* [PATCH v6 0/6] x86/cet: Update CET kernel interface
@ 2023-12-29 16:43 H.J. Lu
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
` (5 more replies)
0 siblings, 6 replies; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
If there are no objections, I will check in this patch series next week.
H.J.
--
Changes in v6.
1. Rebase.
2. Formatting changes.
3. Misc changes.
Changes in v5.
1. Rebase.
2. Move allocate-shadow-stack.[ch] to sysdeps/unix/sysv/linux/x86_64.
Changes in v4.
1. Rebase.
2. Remove 3 patches which have been checked into master branch.
Changes in v3:
1. Remove 7 test patches which have been checked into master branch.
Changes in v2:
1. Add add extra 20 stack frames in shadow stack for signal handlers
when allocating shadow stack for ucontexts.
2. Remove the "x86: Check PT_GNU_PROPERTY early" patch which has been
checked into master branch.
Linux kernel 6.6 added SHSTK support for x86-64. This patch set updates
CET kernel interface to Linux kernel 6.6. The main difference from the
current glibc assumption is that SHSTK is enabled by glibc, instead of
kernel. Glibc enables SHSTK after verifying that the application and
all dependency libraries are CET enabled. SHSTK can only be enabled in a
function which will never return. Otherwise, shadow stack will underflow
at the function return.
Not all CET enabled applications and libraries have been properly tested
in CET enabled environments. Some CET enabled applications or libraries
will crash or misbehave when CET is enabled. Don't set CET active by
default so that all applications and libraries will run normally regardless
of whether CET is active or not. Shadow stack can be enabled by
$ export GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
at run-time if shadow stack can be enabled by kernel.
Since only x86-64 is supported, i386 shadow stack codes are unchanged
and CET shouldn't be enabled for i386.
NB: This change can be reverted if it is OK to enable CET by default for
all applications and libraries.
Tested on Intel Tiger Lake under Linux kernel 6.6.7.
*** BLURB HERE ***
H.J. Lu (6):
x86/cet: Sync with Linux kernel 6.6 shadow stack interface
elf: Always provide _dl_get_dl_main_map in libc.a
x86/cet: Enable shadow stack during startup
x86/cet: Check feature_1 in TCB for active IBT and SHSTK
x86/cet: Don't set CET active by default
x86/cet: Run some CET tests with shadow stack
elf/dl-support.c | 2 -
sysdeps/generic/ldsodefs.h | 7 +-
sysdeps/unix/sysv/linux/x86/bits/mman.h | 5 ++
sysdeps/unix/sysv/linux/x86/dl-cet.h | 43 +++++++++--
.../unix/sysv/linux/x86/include/asm/prctl.h | 37 ++++-----
.../sysv/linux/x86/tst-cet-setcontext-1.c | 17 ++---
sysdeps/unix/sysv/linux/x86_64/Makefile | 2 +-
.../unix/sysv/linux/x86_64/__start_context.S | 38 ++--------
.../sysv/linux/x86_64/allocate-shadow-stack.c | 55 ++++++++++++++
.../allocate-shadow-stack.h} | 32 ++------
sysdeps/unix/sysv/linux/x86_64/dl-cet.h | 47 ++++++++++++
sysdeps/unix/sysv/linux/x86_64/getcontext.S | 30 ++------
sysdeps/unix/sysv/linux/x86_64/makecontext.c | 28 +++----
sysdeps/unix/sysv/linux/x86_64/swapcontext.S | 22 +-----
sysdeps/x86/Makefile | 14 ++++
sysdeps/x86/bits/platform/x86.h | 8 ++
sysdeps/x86/cpu-features-offsets.sym | 1 +
sysdeps/x86/cpu-features.c | 48 +-----------
sysdeps/x86/cpu-tunables.c | 15 +++-
sysdeps/x86/dl-cet.c | 76 +++++++++----------
sysdeps/x86/get-cpuid-feature-leaf.c | 13 +++-
sysdeps/x86/include/cpu-features.h | 3 +
sysdeps/x86/libc-start.h | 53 ++++++++++++-
sysdeps/x86/sys/platform/x86.h | 17 +++++
sysdeps/x86/tst-shstk-legacy-1e-static.sh | 1 +
sysdeps/x86/tst-shstk-legacy-1e.sh | 1 +
sysdeps/x86/tst-shstk-legacy-1g.sh | 1 +
sysdeps/x86_64/dl-machine.h | 12 ++-
sysdeps/x86_64/nptl/tls.h | 2 +-
29 files changed, 380 insertions(+), 250 deletions(-)
create mode 100644 sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.c
rename sysdeps/unix/sysv/linux/{x86/cpu-features.c => x86_64/allocate-shadow-stack.h} (53%)
create mode 100644 sysdeps/unix/sysv/linux/x86_64/dl-cet.h
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
2024-01-03 19:21 ` Adhemerval Zanella Netto
2024-01-10 23:49 ` Edgecombe, Rick P
2023-12-29 16:43 ` [PATCH v6 2/6] elf: Always provide _dl_get_dl_main_map in libc.a H.J. Lu
` (4 subsequent siblings)
5 siblings, 2 replies; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
Sync with Linux kernel 6.6 shadow stack interface. Since only x86-64 is
supported, i386 shadow stack codes are unchanged and CET shouldn't be
enabled for i386.
1. When the shadow stack base in TCB is unset, the default shadow stack
is in use. Use the current shadow stack pointer as the marker for the
default shadow stack. It is used to identify if the current shadow stack
is the same as the target shadow stack when switching ucontexts. If yes,
INCSSP will be used to unwind shadow stack. Otherwise, shadow stack
restore token will be used.
2. Allocate shadow stack with the map_shadow_stack syscall. Since there
is no function to explicitly release ucontext, there is no place to
release shadow stack allocated by map_shadow_stack in ucontext functions.
Such shadow stacks will be leaked.
3. Rename arch_prctl CET commands to ARCH_SHSTK_XXX.
4. Rewrite the CET control functions with the current kernel shadow stack
interface.
Since CET is no longer enabled by kernel, a separate patch will enable
shadow stack during startup.
---
sysdeps/unix/sysv/linux/x86/bits/mman.h | 5 ++
sysdeps/unix/sysv/linux/x86/cpu-features.c | 13 +++--
sysdeps/unix/sysv/linux/x86/dl-cet.h | 16 ++++--
.../unix/sysv/linux/x86/include/asm/prctl.h | 37 ++++++-------
.../sysv/linux/x86/tst-cet-setcontext-1.c | 17 +++---
sysdeps/unix/sysv/linux/x86_64/Makefile | 2 +-
.../unix/sysv/linux/x86_64/__start_context.S | 38 +++----------
.../sysv/linux/x86_64/allocate-shadow-stack.c | 55 +++++++++++++++++++
.../sysv/linux/x86_64/allocate-shadow-stack.h | 24 ++++++++
sysdeps/unix/sysv/linux/x86_64/getcontext.S | 30 ++--------
sysdeps/unix/sysv/linux/x86_64/makecontext.c | 28 +++++-----
sysdeps/unix/sysv/linux/x86_64/swapcontext.S | 22 ++------
sysdeps/x86/cpu-features.c | 15 +++--
sysdeps/x86/dl-cet.c | 2 +-
sysdeps/x86_64/nptl/tls.h | 2 +-
15 files changed, 173 insertions(+), 133 deletions(-)
create mode 100644 sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.c
create mode 100644 sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.h
diff --git a/sysdeps/unix/sysv/linux/x86/bits/mman.h b/sysdeps/unix/sysv/linux/x86/bits/mman.h
index 3d356e86a0..232b55a13d 100644
--- a/sysdeps/unix/sysv/linux/x86/bits/mman.h
+++ b/sysdeps/unix/sysv/linux/x86/bits/mman.h
@@ -27,6 +27,11 @@
#define MAP_32BIT 0x40 /* Only give out 32-bit addresses. */
#define MAP_ABOVE4G 0x80 /* Only map above 4GB. */
+#ifdef __USE_MISC
+/* Set up a restore token in the newly allocated shadow stack */
+# define SHADOW_STACK_SET_TOKEN 0x1
+#endif
+
#include <bits/mman-map-flags-generic.h>
/* Include generic Linux declarations. */
diff --git a/sysdeps/unix/sysv/linux/x86/cpu-features.c b/sysdeps/unix/sysv/linux/x86/cpu-features.c
index 41e7600668..0e6e2bf855 100644
--- a/sysdeps/unix/sysv/linux/x86/cpu-features.c
+++ b/sysdeps/unix/sysv/linux/x86/cpu-features.c
@@ -23,10 +23,15 @@
static inline int __attribute__ ((always_inline))
get_cet_status (void)
{
- unsigned long long cet_status[3];
- if (INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_CET_STATUS, cet_status) == 0)
- return cet_status[0];
- return 0;
+ unsigned long long kernel_feature;
+ unsigned int status = 0;
+ if (INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_STATUS,
+ &kernel_feature) == 0)
+ {
+ if ((kernel_feature & ARCH_SHSTK_SHSTK) != 0)
+ status = GNU_PROPERTY_X86_FEATURE_1_SHSTK;
+ }
+ return status;
}
# ifndef SHARED
diff --git a/sysdeps/unix/sysv/linux/x86/dl-cet.h b/sysdeps/unix/sysv/linux/x86/dl-cet.h
index c885bf1323..da220ac627 100644
--- a/sysdeps/unix/sysv/linux/x86/dl-cet.h
+++ b/sysdeps/unix/sysv/linux/x86/dl-cet.h
@@ -21,12 +21,20 @@
static inline int __attribute__ ((always_inline))
dl_cet_disable_cet (unsigned int cet_feature)
{
- return (int) INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_CET_DISABLE,
- cet_feature);
+ if (cet_feature != GNU_PROPERTY_X86_FEATURE_1_SHSTK)
+ return -1;
+ long long int kernel_feature = ARCH_SHSTK_SHSTK;
+ return (int) INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_DISABLE,
+ kernel_feature);
}
static inline int __attribute__ ((always_inline))
-dl_cet_lock_cet (void)
+dl_cet_lock_cet (unsigned int cet_feature)
{
- return (int) INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_CET_LOCK, 0);
+ if (cet_feature != GNU_PROPERTY_X86_FEATURE_1_SHSTK)
+ return -1;
+ /* Lock all SHSTK features. */
+ long long int kernel_feature = -1;
+ return (int) INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_LOCK,
+ kernel_feature);
}
diff --git a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
index 45ad0b052f..2f511321ad 100644
--- a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
+++ b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
@@ -4,24 +4,19 @@
#include_next <asm/prctl.h>
-#ifndef ARCH_CET_STATUS
-/* CET features:
- IBT: GNU_PROPERTY_X86_FEATURE_1_IBT
- SHSTK: GNU_PROPERTY_X86_FEATURE_1_SHSTK
- */
-/* Return CET features in unsigned long long *addr:
- features: addr[0].
- shadow stack base address: addr[1].
- shadow stack size: addr[2].
- */
-# define ARCH_CET_STATUS 0x3001
-/* Disable CET features in unsigned int features. */
-# define ARCH_CET_DISABLE 0x3002
-/* Lock all CET features. */
-# define ARCH_CET_LOCK 0x3003
-/* Allocate a new shadow stack with unsigned long long *addr:
- IN: requested shadow stack size: *addr.
- OUT: allocated shadow stack address: *addr.
- */
-# define ARCH_CET_ALLOC_SHSTK 0x3004
-#endif /* ARCH_CET_STATUS */
+#ifndef ARCH_SHSTK_ENABLE
+/* Enable SHSTK features in unsigned long int features. */
+# define ARCH_SHSTK_ENABLE 0x5001
+/* Disable SHSTK features in unsigned long int features. */
+# define ARCH_SHSTK_DISABLE 0x5002
+/* Lock SHSTK features in unsigned long int features. */
+# define ARCH_SHSTK_LOCK 0x5003
+/* Unlock SHSTK features in unsigned long int features. */
+# define ARCH_SHSTK_UNLOCK 0x5004
+/* Return SHSTK features in unsigned long int features. */
+# define ARCH_SHSTK_STATUS 0x5005
+
+/* ARCH_SHSTK_ features bits */
+# define ARCH_SHSTK_SHSTK 0x1
+# define ARCH_SHSTK_WRSS 0x2
+#endif
diff --git a/sysdeps/unix/sysv/linux/x86/tst-cet-setcontext-1.c b/sysdeps/unix/sysv/linux/x86/tst-cet-setcontext-1.c
index 837a9fd0eb..2ea66c803b 100644
--- a/sysdeps/unix/sysv/linux/x86/tst-cet-setcontext-1.c
+++ b/sysdeps/unix/sysv/linux/x86/tst-cet-setcontext-1.c
@@ -87,15 +87,14 @@ do_test (void)
ctx[4].uc_link = &ctx[0];
makecontext (&ctx[4], (void (*) (void)) f1, 0);
- /* NB: When shadow stack is enabled, makecontext calls arch_prctl
- with ARCH_CET_ALLOC_SHSTK to allocate a new shadow stack which
- can be unmapped. The base address and size of the new shadow
- stack are returned in __ssp[1] and __ssp[2]. makecontext is
- called for CTX1, CTX3 and CTX4. But only CTX1 is used. New
- shadow stacks are allocated in the order of CTX3, CTX1, CTX4.
