Skip to contentSkip to conversation

Intel SGX1 support

PATCH Superseded Jarkko Sakkinen 11 participants 76 messages Sep 25, 2018 – Nov 1, 2018

Subsystems (7)

v14 of 40 Superseded Latest is v40 →

lore
Share & tools
lore
#0 lore raw

v14 cover Intel SGX1 support

Intel(R) SGX is a set of CPU instructions that can be used by applications to set aside private regions of code and data. The code outside the enclave is disallowed to access the memory inside the enclave by the CPU access control. In a way you can think that SGX provides inverted sandbox. It protects the application from a malicious host.

There is a new hardware unit in the processor called Memory Encryption Engine (MEE) starting from the Skylake microacrhitecture. BIOS can define one or many MEE regions that can hold enclave data by configuring them with PRMRR registers.

The MEE automatically encrypts the data leaving the processor package to the MEE regions. The data is encrypted using a random key whose life-time is exactly one power cycle.

The current implementation requires that the firmware sets IA32_SGXLEPUBKEYHASH* MSRs as writable so that ultimately the kernel can decide what enclaves it wants run. The implementation does not create any bottlenecks to support read-only MSRs later on.

You can tell if your CPU supports SGX by looking into /proc/cpuinfo:

cat /proc/cpuinfo | grep sgx

v14:
* Change the comment about X86_FEATURE_SGX_LC from “SGX launch configuration” to “SGX launch control”.
* Move the SGX-related CPU feature flags as part of the Linux defined virtual leaf 8.
* Add SGX_ prefix to the constants defining the ENCLS leaf functions.
* Use GENMASK*() and BIT*() in sgx_arch.h instead of raw hex numbers.
* Refine the long description for CONFIG_INTEL_SGX_CORE.
* Do not use pr_*_ratelimited() in the driver. The use of the rate limited versions is legacy cruft from the prototyping phase.
* Detect sleep with SGX_INVALID_EINIT_TOKEN instead of counting power cycles.
* Manually prefix with “sgx:” in the core SGX code instead of redefining pr_fmt.
* Report if IA32_SGXLEPUBKEYHASHx MSRs are not writable in the driver instead of core because it is a driver requirement.
* Change prompt to bool in the entry for CONFIG_INTEL_SGX_CORE because the default is ‘n’.
* Rename struct sgx_epc_bank as struct sgx_epc_section in order to match the SDM.
* Allocate struct sgx_epc_page instances one at a time.
* Use “__iomem void *” pointers for the mapped EPC memory consistently.
* Retry once on SGX_INVALID_TOKEN in sgx_einit() instead of counting power cycles.
* Call enclave swapping operations directly from the driver instead of calling them .indirectly through struct sgx_epc_page_ops because indirect calls are not required yet as the patch set does not contain the KVM support.
* Added special signal SEGV_SGXERR to notify about SGX EPCM violation errors.

v13:
* Always use SGX_CPUID constant instead of a hardcoded value.
* Simplified and documented the macros and functions for ENCLS leaves.
* Enable sgx_free_page() to free active enclave pages on demand in order to allow sgx_invalidate() to delete enclave pages. It no longer performs EREMOVE if a page is in the process of being reclaimed.
* Use PM notifier per enclave so that we don't have to traverse the global list of active EPC pages to find enclaves.
* Removed unused SGX_LE_ROLLBACK constant from uapi/asm/sgx.h
* Always use ioremap() to map EPC banks as we only support 64-bit kernel.
* Invalidate IA32_SGXLEPUBKEYHASH cache used by sgx_einit() when going to sleep.

v12:
* Split to more narrow scoped commits in order to ease the review process and use co-developed-by tag for co-authors of commits instead of listing them in the source files.
* Removed cruft EXPORT_SYMBOL() declarations and converted to static variables.
* Removed in-kernel LE i.e. this version of the SGX software stack only supports unlocked IA32_SGXLEPUBKEYHASHx MSRs.
* Refined documentation on launching enclaves, swapping and enclave construction.
* Refined sgx_arch.h to include alignment information for every struct that requires it and removed structs that are not needed without an LE.
* Got rid of SGX_CPUID.
* SGX detection now prints log messages about firmware configuration issues.

v11:
* Polished ENCLS wrappers with refined exception handling.
* ksgxswapd was not stopped (regression in v5) in sgx_page_cache_teardown(), which causes a leaked kthread after driver deinitialization.
* Shutdown sgx_le_proxy when going to suspend because its EPC pages will be invalidated when resuming, which will cause it not function properly anymore.
* Set EINITTOKEN.VALID to zero for a token that is passed when SGXLEPUBKEYHASH matches MRSIGNER as alloc_page() does not give a zero page.
* Fixed the check in sgx_edbgrd() for a TCS page. Allowed to read offsets around the flags field, which causes a #GP. Only flags read is readable.
* On read access memcpy() call inside sgx_vma_access() had src and dest parameters in wrong order.
* The build issue with CONFIG_KASAN is now fixed. Added undefined symbols to LE even if “KASAN_SANITIZE := false” was set in the makefile.
* Fixed a regression in the #PF handler. If a page has SGX_ENCL_PAGE_RESERVED flag the #PF handler should unconditionally fail. It did not, which caused weird races when trying to change other parts of swapping code.
* EPC management has been refactored to a flat LRU cache and moved to arch/x86. The swapper thread reads a cluster of EPC pages and swaps all of them. It can now swap from multiple enclaves in the same round.
* For the sake of consistency with SGX_IOC_ENCLAVE_ADD_PAGE, return -EINVAL when an enclave is already initialized or dead instead of zero.

v10:
* Cleaned up anon inode based IPC between the ring-0 and ring-3 parts of the driver.
* Unset the reserved flag from an enclave page if EDBGRD/WR fails (regression in v6).
* Close the anon inode when LE is stopped (regression in v9).
* Update the documentation with a more detailed description of SGX.

v9:
* Replaced kernel-LE IPC based on pipes with an anonymous inode. The driver does not require anymore new exports.

v8:
* Check that public key MSRs match the LE public key hash in the driver initialization when the MSRs are read-only.
* Fix the race in VA slot allocation by checking the fullness immediately after succeesful allocation.
* Fix the race in hash mrsigner calculation between the launch enclave and user enclaves by having a separate lock for hash calculation.

v7:
* Fixed offset calculation in sgx_edbgr/wr(). Address was masked with PAGE_MASK when it should have been masked with ~PAGE_MASK.
* Fixed a memory leak in sgx_ioc_enclave_create().
* Simplified swapping code by using a pointer array for a cluster instead of a linked list.
* Squeezed struct sgx_encl_page to 32 bytes.
* Fixed deferencing of an RSA key on OpenSSL 1.1.0.
* Modified TC's CMAC to use kernel AES-NI. Restructured the code
a bit in order to better align with kernel conventions.

v6:
* Fixed semaphore underrun when accessing /dev/sgx from the launch enclave.
* In sgx_encl_create() s/IS_ERR(secs)/IS_ERR(encl)/.
* Removed virtualization chapter from the documentation.
* Changed the default filename for the signing key as signing_key.pem.
* Reworked EPC management in a way that instead of a linked list of struct sgx_epc_page instances there is an array of integers that encodes address and bank of an EPC page (the same data as 'pa' field earlier). The locking has been moved to the EPC bank level instead of a global lock.
* Relaxed locking requirements for EPC management. EPC pages can be released back to the EPC bank concurrently.
* Cleaned up ptrace() code.
* Refined commit messages for new architectural constants.
* Sorted includes in every source file.
* Sorted local variable declarations according to the line length in every function.
* Style fixes based on Darren's comments to sgx_le.c.

v5:
* Described IPC between the Launch Enclave and kernel in the commit messages.
* Fixed all relevant checkpatch.pl issues that I have forgot fix in earlier versions except those that exist in the imported TinyCrypt code.
* Fixed spelling mistakes in the documentation.
* Forgot to check the return value of sgx_drv_subsys_init().
* Encapsulated properly page cache init and teardown.
* Collect epc pages to a temp list in sgx_add_epc_bank
* Removed SGX_ENCLAVE_INIT_ARCH constant.

v4:
* Tied life-cycle of the sgx_le_proxy process to /dev/sgx.
* Removed __exit annotation from sgx_drv_subsys_exit().
* Fixed a leak of a backing page in sgx_process_add_page_req() in the case when vm_insert_pfn() fails.
* Removed unused symbol exports for sgx_page_cache.c.
* Updated sgx_alloc_page() to require encl parameter and documented the behavior (Sean Christopherson).
* Refactored a more lean API for sgx_encl_find() and documented the behavior.
* Moved #PF handler to sgx_fault.c.
* Replaced subsys_system_register() with plain bus_register().
* Retry EINIT 2nd time only if MSRs are not locked.

v3:
* Check that FEATURE_CONTROL_LOCKED and FEATURE_CONTROL_SGX_ENABLE are set.
* Return -ERESTARTSYS in __sgx_encl_add_page() when sgx_alloc_page() fails.
* Use unused bits in epc_page->pa to store the bank number.
* Removed #ifdef for WQ_NONREENTRANT.
* If mmu_notifier_register() fails with -EINTR, return -ERESTARTSYS.
* Added --remove-section=.got.plt to objcopy flags in order to prevent a dummy .got.plt, which will cause an inconsistent size for the LE.
* Documented sgx_encl_* functions.
* Added remark about AES implementation used inside the LE.
* Removed redundant sgx_sys_exit() from le/main.c.
* Fixed struct sgx_secinfo alignment from 128 to 64 bytes.
* Validate miscselect in sgx_encl_create().
* Fixed SSA frame size calculation to take the misc region into account.
* Implemented consistent exception handling to __encls() and __encls_ret().
* Implemented a proper device model in order to allow sysfs attributes and in-kernel API.
* Cleaned up various "find enclave" implementations to the unified sgx_encl_find().
* Validate that vm_pgoff is zero.
* Discard backing pages with shmem_truncate_range() after EADD.
* Added missing EEXTEND operations to LE signing and launch.
* Fixed SSA size for GPRS region from 168 to 184 bytes.
* Fixed the checks for TCS flags. Now DBGOPTIN is allowed.
* Check that TCS addresses are in ELRANGE and not just page aligned.
* Require kernel to be compiled with X64_64 and CPU_SUP_INTEL.
* Fixed an incorrect value for SGX_ATTR_DEBUG from 0x01 to 0x02.

v2:
* get_rand_uint32() changed the value of the pointer instead of value where it is pointing at.
* Launch enclave incorrectly used sigstruct attributes-field instead of enclave attributes-field.
* Removed unused struct sgx_add_page_req from sgx_ioctl.c
* Removed unused sgx_has_sgx2.
* Updated arch/x86/include/asm/sgx.h so that it provides stub implementations when sgx in not enabled.
* Removed cruft rdmsr-calls from sgx_set_pubkeyhash_msrs().
* return -ENOMEM in sgx_alloc_page() when VA pages consume too much space
* removed unused global sgx_nr_pids
* moved sgx_encl_release to sgx_encl.c
* return -ERESTARTSYS instead of -EINTR in sgx_encl_init() Jarkko Sakkinen (10):
x86/sgx: Update MAINTAINERS
x86/sgx: Architectural structures
x86/sgx: Add wrappers for ENCLS leaf functions
x86/sgx: Add data structures for tracking the EPC pages x86/sgx: Enclave Page Cache (EPC) memory manager platform/x86: Intel SGX driver
platform/x86: Add swapping functionality to the Intel SGX driver x86/sgx: Add a simple swapper for the EPC memory manager platform/x86: ptrace() support for the SGX driver
x86/sgx: Driver documentation

Kai Huang (1):
x86/cpufeature: Add SGX and SGX_LC CPU features

Sean Christopherson (8):
x86/cpufeatures: Add SGX feature bits
x86/msr: Add SGX definitions to msr-index.h
x86/mm: x86/sgx: Add new 'PF_SGX' page fault error code bit x86/fault: x86/mm/pkeys: relocate stale comment regarding OSPKE
signal: x86/sgx: Add SIGSEGV siginfo code for SGX EPCM fault x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX x86/sgx: Detect Intel SGX
x86/sgx: Add sgx_einit() for initializing enclaves

 Documentation/index.rst                       |   1 +
 Documentation/x86/intel_sgx.rst               | 185 ++++
 MAINTAINERS                                   |   7 +
 arch/x86/Kconfig                              |  18 +
 arch/x86/include/asm/cpufeatures.h            |   8 +
 arch/x86/include/asm/msr-index.h              |   8 +
 arch/x86/include/asm/sgx.h                    | 326 ++++++
 arch/x86/include/asm/sgx_arch.h               | 403 ++++++++
 arch/x86/include/asm/traps.h                  |   1 +
 arch/x86/include/uapi/asm/sgx.h               |  60 ++
 arch/x86/include/uapi/asm/sgx_errno.h         |  91 ++
 arch/x86/kernel/cpu/Makefile                  |   1 +
 arch/x86/kernel/cpu/common.c                  |   1 +
 arch/x86/kernel/cpu/intel.c                   |  40 +
 arch/x86/kernel/cpu/intel_sgx.c               | 502 +++++++++
 arch/x86/mm/fault.c                           |  30 +-
 drivers/platform/x86/Kconfig                  |   2 +
 drivers/platform/x86/Makefile                 |   1 +
 drivers/platform/x86/intel_sgx/Kconfig        |  20 +
 drivers/platform/x86/intel_sgx/Makefile       |  14 +
 drivers/platform/x86/intel_sgx/sgx.h          | 212 ++++
 drivers/platform/x86/intel_sgx/sgx_encl.c     | 954 ++++++++++++++++++
 .../platform/x86/intel_sgx/sgx_encl_page.c    | 177 ++++
 drivers/platform/x86/intel_sgx/sgx_fault.c    | 108 ++
 drivers/platform/x86/intel_sgx/sgx_ioctl.c    | 234 +++++
 drivers/platform/x86/intel_sgx/sgx_main.c     | 267 +++++
 drivers/platform/x86/intel_sgx/sgx_util.c     | 156 +++
 drivers/platform/x86/intel_sgx/sgx_vma.c      | 167 +++
 include/uapi/asm-generic/siginfo.h            |   4 +
 29 files changed, 3990 insertions(+), 8 deletions(-)
 create mode 100644 Documentation/x86/intel_sgx.rst
 create mode 100644 arch/x86/include/asm/sgx.h
 create mode 100644 arch/x86/include/asm/sgx_arch.h
 create mode 100644 arch/x86/include/uapi/asm/sgx.h
 create mode 100644 arch/x86/include/uapi/asm/sgx_errno.h
 create mode 100644 arch/x86/kernel/cpu/intel_sgx.c
 create mode 100644 drivers/platform/x86/intel_sgx/Kconfig
 create mode 100644 drivers/platform/x86/intel_sgx/Makefile
 create mode 100644 drivers/platform/x86/intel_sgx/sgx.h
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_encl.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_encl_page.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_fault.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_ioctl.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_main.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_util.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_vma.c
Signature · 1 line
2.17.1
#1 lore raw

PATCH v14 1/19 x86/sgx: Update MAINTAINERS

Add the maintainer information for the SGX subsystem.

