diff mbox series

[v3,2/9] selftests/sgx: Produce static-pie executable for test enclave

Message ID 20230819094332.8535-3-jo.vanbulck@cs.kuleuven.be (mailing list archive)
State New, archived
Headers show
Series selftests/sgx: Fix compilation errors | expand

Commit Message

Jo Van Bulck Aug. 19, 2023, 9:43 a.m. UTC
The current combination of -static and -fPIC creates a static executable
with position-dependent addresses for global variables. Use -static-pie
and -fPIE to create a proper static position independent executable that
can be loaded at any address without a dynamic linker.

Link: https://lore.kernel.org/all/f9c24d89-ed72-7d9e-c650-050d722c6b04@cs.kuleuven.be/
Signed-off-by: Jo Van Bulck <jo.vanbulck@cs.kuleuven.be>
Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
---
 tools/testing/selftests/sgx/Makefile              |  2 +-
 tools/testing/selftests/sgx/test_encl.lds         |  1 +
 tools/testing/selftests/sgx/test_encl_bootstrap.S | 12 ++++++------
 3 files changed, 8 insertions(+), 7 deletions(-)

Comments

Huang, Kai Aug. 22, 2023, 12:26 a.m. UTC | #1
On Sat, 2023-08-19 at 11:43 +0200, Jo Van Bulck wrote:
> The current combination of -static and -fPIC creates a static executable
> with position-dependent addresses for global variables. Use -static-pie
> and -fPIE to create a proper static position independent executable that
> can be loaded at any address without a dynamic linker.

IMHO, this patch is also mixing different things together.

Specifically, ...
> 
> Link: https://lore.kernel.org/all/f9c24d89-ed72-7d9e-c650-050d722c6b04@cs.kuleuven.be/
> Signed-off-by: Jo Van Bulck <jo.vanbulck@cs.kuleuven.be>
> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
> ---
>  tools/testing/selftests/sgx/Makefile              |  2 +-
>  tools/testing/selftests/sgx/test_encl.lds         |  1 +
>  tools/testing/selftests/sgx/test_encl_bootstrap.S | 12 ++++++------
>  3 files changed, 8 insertions(+), 7 deletions(-)
> 
> diff --git a/tools/testing/selftests/sgx/Makefile b/tools/testing/selftests/sgx/Makefile
> index 50aab6b57..1d6315a2e 100644
> --- a/tools/testing/selftests/sgx/Makefile
> +++ b/tools/testing/selftests/sgx/Makefile
> @@ -13,7 +13,7 @@ endif
>  
>  INCLUDES := -I$(top_srcdir)/tools/include
>  HOST_CFLAGS := -Wall -Werror -g $(INCLUDES) -fPIC -z noexecstack
> -ENCL_CFLAGS := -Wall -Werror -static -nostdlib -nostartfiles -fPIC \
> +ENCL_CFLAGS := -Wall -Werror -static-pie -nostdlib -nostartfiles -fPIE \
>  	       -fno-stack-protector -mrdrnd $(INCLUDES)

... I think only this build flag change should be done in  this patch, as
described in the changelog.

Because ...
>  
>  TEST_CUSTOM_PROGS := $(OUTPUT)/test_sgx
> diff --git a/tools/testing/selftests/sgx/test_encl.lds b/tools/testing/selftests/sgx/test_encl.lds
> index a1ec64f7d..62d37160f 100644
> --- a/tools/testing/selftests/sgx/test_encl.lds
> +++ b/tools/testing/selftests/sgx/test_encl.lds
> @@ -10,6 +10,7 @@ PHDRS
>  SECTIONS
>  {
>  	. = 0;
> +        __encl_base = .;
>  	.tcs : {
>  		*(.tcs*)
>  	} : tcs
> diff --git a/tools/testing/selftests/sgx/test_encl_bootstrap.S b/tools/testing/selftests/sgx/test_encl_bootstrap.S
> index 03ae0f57e..28fe5d2ac 100644
> --- a/tools/testing/selftests/sgx/test_encl_bootstrap.S
> +++ b/tools/testing/selftests/sgx/test_encl_bootstrap.S
> @@ -42,9 +42,12 @@
>  encl_entry:
>  	# RBX contains the base address for TCS, which is the first address
>  	# inside the enclave for TCS #1 and one page into the enclave for
> -	# TCS #2. By adding the value of encl_stack to it, we get
> -	# the absolute address for the stack.
> -	lea	(encl_stack)(%rbx), %rax
> +	# TCS #2. First make it relative by substracting __encl_base and
> +	# then add the address of encl_stack to get the address for the stack.
> +	lea __encl_base(%rip), %rax
> +	sub %rax, %rbx
> +	lea encl_stack(%rip), %rax
> +	add %rbx, %rax

... if I am not missing anything, this chunk isn't needed for _this_ patch.  The
old code can still produce the correct stack address. __encl_base is only needed
for the next patch, thus the relevant change should be moved to the next patch.