- It is very likely that CTX1's shadow stack is placed between
- CTX3 and CTX4. We munmap CTX3's and CTX4's shadow stacks to
- create gaps above and below CTX1's shadow stack. We check
+ /* NB: When shadow stack is enabled, makecontext calls map_shadow_stack
+ to allocate a new shadow stack which can be unmapped. The base
+ address and size of the new shadow stack are returned in __ssp[1]
+ and __ssp[2]. makecontext is called for CTX1, CTX3 and CTX4. But
+ only CTX1 is used. New shadow stacks are allocated in the order
+ of CTX3, CTX1, CTX4. It is very likely that CTX1's shadow stack is
+ placed between CTX3 and CTX4. We munmap CTX3's and CTX4's shadow
+ stacks to create gaps above and below CTX1's shadow stack. We check
that setcontext CTX1 works correctly in this case. */
if (_get_ssp () != 0)
{
diff --git a/sysdeps/unix/sysv/linux/x86_64/Makefile b/sysdeps/unix/sysv/linux/x86_64/Makefile
index 5e19202ebf..06b873949e 100644
--- a/sysdeps/unix/sysv/linux/x86_64/Makefile
+++ b/sysdeps/unix/sysv/linux/x86_64/Makefile
@@ -3,7 +3,7 @@ sysdep_routines += ioperm iopl
endif
ifeq ($(subdir),stdlib)
-sysdep_routines += __start_context
+sysdep_routines += __start_context allocate-shadow-stack
endif
ifeq ($(subdir),csu)
diff --git a/sysdeps/unix/sysv/linux/x86_64/__start_context.S b/sysdeps/unix/sysv/linux/x86_64/__start_context.S
index f6436dd6bb..ae04203c90 100644
--- a/sysdeps/unix/sysv/linux/x86_64/__start_context.S
+++ b/sysdeps/unix/sysv/linux/x86_64/__start_context.S
@@ -24,20 +24,14 @@
/* Use CALL to push __start_context onto the new stack as well as the new
shadow stack. RDI points to ucontext:
Incoming:
- __ssp[0]: The original caller's shadow stack pointer.
- __ssp[1]: The size of the new shadow stack.
- __ssp[2]: The size of the new shadow stack.
- Outgoing:
__ssp[0]: The new shadow stack pointer.
__ssp[1]: The base address of the new shadow stack.
__ssp[2]: The size of the new shadow stack.
*/
ENTRY(__push___start_context)
- /* Save the pointer to ucontext. */
- movq %rdi, %r9
/* Get the original shadow stack pointer. */
- rdsspq %r8
+ rdsspq %rcx
/* Save the original stack pointer. */
movq %rsp, %rdx
/* Load the top of the new stack into RSI. */
@@ -45,24 +39,12 @@ ENTRY(__push___start_context)
/* Add 8 bytes to RSI since CALL will push the 8-byte return
address onto stack. */
leaq 8(%rsi), %rsp
- /* Allocate the new shadow stack. The size of the new shadow
- stack is passed in __ssp[1]. */
- lea (oSSP + 8)(%rdi), %RSI_LP
- movl $ARCH_CET_ALLOC_SHSTK, %edi
- movl $__NR_arch_prctl, %eax
- /* The new shadow stack base is returned in __ssp[1]. */
- syscall
- testq %rax, %rax
- jne L(hlt) /* This should never happen. */
-
- /* Get the size of the new shadow stack. */
- movq 8(%rsi), %rdi
-
- /* Get the base address of the new shadow stack. */
- movq (%rsi), %rsi
-
+ /* The size of the new shadow stack is stored in __ssp[2]. */
+ mov (oSSP + 16)(%rdi), %RSI_LP
+ /* The new shadow stack base is stored in __ssp[1]. */
+ mov (oSSP + 8)(%rdi), %RAX_LP
/* Use the restore stoken to restore the new shadow stack. */
- rstorssp -8(%rsi, %rdi)
+ rstorssp -8(%rax, %rsi)
/* Save the restore token on the original shadow stack. */
saveprevssp
@@ -73,18 +55,12 @@ ENTRY(__push___start_context)
jmp __start_context
1:
- /* Get the new shadow stack pointer. */
- rdsspq %rdi
-
/* Use the restore stoken to restore the original shadow stack. */
- rstorssp -8(%r8)
+ rstorssp -8(%rcx)
/* Save the restore token on the new shadow stack. */
saveprevssp
- /* Store the new shadow stack pointer in __ssp[0]. */
- movq %rdi, oSSP(%r9)
-
/* Restore the original stack. */
mov %rdx, %rsp
ret
diff --git a/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.c b/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.c
new file mode 100644
index 0000000000..f2e1d03b96
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.c
@@ -0,0 +1,55 @@
+/* Helper function to allocate shadow stack.
+ Copyright (C) 2023 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
+
+#include <sysdep.h>
+#include <stdint.h>
+#include <errno.h>
+#include <sys/mman.h>
+#include <libc-pointer-arith.h>
+#include <allocate-shadow-stack.h>
+
+/* NB: This can be treated as a syscall by caller. */
+
+long int
+__allocate_shadow_stack (size_t stack_size,
+ shadow_stack_size_t *child_stack)
+{
+#ifdef __NR_map_shadow_stack
+ size_t shadow_stack_size
+ = stack_size >> STACK_SIZE_TO_SHADOW_STACK_SIZE_SHIFT;
+ /* Align shadow stack to 8 bytes. */
+ shadow_stack_size = ALIGN_UP (shadow_stack_size, 8);
+ /* Since sigaltstack shares shadow stack with the current context in
+ the thread, add extra 20 stack frames in shadow stack for signal
+ handlers. */
+ shadow_stack_size += 20 * 8;
+ void *shadow_stack = (void *)INLINE_SYSCALL_CALL
+ (map_shadow_stack, NULL, shadow_stack_size, SHADOW_STACK_SET_TOKEN);
+ /* Report the map_shadow_stack error. */
+ if (shadow_stack == MAP_FAILED)
+ return -errno;
+
+ /* Save the shadow stack base and size on child stack. */
+ child_stack[0] = (uintptr_t) shadow_stack;
+ child_stack[1] = shadow_stack_size;
+
+ return 0;
+#else
+ return -ENOSYS;
+#endif
+}
diff --git a/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.h b/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.h
new file mode 100644
index 0000000000..d05aaf16e5
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/x86_64/allocate-shadow-stack.h
@@ -0,0 +1,24 @@
+/* Helper function to allocate shadow stack.
+ Copyright (C) 2023 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
+
+#include <ucontext.h>
+
+typedef __typeof (((ucontext_t *) 0)->__ssp[0]) shadow_stack_size_t;
+
+extern long int __allocate_shadow_stack (size_t, shadow_stack_size_t *)
+ attribute_hidden;
diff --git a/sysdeps/unix/sysv/linux/x86_64/getcontext.S b/sysdeps/unix/sysv/linux/x86_64/getcontext.S
index a00e2f6290..71f3802dca 100644
--- a/sysdeps/unix/sysv/linux/x86_64/getcontext.S
+++ b/sysdeps/unix/sysv/linux/x86_64/getcontext.S
@@ -58,35 +58,15 @@ ENTRY(__getcontext)
testl $X86_FEATURE_1_SHSTK, %fs:FEATURE_1_OFFSET
jz L(no_shstk)
- /* Save RDI in RDX which won't be clobbered by syscall. */
- movq %rdi, %rdx
-
xorl %eax, %eax
cmpq %fs:SSP_BASE_OFFSET, %rax
jnz L(shadow_stack_bound_recorded)
- /* Get the base address and size of the default shadow stack
- which must be the current shadow stack since nothing has
- been recorded yet. */
- sub $24, %RSP_LP
- mov %RSP_LP, %RSI_LP
- movl $ARCH_CET_STATUS, %edi
- movl $__NR_arch_prctl, %eax
- syscall
- testq %rax, %rax
- jz L(continue_no_err)
-
- /* This should never happen. */
- hlt
-
-L(continue_no_err):
- /* Record the base of the current shadow stack. */
- movq 8(%rsp), %rax
+ /* When the shadow stack base is unset, the default shadow
+ stack is in use. Use the current shadow stack pointer
+ as the marker for the default shadow stack. */
+ rdsspq %rax
movq %rax, %fs:SSP_BASE_OFFSET
- add $24, %RSP_LP
-
- /* Restore RDI. */
- movq %rdx, %rdi
L(shadow_stack_bound_recorded):
/* Get the current shadow stack pointer. */
@@ -94,7 +74,7 @@ L(shadow_stack_bound_recorded):
/* NB: Save the caller's shadow stack so that we can jump back
to the caller directly. */
addq $8, %rax
- movq %rax, oSSP(%rdx)
+ movq %rax, oSSP(%rdi)
/* Save the current shadow stack base in ucontext. */
movq %fs:SSP_BASE_OFFSET, %rax
diff --git a/sysdeps/unix/sysv/linux/x86_64/makecontext.c b/sysdeps/unix/sysv/linux/x86_64/makecontext.c
index de9e03eb81..e4f025bd50 100644
--- a/sysdeps/unix/sysv/linux/x86_64/makecontext.c
+++ b/sysdeps/unix/sysv/linux/x86_64/makecontext.c
@@ -24,6 +24,7 @@
# include <pthread.h>
# include <libc-pointer-arith.h>
# include <sys/prctl.h>
+# include <allocate-shadow-stack.h>
#endif
#include "ucontext_i.h"
@@ -88,23 +89,24 @@ __makecontext (ucontext_t *ucp, void (*func) (void), int argc, ...)
if ((feature_1 & X86_FEATURE_1_SHSTK) != 0)
{
/* Shadow stack is enabled. We need to allocate a new shadow
- stack. */
- unsigned long ssp_size = (((uintptr_t) sp
- - (uintptr_t) ucp->uc_stack.ss_sp)
- >> STACK_SIZE_TO_SHADOW_STACK_SIZE_SHIFT);
- /* Align shadow stack to 8 bytes. */
- ssp_size = ALIGN_UP (ssp_size, 8);
-
- ucp->__ssp[1] = ssp_size;
- ucp->__ssp[2] = ssp_size;
-
- /* Call __push___start_context to allocate a new shadow stack,
- push __start_context onto the new stack as well as the new
- shadow stack. NB: After __push___start_context returns,
+ stack. NB:
ucp->__ssp[0]: The new shadow stack pointer.
ucp->__ssp[1]: The base address of the new shadow stack.
ucp->__ssp[2]: The size of the new shadow stack.