---
 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)
MAINTAINERS +7 -0
index 02a39617ec82..a5597a89c7a9 100644
@@ -7595,6 +7595,13 @@ L:	[email protected]
 S:	Maintained
 F:	drivers/gpio/gpio-intel-mid.c
 
+INTEL SGX
+M:	Jarkko Sakkinen <[email protected]>
+L:	[email protected]
+Q:	https://patchwork.kernel.org/project/intel-sgx/list/
+F:	drivers/platform/x86/intel_sgx/
+K:	\bSGX_
+
 INVENSENSE MPU-3050 GYROSCOPE DRIVER
 M:	Linus Walleij <[email protected]>
 L:	[email protected]
Signature · 1 line
2.17.1
#2 lore raw

PATCH v14 2/19 x86/sgx: Architectural structures

In reply to Jarkko Sakkinen

Add arch/x86/include/asm/sgx_arch.h, which contains definitions for the architectural data structures used by the CPU to implement the SGX.

---
 arch/x86/include/asm/sgx_arch.h       | 403 ++++++++++++++++++++++++++
 arch/x86/include/uapi/asm/sgx_errno.h |  91 ++++++
 2 files changed, 494 insertions(+)
 create mode 100644 arch/x86/include/asm/sgx_arch.h
 create mode 100644 arch/x86/include/uapi/asm/sgx_errno.h
arch/x86/include/asm/sgx_arch.h new file +403 -0
new file mode 100644
index 000000000000..f2a58ac0843e
@@ -0,0 +1,403 @@
+/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */
+/**
+ * Copyright(c) 2016-18 Intel Corporation.
+ *
+ * Contains the architectural data structures used by the CPU to implement SGX.
+ * The data structures defined by the Linux software stack should not be placed
+ * here.
+ */
+#ifndef _ASM_X86_SGX_ARCH_H
+#define _ASM_X86_SGX_ARCH_H
+
+#include <linux/types.h>
+#include <uapi/asm/sgx_errno.h>
+
+#define SGX_CPUID		0x12
+#define SGX_CPUID_SECTION	2
+#define SGX_MAX_EPC_SECTIONS	8
+
+/**
+ * enum sgx_section_types - SGX cpuid section subleaf types
+ * %SGX_CPUID_SECTION_INVALID:	subleaf for the section is not valid
+ * %SGX_CPUID_SECTION_VALID:	subleaf for the section is valid
+ * %SGX_CPUID_SECTION_MASK:	bits that define the subleaf type
+ *
+ * The section subleaves are traversed in order up until an invalid subleaf is
+ * encountered. The information about the section type is contained in the
+ * first four bits of eax.
+ */
+enum sgx_section_types {
+	SGX_CPUID_SECTION_INVALID	= 0x0,
+	SGX_CPUID_SECTION_VALID		= 0x1,
+	SGX_CPUID_SECTION_MASK		= 0xF,
+};
+
+/**
+ * enum sgx_encls_leaves - ENCLS leaf functions
+ * %SGX_ECREATE:	Create an enclave.
+ * %SGX_EADD:		Add a page to an enclave.
+ * %SGX_EINIT:		Launch an enclave.
+ * %SGX_EREMOVE:	Remove a page from an enclave.
+ * %SGX_EDBGRD:		Read a word from an enclve (peek).
+ * %SGX_EDBGWR:		Write a word to an enclave (poke).
+ * %SGX_EEXTEND:	Measure 256 bytes of an added enclave page.
+ * %SGX_ELDB:		Load a swapped page in blocked state.
+ * %SGX_ELDU:		Load a swapped page in unblocked state.
+ * %SGX_EBLOCK:		Change page state to blocked i.e. entering hardware
+ *			threads cannot access it and create new TLB entries.
+ * %SGX_EPA:		Create a Version Array (VA) page used to store isvsvn
+ *			number for a swapped EPC page.
+ * %SGX_EWB:		Swap an enclave page to the regular memory. Checks that
+ *			all threads have exited that were in the previous
+ *			shoot-down sequence.
+ * %SGX_ETRACK:		Start a new shoot down sequence. Used to together with
+ *			EBLOCK to make sure that a page is safe to swap.
+ */
+enum sgx_encls_leaves {
+	SGX_ECREATE	= 0x00,
+	SGX_EADD	= 0x01,
+	SGX_EINIT	= 0x02,
+	SGX_EREMOVE	= 0x03,
+	SGX_EDGBRD	= 0x04,
+	SGX_EDGBWR	= 0x05,
+	SGX_EEXTEND	= 0x06,
+	SGX_ELDB	= 0x07,
+	SGX_ELDU	= 0x08,
+	SGX_EBLOCK	= 0x09,
+	SGX_EPA		= 0x0A,
+	SGX_EWB		= 0x0B,
+	SGX_ETRACK	= 0x0C,
+	SGX_EAUG	= 0x0D,
+	SGX_EMODPR	= 0x0E,
+	SGX_EMODT	= 0x0F,
+};
+
+#define SGX_MODULUS_SIZE 384
+
+/**
+ * enum sgx_miscselect - additional information to an SSA frame
+ * %SGX_MISC_EXINFO:	Report #PF or #GP to the SSA frame.
+ *
+ * Save State Area (SSA) is a stack inside the enclave used to store processor
+ * state when an exception or interrupt occurs. This enum defines additional
+ * information stored to an SSA frame.
+ */
+enum sgx_miscselect {
+	SGX_MISC_EXINFO		= BIT(0),
+	SGX_MISC_RESERVED_MASK	= GENMASK_ULL(63, 1)
+};
+
+#define SGX_SSA_GPRS_SIZE		182
+#define SGX_SSA_MISC_EXINFO_SIZE	16
+
+/**
+ * enum sgx_attributes - the attributes field in &struct sgx_secs
+ * %SGX_ATTR_INIT:		Enclave can be entered (is initialized).
+ * %SGX_ATTR_DEBUG:		Allow ENCLS(EDBGRD) and ENCLS(EDBGWR).
+ * %SGX_ATTR_MODE64BIT:		Tell that this a 64-bit enclave.
+ * %SGX_ATTR_PROVISIONKEY:      Allow to use provisioning keys for remote
+ *				attestation.
+ * %SGX_ATTR_EINITTOKENKEY:	Allow to use token signing key that is used to
+ *				sign cryptographic tokens that can be passed to
+ *				EINIT as an authorization to run an enclave.
+ */
+enum sgx_attribute {
+	SGX_ATTR_INIT		= BIT(0),
+	SGX_ATTR_DEBUG		= BIT(1),
+	SGX_ATTR_MODE64BIT	= BIT(2),
+	SGX_ATTR_PROVISIONKEY	= BIT(4),
+	SGX_ATTR_EINITTOKENKEY	= BIT(5),
+	SGX_ATTR_RESERVED_MASK  = BIT_ULL(3) | GENMASK_ULL(63, 6)
+};
+
+#define SGX_SECS_RESERVED1_SIZE 24
+#define SGX_SECS_RESERVED2_SIZE 32
+#define SGX_SECS_RESERVED3_SIZE 96
+#define SGX_SECS_RESERVED4_SIZE 3836
+
+/**
+ * struct sgx_secs - SGX Enclave Control Structure (SECS)
+ * @size:		size of the address space
+ * @base:		base address of the  address space
+ * @ssa_frame_size:	size of an SSA frame
+ * @miscselect:		additional information stored to an SSA frame
+ * @attributes:		attributes for enclave
+ * @xfrm:		XSave-Feature Request Mask (subset of XCR0)
+ * @mrenclave:		SHA256-hash of the enclave contents
+ * @mrsigner:		SHA256-hash of the public key used to sign the SIGSTRUCT
+ * @isvprodid:		a user-defined value that is used in key derivation
+ * @isvsvn:		a user-defined value that is used in key derivation
+ *
+ * SGX Enclave Control Structure (SECS) is a special enclave page that is not
+ * visible in the address space. In fact, this structure defines the address
+ * range and other global attributes for the enclave and it is the first EPC
+ * page created for any enclave. It is moved from a temporary buffer to an EPC
+ * by the means of ENCLS(ECREATE) leaf.
+ */
+struct sgx_secs {
+	u64 size;
+	u64 base;
+	u32 ssa_frame_size;
+	u32 miscselect;
+	u8  reserved1[SGX_SECS_RESERVED1_SIZE];
+	u64 attributes;
+	u64 xfrm;
+	u32 mrenclave[8];
+	u8  reserved2[SGX_SECS_RESERVED2_SIZE];
+	u32 mrsigner[8];
+	u8  reserved3[SGX_SECS_RESERVED3_SIZE];
+	u16 isvprodid;
+	u16 isvsvn;
+	u8  reserved4[SGX_SECS_RESERVED4_SIZE];
+} __packed;
+
+/**
+ * enum sgx_tcs_flags - execution flags for TCS
+ * %SGX_TCS_DBGOPTIN:	If enabled allows single-stepping and breakpoints
+ *			inside an enclave. It is cleared by EADD but can
+ *			be set later with EDBGWR.
+ */
+enum sgx_tcs_flags {
+	SGX_TCS_DBGOPTIN	= 0x01,
+	SGX_TCS_RESERVED_MASK	= GENMASK_ULL(63, 1)
+};
+
+#define SGX_TCS_RESERVED_SIZE 503
+
+/**
+ * struct sgx_tcs - Thread Control Structure (TCS)
+ * @state:		used to mark an entered TCS
+ * @flags:		execution flags (cleared by EADD)
+ * @ssa_offset:		SSA stack offset relative to the enclave base
+ * @ssa_index:		the current SSA frame index (cleard by EADD)
+ * @nr_ssa_frames:	the number of frame in the SSA stack
+ * @entry_offset:	entry point offset relative to the enclave base
+ * @exit_addr:		address outside the enclave to exit on an exception or
+ *			interrupt
+ * @fs_offset:		offset relative to the enclave base to become FS
+ *			segment inside the enclave
+ * @gs_offset:		offset relative to the enclave base to become GS
+ *			segment inside the enclave
+ * @fs_limit:		size to become a new FS-limit (only 32-bit enclaves)
+ * @gs_limit:		size to become a new GS-limit (only 32-bit enclaves)
+ *
+ * Thread Control Structure (TCS) is an enclave page visible in its address
+ * space that defines an entry point inside the enclave. A thread enters inside
+ * an enclave by supplying address of TCS to ENCLU(EENTER). A TCS can be entered
+ * by only one thread at a time.
+ */
+struct sgx_tcs {
+	u64 state;
+	u64 flags;
+	u64 ssa_offset;
+	u32 ssa_index;
+	u32 nr_ssa_frames;
+	u64 entry_offset;
+	u64 exit_addr;
+	u64 fs_offset;
+	u64 gs_offset;
+	u32 fs_limit;
+	u32 gs_limit;
+	u64 reserved[SGX_TCS_RESERVED_SIZE];
+} __packed;
+
+/**
+ * struct sgx_pageinfo - an enclave page descriptor
+ * @addr:	address of the enclave page
+ * @contents:	pointer to the page contents
+ * @metadata:	pointer either to a SECINFO or PCMD instance
+ * @secs:	address of the SECS page
+ */
+struct sgx_pageinfo {
+	u64 addr;
+	u64 contents;
+	u64 metadata;
+	u64 secs;
+} __packed __aligned(32);
+
+
+/**
+ * enum sgx_page_type - bits in the SECINFO flags defining the page type
+ * %SGX_PAGE_TYPE_SECS:	a SECS page
+ * %SGX_PAGE_TYPE_TCS:	a TCS page
+ * %SGX_PAGE_TYPE_REG:	a regular page
+ * %SGX_PAGE_TYPE_VA:	a VA page
+ * %SGX_PAGE_TYPE_TRIM:	a page in trimmed state
+ */
+enum sgx_page_type {
+	SGX_PAGE_TYPE_SECS	= 0,
+	SGX_PAGE_TYPE_TCS	= 1,
+	SGX_PAGE_TYPE_REG	= 2,
+	SGX_PAGE_TYPE_VA	= 3,
+	SGX_PAGE_TYPE_TRIM	= 4,
+	SGX_PAGE_TYPE_MASK	= GENMASK(7, 0)
+};
+
+/**
+ * enum sgx_secinfo_flags - the flags field in &struct sgx_secinfo
+ * %SGX_SECINFO_R:	allow read
+ * %SGX_SECINFO_W:	allow write
+ * %SGX_SECINFO_X:	allow execution
+ * %SGX_SECINFO_SECS:	a SECS page
+ * %SGX_SECINFO_TCS:	a TCS page
+ * %SGX_SECINFO_REG:	a regular page
+ * %SGX_SECINFO_VA:	a VA page
+ * %SGX_SECINFO_TRIM:	a page in trimmed state
+ */
+enum sgx_secinfo_flags {
+	SGX_SECINFO_R			= BIT(0),
+	SGX_SECINFO_W			= BIT(1),
+	SGX_SECINFO_X			= BIT(2),
+	SGX_SECINFO_PERMISSION_MASK	= GENMASK_ULL(2, 0),
+	SGX_SECINFO_SECS		= (SGX_PAGE_TYPE_SECS << 8),
+	SGX_SECINFO_TCS			= (SGX_PAGE_TYPE_TCS << 8),
+	SGX_SECINFO_REG			= (SGX_PAGE_TYPE_REG << 8),
+	SGX_SECINFO_VA			= (SGX_PAGE_TYPE_VA << 8),
+	SGX_SECINFO_TRIM		= (SGX_PAGE_TYPE_TRIM << 8),
+	SGX_SECINFO_PAGE_TYPE_MASK	= (SGX_PAGE_TYPE_MASK << 8),
+	SGX_SECINFO_RESERVED_MASK	= ~(SGX_SECINFO_PERMISSION_MASK |
+					    SGX_SECINFO_PAGE_TYPE_MASK)
+};
+
+#define SGX_SECINFO_RESERVED_SIZE 56
+
+/**
+ * struct sgx_secinfo - describes attributes of an EPC page
+ * @flags:	permissions and type
+ *
+ * Used together with ENCLS leaves that add or modify an EPC page to an
+ * enclave to define page permissions and type.
+ */
+struct sgx_secinfo {
+	u64 flags;
+	u8  reserved[SGX_SECINFO_RESERVED_SIZE];
+} __packed __aligned(64);
+
+#define SGX_PCMD_RESERVED_SIZE 40
+
+/**
+ * struct sgx_pcmd - Paging Crypto Metadata (PCMD)
+ * @enclave_id:	enclave identifier
+ * @mac:	MAC over PCMD, page contents and isvsvn
+ *
+ * PCMD is stored for every swapped page to the regular memory. When ELDU loads
+ * the page back it recalculates the MAC by using a isvsvn number stored in a
+ * VA page. Together these two structures bring integrity and rollback
+ * protection.
+ */
+struct sgx_pcmd {
+	struct sgx_secinfo secinfo;
+	u64 enclave_id;
+	u8  reserved[SGX_PCMD_RESERVED_SIZE];
+	u8  mac[16];
+} __packed __aligned(128);
+
+#define SGX_SIGSTRUCT_RESERVED1_SIZE 84
+#define SGX_SIGSTRUCT_RESERVED2_SIZE 20
+#define SGX_SIGSTRUCT_RESERVED3_SIZE 32
+#define SGX_SIGSTRUCT_RESERVED4_SIZE 12
+
+/**
+ * struct sgx_sigstruct - an enclave signature
+ * @header1:		a constant byte string
+ * @vendor:		must be either 0x0000 or 0x8086
+ * @date:		YYYYMMDD in BCD
+ * @header2:		a costant byte string
+ * @application:	an application defined value
+ * @modulus:		the modulus of the public key
+ * @exponent:		the exponent of the public key
+ * @signature:		the signature calculated over the fields except modulus,
+ *			exponent, signature, reserved4, q1 and q2
+ * @miscselect:		additional information stored to an SSA frame
+ * @misc_mask:		required miscselect in SECS
+ * @attributes:		attributes for enclave
+ * @xfrm:		XSave-Feature Request Mask (subset of XCR0)
+ * @attributes_mask:	required attributes in SECS
+ * @xfrm_mask:		required XFRM in SECS
+ * @mrenclave:		SHA256-hash of the enclave contents
+ * @isvprodid:		a user-defined value that is used in key derivation
+ * @isvsvn:		a user-defined value that is used in key derivation
+ * @q1:			a value used in RSA signature verification
+ * @q2:			a value used in RSA signature verification
+ */
+struct sgx_sigstruct {
+	u64 header1[2];
+	u32 vendor;
+	u32 date;
+	u64 header2[2];
+	u32 application;
+	u8  reserved1[SGX_SIGSTRUCT_RESERVED1_SIZE];
+	u8  modulus[SGX_MODULUS_SIZE];
+	u32 exponent;
+	u8  signature[SGX_MODULUS_SIZE];
+	u32 miscselect;
+	u32 misc_mask;
+	u8  reserved2[SGX_SIGSTRUCT_RESERVED2_SIZE];
+	u64 attributes;
+	u64 xfrm;
+	u64 attributes_mask;
+	u64 xfrm_mask;
+	u8  mrenclave[32];
+	u8  reserved3[SGX_SIGSTRUCT_RESERVED3_SIZE];
+	u16 isvprodid;
+	u16 isvsvn;
+	u8  reserved4[SGX_SIGSTRUCT_RESERVED4_SIZE];
+	u8  q1[SGX_MODULUS_SIZE];
+	u8  q2[SGX_MODULUS_SIZE];
+} __packed __aligned(4096);
+
+#define SGX_EINITTOKEN_RESERVED1_SIZE 11
+#define SGX_EINITTOKEN_RESERVED2_SIZE 32
+#define SGX_EINITTOKEN_RESERVED3_SIZE 32
+#define SGX_EINITTOKEN_RESERVED4_SIZE 24
+
+/**
+ * struct sgx_einittoken - a token permitting to launch an enclave
+ * @valid:			one if valid and zero if invalid
+ * @attributes:			attributes for enclave
+ * @xfrm:			XSave-Feature Request Mask (subset of XCR0)
+ * @mrenclave:			SHA256-hash of the enclave contents
+ * @mrsigner:			SHA256-hash of the public key used to sign the
+ *				SIGSTRUCT
+ * @le_cpusvn:			a value that reflects the SGX implementation
+ *				running in in the CPU
+ * @le_isvprodid:		a user-defined value that is used in key
+ *				derivation
+ * @le_isvsvn:			a user-defined value that is used in key
+ *				derivation
+ * @le_keyed_miscselect:	LE's miscselect masked with the token keys
+ *				miscselect
+ * @le_keyed_attributes:	LE's attributes masked with the token keys
+ *				attributes
+ * @le_keyed_xfrm:		LE's XFRM masked with the token keys xfrm
+ * @salt:			random salt for wear-out protection
+ * @mac:			CMAC over the preceding fields
+ *
+ * An enclave with EINITTOKENKEY attribute can access a key with the same name
+ * by using ENCLS(EGETKEY) and use this to sign cryptographic tokens that can
+ * be passed to ENCLS(EINIT) to permit the launch of other enclaves. This is
+ * the only viable way to launch enclaves if IA32_SGXLEPUBKEYHASHn MSRs are
+ * locked assuming that there is a Launch Enclave (LE) available that can be
+ * used for generating these tokens.
+ */
+struct sgx_einittoken {
+	u32 valid;
+	u32 reserved1[SGX_EINITTOKEN_RESERVED1_SIZE];
+	u64 attributes;
+	u64 xfrm;
+	u8  mrenclave[32];
+	u8  reserved2[SGX_EINITTOKEN_RESERVED2_SIZE];
+	u8  mrsigner[32];
+	u8  reserved3[SGX_EINITTOKEN_RESERVED3_SIZE];
+	u8  le_cpusvn[16];
+	u16 le_isvprodid;
+	u16 le_isvsvn;
+	u8  reserved4[SGX_EINITTOKEN_RESERVED4_SIZE];
+	u32 le_keyed_miscselect;
+	u64 le_keyed_attributes;
+	u64 le_keyed_xfrm;
+	u8  salt[32];
+	u8  mac[16];
+} __packed __aligned(512);
+
+#endif /* _ASM_X86_SGX_ARCH_H */
arch/x86/include/uapi/asm/sgx_errno.h new file +91 -0