>  	jmp encl_entry_core
>  encl_dyn_entry:
>  	# Entry point for dynamically created TCS page expected to follow
> @@ -55,12 +58,9 @@ encl_entry_core:
>  	push	%rax
>  
>  	push	%rcx # push the address after EENTER
> -	push	%rbx # push the enclave base address
>  
>  	call	encl_body
>  
> -	pop	%rbx # pop the enclave base address
> -
>  	/* Clear volatile GPRs, except RAX (EEXIT function). */
>  	xor     %rcx, %rcx
>  	xor     %rdx, %rdx

I honestly don't understand what's the purpose of this code change.  I believe
this change can be done (because it looks there's no need push/pop %rbx in the
first place), but again it should be in the next patch I suppose.
Jo Van Bulck Aug. 23, 2023, 1:19 p.m. UTC | #2
On 22.08.23 02:26, Huang, Kai wrote:

> ... I think only this build flag change should be done in  this patch, as
> described in the changelog.
> 
> Because ...

> ... if I am not missing anything, this chunk isn't needed for _this_ patch.  The
> old code can still produce the correct stack address. __encl_base is only needed
> for the next patch, thus the relevant change should be moved to the next patch.

I understand the confusion, but the reason I included this small change 
already in this commit is to make sure the commit compiles standalone. 
That is, when building the original assembly statement "lea 
(encl_stack)(%rbx), %rax" with -static-pie -fPIE, the linker complains 
about a relocation it cannot resolve:

/usr/local/bin/ld: /tmp/cchIWyfG.o: relocation R_X86_64_32S against 
`.data' can not be used when making a PIE object; recompile with -fPIE
collect2: error: ld returned 1 exit status

The problem is that only RIP-relative addressing is legit for local 
symbols, ie "encl_stack(%rip)". Hence, __encl_base is already needed 
here to calculate the stack address in the updated asm sequence in this 
patch.

Hope this helps clarifying!

> I honestly don't understand what's the purpose of this code change.  I believe
> this change can be done (because it looks there's no need push/pop %rbx in the
> first place), but again it should be in the next patch I suppose.

Thanks, the purpose indeed was merely to remove redundant code that is 
not needed. I see that it would be better to include this in a separate 
patch, so I'll update this in the next patch revision.

FWIW: if this is okay, while I'm on it, I'll also take a shot at 
removing remaining (unnecessary) assembly register cleansing code to 
make it more obvious that the test enclave is *not* exemplary secure, as 
per our earlier discussions. Ie in response to Dave's earlier comments 
that "The only patches I want for the kernel are to make the test 
enclave more *obviously* insecure." [1].

Best,
Jo

[1] 
https://lore.kernel.org/all/da0cfb1e-e347-f7f2-ac72-aec0ee0d867d@intel.com/
diff mbox series

Patch

diff --git a/tools/testing/selftests/sgx/Makefile b/tools/testing/selftests/sgx/Makefile
index 50aab6b57..1d6315a2e 100644
--- a/tools/testing/selftests/sgx/Makefile
+++ b/tools/testing/selftests/sgx/Makefile
@@ -13,7 +13,7 @@  endif
 
 INCLUDES := -I$(top_srcdir)/tools/include
 HOST_CFLAGS := -Wall -Werror -g $(INCLUDES) -fPIC -z noexecstack
-ENCL_CFLAGS := -Wall -Werror -static -nostdlib -nostartfiles -fPIC \
+ENCL_CFLAGS := -Wall -Werror -static-pie -nostdlib -nostartfiles -fPIE \
 	       -fno-stack-protector -mrdrnd $(INCLUDES)
 
 TEST_CUSTOM_PROGS := $(OUTPUT)/test_sgx
diff --git a/tools/testing/selftests/sgx/test_encl.lds b/tools/testing/selftests/sgx/test_encl.lds
index a1ec64f7d..62d37160f 100644
--- a/tools/testing/selftests/sgx/test_encl.lds
+++ b/tools/testing/selftests/sgx/test_encl.lds
@@ -10,6 +10,7 @@  PHDRS
 SECTIONS
 {
 	. = 0;
+        __encl_base = .;
 	.tcs : {
 		*(.tcs*)
 	} : tcs
diff --git a/tools/testing/selftests/sgx/test_encl_bootstrap.S b/tools/testing/selftests/sgx/test_encl_bootstrap.S
index 03ae0f57e..28fe5d2ac 100644
--- a/tools/testing/selftests/sgx/test_encl_bootstrap.S
+++ b/tools/testing/selftests/sgx/test_encl_bootstrap.S
@@ -42,9 +42,12 @@ 
 encl_entry:
 	# RBX contains the base address for TCS, which is the first address
 	# inside the enclave for TCS #1 and one page into the enclave for
-	# TCS #2. By adding the value of encl_stack to it, we get
-	# the absolute address for the stack.
-	lea	(encl_stack)(%rbx), %rax
+	# TCS #2. First make it relative by substracting __encl_base and
+	# then add the address of encl_stack to get the address for the stack.
+	lea __encl_base(%rip), %rax
+	sub %rax, %rbx
+	lea encl_stack(%rip), %rax
+	add %rbx, %rax
 	jmp encl_entry_core
 encl_dyn_entry:
 	# Entry point for dynamically created TCS page expected to follow
@@ -55,12 +58,9 @@  encl_entry_core:
 	push	%rax
 
 	push	%rcx # push the address after EENTER
-	push	%rbx # push the enclave base address
 
 	call	encl_body
 
-	pop	%rbx # pop the enclave base address
-
 	/* Clear volatile GPRs, except RAX (EEXIT function). */
 	xor     %rcx, %rcx
 	xor     %rdx, %rdx