*/
+ long int ret
+ = __allocate_shadow_stack (((uintptr_t) sp
+ - (uintptr_t) ucp->uc_stack.ss_sp),
+ &ucp->__ssp[1]);
+ if (ret != 0)
+ {
+ /* FIXME: What should we do? */
+ abort ();
+ }
+
+ ucp->__ssp[0] = ucp->__ssp[1] + ucp->__ssp[2] - 8;
+ /* Call __push___start_context to push __start_context onto the new
+ stack as well as the new shadow stack. */
__push___start_context (ucp);
}
else
diff --git a/sysdeps/unix/sysv/linux/x86_64/swapcontext.S b/sysdeps/unix/sysv/linux/x86_64/swapcontext.S
index 5925752164..2f2fe9875b 100644
--- a/sysdeps/unix/sysv/linux/x86_64/swapcontext.S
+++ b/sysdeps/unix/sysv/linux/x86_64/swapcontext.S
@@ -109,25 +109,11 @@ ENTRY(__swapcontext)
cmpq %fs:SSP_BASE_OFFSET, %rax
jnz L(shadow_stack_bound_recorded)
- /* Get the base address and size of the default shadow stack
- which must be the current shadow stack since nothing has
- been recorded yet. */
- sub $24, %RSP_LP
- mov %RSP_LP, %RSI_LP
- movl $ARCH_CET_STATUS, %edi
- movl $__NR_arch_prctl, %eax
- syscall
- testq %rax, %rax
- jz L(continue_no_err)
-
- /* This should never happen. */
- hlt
-
-L(continue_no_err):
- /* Record the base of the current shadow stack. */
- movq 8(%rsp), %rax
+ /* When the shadow stack base is unset, the default shadow
+ stack is in use. Use the current shadow stack pointer
+ as the marker for the default shadow stack. */
+ rdsspq %rax
movq %rax, %fs:SSP_BASE_OFFSET
- add $24, %RSP_LP
L(shadow_stack_bound_recorded):
/* If we unwind the stack, we can't undo stack unwinding. Just
diff --git a/sysdeps/x86/cpu-features.c b/sysdeps/x86/cpu-features.c
index 0bf923d48b..f180f0d9a4 100644
--- a/sysdeps/x86/cpu-features.c
+++ b/sysdeps/x86/cpu-features.c
@@ -1121,8 +1121,9 @@ no_cpuid:
# ifndef SHARED
/* Check if IBT and SHSTK are enabled by kernel. */
- if ((cet_status & GNU_PROPERTY_X86_FEATURE_1_IBT)
- || (cet_status & GNU_PROPERTY_X86_FEATURE_1_SHSTK))
+ if ((cet_status
+ & (GNU_PROPERTY_X86_FEATURE_1_IBT
+ | GNU_PROPERTY_X86_FEATURE_1_SHSTK)))
{
/* Disable IBT and/or SHSTK if they are enabled by kernel, but
disabled by environment variable:
@@ -1131,9 +1132,11 @@ no_cpuid:
*/
unsigned int cet_feature = 0;
if (!CPU_FEATURE_USABLE (IBT))
- cet_feature |= GNU_PROPERTY_X86_FEATURE_1_IBT;
+ cet_feature |= (cet_status
+ & GNU_PROPERTY_X86_FEATURE_1_IBT);
if (!CPU_FEATURE_USABLE (SHSTK))
- cet_feature |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
+ cet_feature |= (cet_status
+ & GNU_PROPERTY_X86_FEATURE_1_SHSTK);
if (cet_feature)
{
@@ -1148,7 +1151,9 @@ no_cpuid:
lock CET if IBT or SHSTK is enabled permissively. */
if (GL(dl_x86_feature_control).ibt != cet_permissive
&& GL(dl_x86_feature_control).shstk != cet_permissive)
- dl_cet_lock_cet ();
+ dl_cet_lock_cet (GL(dl_x86_feature_1)
+ & (GNU_PROPERTY_X86_FEATURE_1_IBT
+ | GNU_PROPERTY_X86_FEATURE_1_SHSTK));
}
# endif
}
diff --git a/sysdeps/x86/dl-cet.c b/sysdeps/x86/dl-cet.c
index e486e549be..66a78244d4 100644
--- a/sysdeps/x86/dl-cet.c
+++ b/sysdeps/x86/dl-cet.c
@@ -202,7 +202,7 @@ dl_cet_check_startup (struct link_map *m, struct dl_cet_info *info)
feature_1_lock |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
if (feature_1_lock != 0
- && dl_cet_lock_cet () != 0)
+ && dl_cet_lock_cet (feature_1_lock) != 0)
_dl_fatal_printf ("%s: can't lock CET\n", info->program);
}
diff --git a/sysdeps/x86_64/nptl/tls.h b/sysdeps/x86_64/nptl/tls.h
index 1403f939f7..4bcc2552a1 100644
--- a/sysdeps/x86_64/nptl/tls.h
+++ b/sysdeps/x86_64/nptl/tls.h
@@ -60,7 +60,7 @@ typedef struct
void *__private_tm[4];
/* GCC split stack support. */
void *__private_ss;
- /* The lowest address of shadow stack, */
+ /* The marker for the current shadow stack. */
unsigned long long int ssp_base;
/* Must be kept even if it is no longer used by glibc since programs,
like AddressSanitizer, depend on the size of tcbhead_t. */
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 2/6] elf: Always provide _dl_get_dl_main_map in libc.a
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
2023-12-29 16:43 ` [PATCH v6 3/6] x86/cet: Enable shadow stack during startup H.J. Lu
` (3 subsequent siblings)
5 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
Always provide _dl_get_dl_main_map in libc.a. It will be used by x86
to process PT_GNU_PROPERTY segment.
---
elf/dl-support.c | 2 --
sysdeps/generic/ldsodefs.h | 7 +++----
2 files changed, 3 insertions(+), 6 deletions(-)
diff --git a/elf/dl-support.c b/elf/dl-support.c
index 837fa1c836..70c5b3599a 100644
--- a/elf/dl-support.c
+++ b/elf/dl-support.c
@@ -344,7 +344,6 @@ _dl_non_dynamic_init (void)
DL_SYSINFO_IMPLEMENTATION
#endif
-#if ENABLE_STATIC_PIE
/* Since relocation to hidden _dl_main_map causes relocation overflow on
aarch64, a function is used to get the address of _dl_main_map. */
@@ -353,7 +352,6 @@ _dl_get_dl_main_map (void)
{
return &_dl_main_map;
}
-#endif
/* This is used by _dl_runtime_profile, not used on static code. */
void
diff --git a/sysdeps/generic/ldsodefs.h b/sysdeps/generic/ldsodefs.h
index 9b50ddd09f..bbf51bc1f5 100644
--- a/sysdeps/generic/ldsodefs.h
+++ b/sysdeps/generic/ldsodefs.h
@@ -1172,10 +1172,6 @@ void __libc_setup_tls (void);
# if ENABLE_STATIC_PIE
/* Relocate static executable with PIE. */
extern void _dl_relocate_static_pie (void) attribute_hidden;
-
-/* Get a pointer to _dl_main_map. */
-extern struct link_map * _dl_get_dl_main_map (void)
- __attribute__ ((visibility ("hidden")));
# else
# define _dl_relocate_static_pie()
# endif
@@ -1217,6 +1213,9 @@ rtld_hidden_proto (_dl_deallocate_tls)
extern void _dl_nothread_init_static_tls (struct link_map *) attribute_hidden;
+/* Get a pointer to _dl_main_map. */
+extern struct link_map * _dl_get_dl_main_map (void) attribute_hidden;
+
/* Find origin of the executable. */
extern const char *_dl_get_origin (void) attribute_hidden;
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 3/6] x86/cet: Enable shadow stack during startup
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
2023-12-29 16:43 ` [PATCH v6 2/6] elf: Always provide _dl_get_dl_main_map in libc.a H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
2024-01-08 19:50 ` Szabolcs Nagy
2023-12-29 16:43 ` [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK H.J. Lu
` (2 subsequent siblings)
5 siblings, 1 reply; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
Previously, CET was enabled by kernel before passing control to user
space and the startup code must disable CET if applications or shared
libraries aren't CET enabled. Since the current kernel only supports
shadow stack and won't enable shadow stack before passing control to
user space, we need to enable shadow stack during startup if the
application and all shared library are shadow stack enabled. There
is no need to disable shadow stack at startup. Shadow stack can only
be enabled in a function which will never return. Otherwise, shadow
stack will underflow at the function return.
1. GL(dl_x86_feature_1) is set to the CET features which are supported
by the processor and are not disabled by the tunable. Only non-zero
features in GL(dl_x86_feature_1) should be enabled. After enabling
shadow stack with ARCH_SHSTK_ENABLE, ARCH_SHSTK_STATUS is used to check
if shadow stack is really enabled.
2. Use ARCH_SHSTK_ENABLE in RTLD_START in dynamic executable. It is
safe since RTLD_START never returns.
3. Call arch_prctl (ARCH_SHSTK_ENABLE) from ARCH_SETUP_TLS in static
executable. Since the start function using ARCH_SETUP_TLS never returns,
it is safe to enable shadow stack in ARCH_SETUP_TLS.
---
sysdeps/unix/sysv/linux/x86/cpu-features.c | 49 --------------
sysdeps/unix/sysv/linux/x86/dl-cet.h | 27 +++++++-
sysdeps/unix/sysv/linux/x86_64/dl-cet.h | 47 +++++++++++++
sysdeps/x86/cpu-features-offsets.sym | 1 +
sysdeps/x86/cpu-features.c | 51 ---------------
sysdeps/x86/dl-cet.c | 76 +++++++++++-----------
sysdeps/x86/get-cpuid-feature-leaf.c | 2 +-
sysdeps/x86/include/cpu-features.h | 3 +
sysdeps/x86/libc-start.h | 53 ++++++++++++++-
sysdeps/x86_64/dl-machine.h | 12 +++-
10 files changed, 175 insertions(+), 146 deletions(-)
delete mode 100644 sysdeps/unix/sysv/linux/x86/cpu-features.c
create mode 100644 sysdeps/unix/sysv/linux/x86_64/dl-cet.h
diff --git a/sysdeps/unix/sysv/linux/x86/cpu-features.c b/sysdeps/unix/sysv/linux/x86/cpu-features.c
deleted file mode 100644
index 0e6e2bf855..0000000000
--- a/sysdeps/unix/sysv/linux/x86/cpu-features.c
+++ /dev/null
@@ -1,49 +0,0 @@
-/* Initialize CPU feature data for Linux/x86.
- This file is part of the GNU C Library.
- Copyright (C) 2018-2023 Free Software Foundation, Inc.
-
- The GNU C Library is free software; you can redistribute it and/or
- modify it under the terms of the GNU Lesser General Public
- License as published by the Free Software Foundation; either
- version 2.1 of the License, or (at your option) any later version.
-
- The GNU C Library is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- Lesser General Public License for more details.
-
- You should have received a copy of the GNU Lesser General Public
- License along with the GNU C Library; if not, see
- <https://www.gnu.org/licenses/>. */
-
-#if CET_ENABLED
-# include <sys/prctl.h>
-# include <asm/prctl.h>
-
-static inline int __attribute__ ((always_inline))
-get_cet_status (void)
-{
- unsigned long long kernel_feature;
- unsigned int status = 0;
- if (INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_STATUS,
- &kernel_feature) == 0)
- {
- if ((kernel_feature & ARCH_SHSTK_SHSTK) != 0)
- status = GNU_PROPERTY_X86_FEATURE_1_SHSTK;
- }
- return status;
-}
-
-# ifndef SHARED
-static inline void
-x86_setup_tls (void)
-{
- __libc_setup_tls ();
- THREAD_SETMEM (THREAD_SELF, header.feature_1, GL(dl_x86_feature_1));
-}
-
-# define ARCH_SETUP_TLS() x86_setup_tls ()
-# endif
-#endif
-
-#include <sysdeps/x86/cpu-features.c>
diff --git a/sysdeps/unix/sysv/linux/x86/dl-cet.h b/sysdeps/unix/sysv/linux/x86/dl-cet.h
index da220ac627..6ad5e03f69 100644
--- a/sysdeps/unix/sysv/linux/x86/dl-cet.h
+++ b/sysdeps/unix/sysv/linux/x86/dl-cet.h
@@ -18,7 +18,7 @@
#include <sys/prctl.h>
#include <asm/prctl.h>
-static inline int __attribute__ ((always_inline))
+static __always_inline int
dl_cet_disable_cet (unsigned int cet_feature)
{
if (cet_feature != GNU_PROPERTY_X86_FEATURE_1_SHSTK)
@@ -28,7 +28,7 @@ dl_cet_disable_cet (unsigned int cet_feature)
kernel_feature);
}
-static inline int __attribute__ ((always_inline))
+static __always_inline int
dl_cet_lock_cet (unsigned int cet_feature)
{
if (cet_feature != GNU_PROPERTY_X86_FEATURE_1_SHSTK)
@@ -38,3 +38,26 @@ dl_cet_lock_cet (unsigned int cet_feature)
return (int) INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_LOCK,
kernel_feature);
}
+
+static __always_inline unsigned int
+dl_cet_get_cet_status (void)
+{
+ unsigned long long kernel_feature;
+ unsigned int status = 0;
+ if (INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_STATUS,
+ &kernel_feature) == 0)
+ {
+ if ((kernel_feature & ARCH_SHSTK_SHSTK) != 0)
+ status = GNU_PROPERTY_X86_FEATURE_1_SHSTK;
+ }
+ return status;
+}
+
+/* Enable shadow stack with a macro to avoid shadow stack underflow. */
+#define ENABLE_X86_CET(cet_feature) \
+ if ((cet_feature & GNU_PROPERTY_X86_FEATURE_1_SHSTK)) \
+ { \
+ long long int kernel_feature = ARCH_SHSTK_SHSTK; \
+ INTERNAL_SYSCALL_CALL (arch_prctl, ARCH_SHSTK_ENABLE, \
+ kernel_feature); \
+ }
diff --git a/sysdeps/unix/sysv/linux/x86_64/dl-cet.h b/sysdeps/unix/sysv/linux/x86_64/dl-cet.h
new file mode 100644
index 0000000000..e23e05c6b8
--- /dev/null
+++ b/sysdeps/unix/sysv/linux/x86_64/dl-cet.h
@@ -0,0 +1,47 @@
+/* Linux/x86-64 CET initializers function.