91 lines not on this page · raw message

Signature · 1 line
2.17.1
#3 lore raw

PATCH v14 3/19 x86/cpufeature: Add SGX and SGX_LC CPU features

In reply to Jarkko Sakkinen

From: Kai Huang <[email protected]>

Add X86_FEATURE_SGX and X86_FEATURE_SGX_LC that define the bits determining whether the CPU supports SGX and user launch configuration i.e. using a custom root key rather the Intel proprietary key for enclave signing.

---
 arch/x86/include/asm/cpufeatures.h | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/include/asm/cpufeatures.h +2 -0
index 89a048c2faec..7bb647f57d42 100644
@@ -236,6 +236,7 @@
 /* Intel-defined CPU features, CPUID level 0x00000007:0 (EBX), word 9 */
 #define X86_FEATURE_FSGSBASE		( 9*32+ 0) /* RDFSBASE, WRFSBASE, RDGSBASE, WRGSBASE instructions*/
 #define X86_FEATURE_TSC_ADJUST		( 9*32+ 1) /* TSC adjustment MSR 0x3B */
+#define X86_FEATURE_SGX			( 9*32+ 2) /* Software Guard Extensions */
 #define X86_FEATURE_BMI1		( 9*32+ 3) /* 1st group bit manipulation extensions */
 #define X86_FEATURE_HLE			( 9*32+ 4) /* Hardware Lock Elision */
 #define X86_FEATURE_AVX2		( 9*32+ 5) /* AVX2 instructions */
@@ -331,6 +332,7 @@
 #define X86_FEATURE_LA57		(16*32+16) /* 5-level page tables */
 #define X86_FEATURE_RDPID		(16*32+22) /* RDPID instruction */
 #define X86_FEATURE_CLDEMOTE		(16*32+25) /* CLDEMOTE instruction */
+#define X86_FEATURE_SGX_LC		(16*32+30) /* supports SGX launch configuration */
 
 /* AMD-defined CPU features, CPUID level 0x80000007 (EBX), word 17 */
 #define X86_FEATURE_OVERFLOW_RECOV	(17*32+ 0) /* MCA overflow recovery support */
Signature · 1 line
2.17.1
#4 lore raw

PATCH v14 4/19 x86/cpufeatures: Add SGX feature bits

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Add SGX feature bits as part of the Linux defined leaf 8, which currently contains virtualization flags. There are currently four documented SGX feature bits, with more expected in the not-too-distant future.

---
 arch/x86/include/asm/cpufeatures.h |  8 +++++-
 arch/x86/kernel/cpu/intel.c        | 40 ++++++++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 1 deletion(-)
arch/x86/include/asm/cpufeatures.h +7 -1
index 7bb647f57d42..d6f4abe6d0b0 100644
@@ -233,6 +233,12 @@
 #define X86_FEATURE_XENPV		( 8*32+16) /* "" Xen paravirtual guest */
 #define X86_FEATURE_EPT_AD		( 8*32+17) /* Intel Extended Page Table access-dirty bit */
 
+/* SGX flags: Linux defined, word 8 */
+#define X86_FEATURE_SGX1		( 8*32+24) /* SGX1 leaf functions */
+#define X86_FEATURE_SGX2		( 8*32+25) /* SGX2 leaf functions */
+#define X86_FEATURE_SGX_ENCLV		( 8*32+26) /* SGX ENCLV instruction, leafs E[INC|DEC]VIRTCHILD, ESETCONTEXT */
+#define X86_FEATURE_SGX_ENCLS_C		( 8*32+27) /* SGX ENCLS leafs ERDINFO, ETRACK, ELDBC and ELDUC */
+
 /* Intel-defined CPU features, CPUID level 0x00000007:0 (EBX), word 9 */
 #define X86_FEATURE_FSGSBASE		( 9*32+ 0) /* RDFSBASE, WRFSBASE, RDGSBASE, WRGSBASE instructions*/
 #define X86_FEATURE_TSC_ADJUST		( 9*32+ 1) /* TSC adjustment MSR 0x3B */
@@ -332,7 +338,7 @@
 #define X86_FEATURE_LA57		(16*32+16) /* 5-level page tables */
 #define X86_FEATURE_RDPID		(16*32+22) /* RDPID instruction */
 #define X86_FEATURE_CLDEMOTE		(16*32+25) /* CLDEMOTE instruction */
-#define X86_FEATURE_SGX_LC		(16*32+30) /* supports SGX launch configuration */
+#define X86_FEATURE_SGX_LC		(16*32+30) /* supports SGX launch control */
 
 /* AMD-defined CPU features, CPUID level 0x80000007 (EBX), word 17 */
 #define X86_FEATURE_OVERFLOW_RECOV	(17*32+ 0) /* MCA overflow recovery support */
arch/x86/kernel/cpu/intel.c +40 -0
index fc3c07fe7df5..fcf188d5f9df 100644
@@ -19,6 +19,7 @@
 #include <asm/microcode_intel.h>
 #include <asm/hwcap2.h>
 #include <asm/elf.h>
+#include <asm/sgx_arch.h>
 
 #ifdef CONFIG_X86_64
 #include <linux/topology.h>
@@ -512,6 +513,42 @@ static void detect_vmx_virtcap(struct cpuinfo_x86 *c)
 	}
 }
 
+static void detect_sgx(struct cpuinfo_x86 *c)
+{
+#define _X86_FEATURE_SGX1		BIT(0)
+#define _X86_FEATURE_SGX2		BIT(1)
+#define _X86_FEATURE_SGX_ENCLV		BIT(5)
+#define _X86_FEATURE_SGX_ENCLS_C	BIT(6)
+
+	unsigned int eax;
+
+	clear_cpu_cap(c, X86_FEATURE_SGX1);
+	clear_cpu_cap(c, X86_FEATURE_SGX2);
+	clear_cpu_cap(c, X86_FEATURE_SGX_ENCLV);
+	clear_cpu_cap(c, X86_FEATURE_SGX_ENCLS_C);
+
+	if (c->cpuid_level < SGX_CPUID) {
+		pr_err_once("x86/sgx: cannot enumerate CPUID leaf (0x%x)\n",
+			    SGX_CPUID);
+		clear_cpu_cap(c, X86_FEATURE_SGX);
+		return;
+	}
+
+	eax = cpuid_eax(SGX_CPUID);
+
+	if (eax & _X86_FEATURE_SGX1)
+		set_cpu_cap(c, X86_FEATURE_SGX1);
+
+	if (eax & _X86_FEATURE_SGX2)
+		set_cpu_cap(c, X86_FEATURE_SGX2);
+
+	if (eax & _X86_FEATURE_SGX_ENCLV)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLV);
+
+	if (eax & _X86_FEATURE_SGX_ENCLS_C)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLS_C);
+}
+
 #define MSR_IA32_TME_ACTIVATE		0x982
 
 /* Helpers to access TME_ACTIVATE MSR */
@@ -760,6 +797,9 @@ static void init_intel(struct cpuinfo_x86 *c)
 	if (cpu_has(c, X86_FEATURE_VMX))
 		detect_vmx_virtcap(c);
 
+	if (cpu_has(c, X86_FEATURE_SGX))
+		detect_sgx(c);
+
 	if (cpu_has(c, X86_FEATURE_TME))
 		detect_tme(c);
 
Signature · 1 line
2.17.1
maintainer
#5 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Quoted 6 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Add SGX feature bits as part of the Linux defined leaf 8, which currently contains virtualization flags. There are currently four documented SGX feature bits, with more expected in the not-too-distant future.

...

Quoted 24 lines
arch/x86/kernel/cpu/intel.c +24 -0
+static void detect_sgx(struct cpuinfo_x86 *c)
+{
+#define _X86_FEATURE_SGX1		BIT(0)
+#define _X86_FEATURE_SGX2		BIT(1)
+#define _X86_FEATURE_SGX_ENCLV		BIT(5)
+#define _X86_FEATURE_SGX_ENCLS_C	BIT(6)
+
+	unsigned int eax;
+
+	clear_cpu_cap(c, X86_FEATURE_SGX1);
+	clear_cpu_cap(c, X86_FEATURE_SGX2);
+	clear_cpu_cap(c, X86_FEATURE_SGX_ENCLV);
+	clear_cpu_cap(c, X86_FEATURE_SGX_ENCLS_C);
+
+	if (c->cpuid_level < SGX_CPUID) {
+		pr_err_once("x86/sgx: cannot enumerate CPUID leaf (0x%x)\n",
+			    SGX_CPUID);
+		clear_cpu_cap(c, X86_FEATURE_SGX);
+		return;
+	}
+
+	eax = cpuid_eax(SGX_CPUID);
+
+	if (eax & _X86_FEATURE_SGX1)

arch/x86/kernel/cpu/intel.c static void detect_vmx_virtcap(struct cpuinfo_x86 *c) · line 549

Quoted 10 lines
arch/x86/kernel/cpu/intel.c +10 -0
+		set_cpu_cap(c, X86_FEATURE_SGX1);
+
+	if (eax & _X86_FEATURE_SGX2)
+		set_cpu_cap(c, X86_FEATURE_SGX2);
+
+	if (eax & _X86_FEATURE_SGX_ENCLV)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLV);
+
+	if (eax & _X86_FEATURE_SGX_ENCLS_C)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLS_C);

Look at arch/x86/kernel/cpu/scattered.c for how to do this properly.

