -------- 2 files changed, 5 insertions(+), 10 deletions(-)