+ Copyright (C) 2023 Free Software Foundation, Inc.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
+
+#include <cpu-features-offsets.h>
+#include_next <dl-cet.h>
+
+#define X86_STRINGIFY_1(x) #x
+#define X86_STRINGIFY(x) X86_STRINGIFY_1 (x)
+
+/* Enable shadow stack before calling _dl_init if it is enabled in
+ GL(dl_x86_feature_1). Call _dl_setup_x86_features to setup shadow
+ stack. */
+#define RTLD_START_ENABLE_X86_FEATURES \
+"\
+ # Check if shadow stack is enabled in GL(dl_x86_feature_1).\n\
+ movl _rtld_local+" X86_STRINGIFY (RTLD_GLOBAL_DL_X86_FEATURE_1_OFFSET) "(%rip), %edx\n\
+ testl $" X86_STRINGIFY (X86_FEATURE_1_SHSTK) ", %edx\n\
+ jz 1f\n\
+ # Enable shadow stack if enabled in GL(dl_x86_feature_1).\n\
+ movl $" X86_STRINGIFY (ARCH_SHSTK_SHSTK) ", %esi\n\
+ movl $" X86_STRINGIFY (ARCH_SHSTK_ENABLE) ", %edi\n\
+ movl $" X86_STRINGIFY (__NR_arch_prctl) ", %eax\n\
+ syscall\n\
+1:\n\
+ # Pass GL(dl_x86_feature_1) to _dl_cet_setup_features.\n\
+ movl %edx, %edi\n\
+ # Align stack for the _dl_cet_setup_features call.\n\
+ andq $-16, %rsp\n\
+ call _dl_cet_setup_features\n\
+ # Restore %rax and %rsp from %r12 and %r13.\n\
+ movq %r12, %rax\n\
+ movq %r13, %rsp\n\
+"
diff --git a/sysdeps/x86/cpu-features-offsets.sym b/sysdeps/x86/cpu-features-offsets.sym
index 6d03cea8e8..5429f60632 100644
--- a/sysdeps/x86/cpu-features-offsets.sym
+++ b/sysdeps/x86/cpu-features-offsets.sym
@@ -4,3 +4,4 @@
RTLD_GLOBAL_RO_DL_X86_CPU_FEATURES_OFFSET offsetof (struct rtld_global_ro, _dl_x86_cpu_features)
XSAVE_STATE_SIZE_OFFSET offsetof (struct cpu_features, xsave_state_size)
+RTLD_GLOBAL_DL_X86_FEATURE_1_OFFSET offsetof (struct rtld_global, _dl_x86_feature_1)
diff --git a/sysdeps/x86/cpu-features.c b/sysdeps/x86/cpu-features.c
index f180f0d9a4..097868c1d9 100644
--- a/sysdeps/x86/cpu-features.c
+++ b/sysdeps/x86/cpu-features.c
@@ -1106,57 +1106,6 @@ no_cpuid:
TUNABLE_CALLBACK (set_x86_ibt));
TUNABLE_GET (x86_shstk, tunable_val_t *,
TUNABLE_CALLBACK (set_x86_shstk));
-
- /* Check CET status. */
- unsigned int cet_status = get_cet_status ();
-
- if ((cet_status & GNU_PROPERTY_X86_FEATURE_1_IBT) == 0)
- CPU_FEATURE_UNSET (cpu_features, IBT)
- if ((cet_status & GNU_PROPERTY_X86_FEATURE_1_SHSTK) == 0)
- CPU_FEATURE_UNSET (cpu_features, SHSTK)
-
- if (cet_status)
- {
- GL(dl_x86_feature_1) = cet_status;
-
-# ifndef SHARED
- /* Check if IBT and SHSTK are enabled by kernel. */
- if ((cet_status
- & (GNU_PROPERTY_X86_FEATURE_1_IBT
- | GNU_PROPERTY_X86_FEATURE_1_SHSTK)))
- {
- /* Disable IBT and/or SHSTK if they are enabled by kernel, but
- disabled by environment variable:
-
- GLIBC_TUNABLES=glibc.cpu.hwcaps=-IBT,-SHSTK
- */
- unsigned int cet_feature = 0;
- if (!CPU_FEATURE_USABLE (IBT))
- cet_feature |= (cet_status
- & GNU_PROPERTY_X86_FEATURE_1_IBT);
- if (!CPU_FEATURE_USABLE (SHSTK))
- cet_feature |= (cet_status
- & GNU_PROPERTY_X86_FEATURE_1_SHSTK);
-
- if (cet_feature)
- {
- int res = dl_cet_disable_cet (cet_feature);
-
- /* Clear the disabled bits in dl_x86_feature_1. */
- if (res == 0)
- GL(dl_x86_feature_1) &= ~cet_feature;
- }
-
- /* Lock CET if IBT or SHSTK is enabled in executable. Don't
- lock CET if IBT or SHSTK is enabled permissively. */
- if (GL(dl_x86_feature_control).ibt != cet_permissive
- && GL(dl_x86_feature_control).shstk != cet_permissive)
- dl_cet_lock_cet (GL(dl_x86_feature_1)
- & (GNU_PROPERTY_X86_FEATURE_1_IBT
- | GNU_PROPERTY_X86_FEATURE_1_SHSTK));
- }
-# endif
- }
#endif
#ifndef SHARED
diff --git a/sysdeps/x86/dl-cet.c b/sysdeps/x86/dl-cet.c
index 66a78244d4..82167ebf50 100644
--- a/sysdeps/x86/dl-cet.c
+++ b/sysdeps/x86/dl-cet.c
@@ -173,40 +173,11 @@ dl_cet_check_startup (struct link_map *m, struct dl_cet_info *info)
= info->enable_feature_1 ^ info->feature_1_enabled;
if (disable_feature_1 != 0)
{
- /* Disable features in the kernel because of legacy objects or
- cet_always_off. */
- if (dl_cet_disable_cet (disable_feature_1) != 0)
- _dl_fatal_printf ("%s: can't disable x86 Features\n",
- info->program);
-
/* Clear the disabled bits. Sync dl_x86_feature_1 and
info->feature_1_enabled with info->enable_feature_1. */
info->feature_1_enabled = info->enable_feature_1;
GL(dl_x86_feature_1) = info->enable_feature_1;
}
-
- if (HAS_CPU_FEATURE (IBT) || HAS_CPU_FEATURE (SHSTK))
- {
- /* Lock CET features only if IBT or SHSTK are enabled and are not
- enabled permissively. */
- unsigned int feature_1_lock = 0;
-
- if (((info->feature_1_enabled & GNU_PROPERTY_X86_FEATURE_1_IBT)
- != 0)
- && info->enable_ibt_type != cet_permissive)
- feature_1_lock |= GNU_PROPERTY_X86_FEATURE_1_IBT;
-
- if (((info->feature_1_enabled & GNU_PROPERTY_X86_FEATURE_1_SHSTK)
- != 0)
- && info->enable_shstk_type != cet_permissive)
- feature_1_lock |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
-
- if (feature_1_lock != 0
- && dl_cet_lock_cet (feature_1_lock) != 0)
- _dl_fatal_printf ("%s: can't lock CET\n", info->program);
- }
-
- THREAD_SETMEM (THREAD_SELF, header.feature_1, GL(dl_x86_feature_1));
}
#endif
@@ -298,6 +269,15 @@ dl_cet_check (struct link_map *m, const char *program)
{
struct dl_cet_info info;
+ /* CET is enabled only if RTLD_START_ENABLE_X86_FEATURES is defined. */
+#if defined SHARED && defined RTLD_START_ENABLE_X86_FEATURES
+ /* Set dl_x86_feature_1 to features enabled in the executable. */
+ if (program != NULL)
+ GL(dl_x86_feature_1) = (m->l_x86_feature_1_and
+ & (X86_FEATURE_1_IBT
+ | X86_FEATURE_1_SHSTK));
+#endif
+
/* Check how IBT and SHSTK should be enabled. */
info.enable_ibt_type = GL(dl_x86_feature_control).ibt;
info.enable_shstk_type = GL(dl_x86_feature_control).shstk;
@@ -307,17 +287,9 @@ dl_cet_check (struct link_map *m, const char *program)
/* No legacy object check if IBT and SHSTK are always on. */
if (info.enable_ibt_type == cet_always_on
&& info.enable_shstk_type == cet_always_on)
- {
-#ifdef SHARED
- /* Set it only during startup. */
- if (program != NULL)
- THREAD_SETMEM (THREAD_SELF, header.feature_1,
- info.feature_1_enabled);
-#endif
- return;
- }
+ return;
- /* Check if IBT and SHSTK were enabled by kernel. */
+ /* Check if IBT and SHSTK were enabled. */
if (info.feature_1_enabled == 0)
return;
@@ -351,6 +323,32 @@ _dl_cet_open_check (struct link_map *l)
dl_cet_check (l, NULL);
}
+/* Set GL(dl_x86_feature_1) to the enabled features and clear the
+ active bits of the disabled features. */
+
+attribute_hidden void
+_dl_cet_setup_features (unsigned int cet_feature)
+{
+ /* NB: cet_feature == GL(dl_x86_feature_1) which is set to features
+ enabled from executable, not necessarily supported by kernel. */
+ if (cet_feature != 0)
+ {
+ cet_feature = dl_cet_get_cet_status ();
+ if (cet_feature != 0)
+ {
+ THREAD_SETMEM (THREAD_SELF, header.feature_1, cet_feature);
+
+ /* Lock CET if IBT or SHSTK is enabled in executable. Don't
+ lock CET if IBT or SHSTK is enabled permissively. */
+ if (GL(dl_x86_feature_control).ibt != cet_permissive
+ && (GL(dl_x86_feature_control).shstk != cet_permissive))
+ dl_cet_lock_cet (cet_feature);
+ }
+ /* Sync GL(dl_x86_feature_1) with kernel. */
+ GL(dl_x86_feature_1) = cet_feature;
+ }
+}
+
#ifdef SHARED
# ifndef LINKAGE
diff --git a/sysdeps/x86/get-cpuid-feature-leaf.c b/sysdeps/x86/get-cpuid-feature-leaf.c
index 40a46cc79c..9317a6b494 100644
--- a/sysdeps/x86/get-cpuid-feature-leaf.c
+++ b/sysdeps/x86/get-cpuid-feature-leaf.c
@@ -24,7 +24,7 @@ __x86_get_cpuid_feature_leaf (unsigned int leaf)
static const struct cpuid_feature feature = {};
if (leaf < CPUID_INDEX_MAX)
return ((const struct cpuid_feature *)
- &GLRO(dl_x86_cpu_features).features[leaf]);
+ &GLRO(dl_x86_cpu_features).features[leaf]);
else
return &feature;
}
diff --git a/sysdeps/x86/include/cpu-features.h b/sysdeps/x86/include/cpu-features.h
index 2d7427a6c0..23bd8146a2 100644
--- a/sysdeps/x86/include/cpu-features.h
+++ b/sysdeps/x86/include/cpu-features.h
@@ -990,6 +990,9 @@ extern const struct cpu_features *_dl_x86_get_cpu_features (void)
# define INIT_ARCH()
# define _dl_x86_get_cpu_features() (&GLRO(dl_x86_cpu_features))
extern void _dl_x86_init_cpu_features (void) attribute_hidden;
+
+extern void _dl_cet_setup_features (unsigned int)
+ attribute_hidden;
#endif
#ifdef __x86_64__
diff --git a/sysdeps/x86/libc-start.h b/sysdeps/x86/libc-start.h
index e93da6ef3d..7a91dd192a 100644
--- a/sysdeps/x86/libc-start.h
+++ b/sysdeps/x86/libc-start.h
@@ -19,7 +19,56 @@
#ifndef SHARED
# define ARCH_SETUP_IREL() apply_irel ()
# define ARCH_APPLY_IREL()
-# ifndef ARCH_SETUP_TLS
-# define ARCH_SETUP_TLS() __libc_setup_tls ()
+# ifdef __CET__
+/* Get CET features enabled in the static executable. */
+