Signature · 4 lines
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.
#6 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Quoted 8 lines · Borislav Petkov
Quoted 8 lines
arch/x86/kernel/cpu/intel.c +8 -0
+	eax = cpuid_eax(SGX_CPUID);
+
+	if (eax & _X86_FEATURE_SGX1)
+		set_cpu_cap(c, X86_FEATURE_SGX1);
+
+	if (eax & _X86_FEATURE_SGX2)
+		set_cpu_cap(c, X86_FEATURE_SGX2);
+
Quoted 8 lines

arch/x86/kernel/cpu/intel.c static void detect_vmx_virtcap(struct cpuinfo_x86 *c) · line 549

Quoted 5 lines
arch/x86/kernel/cpu/intel.c +5 -0
+	if (eax & _X86_FEATURE_SGX_ENCLV)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLV);
+
+	if (eax & _X86_FEATURE_SGX_ENCLS_C)
+		set_cpu_cap(c, X86_FEATURE_SGX_ENCLS_C);

Look at arch/x86/kernel/cpu/scattered.c for how to do this properly.

Thank you. I guess I understand what you want me to do i.e.

+       { X86_FEATURE_SGX1,             CPUID_EAX,  0, 0x00000012, 0 },
+       { X86_FEATURE_SGX2,             CPUID_EAX,  1, 0x00000012, 0 },
+       { X86_FEATURE_ENCLV,            CPUID_EAX,  5, 0x00000012, 0 },
+       { X86_FEATURE_ENCLS_C,          CPUID_EAX,  6, 0x00000012, 0 },

What puzzles me is that I cannot find any call site for get_scattered_cpuid_leaf():

$ git grep get_scattered
arch/x86/kernel/cpu/cpu.h:extern u32 get_scattered_cpuid_leaf(unsigned int level,
arch/x86/kernel/cpu/scattered.c:u32 get_scattered_cpuid_leaf(unsigned int level, unsigned int sub_leaf,
arch/x86/kernel/cpu/scattered.c:EXPORT_SYMBOL_GPL(get_scattered_cpuid_leaf);

/Jarkko

maintainer
#7 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Review

Quoted 6 lines · Jarkko Sakkinen

Thank you. I guess I understand what you want me to do i.e.

(quoted diff) +4 -0
+       { X86_FEATURE_SGX1,             CPUID_EAX,  0, 0x00000012, 0 },
+       { X86_FEATURE_SGX2,             CPUID_EAX,  1, 0x00000012, 0 },
+       { X86_FEATURE_ENCLV,            CPUID_EAX,  5, 0x00000012, 0 },
+       { X86_FEATURE_ENCLS_C,          CPUID_EAX,  6, 0x00000012, 0 },

Yap.

Quoted 7 lines

What puzzles me is that I cannot find any call site for get_scattered_cpuid_leaf():

$ git grep get_scattered
arch/x86/kernel/cpu/cpu.h:extern u32 get_scattered_cpuid_leaf(unsigned int level,
arch/x86/kernel/cpu/scattered.c:u32 get_scattered_cpuid_leaf(unsigned int level, unsigned int sub_leaf,
arch/x86/kernel/cpu/scattered.c:EXPORT_SYMBOL_GPL(get_scattered_cpuid_leaf);

That used to be used in kvm last, see b7b27aa011a1df42728d1768fc181d9ce69e6911, which removed it.

And it got added by

47bdf3378d62 ("x86/cpuid: Provide get_scattered_cpuid_leaf()")

Lemme add those people to CC.

If no one wants it, feel free to decomission it.

Signature · 4 lines
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.
#8 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Quoted 21 lines · Borislav Petkov

Review

Quoted 6 lines · Jarkko Sakkinen

Thank you. I guess I understand what you want me to do i.e.

(quoted diff) +4 -0
+       { X86_FEATURE_SGX1,             CPUID_EAX,  0, 0x00000012, 0 },
+       { X86_FEATURE_SGX2,             CPUID_EAX,  1, 0x00000012, 0 },
+       { X86_FEATURE_ENCLV,            CPUID_EAX,  5, 0x00000012, 0 },
+       { X86_FEATURE_ENCLS_C,          CPUID_EAX,  6, 0x00000012, 0 },

Yap.

Quoted 7 lines

What puzzles me is that I cannot find any call site for get_scattered_cpuid_leaf():

$ git grep get_scattered
arch/x86/kernel/cpu/cpu.h:extern u32 get_scattered_cpuid_leaf(unsigned int level,
arch/x86/kernel/cpu/scattered.c:u32 get_scattered_cpuid_leaf(unsigned int level, unsigned int sub_leaf,
arch/x86/kernel/cpu/scattered.c:EXPORT_SYMBOL_GPL(get_scattered_cpuid_leaf);

That used to be used in kvm last, see b7b27aa011a1df42728d1768fc181d9ce69e6911, which removed it.

Quoted 7 lines

And it got added by

47bdf3378d62 ("x86/cpuid: Provide get_scattered_cpuid_leaf()")

Lemme add those people to CC.

If no one wants it, feel free to decomission it.

Still kind of leaves me puzzled about the situation :-) Why this wasn't used for example to replace detect_vmx_virtcap()?

/Jarkko

maintainer
#9 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Quoted 2 lines · Jarkko Sakkinen

Still kind of leaves me puzzled about the situation :-) Why this wasn't used for example to replace detect_vmx_virtcap()?

I don't understand - detect_vmx_virtcap() is setting X86_FEATURE bits based on MSR settings. get_scattered_cpuid_leaf() is regenerating a CPUID leaf from the scattered bits.

Looks like apples and oranges to me...

Signature · 4 lines
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.
#10 lore raw

Re: [PATCH v14 04/19] x86/cpufeatures: Add SGX feature bits

Quoted 15 lines · Borislav Petkov
Quoted 2 lines · Jarkko Sakkinen

Still kind of leaves me puzzled about the situation :-) Why this wasn't used for example to replace detect_vmx_virtcap()?

I don't understand - detect_vmx_virtcap() is setting X86_FEATURE bits based on MSR settings. get_scattered_cpuid_leaf() is regenerating a CPUID leaf from the scattered bits.

Looks like apples and oranges to me...

Signature · 4 lines
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.

Ugh, sorry, I was not thinking clearly :-)

/Jarkko

#11 lore raw

PATCH v14 5/19 x86/msr: Add SGX definitions to msr-index.h

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

ENCLS and ENCLU are usable if and only if SGX_ENABLE is set and After SGX is activated the IA32_SGXLEPUBKEYHASHn MSRs are writable if SGX_LC_WR is set and the feature control is locked.

SGX related bits in IA32_FEATURE_CONTROL cannot be set before SGX is activated by the pre-boot firmware. SGX activation is triggered by setting bit 0 in the MSR 0x7a. Until SGX is activated, the LE hash MSRs are writable to allow pre-boot firmware to lock down the LE root key with a non-Intel value.

---
 arch/x86/include/asm/msr-index.h | 8 ++++++++
 arch/x86/kernel/cpu/common.c     | 1 +
 2 files changed, 9 insertions(+)
arch/x86/include/asm/msr-index.h +8 -0
index 4731f0cf97c5..44d7b68e7f52 100644
@@ -486,6 +486,8 @@
 #define FEATURE_CONTROL_LOCKED				(1<<0)
 #define FEATURE_CONTROL_VMXON_ENABLED_INSIDE_SMX	(1<<1)
 #define FEATURE_CONTROL_VMXON_ENABLED_OUTSIDE_SMX	(1<<2)
+#define FEATURE_CONTROL_SGX_ENABLE			(1<<18)
+#define FEATURE_CONTROL_SGX_LE_WR			(1<<17)
 #define FEATURE_CONTROL_LMCE				(1<<20)
 
 #define MSR_IA32_APICBASE		0x0000001b
@@ -552,6 +554,12 @@
 #define PACKAGE_THERM_INT_LOW_ENABLE		(1 << 1)
 #define PACKAGE_THERM_INT_PLN_ENABLE		(1 << 24)
 
+/* Intel SGX MSRs */
+#define MSR_IA32_SGXLEPUBKEYHASH0	0x0000008C
+#define MSR_IA32_SGXLEPUBKEYHASH1	0x0000008D
+#define MSR_IA32_SGXLEPUBKEYHASH2	0x0000008E
+#define MSR_IA32_SGXLEPUBKEYHASH3	0x0000008F
+
 /* Thermal Thresholds Support */
 #define THERM_INT_THRESHOLD0_ENABLE    (1 << 15)
 #define THERM_SHIFT_THRESHOLD0        8
arch/x86/kernel/cpu/common.c +1 -0
index 44c4ef3d989b..aedfd9657a6c 100644
@@ -52,6 +52,7 @@
 #include <asm/microcode_intel.h>
 #include <asm/intel-family.h>
 #include <asm/cpu_device_id.h>
+#include <asm/sgx_arch.h>
 
 #ifdef CONFIG_X86_LOCAL_APIC
 #include <asm/uv/uv.h>
Signature · 1 line
2.17.1
#12 lore raw

PATCH v14 6/19 x86/mm: x86/sgx: Add new 'PF_SGX' page fault error code bit

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

The SGX bit is set in the #PF error code if and only if the fault is detected by the Enclave Page Cache Map (EPCM), a hardware-managed table that enforces the paging permissions defined by the enclave, e.g. to prevent the kernel from changing the permissions of an enclave's page(s).

---
 arch/x86/include/asm/traps.h | 1 +
 1 file changed, 1 insertion(+)
arch/x86/include/asm/traps.h +1 -0
index 3de69330e6c5..165c93dd700e 100644
@@ -162,5 +162,6 @@ enum x86_pf_error_code {
 	X86_PF_RSVD	=		1 << 3,
 	X86_PF_INSTR	=		1 << 4,
 	X86_PF_PK	=		1 << 5,
+	X86_PF_SGX	=		1 << 15,
 };
 #endif /* _ASM_X86_TRAPS_H */
Signature · 1 line
2.17.1
#13 lore raw

PATCH v14 7/19 x86/fault: x86/mm/pkeys: relocate stale comment regarding OSPKE

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

The code being referenced was moved to a separate helper function but the comment was left behind.

Fixes 33a709b25a76 ("mm/gup, x86/mm/pkeys: Check VMAs and PTEs for protection keys")
Cc Dave Hansen <[email protected]>
---
 arch/x86/mm/fault.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
arch/x86/mm/fault.c +5 -5
index 47bebfe6efa7..85d20516b2f3 100644
@@ -940,6 +940,11 @@ static inline bool bad_area_access_from_pkeys(unsigned long error_code,
 	/* This code is always called on the current mm */
 	bool foreign = false;
 
+	/*
+	 * This OSPKE check is not strictly necessary at runtime.
+	 * But, doing it this way allows compiler optimizations
+	 * if pkeys are compiled out.
+	 */
 	if (!boot_cpu_has(X86_FEATURE_OSPKE))
 		return false;
 	if (error_code & X86_PF_PK)
@@ -955,11 +960,6 @@ static noinline void
 bad_area_access_error(struct pt_regs *regs, unsigned long error_code,
 		      unsigned long address, struct vm_area_struct *vma)
 {
-	/*
-	 * This OSPKE check is not strictly necessary at runtime.
-	 * But, doing it this way allows compiler optimizations
-	 * if pkeys are compiled out.
-	 */
 	if (bad_area_access_from_pkeys(error_code, vma))
 		__bad_area(regs, error_code, address, vma, SEGV_PKUERR);
 	else
Signature · 1 line
2.17.1
#14 lore raw

PATCH v14 8/19 signal: x86/sgx: Add SIGSEGV siginfo code for SGX EPCM fault

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

The SGX Enclave Page Cache Map (EPCM) is a hardware-managed table that enforces accesses to an enclave's EPC page in addition to the software-managed kernel page tables, i.e. the effective permissions for an EPC page are a logical AND of the kernel's page tables and the corresponding EPCM entry. The primary purpose of the EPCM is to prevent a malcious or compromised kernel from attacking an enclave by modifying the enclave's page tables. The EPCM entires for an enclave are populated when the enclave is built and verified, using metadata provided by the enclave that is included in the measurement used to verify the enclave.

In normal operation of a properly functioning, non-malicious kernel (and enclave), the EPCM permissions will never trigger a fault, i.e. the kernel may make the permissions for an EPC page more restrictive, e.g. mark it not-present to swap out the EPC page, but the kernel will never make its permissions less restrictive.

But, there is a legitimate scenario in which the kernel's page tables can become less restrictive than the EPCM: on current hardware all enclaves are destroyed (by hardware) on a transition to S3 or lower sleep states, i.e. all EPCM entries are invalid (not-present) after the system resumes from its sleep state.

Unfortunately, on CPUs that support only SGX1, EPCM violations result in a #GP. The upside of the #GP is that no kernel changes are needed to deal with the EPCM being blasted away by hardware, e.g. userspace gets a SIGSEGV, assumes the EPCM was lost and restarts its enclave and everyone is happy. The downside is that userspace has to assume the SIGSEGV was because the EPC was lost (or possibly do some leg work to rule out other causes).

In SGX2, the oddity of delivering a #GP due to what are inherently paging related violations is remedied. CPUs that support SGX2 deliver EPCM violations as #PFs with a new SGX error code bit set. So, now that hardware provides us with a way to unequivocally determine that a fault was due to a EPCM violation, define a signfo code for SIGSEGV so that the information can be passed onto userspace.

---
 include/uapi/asm-generic/siginfo.h | 4 ++++
 1 file changed, 4 insertions(+)
include/uapi/asm-generic/siginfo.h +4 -0
index 80e2a7227205..fdd898e2325b 100644
@@ -225,7 +225,11 @@ typedef struct siginfo {
 #else
 # define SEGV_PKUERR	4	/* failed protection key checks */
 #endif
+#ifdef __x86_64__
+#define SEGV_SGXERR	5	/* SGX Enclave Page Cache Map fault */
+#else
 #define SEGV_ACCADI	5	/* ADI not enabled for mapped object */
+#endif
 #define SEGV_ADIDERR	6	/* Disrupting MCD error */
 #define SEGV_ADIPERR	7	/* Precise MCD exception */
 #define NSIGSEGV	7
Signature · 1 line
2.17.1
#15 lore raw

Re: [PATCH v14 08/19] signal: x86/sgx: Add SIGSEGV siginfo code for SGX EPCM fault

Quoted 52 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

The SGX Enclave Page Cache Map (EPCM) is a hardware-managed table that enforces accesses to an enclave's EPC page in addition to the software-managed kernel page tables, i.e. the effective permissions for an EPC page are a logical AND of the kernel's page tables and the corresponding EPCM entry. The primary purpose of the EPCM is to prevent a malcious or compromised kernel from attacking an enclave by modifying the enclave's page tables. The EPCM entires for an enclave are populated when the enclave is built and verified, using metadata provided by the enclave that is included in the measurement used to verify the enclave.

In normal operation of a properly functioning, non-malicious kernel (and enclave), the EPCM permissions will never trigger a fault, i.e. the kernel may make the permissions for an EPC page more restrictive, e.g. mark it not-present to swap out the EPC page, but the kernel will never make its permissions less restrictive.

