Merge tag 'x86_sgx_for_v5.11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 SGC support from Borislav Petkov:
 "Intel Software Guard eXtensions enablement. This has been long in the
  making, we were one revision number short of 42. :)

  Intel SGX is new hardware functionality that can be used by
  applications to populate protected regions of user code and data
  called enclaves. Once activated, the new hardware protects enclave
  code and data from outside access and modification.

  Enclaves provide a place to store secrets and process data with those
  secrets. SGX has been used, for example, to decrypt video without
  exposing the decryption keys to nosy debuggers that might be used to
  subvert DRM. Software has generally been rewritten specifically to run
  in enclaves, but there are also projects that try to run limited
  unmodified software in enclaves.

  Most of the functionality is concentrated into arch/x86/kernel/cpu/sgx/
  except the addition of a new mprotect() hook to control enclave page
  permissions and support for vDSO exceptions fixup which will is used
  by SGX enclaves.

  All this work by Sean Christopherson, Jarkko Sakkinen and many others"

* tag 'x86_sgx_for_v5.11' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (30 commits)
  x86/sgx: Return -EINVAL on a zero length buffer in sgx_ioc_enclave_add_pages()
  x86/sgx: Fix a typo in kernel-doc markup
  x86/sgx: Fix sgx_ioc_enclave_provision() kernel-doc comment
  x86/sgx: Return -ERESTARTSYS in sgx_ioc_enclave_add_pages()
  selftests/sgx: Use a statically generated 3072-bit RSA key
  x86/sgx: Clarify 'laundry_list' locking
  x86/sgx: Update MAINTAINERS
  Documentation/x86: Document SGX kernel architecture
  x86/sgx: Add ptrace() support for the SGX driver
  x86/sgx: Add a page reclaimer
  selftests/x86: Add a selftest for SGX
  x86/vdso: Implement a vDSO for Intel SGX enclave call
  x86/traps: Attempt to fixup exceptions in vDSO before signaling
  x86/fault: Add a helper function to sanitize error code
  x86/vdso: Add support for exception fixup in vDSO functions
  x86/sgx: Add SGX_IOC_ENCLAVE_PROVISION
  x86/sgx: Add SGX_IOC_ENCLAVE_INIT
  x86/sgx: Add SGX_IOC_ENCLAVE_ADD_PAGES
  x86/sgx: Add SGX_IOC_ENCLAVE_CREATE
  x86/sgx: Add an SGX misc driver interface
  ...

1 files changed, 33 insertions, 12 deletions

diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 82bf37a5c9ec..f1f1b5a0956a 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -30,6 +30,7 @@
 #include <asm/cpu_entry_area.h>		/* exception stack		*/
 #include <asm/pgtable_areas.h>		/* VMALLOC_START, ...		*/
 #include <asm/kvm_para.h>		/* kvm_handle_async_pf		*/
+#include <asm/vdso.h>			/* fixup_vdso_exception()	*/
 
 #define CREATE_TRACE_POINTS
 #include <asm/trace/exceptions.h>
@@ -602,11 +603,9 @@ pgtable_bad(struct pt_regs *regs, unsigned long error_code,
 	oops_end(flags, regs, sig);
 }
 
-static void set_signal_archinfo(unsigned long address,
-				unsigned long error_code)
+static void sanitize_error_code(unsigned long address,
+				unsigned long *error_code)
 {
-	struct task_struct *tsk = current;
-
 	/*
 	 * To avoid leaking information about the kernel page
 	 * table layout, pretend that user-mode accesses to
@@ -617,7 +616,13 @@ static void set_signal_archinfo(unsigned long address,
 	 * information and does not appear to cause any problems.
 	 */
 	if (address >= TASK_SIZE_MAX)
-		error_code |= X86_PF_PROT;
+		*error_code |= X86_PF_PROT;
+}
+
+static void set_signal_archinfo(unsigned long address,
+				unsigned long error_code)
+{
+	struct task_struct *tsk = current;
 
 	tsk->thread.trap_nr = X86_TRAP_PF;
 	tsk->thread.error_code = error_code | X86_PF_USER;
@@ -658,6 +663,8 @@ no_context(struct pt_regs *regs, unsigned long error_code,
 		 * faulting through the emulate_vsyscall() logic.
 		 */
 		if (current->thread.sig_on_uaccess_err && signal) {
+			sanitize_error_code(address, &error_code);
+
 			set_signal_archinfo(address, error_code);
 
 			/* XXX: hwpoison faults will set the wrong code. */
@@ -806,13 +813,10 @@ __bad_area_nosemaphore(struct pt_regs *regs, unsigned long error_code,
 		if (is_errata100(regs, address))
 			return;
 
-		/*
-		 * To avoid leaking information about the kernel page table
-		 * layout, pretend that user-mode accesses to kernel addresses
-		 * are always protection faults.
-		 */
-		if (address >= TASK_SIZE_MAX)
-			error_code |= X86_PF_PROT;
+		sanitize_error_code(address, &error_code);
+
+		if (fixup_vdso_exception(regs, X86_TRAP_PF, error_code, address))
+			return;
 
 		if (likely(show_unhandled_signals))
 			show_signal_msg(regs, error_code, address, tsk);
@@ -931,6 +935,11 @@ do_sigbus(struct pt_regs *regs, unsigned long error_code, unsigned long address,
 	if (is_prefetch(regs, error_code, address))
 		return;
 
+	sanitize_error_code(address, &error_code);
+
+	if (fixup_vdso_exception(regs, X86_TRAP_PF, error_code, address))
+		return;
+
 	set_signal_archinfo(address, error_code);
 
 #ifdef CONFIG_MEMORY_FAILURE
@@ -1102,6 +1111,18 @@ access_error(unsigned long error_code, struct vm_area_struct *vma)
 		return 1;
 
 	/*
+	 * SGX hardware blocked the access.  This usually happens
+	 * when the enclave memory contents have been destroyed, like
+	 * after a suspend/resume cycle. In any case, the kernel can't
+	 * fix the cause of the fault.  Handle the fault as an access
+	 * error even in cases where no actual access violation
+	 * occurred.  This allows userspace to rebuild the enclave in
+	 * response to the signal.
+	 */
+	if (unlikely(error_code & X86_PF_SGX))
+		return 1;
+
+	/*
 	 * Make sure to check the VMA so that we do not perform
 	 * faults just to hit a X86_PF_PK as soon as we fill in a
 	 * page.