+static inline unsigned int
+get_cet_feature (void)
+{
+ /* Check if CET is supported and not disabled by tunables. */
+ const struct cpu_features *cpu_features = __get_cpu_features ();
+ unsigned int cet_feature = 0;
+ if (CPU_FEATURE_USABLE_P (cpu_features, IBT))
+ cet_feature |= GNU_PROPERTY_X86_FEATURE_1_IBT;
+ if (CPU_FEATURE_USABLE_P (cpu_features, SHSTK))
+ cet_feature |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
+ if (!cet_feature)
+ return cet_feature;
+
+ struct link_map *main_map = _dl_get_dl_main_map ();
+
+ /* Scan program headers backward to check PT_GNU_PROPERTY early for
+ x86 feature bits on static executable. */
+ const ElfW(Phdr) *phdr = GL(dl_phdr);
+ const ElfW(Phdr) *ph;
+ for (ph = phdr + GL(dl_phnum); ph != phdr; ph--)
+ if (ph[-1].p_type == PT_GNU_PROPERTY)
+ {
+ _dl_process_pt_gnu_property (main_map, -1, &ph[-1]);
+ /* Enable IBT and SHSTK only if they are enabled on static
+ executable. */
+ cet_feature &= (main_map->l_x86_feature_1_and
+ & (GNU_PROPERTY_X86_FEATURE_1_IBT
+ | GNU_PROPERTY_X86_FEATURE_1_SHSTK));
+ /* Set GL(dl_x86_feature_1) to the enabled CET features. */
+ GL(dl_x86_feature_1) = cet_feature;
+ break;
+ }
+
+ return cet_feature;
+}
+
+/* The function using this macro to enable shadow stack must not return
+ to avoid shadow stack underflow. */
+# define ARCH_SETUP_TLS() \
+ { \
+ __libc_setup_tls (); \
+ \
+ unsigned int cet_feature = get_cet_feature (); \
+ ENABLE_X86_CET (cet_feature); \
+ _dl_cet_setup_features (cet_feature); \
+ }
+# else
+# define ARCH_SETUP_TLS() __libc_setup_tls ()
# endif
#endif /* !SHARED */
diff --git a/sysdeps/x86_64/dl-machine.h b/sysdeps/x86_64/dl-machine.h
index 581a2f1a9e..faeae723cb 100644
--- a/sysdeps/x86_64/dl-machine.h
+++ b/sysdeps/x86_64/dl-machine.h
@@ -29,6 +29,11 @@
#include <dl-static-tls.h>
#include <dl-machine-rel.h>
#include <isa-level.h>
+#ifdef __CET__
+# include <dl-cet.h>
+#else
+# define RTLD_START_ENABLE_X86_FEATURES
+#endif
/* Return nonzero iff ELF header is compatible with the running host. */
static inline int __attribute__ ((unused))
@@ -146,13 +151,16 @@ _start:\n\
_dl_start_user:\n\
# Save the user entry point address in %r12.\n\
movq %rax, %r12\n\
+ # Save %rsp value in %r13.\n\
+ movq %rsp, %r13\n\
+"\
+ RTLD_START_ENABLE_X86_FEATURES \
+"\
# Read the original argument count.\n\
movq (%rsp), %rdx\n\
# Call _dl_init (struct link_map *main_map, int argc, char **argv, char **env)\n\
# argc -> rsi\n\
movq %rdx, %rsi\n\
- # Save %rsp value in %r13.\n\
- movq %rsp, %r13\n\
# And align stack for the _dl_init call. \n\
andq $-16, %rsp\n\
# _dl_loaded -> rdi\n\
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
` (2 preceding siblings ...)
2023-12-29 16:43 ` [PATCH v6 3/6] x86/cet: Enable shadow stack during startup H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
2024-01-10 13:17 ` Joseph Myers
2023-12-29 16:43 ` [PATCH v6 5/6] x86/cet: Don't set CET active by default H.J. Lu
2023-12-29 16:43 ` [PATCH v6 6/6] x86/cet: Run some CET tests with shadow stack H.J. Lu
5 siblings, 1 reply; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
Initially, IBT and SHSTK are marked as active when CPU supports them
and CET are enabled in glibc. They can be disabled early by tunables
before relocation. Since after relocation, GLRO(dl_x86_cpu_features)
becomes read-only, we can't update GLRO(dl_x86_cpu_features) to mark
IBT and SHSTK as inactive. Instead, check the feature_1 field in TCB
to decide if IBT and SHST are active.
---
sysdeps/x86/bits/platform/x86.h | 8 ++++++++
sysdeps/x86/get-cpuid-feature-leaf.c | 11 ++++++++++-
sysdeps/x86/sys/platform/x86.h | 17 +++++++++++++++++
3 files changed, 35 insertions(+), 1 deletion(-)
diff --git a/sysdeps/x86/bits/platform/x86.h b/sysdeps/x86/bits/platform/x86.h
index 1e23d53ba2..1575ae53fb 100644
--- a/sysdeps/x86/bits/platform/x86.h
+++ b/sysdeps/x86/bits/platform/x86.h
@@ -337,3 +337,11 @@ enum
x86_cpu_AVX10_YMM = x86_cpu_index_24_ecx_0_ebx + 17,
x86_cpu_AVX10_ZMM = x86_cpu_index_24_ecx_0_ebx + 18,
};
+
+/* Bits in the feature_1 field in TCB. */
+
+enum
+{
+ x86_feature_1_ibt = 1U << 0,
+ x86_feature_1_shstk = 1U << 1
+};
diff --git a/sysdeps/x86/get-cpuid-feature-leaf.c b/sysdeps/x86/get-cpuid-feature-leaf.c
index 9317a6b494..f69936b31e 100644
--- a/sysdeps/x86/get-cpuid-feature-leaf.c
+++ b/sysdeps/x86/get-cpuid-feature-leaf.c
@@ -15,9 +15,18 @@
License along with the GNU C Library; if not, see
<https://www.gnu.org/licenses/>. */
-
+#include <assert.h>
+#include <tcb-offsets.h>
#include <ldsodefs.h>
+#ifdef __x86_64__
+# ifdef __LP64__
+_Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
+# else
+_Static_assert (FEATURE_1_OFFSET == 40, "FEATURE_1_OFFSET != 40");
+# endif
+#endif
+
const struct cpuid_feature *
__x86_get_cpuid_feature_leaf (unsigned int leaf)
{
diff --git a/sysdeps/x86/sys/platform/x86.h b/sysdeps/x86/sys/platform/x86.h
index 1ea2c5fc0b..89b1b16f22 100644
--- a/sysdeps/x86/sys/platform/x86.h
+++ b/sysdeps/x86/sys/platform/x86.h
@@ -45,6 +45,23 @@ x86_cpu_present (unsigned int __index)
static __inline__ _Bool
x86_cpu_active (unsigned int __index)
{
+ if (__index == x86_cpu_IBT || __index == x86_cpu_SHSTK)
+ {
+#ifdef __x86_64__
+ unsigned int __feature_1;
+# ifdef __LP64__
+ __asm__ ("mov %%fs:72, %0" : "=r" (__feature_1));
+# else
+ __asm__ ("mov %%fs:40, %0" : "=r" (__feature_1));
+# endif
+ if (__index == x86_cpu_IBT)
+ return __feature_1 & x86_feature_1_ibt;
+ else
+ return __feature_1 & x86_feature_1_shstk;
+#else
+ return false;
+#endif
+ }
const struct cpuid_feature *__ptr = __x86_get_cpuid_feature_leaf
(__index / (8 * sizeof (unsigned int) * 4));
unsigned int __reg
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 5/6] x86/cet: Don't set CET active by default
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
` (3 preceding siblings ...)
2023-12-29 16:43 ` [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
2023-12-29 16:43 ` [PATCH v6 6/6] x86/cet: Run some CET tests with shadow stack H.J. Lu
5 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
Not all CET enabled applications and libraries have been properly tested
in CET enabled environments. Some CET enabled applications or libraries
will crash or misbehave when CET is enabled. Don't set CET active by
default so that all applications and libraries will run normally regardless
of whether CET is active or not. Shadow stack can be enabled by
$ export GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
at run-time if shadow stack can be enabled by kernel.
NB: This commit can be reverted if it is OK to enable CET by default for
all applications and libraries.
---
sysdeps/x86/cpu-features.c | 2 +-
sysdeps/x86/cpu-tunables.c | 15 ++++++++++++++-
2 files changed, 15 insertions(+), 2 deletions(-)
diff --git a/sysdeps/x86/cpu-features.c b/sysdeps/x86/cpu-features.c
index 097868c1d9..80a07ac589 100644
--- a/sysdeps/x86/cpu-features.c
+++ b/sysdeps/x86/cpu-features.c
@@ -110,7 +110,7 @@ update_active (struct cpu_features *cpu_features)
if (!CPU_FEATURES_CPU_P (cpu_features, RTM_ALWAYS_ABORT))
CPU_FEATURE_SET_ACTIVE (cpu_features, RTM);
-#if CET_ENABLED
+#if CET_ENABLED && 0
CPU_FEATURE_SET_ACTIVE (cpu_features, IBT);
CPU_FEATURE_SET_ACTIVE (cpu_features, SHSTK);
#endif
diff --git a/sysdeps/x86/cpu-tunables.c b/sysdeps/x86/cpu-tunables.c
index 142c6b9240..1742400525 100644
--- a/sysdeps/x86/cpu-tunables.c
+++ b/sysdeps/x86/cpu-tunables.c
@@ -35,6 +35,17 @@
break; \
}
+#define CHECK_GLIBC_IFUNC_CPU_BOTH(f, cpu_features, name, len) \
+ _Static_assert (sizeof (#name) - 1 == len, #name " != " #len); \
+ if (tunable_str_comma_strcmp_cte (&f, #name)) \
+ { \
+ if (f.disable) \
+ CPU_FEATURE_UNSET (cpu_features, name) \
+ else \
+ CPU_FEATURE_SET_ACTIVE (cpu_features, name) \
+ break; \
+ }
+
/* Disable a preferred feature NAME. We don't enable a preferred feature
which isn't available. */
#define CHECK_GLIBC_IFUNC_PREFERRED_OFF(f, cpu_features, name, len) \
@@ -131,11 +142,13 @@ TUNABLE_CALLBACK (set_hwcaps) (tunable_val_t *valp)
}
break;
case 5:
+ {
+ CHECK_GLIBC_IFUNC_CPU_BOTH (n, cpu_features, SHSTK, 5);
+ }
if (n.disable)
{
CHECK_GLIBC_IFUNC_CPU_OFF (n, cpu_features, LZCNT, 5);
CHECK_GLIBC_IFUNC_CPU_OFF (n, cpu_features, MOVBE, 5);
- CHECK_GLIBC_IFUNC_CPU_OFF (n, cpu_features, SHSTK, 5);
CHECK_GLIBC_IFUNC_CPU_OFF (n, cpu_features, SSSE3, 5);
CHECK_GLIBC_IFUNC_CPU_OFF (n, cpu_features, XSAVE, 5);
}
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* [PATCH v6 6/6] x86/cet: Run some CET tests with shadow stack
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
` (4 preceding siblings ...)