But, there is a legitimate scenario in which the kernel's page tables can become less restrictive than the EPCM: on current hardware all enclaves are destroyed (by hardware) on a transition to S3 or lower sleep states, i.e. all EPCM entries are invalid (not-present) after the system resumes from its sleep state.

Unfortunately, on CPUs that support only SGX1, EPCM violations result in a #GP. The upside of the #GP is that no kernel changes are needed to deal with the EPCM being blasted away by hardware, e.g. userspace gets a SIGSEGV, assumes the EPCM was lost and restarts its enclave and everyone is happy. The downside is that userspace has to assume the SIGSEGV was because the EPC was lost (or possibly do some leg work to rule out other causes).

In SGX2, the oddity of delivering a #GP due to what are inherently paging related violations is remedied. CPUs that support SGX2 deliver EPCM violations as #PFs with a new SGX error code bit set. So, now that hardware provides us with a way to unequivocally determine that a fault was due to a EPCM violation, define a signfo code for SIGSEGV so that the information can be passed onto userspace.

---
 include/uapi/asm-generic/siginfo.h | 4 ++++
 1 file changed, 4 insertions(+)
include/uapi/asm-generic/siginfo.h +0 -0
index 80e2a7227205..fdd898e2325b 100644

include/uapi/asm-generic/siginfo.h typedef struct siginfo { · line 228

Quoted 4 lines
include/uapi/asm-generic/siginfo.h +1 -0
@@ -225,7 +225,11 @@ typedef struct siginfo {
 #else
 # define SEGV_PKUERR	4	/* failed protection key checks */
 #endif
+#ifdef __x86_64__

Argh, this needs to be "#if defined(__i386__) || defined(__x86_64__)" otherwise 32-bit builds break on later patches that use SEGV_SGXERR, e.g. the errors flagged by the 0-DAY bot.

Quoted 10 lines
include/uapi/asm-generic/siginfo.h +3 -0
+#define SEGV_SGXERR	5	/* SGX Enclave Page Cache Map fault */
+#else
 #define SEGV_ACCADI	5	/* ADI not enabled for mapped object */
+#endif
 #define SEGV_ADIDERR	6	/* Disrupting MCD error */
 #define SEGV_ADIPERR	7	/* Precise MCD exception */
 #define NSIGSEGV	7
Signature · 1 line
2.17.1
#16 lore raw

Re: [PATCH v14 08/19] signal: x86/sgx: Add SIGSEGV siginfo code for SGX EPCM fault

In reply to Jarkko Sakkinen
Quoted 52 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

The SGX Enclave Page Cache Map (EPCM) is a hardware-managed table that enforces accesses to an enclave's EPC page in addition to the software-managed kernel page tables, i.e. the effective permissions for an EPC page are a logical AND of the kernel's page tables and the corresponding EPCM entry. The primary purpose of the EPCM is to prevent a malcious or compromised kernel from attacking an enclave by modifying the enclave's page tables. The EPCM entires for an enclave are populated when the enclave is built and verified, using metadata provided by the enclave that is included in the measurement used to verify the enclave.

In normal operation of a properly functioning, non-malicious kernel (and enclave), the EPCM permissions will never trigger a fault, i.e. the kernel may make the permissions for an EPC page more restrictive, e.g. mark it not-present to swap out the EPC page, but the kernel will never make its permissions less restrictive.

But, there is a legitimate scenario in which the kernel's page tables can become less restrictive than the EPCM: on current hardware all enclaves are destroyed (by hardware) on a transition to S3 or lower sleep states, i.e. all EPCM entries are invalid (not-present) after the system resumes from its sleep state.

Unfortunately, on CPUs that support only SGX1, EPCM violations result in a #GP. The upside of the #GP is that no kernel changes are needed to deal with the EPCM being blasted away by hardware, e.g. userspace gets a SIGSEGV, assumes the EPCM was lost and restarts its enclave and everyone is happy. The downside is that userspace has to assume the SIGSEGV was because the EPC was lost (or possibly do some leg work to rule out other causes).

In SGX2, the oddity of delivering a #GP due to what are inherently paging related violations is remedied. CPUs that support SGX2 deliver EPCM violations as #PFs with a new SGX error code bit set. So, now that hardware provides us with a way to unequivocally determine that a fault was due to a EPCM violation, define a signfo code for SIGSEGV so that the information can be passed onto userspace.

---
 include/uapi/asm-generic/siginfo.h | 4 ++++
 1 file changed, 4 insertions(+)
include/uapi/asm-generic/siginfo.h +0 -0
index 80e2a7227205..fdd898e2325b 100644

include/uapi/asm-generic/siginfo.h typedef struct siginfo { · line 232

Quoted 8 lines
include/uapi/asm-generic/siginfo.h +4 -0
@@ -225,7 +225,11 @@ typedef struct siginfo {
 #else
 # define SEGV_PKUERR	4	/* failed protection key checks */
 #endif
+#ifdef __x86_64__
+#define SEGV_SGXERR	5	/* SGX Enclave Page Cache Map fault */
+#else
 #define SEGV_ACCADI	5	/* ADI not enabled for mapped object */
+#endif

Don't do this crazy ifdef thing. si_codes are not supposed to be per architecture. There are a few historical bugs but with a 32bit space it is just stupid to add #ifdefs.

Just set.
#define SEGV_SGXERR 8 and increase NSIGSEGV

Anything else is just asking for trouble. Especially when you want to get SGX working on itaninum.

include/uapi/asm-generic/siginfo.h typedef struct siginfo { · line 235

Quoted 3 lines
include/uapi/asm-generic/siginfo.h +0 -0
 #define SEGV_ADIDERR	6	/* Disrupting MCD error */
 #define SEGV_ADIPERR	7	/* Precise MCD exception */
 #define NSIGSEGV	7

Eric

#17 lore raw

PATCH v14 9/19 x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Signal SIGSEGV(SEGV_SGXERR) for all faults with PF_SGX set in the error code. The PF_SGX bit is set if and only if the #PF is detected by the Enclave Page Cache Map (EPCM), which is consulted only after an access walks the kernel's page tables, i.e.:

a. the access was allowed by the kernel
b. the kernel's tables have become less restrictive than the EPCM c. the kernel cannot fixup the cause of the fault

Noteably, (b) implies that either the kernel has botched the EPC mappings or the EPCM has been invalidated due to a power event. In either case, userspace needs to be alerted so that it can take appropriate action, e.g. restart the enclave. This is reinforced by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

---
 arch/x86/mm/fault.c | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)
arch/x86/mm/fault.c +17 -3
index 85d20516b2f3..3fb2b2838d6c 100644
@@ -960,10 +960,13 @@ static noinline void
 bad_area_access_error(struct pt_regs *regs, unsigned long error_code,
 		      unsigned long address, struct vm_area_struct *vma)
 {
+	int si_code = SEGV_ACCERR;
+
 	if (bad_area_access_from_pkeys(error_code, vma))
-		__bad_area(regs, error_code, address, vma, SEGV_PKUERR);
-	else
-		__bad_area(regs, error_code, address, vma, SEGV_ACCERR);
+		si_code = SEGV_PKUERR;
+	else if (unlikely(error_code & X86_PF_SGX))
+		si_code = SEGV_SGXERR;
+	__bad_area(regs, error_code, address, vma, si_code);
 }
 
 static void
@@ -1153,6 +1156,17 @@ access_error(unsigned long error_code, struct vm_area_struct *vma)
 	if (error_code & X86_PF_PK)
 		return 1;
 
+	/*
+	 * Access is blocked by the Enclave Page Cache Map (EPCM),
+	 * i.e. the access is allowed by the PTE but not the EPCM.
+	 * This usually happens when the EPCM is yanked out from
+	 * under us, e.g. by hardware after a suspend/resume cycle.
+	 * In any case, there is nothing that can be done by the
+	 * kernel to resolve the fault (short of killing the task).
+	 */
+	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
Signature · 1 line
2.17.1
maintainer
#18 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Minor nit:

Quoted 6 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

Not killing the task is quite nice, but...

arch/x86/mm/fault.c access_error(unsigned long error_code, struct vm_area_struct *vma) · line 1165

Quoted 7 lines
arch/x86/mm/fault.c +7 -0
+       /*
+        * Access is blocked by the Enclave Page Cache Map (EPCM),
+        * i.e. the access is allowed by the PTE but not the EPCM.
+        * This usually happens when the EPCM is yanked out from
+        * under us, e.g. by hardware after a suspend/resume cycle.
+        * In any case, there is nothing that can be done by the
+        * kernel to resolve the fault (short of killing the task).

Maybe s/killing the task/sending a signal/?

--Andy

#19 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 21 lines · Andy Lutomirski

Minor nit:

Quoted 6 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

Not killing the task is quite nice, but...

arch/x86/mm/fault.c access_error(unsigned long error_code, struct vm_area_struct *vma) · line 1165

Quoted 7 lines
arch/x86/mm/fault.c +7 -0
+       /*
+        * Access is blocked by the Enclave Page Cache Map (EPCM),
+        * i.e. the access is allowed by the PTE but not the EPCM.
+        * This usually happens when the EPCM is yanked out from
+        * under us, e.g. by hardware after a suspend/resume cycle.
+        * In any case, there is nothing that can be done by the
+        * kernel to resolve the fault (short of killing the task).
Quoted 1 line

Maybe s/killing the task/sending a signal/?

My intent was to document that, unlike all other page faults, the kernel can't fix the source of the fault even if it were omniscient.

How about this? With formatting changes since it's long-winded...

/*
* Access is blocked by the Enclave Page Cache Map (EPCM), i.e. the
* access is allowed by the PTE but not the EPCM. This usually happens
* when the EPCM is yanked out from under us, e.g. by hardware after a
* suspend/resume cycle. In any case, software, i.e. the kernel, can't
* fix the source of the fault as the EPCM can't be directly modified
* by software. Handle the fault as an access error in order to signal
* userspace, e.g. so that userspace can rebuild their enclave(s), even
* though userspace may not have actually violated access permissions. */

#20 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 2 lines

On Sep 26, 2018, at 10:35 AM, Sean Christopherson <sean.j.christopherson@=

Quoted 24 lines · intel.com> wrote:

=20

Quoted 21 lines · Andy Lutomirski

Minor nit:
=20

Quoted 3 lines · Jarkko Sakkinen

=20
From: Sean Christopherson <[email protected]> =20

=20

Quoted 2 lines

by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

=20
Not killing the task is quite nice, but...
=20

arch/x86/mm/fault.c access_error(unsigned long error_code, struct vm_area_struct *vma) · line 1165

Quoted 7 lines
arch/x86/mm/fault.c +7 -0
+       /*
+        * Access is blocked by the Enclave Page Cache Map (EPCM),
+        * i.e. the access is allowed by the PTE but not the EPCM.
+        * This usually happens when the EPCM is yanked out from
+        * under us, e.g. by hardware after a suspend/resume cycle.
+        * In any case, there is nothing that can be done by the
+        * kernel to resolve the fault (short of killing the task).
Quoted 2 lines

=20
Maybe s/killing the task/sending a signal/?

Quoted 8 lines

=20
My intent was to document that, unlike all other page faults, the kernel can't fix the source of the fault even if it were omniscient. =20
How about this? With formatting changes since it's long-winded... =20
/*
* Access is blocked by the Enclave Page Cache Map (EPCM), i.e. th=

e

Quoted 1 line

* access is allowed by the PTE but not the EPCM. This usually ha=

ppens

Quoted 1 line

* when the EPCM is yanked out from under us, e.g. by hardware aft=

er a

Quoted 1 line

* suspend/resume cycle. In any case, software, i.e. the kernel, =

can't

Quoted 1 line

* fix the source of the fault as the EPCM can't be directly modif=

ied

Quoted 1 line

* by software. Handle the fault as an access error in order to s=

ignal

Quoted 1 line

* userspace, e.g. so that userspace can rebuild their enclave(s),=

even

Quoted 1 line

* though userspace may not have actually violated access permissi=

ons.

Quoted 2 lines
        */

=20

Looks good to me.=

#21 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 15 lines · Andy Lutomirski
Quoted 4 lines

e omniscient.

How about this? With formatting changes since it's long-winded...

Quoted 11 lines

/*
* Access is blocked by the Enclave Page Cache Map (EPCM), i.e. the
* access is allowed by the PTE but not the EPCM. This usually happens
* when the EPCM is yanked out from under us, e.g. by hardware after a
* suspend/resume cycle. In any case, software, i.e. the kernel, can't
* fix the source of the fault as the EPCM can't be directly modified
* by software. Handle the fault as an access error in order to signal
* userspace, e.g. so that userspace can rebuild their enclave(s), even
* though userspace may not have actually violated access permissions. */

Quoted 1 line

Looks good to me.

Including the actual architectural definition of the bit might add some clarity. The SDM explicitly says (Vol 3a section 4.7):

The fault resulted from violation of SGX-specific access-control requirements.

Which totally squares with returning true from access_error().

There's also a tidbit that says:

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

#22 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 33 lines · Dave Hansen
Quoted 15 lines · Andy Lutomirski
Quoted 4 lines

e omniscient.

How about this? With formatting changes since it's long-winded...

Quoted 11 lines

/*
* Access is blocked by the Enclave Page Cache Map (EPCM), i.e. the
* access is allowed by the PTE but not the EPCM. This usually happens
* when the EPCM is yanked out from under us, e.g. by hardware after a
* suspend/resume cycle. In any case, software, i.e. the kernel, can't
* fix the source of the fault as the EPCM can't be directly modified
* by software. Handle the fault as an access error in order to signal
* userspace, e.g. so that userspace can rebuild their enclave(s), even
* though userspace may not have actually violated access permissions. */

Quoted 1 line

Looks good to me.

Including the actual architectural definition of the bit might add some clarity. The SDM explicitly says (Vol 3a section 4.7):

The fault resulted from violation of SGX-specific access-control requirements.

Which totally squares with returning true from access_error().

There's also a tidbit that says:

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

Quoted 6 lines

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

Are you looking for something in the changelog, the comment, or just a response? If it's the latter...

On bare metal with a bug-free kernel, the only scenario I'm aware of where we'll encounter these faults is when hardware pulls the rug out from under us. In a virtualized environment all bets are off because the architecture allows VMMs to silently "destroy" the EPC at will, e.g. KVM, and I believe Hyper-V, will take advantage of this behavior to support live migration. Post migration, the destination system will generate PF_SGX because the EPC{M} can't be migrated between system, i.e. the destination EPCM sees all EPC pages as invalid.

#23 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 7 lines · Sean Christopherson
Quoted 3 lines · Dave Hansen

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

Are you looking for something in the changelog, the comment, or just a response? If it's the latter...

Comments, please.

Quoted 8 lines

On bare metal with a bug-free kernel, the only scenario I'm aware of where we'll encounter these faults is when hardware pulls the rug out from under us. In a virtualized environment all bets are off because the architecture allows VMMs to silently "destroy" the EPC at will, e.g. KVM, and I believe Hyper-V, will take advantage of this behavior to support live migration. Post migration, the destination system will generate PF_SGX because the EPC{M} can't be migrated between system, i.e. the destination EPCM sees all EPC pages as invalid.