2023-12-29 16:43 ` [PATCH v6 5/6] x86/cet: Don't set CET active by default H.J. Lu
@ 2023-12-29 16:43 ` H.J. Lu
5 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2023-12-29 16:43 UTC (permalink / raw)
To: libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
When CET is disabled by default, run some CET tests with shadow stack
enabled using
$ export GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
---
sysdeps/x86/Makefile | 14 ++++++++++++++
sysdeps/x86/tst-shstk-legacy-1e-static.sh | 1 +
sysdeps/x86/tst-shstk-legacy-1e.sh | 1 +
sysdeps/x86/tst-shstk-legacy-1g.sh | 1 +
4 files changed, 17 insertions(+)
diff --git a/sysdeps/x86/Makefile b/sysdeps/x86/Makefile
index 0aafd2afeb..a49b13c595 100644
--- a/sysdeps/x86/Makefile
+++ b/sysdeps/x86/Makefile
@@ -249,6 +249,13 @@ CFLAGS-tst-cet-legacy-10-static.c += -mshstk
CFLAGS-tst-cet-legacy-10a.c += -fcf-protection=none
CFLAGS-tst-cet-legacy-10a-static.c += -fcf-protection=none
+tst-cet-legacy-4-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-cet-legacy-6-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-cet-legacy-10-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-cet-legacy-10-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-cet-legacy-10a-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-cet-legacy-10a-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+
CFLAGS-tst-shstk-legacy-1a.c += -fcf-protection=none
CFLAGS-tst-shstk-legacy-1a-static.c += -fcf-protection=none
CFLAGS-tst-shstk-legacy-1d.c += -fcf-protection=none
@@ -288,14 +295,20 @@ tst-cet-legacy-6b-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=-IBT,-SHSTK
tst-cet-legacy-9-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=-IBT,-SHSTK
tst-cet-legacy-9-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=-IBT,-SHSTK
+tst-shstk-legacy-1a-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-shstk-legacy-1a-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
$(objpfx)tst-shstk-legacy-1a: $(objpfx)tst-shstk-legacy-1-extra.o
$(objpfx)tst-shstk-legacy-1a-static: $(objpfx)tst-shstk-legacy-1-extra.o
+tst-shstk-legacy-1b-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-shstk-legacy-1b-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
$(objpfx)tst-shstk-legacy-1b: $(objpfx)tst-shstk-legacy-1-extra.o
$(objpfx)tst-shstk-legacy-1b-static: $(objpfx)tst-shstk-legacy-1-extra.o
tst-shstk-legacy-1c-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=-SHSTK
tst-shstk-legacy-1c-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=-SHSTK
$(objpfx)tst-shstk-legacy-1c: $(objpfx)tst-shstk-legacy-1-extra.o
$(objpfx)tst-shstk-legacy-1c-static: $(objpfx)tst-shstk-legacy-1-extra.o
+tst-shstk-legacy-1d-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
+tst-shstk-legacy-1d-static-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
$(objpfx)tst-shstk-legacy-1d: $(objpfx)tst-shstk-legacy-1-extra.o
$(objpfx)tst-shstk-legacy-1d-static: $(objpfx)tst-shstk-legacy-1-extra.o
$(objpfx)tst-shstk-legacy-1e: $(objpfx)tst-shstk-legacy-1-extra.o
@@ -309,6 +322,7 @@ $(objpfx)tst-shstk-legacy-1e-static.out: \
$(objpfx)tst-shstk-legacy-1e-static
$(SHELL) $< $(common-objpfx) 2> $@; \
$(evaluate-test)
+tst-shstk-legacy-1f-ENV = GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK
$(objpfx)tst-shstk-legacy-1f: $(objpfx)tst-shstk-legacy-mod-1.so
$(objpfx)tst-shstk-legacy-mod-1.so: \
$(objpfx)tst-shstk-legacy-mod-1.os \
diff --git a/sysdeps/x86/tst-shstk-legacy-1e-static.sh b/sysdeps/x86/tst-shstk-legacy-1e-static.sh
index e943aec70e..008c50dae3 100755
--- a/sysdeps/x86/tst-shstk-legacy-1e-static.sh
+++ b/sysdeps/x86/tst-shstk-legacy-1e-static.sh
@@ -20,6 +20,7 @@
common_objpfx=$1; shift
+GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK \
${common_objpfx}elf/tst-shstk-legacy-1e-static
# The exit status should only be unsupported (77) or segfault (139).
status=$?
diff --git a/sysdeps/x86/tst-shstk-legacy-1e.sh b/sysdeps/x86/tst-shstk-legacy-1e.sh
index b0467aa899..82f2acbf75 100755
--- a/sysdeps/x86/tst-shstk-legacy-1e.sh
+++ b/sysdeps/x86/tst-shstk-legacy-1e.sh
@@ -21,6 +21,7 @@
common_objpfx=$1; shift
test_program_prefix=$1; shift
+GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK \
${test_program_prefix} \
${common_objpfx}elf/tst-shstk-legacy-1e
# The exit status should only be unsupported (77) or segfault (139).
diff --git a/sysdeps/x86/tst-shstk-legacy-1g.sh b/sysdeps/x86/tst-shstk-legacy-1g.sh
index c112bf6d8d..261eef7cac 100755
--- a/sysdeps/x86/tst-shstk-legacy-1g.sh
+++ b/sysdeps/x86/tst-shstk-legacy-1g.sh
@@ -21,6 +21,7 @@
common_objpfx=$1; shift
test_program_prefix=$1; shift
+GLIBC_TUNABLES=glibc.cpu.hwcaps=SHSTK \
${test_program_prefix} \
${common_objpfx}elf/tst-shstk-legacy-1g
# The exit status should only be unsupported (77) or segfault (139).
--
2.43.0
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
@ 2024-01-03 19:21 ` Adhemerval Zanella Netto
2024-01-03 19:32 ` H.J. Lu
2024-01-10 23:49 ` Edgecombe, Rick P
1 sibling, 1 reply; 16+ messages in thread
From: Adhemerval Zanella Netto @ 2024-01-03 19:21 UTC (permalink / raw)
To: H.J. Lu, libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
On 29/12/23 13:43, H.J. Lu wrote:
> Sync with Linux kernel 6.6 shadow stack interface. Since only x86-64 is
> supported, i386 shadow stack codes are unchanged and CET shouldn't be
> enabled for i386.
>
> 1. When the shadow stack base in TCB is unset, the default shadow stack
> is in use. Use the current shadow stack pointer as the marker for the
> default shadow stack. It is used to identify if the current shadow stack
> is the same as the target shadow stack when switching ucontexts. If yes,
> INCSSP will be used to unwind shadow stack. Otherwise, shadow stack
> restore token will be used.
> 2. Allocate shadow stack with the map_shadow_stack syscall. Since there
> is no function to explicitly release ucontext, there is no place to
> release shadow stack allocated by map_shadow_stack in ucontext functions.
> Such shadow stacks will be leaked.
> 3. Rename arch_prctl CET commands to ARCH_SHSTK_XXX.
> 4. Rewrite the CET control functions with the current kernel shadow stack
> interface.
>
> Since CET is no longer enabled by kernel, a separate patch will enable
> shadow stack during startup.
> ---
> diff --git a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> index 45ad0b052f..2f511321ad 100644
> --- a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> +++ b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> @@ -4,24 +4,19 @@
>
> #include_next <asm/prctl.h>
>
> -#ifndef ARCH_CET_STATUS
> -/* CET features:
> - IBT: GNU_PROPERTY_X86_FEATURE_1_IBT
> - SHSTK: GNU_PROPERTY_X86_FEATURE_1_SHSTK
> - */
> -/* Return CET features in unsigned long long *addr:
> - features: addr[0].
> - shadow stack base address: addr[1].
> - shadow stack size: addr[2].
> - */
> -# define ARCH_CET_STATUS 0x3001
> -/* Disable CET features in unsigned int features. */
> -# define ARCH_CET_DISABLE 0x3002
> -/* Lock all CET features. */
> -# define ARCH_CET_LOCK 0x3003
> -/* Allocate a new shadow stack with unsigned long long *addr:
> - IN: requested shadow stack size: *addr.
> - OUT: allocated shadow stack address: *addr.
> - */
> -# define ARCH_CET_ALLOC_SHSTK 0x3004
This broke i686 build with --enable-cet:
ld: [...]libc_pic.os.clean: warning: relocation against `ARCH_CET_ALLOC_SHSTK' in read-only section `.text'
[...]
ld: /home/azanella/Projects/glibc/build/i686-linux-gnu/libc_pic.os.clean: in function `getcontext':
[...]/glibc-git/stdlib/../sysdeps/unix/sysv/linux/i386/getcontext.S:65:(.text+0x1e12c): undefined reference to `ARCH_CET_STATUS'
[...]
Which leads the question whether CET is really supported on i386. The
commit message is not clear, the 'CET shouldn't be enabled for i386'
leads to the idea that --enable-cet should not be used for i686 and/or
is not supported. Is this the case?
If so, should we make --eanble-cet x86_64 only and remove the previous
i686 support (such as 5d844e1b72513cf59b5e7c14295644efdcc66e44)?
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface
2024-01-03 19:21 ` Adhemerval Zanella Netto
@ 2024-01-03 19:32 ` H.J. Lu
2024-01-03 20:14 ` H.J. Lu
0 siblings, 1 reply; 16+ messages in thread
From: H.J. Lu @ 2024-01-03 19:32 UTC (permalink / raw)
To: Adhemerval Zanella Netto; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
On Wed, Jan 3, 2024 at 11:21 AM Adhemerval Zanella Netto
<adhemerval.zanella@linaro.org> wrote:
>
>
>
> On 29/12/23 13:43, H.J. Lu wrote:
> > Sync with Linux kernel 6.6 shadow stack interface. Since only x86-64 is
> > supported, i386 shadow stack codes are unchanged and CET shouldn't be
> > enabled for i386.
> >
> > 1. When the shadow stack base in TCB is unset, the default shadow stack
> > is in use. Use the current shadow stack pointer as the marker for the
> > default shadow stack. It is used to identify if the current shadow stack
> > is the same as the target shadow stack when switching ucontexts. If yes,
> > INCSSP will be used to unwind shadow stack. Otherwise, shadow stack
> > restore token will be used.
> > 2. Allocate shadow stack with the map_shadow_stack syscall. Since there
> > is no function to explicitly release ucontext, there is no place to
> > release shadow stack allocated by map_shadow_stack in ucontext functions.
> > Such shadow stacks will be leaked.
> > 3. Rename arch_prctl CET commands to ARCH_SHSTK_XXX.
> > 4. Rewrite the CET control functions with the current kernel shadow stack
> > interface.
> >
> > Since CET is no longer enabled by kernel, a separate patch will enable
> > shadow stack during startup.
> > ---
>
> > diff --git a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > index 45ad0b052f..2f511321ad 100644
> > --- a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > +++ b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > @@ -4,24 +4,19 @@
> >
> > #include_next <asm/prctl.h>
> >
> > -#ifndef ARCH_CET_STATUS
> > -/* CET features:
> > - IBT: GNU_PROPERTY_X86_FEATURE_1_IBT
> > - SHSTK: GNU_PROPERTY_X86_FEATURE_1_SHSTK
> > - */
> > -/* Return CET features in unsigned long long *addr:
> > - features: addr[0].
> > - shadow stack base address: addr[1].
> > - shadow stack size: addr[2].
> > - */
> > -# define ARCH_CET_STATUS 0x3001
> > -/* Disable CET features in unsigned int features. */
> > -# define ARCH_CET_DISABLE 0x3002
> > -/* Lock all CET features. */
> > -# define ARCH_CET_LOCK 0x3003
> > -/* Allocate a new shadow stack with unsigned long long *addr:
> > - IN: requested shadow stack size: *addr.
> > - OUT: allocated shadow stack address: *addr.
> > - */
> > -# define ARCH_CET_ALLOC_SHSTK 0x3004
>
> This broke i686 build with --enable-cet:
>
> ld: [...]libc_pic.os.clean: warning: relocation against `ARCH_CET_ALLOC_SHSTK' in read-only section `.text'
> [...]
> ld: /home/azanella/Projects/glibc/build/i686-linux-gnu/libc_pic.os.clean: in function `getcontext':
> [...]/glibc-git/stdlib/../sysdeps/unix/sysv/linux/i386/getcontext.S:65:(.text+0x1e12c): undefined reference to `ARCH_CET_STATUS'
> [...]
>
> Which leads the question whether CET is really supported on i386. The
> commit message is not clear, the 'CET shouldn't be enabled for i386'
> leads to the idea that --enable-cet should not be used for i686 and/or
> is not supported. Is this the case?
Yes.
> If so, should we make --eanble-cet x86_64 only and remove the previous
> i686 support (such as 5d844e1b72513cf59b5e7c14295644efdcc66e44)?
We can move --enable-cet to x86_64. There are many i386 CET bits.
I think we should leave them alone for now.
--
H.J.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface
2024-01-03 19:32 ` H.J. Lu
@ 2024-01-03 20:14 ` H.J. Lu
0 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2024-01-03 20:14 UTC (permalink / raw)
To: Adhemerval Zanella Netto; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
On Wed, Jan 3, 2024 at 11:32 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Wed, Jan 3, 2024 at 11:21 AM Adhemerval Zanella Netto
> <adhemerval.zanella@linaro.org> wrote:
> >
> >
> >
> > On 29/12/23 13:43, H.J. Lu wrote:
> > > Sync with Linux kernel 6.6 shadow stack interface. Since only x86-64 is
> > > supported, i386 shadow stack codes are unchanged and CET shouldn't be
> > > enabled for i386.
> > >
> > > 1. When the shadow stack base in TCB is unset, the default shadow stack
> > > is in use. Use the current shadow stack pointer as the marker for the
> > > default shadow stack. It is used to identify if the current shadow stack
> > > is the same as the target shadow stack when switching ucontexts. If yes,
> > > INCSSP will be used to unwind shadow stack. Otherwise, shadow stack
> > > restore token will be used.
> > > 2. Allocate shadow stack with the map_shadow_stack syscall. Since there
> > > is no function to explicitly release ucontext, there is no place to
> > > release shadow stack allocated by map_shadow_stack in ucontext functions.
> > > Such shadow stacks will be leaked.
> > > 3. Rename arch_prctl CET commands to ARCH_SHSTK_XXX.
> > > 4. Rewrite the CET control functions with the current kernel shadow stack
> > > interface.
> > >
> > > Since CET is no longer enabled by kernel, a separate patch will enable
> > > shadow stack during startup.
> > > ---
> >
> > > diff --git a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > > index 45ad0b052f..2f511321ad 100644
> > > --- a/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > > +++ b/sysdeps/unix/sysv/linux/x86/include/asm/prctl.h
> > > @@ -4,24 +4,19 @@
> > >
> > > #include_next <asm/prctl.h>
> > >
> > > -#ifndef ARCH_CET_STATUS
> > > -/* CET features:
> > > - IBT: GNU_PROPERTY_X86_FEATURE_1_IBT
> > > - SHSTK: GNU_PROPERTY_X86_FEATURE_1_SHSTK
> > > - */
> > > -/* Return CET features in unsigned long long *addr:
> > > - features: addr[0].
> > > - shadow stack base address: addr[1].
> > > - shadow stack size: addr[2].
> > > - */
> > > -# define ARCH_CET_STATUS 0x3001
> > > -/* Disable CET features in unsigned int features. */
> > > -# define ARCH_CET_DISABLE 0x3002
> > > -/* Lock all CET features. */
> > > -# define ARCH_CET_LOCK 0x3003
> > > -/* Allocate a new shadow stack with unsigned long long *addr:
> > > - IN: requested shadow stack size: *addr.
> > > - OUT: allocated shadow stack address: *addr.
> > > - */
> > > -# define ARCH_CET_ALLOC_SHSTK 0x3004
> >
> > This broke i686 build with --enable-cet:
> >
> > ld: [...]libc_pic.os.clean: warning: relocation against `ARCH_CET_ALLOC_SHSTK' in read-only section `.text'
> > [...]
> > ld: /home/azanella/Projects/glibc/build/i686-linux-gnu/libc_pic.os.clean: in function `getcontext':
> > [...]/glibc-git/stdlib/../sysdeps/unix/sysv/linux/i386/getcontext.S:65:(.text+0x1e12c): undefined reference to `ARCH_CET_STATUS'
> > [...]
> >
> > Which leads the question whether CET is really supported on i386. The
> > commit message is not clear, the 'CET shouldn't be enabled for i386'
> > leads to the idea that --enable-cet should not be used for i686 and/or
> > is not supported. Is this the case?
>
> Yes.
>
> > If so, should we make --eanble-cet x86_64 only and remove the previous
> > i686 support (such as 5d844e1b72513cf59b5e7c14295644efdcc66e44)?
>
> We can move --enable-cet to x86_64. There are many i386 CET bits.
> I think we should leave them alone for now.
>
Here is the patch to ignore --enable-cet for i386:
https://patchwork.sourceware.org/project/glibc/list/?series=29098
--
H.J.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 3/6] x86/cet: Enable shadow stack during startup
2023-12-29 16:43 ` [PATCH v6 3/6] x86/cet: Enable shadow stack during startup H.J. Lu
@ 2024-01-08 19:50 ` Szabolcs Nagy
2024-01-08 22:29 ` H.J. Lu
0 siblings, 1 reply; 16+ messages in thread
From: Szabolcs Nagy @ 2024-01-08 19:50 UTC (permalink / raw)
To: H.J. Lu, libc-alpha; +Cc: rick.p.edgecombe, goldstein.w.n
The 12/29/2023 08:43, H.J. Lu wrote:
> Previously, CET was enabled by kernel before passing control to user
> space and the startup code must disable CET if applications or shared
> libraries aren't CET enabled. Since the current kernel only supports
> shadow stack and won't enable shadow stack before passing control to
> user space, we need to enable shadow stack during startup if the
> application and all shared library are shadow stack enabled.
not all shared libraries are checked for shadow stack compat: ld.so
and vdso are not checked (i.e. kernel mapped dsos other than the
main.exe) and now static linking is another special case that is
handled differently.
_dl_process_property_note is called for property notes in dsos that
are loaded via _dl_map_object_from_fd and for the main.exe. this
happens early at phdr processing time:
- it is not called for ld.so or vdso.
- x86 libc start code calls it for static main.exe.
- it is called twice for main.exe when explicitly loaded via ld.so.
- there may be special link maps (e.g. fake ld.so one for dlmopen)
those are not processed.
_rtld_main_check is called for main.exe and _dl_open_check is called
for a dlopened library after deps are loaded, but before relocation:
- these callbacks decide if loading the module is ok or not, so they
have to iterate over dependencies and check the result of the
previous property note processing. it's not clear what's the right
way iterate over deps (l_initfini have redundant entries and it is
missing in case of static linking, what's wrong with using
l_searchlist.r_list instead?).
- deps include ld.so which is not processed (x86 currently special
cases it during checks, but that means ld.so may be missing the
property note so e.g. we would not catch toolchain inconsistency
and one cannot force cet off by removing the marking from ld.so.)
- is it guaranteed that other than ld.so all deps are processed?
if not then map->l_x86_feature_1_and may be wrong when checked.
- x86 libc start code does not call these for static main.exe
instead replicates the logic (see my comment below).
i think cleaning this up would be useful: on aarch64 the note
processing has to do syscalls (to change BTI mappings) so if it
is called redundantly or missing that has significant impact.
and obviously it is useful to know which dsos are checked when
looking for security / reliability guarantees.
> --- a/sysdeps/x86/libc-start.h
> +++ b/sysdeps/x86/libc-start.h
> @@ -19,7 +19,56 @@
> #ifndef SHARED
> # define ARCH_SETUP_IREL() apply_irel ()
> # define ARCH_APPLY_IREL()
> -# ifndef ARCH_SETUP_TLS
> -# define ARCH_SETUP_TLS() __libc_setup_tls ()
> +# ifdef __CET__
> +/* Get CET features enabled in the static executable. */
> +
> +static inline unsigned int
> +get_cet_feature (void)
> +{
> + /* Check if CET is supported and not disabled by tunables. */
> + const struct cpu_features *cpu_features = __get_cpu_features ();
> + unsigned int cet_feature = 0;
> + if (CPU_FEATURE_USABLE_P (cpu_features, IBT))
> + cet_feature |= GNU_PROPERTY_X86_FEATURE_1_IBT;
> + if (CPU_FEATURE_USABLE_P (cpu_features, SHSTK))
> + cet_feature |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
> + if (!cet_feature)
> + return cet_feature;
> +
> + struct link_map *main_map = _dl_get_dl_main_map ();
> +
> + /* Scan program headers backward to check PT_GNU_PROPERTY early for
> + x86 feature bits on static executable. */
> + const ElfW(Phdr) *phdr = GL(dl_phdr);
> + const ElfW(Phdr) *ph;
> + for (ph = phdr + GL(dl_phnum); ph != phdr; ph--)
> + if (ph[-1].p_type == PT_GNU_PROPERTY)
> + {
> + _dl_process_pt_gnu_property (main_map, -1, &ph[-1]);
> + /* Enable IBT and SHSTK only if they are enabled on static
> + executable. */
> + cet_feature &= (main_map->l_x86_feature_1_and
> + & (GNU_PROPERTY_X86_FEATURE_1_IBT
> + | GNU_PROPERTY_X86_FEATURE_1_SHSTK));
> + /* Set GL(dl_x86_feature_1) to the enabled CET features. */
> + GL(dl_x86_feature_1) = cet_feature;
in theory this logic can be
_rtld_main_check (main_map, _dl_argv[0]);
except l_initfini is currently not set up for static main.exe
which may be an oversight.
> + break;
> + }
> +
> + return cet_feature;
> +}
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 3/6] x86/cet: Enable shadow stack during startup
2024-01-08 19:50 ` Szabolcs Nagy
@ 2024-01-08 22:29 ` H.J. Lu
0 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2024-01-08 22:29 UTC (permalink / raw)
To: Szabolcs Nagy; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
On Mon, Jan 8, 2024 at 11:50 AM Szabolcs Nagy <szabolcs.nagy@arm.com> wrote:
>
> The 12/29/2023 08:43, H.J. Lu wrote:
> > Previously, CET was enabled by kernel before passing control to user
> > space and the startup code must disable CET if applications or shared
> > libraries aren't CET enabled. Since the current kernel only supports
> > shadow stack and won't enable shadow stack before passing control to
> > user space, we need to enable shadow stack during startup if the
> > application and all shared library are shadow stack enabled.
>
> not all shared libraries are checked for shadow stack compat: ld.so
> and vdso are not checked (i.e. kernel mapped dsos other than the
Shadow stack enabled glibc must have shadow stack enabled
ld.so and shadow stack enabled kernel must have shadow stack
enabled vdso.
> main.exe) and now static linking is another special case that is
> handled differently.
>
>
> _dl_process_property_note is called for property notes in dsos that
> are loaded via _dl_map_object_from_fd and for the main.exe. this
> happens early at phdr processing time:
>
> - it is not called for ld.so or vdso.
> - x86 libc start code calls it for static main.exe.
> - it is called twice for main.exe when explicitly loaded via ld.so.
> - there may be special link maps (e.g. fake ld.so one for dlmopen)
> those are not processed.
>
> _rtld_main_check is called for main.exe and _dl_open_check is called
> for a dlopened library after deps are loaded, but before relocation:
>
> - these callbacks decide if loading the module is ok or not, so they
> have to iterate over dependencies and check the result of the
> previous property note processing. it's not clear what's the right
> way iterate over deps (l_initfini have redundant entries and it is
> missing in case of static linking, what's wrong with using
> l_searchlist.r_list instead?).
The current logic for dynamic executables is
1. Loading everything first.
2. Call _dl_process_property_note on each object and set
GL(dl_x86_feature_1) accordingly.
3. Enable features in GL(dl_x86_feature_1) in _dl_start_user.
4. Check if shadow stack is enabled and update
GL(dl_x86_feature_1) accordingly.
> - deps include ld.so which is not processed (x86 currently special
> cases it during checks, but that means ld.so may be missing the
> property note so e.g. we would not catch toolchain inconsistency
> and one cannot force cet off by removing the marking from ld.so.)
ld.so must have shadow stack enabled. If not, it is a glibc or toolchain
bug.
> - is it guaranteed that other than ld.so all deps are processed?
Yes.
> if not then map->l_x86_feature_1_and may be wrong when checked.
> - x86 libc start code does not call these for static main.exe
> instead replicates the logic (see my comment below).
Since sysdeps/x86/libc-start.h is only for static executables,
it only needs to check PT_GNU_PROPERTY in executables.
> i think cleaning this up would be useful: on aarch64 the note
> processing has to do syscalls (to change BTI mappings) so if it
> is called redundantly or missing that has significant impact.
> and obviously it is useful to know which dsos are checked when
> looking for security / reliability guarantees.