OK, cool.

That's good background fodder for the changelog.

But, for the comment, I'm happy with something like this:

/*
* The fault resulted from violation of SGX-specific access-
* controls. This is expected to be the result of some lower
* layer action (CPU suspend/resume, VM migration) and is
* not related to anything the OS did. Treat it as an access
* error to ensure it is passed up to the app via a signal where
* it can be handled.
*/

I really don't think we need to delve too deeply into the relationship between EPCM and PTEs or anything. Let's just say, "it's not the kernel's fault, it's not the app's fault, so throw up our hands".

#24 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 36 lines · Dave Hansen
Quoted 7 lines · Sean Christopherson
Quoted 3 lines · Dave Hansen

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

Are you looking for something in the changelog, the comment, or just a response? If it's the latter...

Comments, please.

Quoted 8 lines

On bare metal with a bug-free kernel, the only scenario I'm aware of where we'll encounter these faults is when hardware pulls the rug out from under us. In a virtualized environment all bets are off because the architecture allows VMMs to silently "destroy" the EPC at will, e.g. KVM, and I believe Hyper-V, will take advantage of this behavior to support live migration. Post migration, the destination system will generate PF_SGX because the EPC{M} can't be migrated between system, i.e. the destination EPCM sees all EPC pages as invalid.

OK, cool.

That's good background fodder for the changelog.

But, for the comment, I'm happy with something like this:

/*
* The fault resulted from violation of SGX-specific access-
* controls. This is expected to be the result of some lower
* layer action (CPU suspend/resume, VM migration) and is
* not related to anything the OS did. Treat it as an access
* error to ensure it is passed up to the app via a signal where
* it can be handled.
*/

Quoted 3 lines

I really don't think we need to delve too deeply into the relationship between EPCM and PTEs or anything. Let's just say, "it's not the kernel's fault, it's not the app's fault, so throw up our hands".

There is a non-nitpicky consideration here. Logically, user code is going to do this (totally made-up pseudocode):

enclave_t enclave = load_and_init_enclave(...);
int ret = sgx_run(enclave, some pointers to non-enclave-memory buffers, ...);

and, with the code in this patch, a correct implementation of sgx_run() requires installing a signal handler. This is nasty, since signal handlers, expecially for something like SIGSEGV or SIGBUS, are not fantastic to say the least in libraries.

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

On Windows, this is much less of a concern, because Windows has real scoped fault handling. But Linux doesn't, at least not yet.

--
Andy Lutomirski
AMA Capital Management, LLC

#25 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 4 lines · Andy Lutomirski

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

Yeah, signals suck.

So, instead of doing the enclave entry instruction (EENTER is it?), the app would do the vDSO call. It would have some calling convention, like "set %rax to 0 before entering". Then, we just teach the page fault handler about the %RIP in the vDSO that can fault and how to move one instruction later, munge %RIP to a value that tells about the error, then return from the fault. It would basically be like the kernel exception tables, but for userspace. Right?

How would a syscall work, though? I assume we can't just enter the enclave from ring0.

#26 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 9 lines · Dave Hansen
Quoted 4 lines · Andy Lutomirski

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

Yeah, signals suck.

Quoted 7 lines

So, instead of doing the enclave entry instruction (EENTER is it?), the app would do the vDSO call. It would have some calling convention, like "set %rax to 0 before entering". Then, we just teach the page fault handler about the %RIP in the vDSO that can fault and how to move one instruction later, munge %RIP to a value that tells about the error, then return from the fault. It would basically be like the kernel exception tables, but for userspace. Right?

Yeah. Maybe like this:

xorl %eax,%eax
eenter_insn:
ENCLU[whatever]
eenter_landing_pad:
ret

And the kernel would use the existing vdso2c vdso-symbol-finding mechanism to do the fixup.

Quoted 3 lines

How would a syscall work, though? I assume we can't just enter the enclave from ring0.

My understanding of how AEX works is a bit vague, but maybe a syscall could reuse the mechanism? The vDSO approach seems considerably simpler.

We do need to make sure that a fault that happens on or after return from an AEX event does the right thing. But I'm still vague on how that works, sigh.

--Andy

#27 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 16 lines · Andy Lutomirski

Yeah. Maybe like this: > > xorl %eax,%eax > eenter_insn:
ENCLU[whatever]
eenter_landing_pad:
ret

And the kernel would use the existing vdso2c vdso-symbol-finding mechanism to do the fixup.

Quoted 3 lines

How would a syscall work, though? I assume we can't just enter the enclave from ring0.

My understanding of how AEX works is a bit vague, but maybe a syscall could reuse the mechanism? The vDSO approach seems considerably simpler.

Quoted 5 lines

We do need to make sure that a fault that happens on or after return from an AEX event does the right thing. But I'm still vague on how that works, sigh.

--Andy

Returning from AEX does not differ from any other memory access event so AFAIK it should be handled right with the proposed solution already. For convenience I think we could have a fixed trampoline for AEX e.g. this how it is implemented in the open source LE that I did:

sgx_get_token:

	push	%rbx
	mov	$0x02, %rax
	mov	%rsi, %rbx
	mov	%rdx, %rsi
	mov	$sgx_async_exit, %rcx

sgx_async_exit:
ENCLU
pop %rbx
ret

BTW, if I converted the in-kernel LE as a standalone test program, would that be useful for basic testing of the series?

/Jarkko

#28 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 3 lines

On Sep 27, 2018, at 7:21 AM, Jarkko Sakkinen <[email protected]=

Quoted 35 lines · .com> wrote:

=20

Quoted 12 lines · Andy Lutomirski

Yeah. Maybe like this: > > xorl %eax,%eax > eenter_insn:
ENCLU[whatever]
eenter_landing_pad:
ret
=20
And the kernel would use the existing vdso2c vdso-symbol-finding mechanism to do the fixup.
=20

Quoted 3 lines

=20
How would a syscall work, though? I assume we can't just enter the enclave from ring0.

Quoted 10 lines

=20
My understanding of how AEX works is a bit vague, but maybe a syscall could reuse the mechanism? The vDSO approach seems considerably simpler.
=20
We do need to make sure that a fault that happens on or after return from an AEX event does the right thing. But I'm still vague on how that works, sigh.
=20
--Andy

=20
Returning from AEX does not differ from any other memory access event so AFAIK it should be handled right with the proposed solution already. For convenience I think we could have a fixed trampoline for AEX e.g. this how it is implemented in the open source LE that I did:
=20
sgx_get_token:
push %rbx

   mov    $0x02, %rax
   mov    %rsi, %rbx
   mov    %rdx, %rsi
   mov    $sgx_async_exit, %rcx
Quoted 9 lines

sgx_async_exit:
ENCLU
pop %rbx
ret
=20
BTW, if I converted the in-kernel LE as a standalone test program, would that be useful for basic testing of the series?
=20
=20

Definitely. Especially if you stick it in selftests/x86 and make it exit cl= eanly (error code 0) on unsupported hardware.=

#29 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Dave Hansen
Quoted 16 lines · Dave Hansen
Quoted 4 lines · Andy Lutomirski

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

Yeah, signals suck.

So, instead of doing the enclave entry instruction (EENTER is it?), the app would do the vDSO call. It would have some calling convention, like "set %rax to 0 before entering". Then, we just teach the page fault handler about the %RIP in the vDSO that can fault and how to move one instruction later, munge %RIP to a value that tells about the error, then return from the fault. It would basically be like the kernel exception tables, but for userspace. Right?

Quoted 2 lines

How would a syscall work, though? I assume we can't just enter the enclave from ring0.

Enclave cannot be entered from ring-0.

For me this plan sounds simple and sound.

/Jarkko

#30 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 8 lines · Jarkko Sakkinen
Quoted 5 lines · Dave Hansen
Quoted 4 lines · Andy Lutomirski

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

For me this plan sounds simple and sound.

I support avoiding the need for a signal handler for various SGX-specific operations. Looking forward to v15.

I have some thoughts regarding the design of the vDSO function. Please consider the following as you work on the next patch set.

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

2) All the information that would normally be provided through the signal handler (x86 fault number, reason) should be provided to userspace.

3) vDSO functions should be provided for all standard non-enclave ENCLU leafs, and should support most ways that an application might want to use. This includes:

* EENTER with a automatic AEX handler (as in Jarkko's sgx_get_token example)
* EENTER & ERESUME with a user-specified AEX handler, or possibly just a special return value from the ENCLU function on AEX

4) I think the vDSO functions should have a special calling convention (not conforming to the standard SysV ABI), such that most registers are passed between user space and enclave space untouched. Basically, only RAX, RBX, RCX are available as input and output registers.

--
Jethro Beekman | Fortanix

#31 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 4 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

Why?

#32 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 7 lines · Dave Hansen
Quoted 4 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

Why?

Circling back to this question, what if we take the easy way out and simply signal SIGSEGV without an SGX-specific code? I.e. treat #PF with X86_PF_SGX as an access error, no more no less. That should be sufficient for userspace to function, albeit with a little more effort, but presumably no more than would be needed to run on SGX1 hardware.

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported, but that seems like cutting off the nose to spite the face. Supporting userspace EENTER in a limited capacity would allow people to do whatever crazy tricks they're wont to do without having to deal with absurd requests for the vDSO interface.

If we go this route we could also add the vDSO stuff after basic SGX support is in mainline, obviously with approval from the powers that be.

#33 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 7 lines · Sean Christopherson
Quoted 6 lines · Dave Hansen
Quoted 4 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

Quoted 7 lines
Quoted 1 line

Why?

Circling back to this question, what if we take the easy way out and simply signal SIGSEGV without an SGX-specific code? I.e. treat #PF with X86_PF_SGX as an access error, no more no less. That should be sufficient for userspace to function, albeit with a little more effort, but presumably no more than would be needed to run on SGX1 hardware.

There are two sides to this ABI: what the kernel does to support SGX and what userspace does. If we do what you suggest, we remove any (most?) needed kernel changes and foist the burden entirely into userspace. But, we end up with two ABIs: the old one and the new vDSO one.

IOW, once we start doing SIGSEGV, we have to do it forever, despite if we have a newer mechanism.

Quoted 3 lines

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported,

Yep, userspace can call it all it wants, and we can also say that calling it outside the vdso is "undefined".

#34 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 7 lines · Dave Hansen
Quoted 3 lines · Sean Christopherson

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported,

Yep, userspace can call it all it wants, and we can also say that calling it outside the vdso is "undefined".

Is there a precedent for this? Are there any other ring 3 x86 instructions that Linux is claiming to be "undefined" when executed by a user process?

--
Jethro Beekman | Fortanix

#35 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 6 lines · Jethro Beekman
Quoted 5 lines · Dave Hansen
Quoted 3 lines · Sean Christopherson

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported,

Quoted 6 lines
Quoted 2 lines

Yep, userspace can call it all it wants, and we can also say that calling it outside the vdso is "undefined".

Is there a precedent for this? Are there any other ring 3 x86 instructions that Linux is claiming to be "undefined" when executed by a user process?

We did it for MPX. "Don't use MPX unless you first tell the kernel, or we'll eat your puppy."

#36 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 12 lines · Dave Hansen

=20

Quoted 9 lines · Jethro Beekman
Quoted 8 lines · Dave Hansen
Quoted 4 lines · Sean Christopherson

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported,

=20
Yep, userspace can call it all it wants, and we can also say that calling it outside the vdso is "undefined".

Quoted 7 lines
Quoted 4 lines

=20
Is there a precedent for this? Are there any other ring 3 x86 instructions that Linux is claiming to be "undefined" when executed by a user process?

=20
We did it for MPX. "Don't use MPX unless you first tell the kernel, or we'll eat your puppy."

I think EENTER in plain user code should have well defined semantics, and i= t should get regular signals with the appropriate bits set in the error cod= e field in the ucontext. But we should probably simultaneously offer a nic= er API, and the libraries will use it because it=E2=80=99s nicer.=

#37 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 5 lines · Andy Lutomirski

I think EENTER in plain user code should have well defined semantics, and it should get regular signals with the appropriate bits set in the error code field in the ucontext. But we should probably simultaneously offer a nicer API, and the libraries will use it because it’s nicer.

FWIW, if we have a signal-based version and a VDSO-based version, nobody will use the VDSO one.

The Intel libraries are surely going to keep using the approach they've been using for years and I doubt their owners will be tempted even by a simpler interface to change one line of code.

If we want to do the VDSO route, I think we probably need to go whole-hog and toss the signal-based one.

maintainer
#38 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 11 lines · Dave Hansen
Quoted 5 lines · Andy Lutomirski

I think EENTER in plain user code should have well defined semantics, and it should get regular signals with the appropriate bits set in the error code field in the ucontext. But we should probably simultaneously offer a nicer API, and the libraries will use it because it=E2=80=99s nicer.

FWIW, if we have a signal-based version and a VDSO-based version, nobody will use the VDSO one.

Quoted 6 lines

The Intel libraries are surely going to keep using the approach they've been using for years and I doubt their owners will be tempted even by a simpler interface to change one line of code.

If we want to do the VDSO route, I think we probably need to go whole-hog and toss the signal-based one.

That's a fair point. We don't want a situation where a widely-used SGX library registers a SIGSEGV handler.

#39 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Sean Christopherson
Quoted 23 lines · Sean Christopherson
Quoted 7 lines · Dave Hansen
Quoted 4 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

Why?

Circling back to this question, what if we take the easy way out and simply signal SIGSEGV without an SGX-specific code? I.e. treat #PF with X86_PF_SGX as an access error, no more no less. That should be sufficient for userspace to function, albeit with a little more effort, but presumably no more than would be needed to run on SGX1 hardware.

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported, but that seems like cutting off the nose to spite the face. Supporting userspace EENTER in a limited capacity would allow people to do whatever crazy tricks they're wont to do without having to deal with absurd requests for the vDSO interface.

Quoted 4 lines

If we go this route we could also add the vDSO stuff after basic SGX support is in mainline, obviously with approval from the powers that be.

Yeah, this would give stable behavior when vDSO functions are not available.

Here's a question: if we implement this behavior, could be upstream series without vDSO's first and after those changes have been landed we would continue with the vDSO's?

/Jarkko

#40 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 29 lines · Jarkko Sakkinen
Quoted 23 lines · Sean Christopherson
Quoted 7 lines · Dave Hansen
Quoted 4 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

Why?

Circling back to this question, what if we take the easy way out and simply signal SIGSEGV without an SGX-specific code? I.e. treat #PF with X86_PF_SGX as an access error, no more no less. That should be sufficient for userspace to function, albeit with a little more effort, but presumably no more than would be needed to run on SGX1 hardware.

AFAIK there isn't a way to prevent userspace from manually invoking EENTER, short of doing some really nasty text poking or PTE swizzling. We could declare using EENTER as unsupported, but that seems like cutting off the nose to spite the face. Supporting userspace EENTER in a limited capacity would allow people to do whatever crazy tricks they're wont to do without having to deal with absurd requests for the vDSO interface.