>
>
> > --- a/sysdeps/x86/libc-start.h
> > +++ b/sysdeps/x86/libc-start.h
> > @@ -19,7 +19,56 @@
> > #ifndef SHARED
> > # define ARCH_SETUP_IREL() apply_irel ()
> > # define ARCH_APPLY_IREL()
> > -# ifndef ARCH_SETUP_TLS
> > -# define ARCH_SETUP_TLS() __libc_setup_tls ()
> > +# ifdef __CET__
> > +/* Get CET features enabled in the static executable. */
> > +
> > +static inline unsigned int
> > +get_cet_feature (void)
> > +{
> > + /* Check if CET is supported and not disabled by tunables. */
> > + const struct cpu_features *cpu_features = __get_cpu_features ();
> > + unsigned int cet_feature = 0;
> > + if (CPU_FEATURE_USABLE_P (cpu_features, IBT))
> > + cet_feature |= GNU_PROPERTY_X86_FEATURE_1_IBT;
> > + if (CPU_FEATURE_USABLE_P (cpu_features, SHSTK))
> > + cet_feature |= GNU_PROPERTY_X86_FEATURE_1_SHSTK;
> > + if (!cet_feature)
> > + return cet_feature;
> > +
> > + struct link_map *main_map = _dl_get_dl_main_map ();
> > +
> > + /* Scan program headers backward to check PT_GNU_PROPERTY early for
> > + x86 feature bits on static executable. */
> > + const ElfW(Phdr) *phdr = GL(dl_phdr);
> > + const ElfW(Phdr) *ph;
> > + for (ph = phdr + GL(dl_phnum); ph != phdr; ph--)
> > + if (ph[-1].p_type == PT_GNU_PROPERTY)
> > + {
> > + _dl_process_pt_gnu_property (main_map, -1, &ph[-1]);
> > + /* Enable IBT and SHSTK only if they are enabled on static
> > + executable. */
> > + cet_feature &= (main_map->l_x86_feature_1_and
> > + & (GNU_PROPERTY_X86_FEATURE_1_IBT
> > + | GNU_PROPERTY_X86_FEATURE_1_SHSTK));
> > + /* Set GL(dl_x86_feature_1) to the enabled CET features. */
> > + GL(dl_x86_feature_1) = cet_feature;
>
> in theory this logic can be
>
> _rtld_main_check (main_map, _dl_argv[0]);
Does _rtld_main_check work on static executables?
> except l_initfini is currently not set up for static main.exe
> which may be an oversight.
>
> > + break;
> > + }
> > +
> > + return cet_feature;
> > +}
--
H.J.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK
2023-12-29 16:43 ` [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK H.J. Lu
@ 2024-01-10 13:17 ` Joseph Myers
2024-01-10 16:32 ` H.J. Lu
0 siblings, 1 reply; 16+ messages in thread
From: Joseph Myers @ 2024-01-10 13:17 UTC (permalink / raw)
To: H.J. Lu; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
[-- Attachment #1: Type: text/plain, Size: 2012 bytes --]
On Fri, 29 Dec 2023, H.J. Lu wrote:
> diff --git a/sysdeps/x86/get-cpuid-feature-leaf.c b/sysdeps/x86/get-cpuid-feature-leaf.c
> index 9317a6b494..f69936b31e 100644
> --- a/sysdeps/x86/get-cpuid-feature-leaf.c
> +++ b/sysdeps/x86/get-cpuid-feature-leaf.c
> @@ -15,9 +15,18 @@
> License along with the GNU C Library; if not, see
> <https://www.gnu.org/licenses/>. */
>
> -
> +#include <assert.h>
> +#include <tcb-offsets.h>
> #include <ldsodefs.h>
>
> +#ifdef __x86_64__
> +# ifdef __LP64__
> +_Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> +# else
> +_Static_assert (FEATURE_1_OFFSET == 40, "FEATURE_1_OFFSET != 40");
> +# endif
> +#endif
I think this has broken the build for x86_64-gnu (Hurd).
In file included from ../include/features.h:503,
from ../assert/assert.h:35,
from ../include/assert.h:1,
from ../sysdeps/x86/get-cpuid-feature-leaf.c:18:
../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: ‘FEATURE_1_OFFSET’ undeclared here (not in a function)
24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
| ^~~~~~~~~~~~~~~~
../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
| ^~~~
../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: expression in static assertion is not an integer
24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
| ^~~~~~~~~~~~~~~~
../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
| ^~~~
Note that FEATURE_1_OFFSET is defined in
sysdeps/x86_64/nptl/tcb-offsets.sym - which is NPTL-specific, so not used
for Hurd.
--
Joseph S. Myers
josmyers@redhat.com
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK
2024-01-10 13:17 ` Joseph Myers
@ 2024-01-10 16:32 ` H.J. Lu
2024-01-10 16:50 ` H.J. Lu
0 siblings, 1 reply; 16+ messages in thread
From: H.J. Lu @ 2024-01-10 16:32 UTC (permalink / raw)
To: Joseph Myers; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
On Wed, Jan 10, 2024 at 5:20 AM Joseph Myers <josmyers@redhat.com> wrote:
>
> On Fri, 29 Dec 2023, H.J. Lu wrote:
>
> > diff --git a/sysdeps/x86/get-cpuid-feature-leaf.c b/sysdeps/x86/get-cpuid-feature-leaf.c
> > index 9317a6b494..f69936b31e 100644
> > --- a/sysdeps/x86/get-cpuid-feature-leaf.c
> > +++ b/sysdeps/x86/get-cpuid-feature-leaf.c
> > @@ -15,9 +15,18 @@
> > License along with the GNU C Library; if not, see
> > <https://www.gnu.org/licenses/>. */
> >
> > -
> > +#include <assert.h>
> > +#include <tcb-offsets.h>
> > #include <ldsodefs.h>
> >
> > +#ifdef __x86_64__
> > +# ifdef __LP64__
> > +_Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> > +# else
> > +_Static_assert (FEATURE_1_OFFSET == 40, "FEATURE_1_OFFSET != 40");
> > +# endif
> > +#endif
>
> I think this has broken the build for x86_64-gnu (Hurd).
>
> In file included from ../include/features.h:503,
> from ../assert/assert.h:35,
> from ../include/assert.h:1,
> from ../sysdeps/x86/get-cpuid-feature-leaf.c:18:
> ../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: ‘FEATURE_1_OFFSET’ undeclared here (not in a function)
> 24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> | ^~~~~~~~~~~~~~~~
> ../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
> 7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
> | ^~~~
> ../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: expression in static assertion is not an integer
> 24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> | ^~~~~~~~~~~~~~~~
> ../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
> 7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
> | ^~~~
>
> Note that FEATURE_1_OFFSET is defined in
> sysdeps/x86_64/nptl/tcb-offsets.sym - which is NPTL-specific, so not used
> for Hurd.
>
Testing a fix.
Thanks.
--
H.J.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK
2024-01-10 16:32 ` H.J. Lu
@ 2024-01-10 16:50 ` H.J. Lu
0 siblings, 0 replies; 16+ messages in thread
From: H.J. Lu @ 2024-01-10 16:50 UTC (permalink / raw)
To: Joseph Myers; +Cc: libc-alpha, rick.p.edgecombe, goldstein.w.n
On Wed, Jan 10, 2024 at 8:32 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Wed, Jan 10, 2024 at 5:20 AM Joseph Myers <josmyers@redhat.com> wrote:
> >
> > On Fri, 29 Dec 2023, H.J. Lu wrote:
> >
> > > diff --git a/sysdeps/x86/get-cpuid-feature-leaf.c b/sysdeps/x86/get-cpuid-feature-leaf.c
> > > index 9317a6b494..f69936b31e 100644
> > > --- a/sysdeps/x86/get-cpuid-feature-leaf.c
> > > +++ b/sysdeps/x86/get-cpuid-feature-leaf.c
> > > @@ -15,9 +15,18 @@
> > > License along with the GNU C Library; if not, see
> > > <https://www.gnu.org/licenses/>. */
> > >
> > > -
> > > +#include <assert.h>
> > > +#include <tcb-offsets.h>
> > > #include <ldsodefs.h>
> > >
> > > +#ifdef __x86_64__
> > > +# ifdef __LP64__
> > > +_Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> > > +# else
> > > +_Static_assert (FEATURE_1_OFFSET == 40, "FEATURE_1_OFFSET != 40");
> > > +# endif
> > > +#endif
> >
> > I think this has broken the build for x86_64-gnu (Hurd).
> >
> > In file included from ../include/features.h:503,
> > from ../assert/assert.h:35,
> > from ../include/assert.h:1,
> > from ../sysdeps/x86/get-cpuid-feature-leaf.c:18:
> > ../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: ‘FEATURE_1_OFFSET’ undeclared here (not in a function)
> > 24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> > | ^~~~~~~~~~~~~~~~
> > ../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
> > 7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
> > | ^~~~
> > ../sysdeps/x86/get-cpuid-feature-leaf.c:24:17: error: expression in static assertion is not an integer
> > 24 | _Static_assert (FEATURE_1_OFFSET == 72, "FEATURE_1_OFFSET != 72");
> > | ^~~~~~~~~~~~~~~~
> > ../include/sys/cdefs.h:7:59: note: in definition of macro ‘_Static_assert’
> > 7 | # define _Static_assert(expr, diagnostic) _Static_assert (expr, diagnostic)
> > | ^~~~
> >
> > Note that FEATURE_1_OFFSET is defined in
> > sysdeps/x86_64/nptl/tcb-offsets.sym - which is NPTL-specific, so not used
> > for Hurd.
> >
>
> Testing a fix.
>
Here is the patch:
https://patchwork.sourceware.org/project/glibc/patch/20240110164847.3304707-1-hjl.tools@gmail.com/
--
H.J.
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
2024-01-03 19:21 ` Adhemerval Zanella Netto
@ 2024-01-10 23:49 ` Edgecombe, Rick P
1 sibling, 0 replies; 16+ messages in thread
From: Edgecombe, Rick P @ 2024-01-10 23:49 UTC (permalink / raw)
To: libc-alpha, hjl.tools; +Cc: goldstein.w.n
On Fri, 2023-12-29 at 08:43 -0800, H.J. Lu wrote:
> + long int ret
> + = __allocate_shadow_stack (((uintptr_t) sp
> + - (uintptr_t) ucp-
> >uc_stack.ss_sp),
> + &ucp->__ssp[1]);
> + if (ret != 0)
> + {
> + /* FIXME: What should we do? */
> + abort ();
> + }
This is really because makecontext() doesn't normally do any
allocations. The fix should be a new API that can return something for
this failure, and also provides a mechanism to free the shadow stack.
Or take a pointer to a shadow stack token, with the shadow stack
allocated and managed by the caller.
That wouldn't support existing apps seamlessly, but leaking can't be
the long term solution. I even question it as a short term solution.
> +
> + ucp->__ssp[0] = ucp->__ssp[1] + ucp->__ssp[2] - 8;
> + /* Call __push___start_context to push __start_context onto
> the new
> + stack as well as the new shadow stack. */
> __push___start_context (ucp);
^ permalink raw reply [flat|nested] 16+ messages in thread
end of thread, other threads:[~2024-01-10 23:49 UTC | newest]
Thread overview: 16+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-12-29 16:43 [PATCH v6 0/6] x86/cet: Update CET kernel interface H.J. Lu
2023-12-29 16:43 ` [PATCH v6 1/6] x86/cet: Sync with Linux kernel 6.6 shadow stack interface H.J. Lu
2024-01-03 19:21 ` Adhemerval Zanella Netto
2024-01-03 19:32 ` H.J. Lu
2024-01-03 20:14 ` H.J. Lu
2024-01-10 23:49 ` Edgecombe, Rick P
2023-12-29 16:43 ` [PATCH v6 2/6] elf: Always provide _dl_get_dl_main_map in libc.a H.J. Lu
2023-12-29 16:43 ` [PATCH v6 3/6] x86/cet: Enable shadow stack during startup H.J. Lu
2024-01-08 19:50 ` Szabolcs Nagy
2024-01-08 22:29 ` H.J. Lu
2023-12-29 16:43 ` [PATCH v6 4/6] x86/cet: Check feature_1 in TCB for active IBT and SHSTK H.J. Lu
2024-01-10 13:17 ` Joseph Myers
2024-01-10 16:32 ` H.J. Lu
2024-01-10 16:50 ` H.J. Lu
2023-12-29 16:43 ` [PATCH v6 5/6] x86/cet: Don't set CET active by default H.J. Lu
2023-12-29 16:43 ` [PATCH v6 6/6] x86/cet: Run some CET tests with shadow stack H.J. Lu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).