Quoted 4 lines

If we go this route we could also add the vDSO stuff after basic SGX support is in mainline, obviously with approval from the powers that be.

Quoted 6 lines

Yeah, this would give stable behavior when vDSO functions are not available.

Here's a question: if we implement this behavior, could be upstream series without vDSO's first and after those changes have been landed we would continue with the vDSO's?

Right, it was in your last paragraph, sorry. Yeah, I fully support this idea. It will be easier also to work on the vDSO's once we have something landed (instead of working on a moving platform).

/Jarkko

#41 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Jethro Beekman
Quoted 3 lines · Jethro Beekman

1) Even though the vDSO function exists, userspace may still call ENCLU[EENTER] manually, so the fault handling as described in the current patch should also be maintained.

You mean the way it was is in v13 and not the way it is in v14?

Quoted 2 lines

2) All the information that would normally be provided through the signal handler (x86 fault number, reason) should be provided to userspace.

As I've understood it, this should be just a change in the delivery path.

/Jarkko

#42 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Andy Lutomirski
Quoted 52 lines · Andy Lutomirski
Quoted 37 lines · Dave Hansen
Quoted 8 lines · Sean Christopherson
Quoted 4 lines · Dave Hansen

We also need to clarify how this can happen.  Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

Are you looking for something in the changelog, the comment, or just a response?  If it's the latter...

Comments, please.

Quoted 9 lines

On bare metal with a bug-free kernel, the only scenario I'm aware of where we'll encounter these faults is when hardware pulls the rug out from under us.  In a virtualized environment all bets are off because the architecture allows VMMs to silently "destroy" the EPC at will, e.g. KVM, and I believe Hyper-V, will take advantage of this behavior to support live migration.  Post migration, the destination system will generate PF_SGX because the EPC{M} can't be migrated between system, i.e. the destination EPCM sees all EPC pages as invalid.

OK, cool.

That's good background fodder for the changelog.

But, for the comment, I'm happy with something like this:

        /*
         * The fault resulted from violation of SGX-specific access-
         * controls.  This is expected to be the result of some lower
         * layer action (CPU suspend/resume, VM migration) and is
         * not related to anything the OS did.  Treat it as an access
         * error to ensure it is passed up to the app via a signal where
         * it can be handled.
         */

Quoted 3 lines

I really don't think we need to delve too deeply into the relationship between EPCM and PTEs or anything.  Let's just say, "it's not the kernel's fault, it's not the app's fault, so throw up our hands".

There is a non-nitpicky consideration here.  Logically, user code is going to do this (totally made-up pseudocode):

enclave_t enclave = load_and_init_enclave(...);
int ret = sgx_run(enclave, some pointers to non-enclave-memory buffers, ...);

and, with the code in this patch, a correct implementation of sgx_run() requires installing a signal handler.  This is nasty, since signal handlers, expecially for something like SIGSEGV or SIGBUS, are not fantastic to say the least in libraries.

Quoted 4 lines

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

If we want to avoid having to install a signal handler then I'm pretty sure we'd need to fixup all #GPs and "bad access" #PFs that occur on EENTER or in the enclave, not just PF_SGX faults.  SGX1 hardware takes a #GP instead of a #PF on EPCM faults, and SGX2 hardware allows enclaves to allocate/free/adjust EPC pages at runtime, e.g. an enclave runtime might want to intercept #PFs from within the enclave so that the enclave can dynamically grow its stack.

Quoted 7 lines

On Windows, this is much less of a concern, because Windows has real scoped fault handling. But Linux doesn't, at least not yet.

--
Andy Lutomirski
AMA Capital Management, LLC

#43 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 5 lines · Sean Christopherson
Quoted 4 lines

Could we perhaps have a little vDSO entry (or syscall, I suppose) that runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

Quoted 7 lines

If we want to avoid having to install a signal handler then I'm pretty sure we'd need to fixup all #GPs and "bad access" #PFs that occur on EENTER or in the enclave, not just PF_SGX faults.  SGX1 hardware takes a #GP instead of a #PF on EPCM faults, and SGX2 hardware allows enclaves to allocate/free/adjust EPC pages at runtime, e.g. an enclave runtime might want to intercept #PFs from within the enclave so that the enclave can dynamically grow its stack.

I think the technique Andy describes can be used for that as well. It basically works for any case where we know which instructions will take an exception (any exception), call the instruction from a fixed location, and know the fault(s) it can throw.

To me, it's almost like turning these faulting instructions into mini syscall instructions. They enter the kernel only when they need help, though, instead of always.

#44 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Sean Christopherson
Quoted 6 lines · Sean Christopherson
Quoted 3 lines · Andy Lutomirski

runs an enclave an returns an error code, and rig up the #PF handler to check if the error happened in the vDSO entry and fix it up rather than sending a signal?

Quoted 7 lines

If we want to avoid having to install a signal handler then I'm pretty sure we'd need to fixup all #GPs and "bad access" #PFs that occur on EENTER or in the enclave, not just PF_SGX faults.  SGX1 hardware takes a #GP instead of a #PF on EPCM faults, and SGX2 hardware allows enclaves to allocate/free/adjust EPC pages at runtime, e.g. an enclave runtime might want to intercept #PFs from within the enclave so that the enclave can dynamically grow its stack.

If I've understood Andy's proposal correctly, the run-time would get the same information as with a signal. The delivery path for this information would be just different.

/Jarkko

#45 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Dave Hansen
Quoted 33 lines · Dave Hansen
Quoted 15 lines · Andy Lutomirski
Quoted 4 lines

e omniscient.

How about this? With formatting changes since it's long-winded...

Quoted 11 lines

/*
* Access is blocked by the Enclave Page Cache Map (EPCM), i.e. the
* access is allowed by the PTE but not the EPCM. This usually happens
* when the EPCM is yanked out from under us, e.g. by hardware after a
* suspend/resume cycle. In any case, software, i.e. the kernel, can't
* fix the source of the fault as the EPCM can't be directly modified
* by software. Handle the fault as an access error in order to signal
* userspace, e.g. so that userspace can rebuild their enclave(s), even
* though userspace may not have actually violated access permissions. */

Quoted 1 line

Looks good to me.

Including the actual architectural definition of the bit might add some clarity. The SDM explicitly says (Vol 3a section 4.7):

The fault resulted from violation of SGX-specific access-control requirements.

Which totally squares with returning true from access_error().

There's also a tidbit that says:

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

Quoted 6 lines

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

When you change page permissions lets say with mprotect after the and try to do an invalid access according to the EPCM permissions this can happen.

/Jarkko

#46 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 8 lines · Jarkko Sakkinen
Quoted 8 lines

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

Quoted 6 lines
Quoted 3 lines

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

When you change page permissions lets say with mprotect after the and try to do an invalid access according to the EPCM permissions this can happen.

So, there are pages that are non-executable, non-readable, or non-writable both via the page tables and via underlying SGX permissions. Then, we allow an mprotect() and a later access will result in one of these SGX faults?

What permissions are these, exactly? Is it even a good idea to let that mprotect() go through in the first place?

Either way, it sounds like we have some new conditions to spell out in that comment.

#47 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 12 lines · Dave Hansen
Quoted 11 lines · Jarkko Sakkinen
Quoted 11 lines

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

Quoted 8 lines
Quoted 3 lines

When you change page permissions lets say with mprotect after the and try to do an invalid access according to the EPCM permissions this can happen.

So, there are pages that are non-executable, non-readable, or non-writable both via the page tables and via underlying SGX permissions. Then, we allow an mprotect() and a later access will result in one of these SGX faults?

The permissions are intersection of PTE and EPCM permissions.

EPCM permissions are part of the enclave measurement. For SGX1 they are static. For SGX2 they can be changed with EMODPR/EACCEPT protocol (i.e. measurement can be updated after enclave initialization).

Quoted 2 lines

What permissions are these, exactly? Is it even a good idea to let that mprotect() go through in the first place?

You define RWX for each page when you do EADD.

Quoted 2 lines

Either way, it sounds like we have some new conditions to spell out in that comment.

Agreed.

/Jarkko

#48 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 17 lines · Jarkko Sakkinen
Quoted 16 lines · Dave Hansen
Quoted 14 lines · Jarkko Sakkinen
Quoted 11 lines

This flag is 1 if the exception is unrelated to paging and resulted from violation of SGX-specific access-control requirements. ... such a violation can occur only if there is no ordinary page fault...

This is pretty important. It means that *none* of the other paging-related stuff that we're doing applies.

We also need to clarify how this can happen. Is it through something than an app does, or is it solely when the hardware does something under the covers, like suspend/resume.

When you change page permissions lets say with mprotect after the and try to do an invalid access according to the EPCM permissions this can happen.

Quoted 6 lines
Quoted 4 lines

So, there are pages that are non-executable, non-readable, or non-writable both via the page tables and via underlying SGX permissions. Then, we allow an mprotect() and a later access will result in one of these SGX faults?

The permissions are intersection of PTE and EPCM permissions.

Right, but this *fault* bit is not.

Quoted 3 lines

EPCM permissions are part of the enclave measurement. For SGX1 they are static. For SGX2 they can be changed with EMODPR/EACCEPT protocol (i.e. measurement can be updated after enclave initialization).

What does this all have to do with enclave measurement?

Quoted 4 lines
Quoted 2 lines

What permissions are these, exactly? Is it even a good idea to let that mprotect() go through in the first place?

You define RWX for each page when you do EADD.

Are those permissions reflected into the VMAs mapping the enclave memory?

#49 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Andy Lutomirski
Quoted 21 lines · Andy Lutomirski

Minor nit:

Quoted 6 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

Not killing the task is quite nice, but...

arch/x86/mm/fault.c access_error(unsigned long error_code, struct vm_area_struct *vma) · line 1165

Quoted 7 lines
arch/x86/mm/fault.c +7 -0
+       /*
+        * Access is blocked by the Enclave Page Cache Map (EPCM),
+        * i.e. the access is allowed by the PTE but not the EPCM.
+        * This usually happens when the EPCM is yanked out from
+        * under us, e.g. by hardware after a suspend/resume cycle.
+        * In any case, there is nothing that can be done by the
+        * kernel to resolve the fault (short of killing the task).
Quoted 1 line

Maybe s/killing the task/sending a signal/?

Agreed. Thanks.

/Jarkko

#50 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

In reply to Jarkko Sakkinen
Quoted 34 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Signal SIGSEGV(SEGV_SGXERR) for all faults with PF_SGX set in the error code. The PF_SGX bit is set if and only if the #PF is detected by the Enclave Page Cache Map (EPCM), which is consulted only after an access walks the kernel's page tables, i.e.:

a. the access was allowed by the kernel
b. the kernel's tables have become less restrictive than the EPCM c. the kernel cannot fixup the cause of the fault

Noteably, (b) implies that either the kernel has botched the EPC mappings or the EPCM has been invalidated due to a power event. In either case, userspace needs to be alerted so that it can take appropriate action, e.g. restart the enclave. This is reinforced by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

---
 arch/x86/mm/fault.c | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)
arch/x86/mm/fault.c +1 -0
index 85d20516b2f3..3fb2b2838d6c 100644
@@ -960,10 +960,13 @@ static noinline void
 bad_area_access_error(struct pt_regs *regs, unsigned long error_code,
 		      unsigned long address, struct vm_area_struct *vma)
 {
+	int si_code = SEGV_ACCERR;

arch/x86/mm/fault.c static noinline void · line 970

Quoted 10 lines
arch/x86/mm/fault.c +5 -3
+
 	if (bad_area_access_from_pkeys(error_code, vma))
-		__bad_area(regs, error_code, address, vma, SEGV_PKUERR);
-	else
-		__bad_area(regs, error_code, address, vma, SEGV_ACCERR);
+		si_code = SEGV_PKUERR;
+	else if (unlikely(error_code & X86_PF_SGX))
+		si_code = SEGV_SGXERR;
+	__bad_area(regs, error_code, address, vma, si_code);
 }

This conflicts with a cleanup in this area I have sitting in linux-next. It isn't in the x86 tree but you can find my siginfo tree at:

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git siginfo-next

In my tree bad area no longer takes a vma parameter.

If you are going to make changes to the fault handling code this cycle please let's figure out how to build it on top of my clean ups.

Thank you,
Eric Biederman

#51 lore raw

Re: [PATCH v14 09/19] x86/mm: x86/sgx: Signal SEGV_SGXERR for #PFs w/ PF_SGX

Quoted 50 lines · Eric W. Biederman
Quoted 34 lines · Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Signal SIGSEGV(SEGV_SGXERR) for all faults with PF_SGX set in the error code. The PF_SGX bit is set if and only if the #PF is detected by the Enclave Page Cache Map (EPCM), which is consulted only after an access walks the kernel's page tables, i.e.:

a. the access was allowed by the kernel
b. the kernel's tables have become less restrictive than the EPCM c. the kernel cannot fixup the cause of the fault

Noteably, (b) implies that either the kernel has botched the EPC mappings or the EPCM has been invalidated due to a power event. In either case, userspace needs to be alerted so that it can take appropriate action, e.g. restart the enclave. This is reinforced by (c) as the kernel doesn't really have any other reasonable option, e.g. we could kill the task or panic, but neither is warranted.

---
 arch/x86/mm/fault.c | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)
arch/x86/mm/fault.c +1 -0
index 85d20516b2f3..3fb2b2838d6c 100644
@@ -960,10 +960,13 @@ static noinline void
 bad_area_access_error(struct pt_regs *regs, unsigned long error_code,
 		      unsigned long address, struct vm_area_struct *vma)
 {
+	int si_code = SEGV_ACCERR;

arch/x86/mm/fault.c static noinline void · line 970

Quoted 10 lines
arch/x86/mm/fault.c +5 -3
+
 	if (bad_area_access_from_pkeys(error_code, vma))
-		__bad_area(regs, error_code, address, vma, SEGV_PKUERR);
-	else
-		__bad_area(regs, error_code, address, vma, SEGV_ACCERR);
+		si_code = SEGV_PKUERR;
+	else if (unlikely(error_code & X86_PF_SGX))
+		si_code = SEGV_SGXERR;
+	__bad_area(regs, error_code, address, vma, si_code);
 }

This conflicts with a cleanup in this area I have sitting in linux-next. It isn't in the x86 tree but you can find my siginfo tree at:

Quoted 7 lines

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git
siginfo-next

In my tree bad area no longer takes a vma parameter.

If you are going to make changes to the fault handling code this cycle please let's figure out how to build it on top of my clean ups.

We are now going with the proposed vdso solution for v15. Thank you for your feedback. I'll sync with you if that route would show non-feasible (still prototyping the vdso solution).

Quoted 2 lines

Thank you,
Eric Biederman

/Jarkko

#52 lore raw

PATCH v14 10/19 x86/sgx: Detect Intel SGX

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Intel(R) SGX is a set of CPU instructions that can be used by applications to set aside private regions of code and data. The code outside the enclave is disallowed to access the memory inside the enclave by the CPU access control.

Add a check for SGX to arch/x86 and a new config option, INTEL_SGX_CORE. Expose a boolean variable 'sgx_enabled' to query whether or not the SGX support is available.

---
 arch/x86/Kconfig                | 17 +++++++++++++
 arch/x86/include/asm/sgx.h      | 13 ++++++++++
 arch/x86/kernel/cpu/Makefile    |  1 +
 arch/x86/kernel/cpu/intel_sgx.c | 44 +++++++++++++++++++++++++++++++++
 4 files changed, 75 insertions(+)
 create mode 100644 arch/x86/include/asm/sgx.h
 create mode 100644 arch/x86/kernel/cpu/intel_sgx.c
arch/x86/Kconfig +17 -0
index 1a0be022f91d..b47e1a144409 100644
@@ -1913,6 +1913,23 @@ config X86_INTEL_MEMORY_PROTECTION_KEYS
 
 	  If unsure, say y.
 
+config INTEL_SGX_CORE
+	bool "Intel SGX core functionality"
+	depends on X86_64 && CPU_SUP_INTEL
+	help
+	Intel Software Guard eXtensions (SGX) CPU feature that allows ring 3
+	applications to create enclaves: private regions of memory that are
+	architecturally protected from unauthorized access and/or modification.
+
+	This option enables kernel recognition of SGX, high-level management
+	of the Enclave Page Cache (EPC), tracking and writing of SGX Launch
+	Enclave Hash MSRs, and allows for virtualization of SGX via KVM. By
+	itself, this option does not provide SGX support to userspace.
+
+	For details, see Documentation/x86/intel_sgx.rst
+
+	If unsure, say N.
+
 config EFI
 	bool "EFI runtime service support"
 	depends on ACPI
arch/x86/include/asm/sgx.h new file +13 -0
new file mode 100644
index 000000000000..f4f82f0453a9
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */
+/**
+ * Copyright(c) 2016-18 Intel Corporation.
+ */
+#ifndef _ASM_X86_SGX_H
+#define _ASM_X86_SGX_H
+
+#include <linux/types.h>
+
+extern bool sgx_enabled;
+extern bool sgx_lc_enabled;
+
+#endif /* _ASM_X86_SGX_H */
arch/x86/kernel/cpu/Makefile +1 -0
index 347137e80bf5..71876f2b35fc 100644
@@ -38,6 +38,7 @@ obj-$(CONFIG_CPU_SUP_UMC_32)		+= umc.o
 obj-$(CONFIG_INTEL_RDT)	+= intel_rdt.o intel_rdt_rdtgroup.o intel_rdt_monitor.o
 obj-$(CONFIG_INTEL_RDT)	+= intel_rdt_ctrlmondata.o intel_rdt_pseudo_lock.o
 CFLAGS_intel_rdt_pseudo_lock.o = -I$(src)
+obj-$(CONFIG_INTEL_SGX_CORE)		+= intel_sgx.o
 
 obj-$(CONFIG_X86_MCE)			+= mcheck/
 obj-$(CONFIG_MTRR)			+= mtrr/
arch/x86/kernel/cpu/intel_sgx.c new file +44 -0

44 lines not on this page · raw message

Signature · 1 line
2.17.1
reviewer
#53 lore raw

PATCH v14 10/19 x86/sgx: Detect Intel SGX

Quoted 15 lines · Jarkko Sakkinen
arch/x86/Kconfig +7 -0
index 1a0be022f91d..b47e1a144409 100644
@@ -1913,6 +1913,23 @@ config X86_INTEL_MEMORY_PROTECTION_KEYS
 
 	  If unsure, say y.
 
+config INTEL_SGX_CORE
+	bool "Intel SGX core functionality"
+	depends on X86_64 && CPU_SUP_INTEL
+	help
+	Intel Software Guard eXtensions (SGX) CPU feature that allows ring 3
+	applications to create enclaves: private regions of memory that are
+	architecturally protected from unauthorized access and/or modification.

arch/x86/Kconfig config X86_INTEL_MEMORY_PROTECTION_KEYS · line 1932

Quoted 10 lines
arch/x86/Kconfig +10 -0
+
+	This option enables kernel recognition of SGX, high-level management
+	of the Enclave Page Cache (EPC), tracking and writing of SGX Launch
+	Enclave Hash MSRs, and allows for virtualization of SGX via KVM. By
+	itself, this option does not provide SGX support to userspace.
+
+	For details, see Documentation/x86/intel_sgx.rst
+
+	If unsure, say N.
+

Hi,
coding-style.rst says that help text should be indented with one tab + 2 spaces.

thanks.

Signature · 1 line
~Randy
#54 lore raw

Re: [PATCH v14 10/19] x86/sgx: Detect Intel SGX

Quoted 31 lines · Randy Dunlap
Quoted 15 lines · Jarkko Sakkinen
arch/x86/Kconfig +7 -0
index 1a0be022f91d..b47e1a144409 100644
@@ -1913,6 +1913,23 @@ config X86_INTEL_MEMORY_PROTECTION_KEYS
 
 	  If unsure, say y.
 
+config INTEL_SGX_CORE
+	bool "Intel SGX core functionality"
+	depends on X86_64 && CPU_SUP_INTEL
+	help
+	Intel Software Guard eXtensions (SGX) CPU feature that allows ring 3
+	applications to create enclaves: private regions of memory that are
+	architecturally protected from unauthorized access and/or modification.

arch/x86/Kconfig config X86_INTEL_MEMORY_PROTECTION_KEYS · line 1932

Quoted 10 lines
arch/x86/Kconfig +10 -0
+
+	This option enables kernel recognition of SGX, high-level management
+	of the Enclave Page Cache (EPC), tracking and writing of SGX Launch
+	Enclave Hash MSRs, and allows for virtualization of SGX via KVM. By
+	itself, this option does not provide SGX support to userspace.
+
+	For details, see Documentation/x86/intel_sgx.rst
+
+	If unsure, say N.
+

Hi,
coding-style.rst says that help text should be indented with one tab + 2 spaces.

Quoted 1 line

thanks.

Thank you.

/Jarkko

#55 lore raw

PATCH v14 11/19 x86/sgx: Add wrappers for ENCLS leaf functions

In reply to Jarkko Sakkinen

Add wrappers for Intel(R) SGX ENCLS opcode leaf functions except ENCLS(EINIT). ENCLS invokes privileged functions for managing (creation, initialization and swapping) and debugging enclaves.

---
 arch/x86/include/asm/sgx.h | 244 +++++++++++++++++++++++++++++++++++++
 1 file changed, 244 insertions(+)
arch/x86/include/asm/sgx.h +244 -0
index f4f82f0453a9..e66e2572011e 100644
@@ -10,4 +10,248 @@
 extern bool sgx_enabled;
 extern bool sgx_lc_enabled;
 
+/**
+ * ENCLS_FAULT_FLAG - flag signifying an ENCLS return code is a trapnr
+ *
+ * ENCLS has its own (positive value) error codes and also generates
+ * ENCLS specific #GP and #PF faults.  And the ENCLS values get munged
+ * with system error codes as everything percolates back up the stack.
+ * Unfortunately (for us), we need to precisely identify each unique
+ * error code, e.g. the action taken if EWB fails varies based on the
+ * type of fault and on the exact SGX error code, i.e. we can't simply
+ * convert all faults to -EFAULT.
+ *
+ * To make all three error types coexist, we set bit 30 to identify an
+ * ENCLS fault.  Bit 31 (technically bits N:31) is used to differentiate
+ * between positive (faults and SGX error codes) and negative (system
+ * error codes) values.
+ */
+#define ENCLS_FAULT_FLAG 0x40000000
+
+/**
+ * Check for a fault by looking for a postive value with the fault
+ * flag set.  The postive value check is needed to filter out system
+ * error codes since negative values will have all higher order bits
+ * set, including ENCLS_FAULT_FLAG.
+ */
+#define IS_ENCLS_FAULT(r) ((int)(r) > 0 && ((r) & ENCLS_FAULT_FLAG))
+
+/**
+ * Retrieve the encoded trapnr from the specified return code.
+ */
+#define ENCLS_TRAPNR(r) ((r) & ~ENCLS_FAULT_FLAG)
+
+/**
+ * encls_to_err - translate an ENCLS fault or SGX code into a system error code
+ * @ret:	positive value return code
+ *
+ * Translate a postive return code, e.g. from ENCLS, into a system error
+ * code.  Primarily used by functions that cannot return a non-negative
+ * error code, e.g. kernel callbacks.
+ *
+ * Return:
+ *	0 on success,
+ *	-errno on failure
+ */
+static inline int encls_to_err(int ret)
+{
+	if (IS_ENCLS_FAULT(ret))
+		return -EFAULT;
+
+	switch (ret) {
+	case SGX_UNMASKED_EVENT:
+		return -EINTR;
+	case SGX_INVALID_SIG_STRUCT:
+	case SGX_INVALID_ATTRIBUTE:
+	case SGX_INVALID_MEASUREMENT:
+	case SGX_INVALID_EINITTOKEN:
+	case SGX_INVALID_CPUSVN:
+	case SGX_INVALID_ISVSVN:
+	case SGX_INVALID_KEYNAME:
+		return -EINVAL;
+	case SGX_ENCLAVE_ACT:
+	case SGX_CHILD_PRESENT:
+	case SGX_ENTRYEPOCH_LOCKED:
+	case SGX_PREV_TRK_INCMPL:
+	case SGX_PAGE_NOT_MODIFIABLE:
+	case SGX_PAGE_NOT_DEBUGGABLE:
+		return -EBUSY;
+	default:
+		return -EIO;
+	};
+}
+
+/**
+ * __encls_ret_N - encode an ENCLS leaf that returns an error code in EAX
+ * @rax:	leaf number
+ * @inputs:	asm inputs for the leaf
+ *
+ * Emit assembly for an ENCLS leaf that returns an error code, e.g. EREMOVE.
+ * And because SGX isn't complex enough as it is, leafs that return an error
+ * code also modify flags.
+ *
+ * Return:
+ *	0 on success,
+ *	SGX error code on failure
+ */
+#define __encls_ret_N(rax, inputs...)				\
+	({							\
+	int ret;						\
+	asm volatile(						\
+	"1: .byte 0x0f, 0x01, 0xcf;\n\t"			\
+	"2:\n"							\
+	".section .fixup,\"ax\"\n"				\
+	"3: orl $"__stringify(ENCLS_FAULT_FLAG)",%%eax\n"	\
+	"   jmp 2b\n"						\
+	".previous\n"						\
+	_ASM_EXTABLE_FAULT(1b, 3b)				\
+	: "=a"(ret)						\
+	: "a"(rax), inputs					\
+	: "memory", "cc");					\
+	ret;							\
+	})
+
+#define __encls_ret_1(rax, rcx)		\
+	({				\
+	__encls_ret_N(rax, "c"(rcx));	\
+	})
+
+#define __encls_ret_2(rax, rbx, rcx)		\
+	({					\
+	__encls_ret_N(rax, "b"(rbx), "c"(rcx));	\
+	})
+
+#define __encls_ret_3(rax, rbx, rcx, rdx)			\
+	({							\
+	__encls_ret_N(rax, "b"(rbx), "c"(rcx), "d"(rdx));	\
+	})
+
+/**
+ * __encls_N - encode an ENCLS leaf that doesn't return an error code
+ * @rax:	leaf number
+ * @rbx_out:	optional output variable
+ * @inputs:	asm inputs for the leaf
+ *
+ * Emit assembly for an ENCLS leaf that does not return an error code,
+ * e.g. ECREATE.  Leaves without error codes either succeed or fault.
+ * @rbx_out is an optional parameter for use by EDGBRD, which returns
+ * the the requested value in RBX.
+ *
+ * Return:
+ *   0 on success,
+ *   trapnr with ENCLS_FAULT_FLAG set on fault
+ */
+#define __encls_N(rax, rbx_out, inputs...)			\
+	({							\
+	int ret;						\
+	asm volatile(						\
+	"1: .byte 0x0f, 0x01, 0xcf;\n\t"			\
+	"   xor %%eax,%%eax;\n"					\
+	"2:\n"							\
+	".section .fixup,\"ax\"\n"				\
+	"3: orl $"__stringify(ENCLS_FAULT_FLAG)",%%eax\n"	\
+	"   jmp 2b\n"						\
+	".previous\n"						\
+	_ASM_EXTABLE_FAULT(1b, 3b)				\
+	: "=a"(ret), "=b"(rbx_out)				\
+	: "a"(rax), inputs					\
+	: "memory");						\
+	ret;							\
+	})
+
+#define __encls_2(rax, rbx, rcx)				\
+	({							\
+	unsigned long ign_rbx_out;				\
+	__encls_N(rax, ign_rbx_out, "b"(rbx), "c"(rcx));	\
+	})
+
+#define __encls_1_1(rax, data, rcx)			\
+	({						\
+	unsigned long rbx_out;				\
+	int ret = __encls_N(rax, rbx_out, "c"(rcx));	\
+	if (!ret)					\
+		data = rbx_out;				\
+	ret;						\
+	})
+
+static inline int __ecreate(struct sgx_pageinfo *pginfo, void __iomem *secs)
+{
+	return __encls_2(SGX_ECREATE, pginfo, secs);
+}
+
+static inline int __eextend(void __iomem *secs, void __iomem *addr)
+{
+	return __encls_2(SGX_EEXTEND, secs, addr);
+}
+
+static inline int __eadd(struct sgx_pageinfo *pginfo, void __iomem *addr)
+{
+	return __encls_2(SGX_EADD, pginfo, addr);
+}
+
+static inline int __einit(void *sigstruct, struct sgx_einittoken *einittoken,
+			  void __iomem *secs)
+{
+	return __encls_ret_3(SGX_EINIT, sigstruct, secs, einittoken);
+}
+
+static inline int __eremove(void __iomem *addr)
+{
+	return __encls_ret_1(SGX_EREMOVE, addr);
+}
+
+static inline int __edbgwr(void __iomem *addr, unsigned long *data)
+{
+	return __encls_2(SGX_EDGBWR, *data, addr);
+}
+
+static inline int __edbgrd(void __iomem *addr, unsigned long *data)
+{
+	return __encls_1_1(SGX_EDGBRD, *data, addr);
+}
+
+static inline int __etrack(void __iomem *addr)
+{
+	return __encls_ret_1(SGX_ETRACK, addr);
+}
+
+static inline int __eldu(struct sgx_pageinfo *pginfo, void __iomem *addr,
+			 void *va)
+{
+	return __encls_ret_3(SGX_ELDU, pginfo, addr, va);
+}
+
+static inline int __eblock(void __iomem *addr)
+{
+	return __encls_ret_1(SGX_EBLOCK, addr);
+}
+
+static inline int __epa(void __iomem *addr)
+{
+	unsigned long rbx = SGX_PAGE_TYPE_VA;
+
+	return __encls_2(SGX_EPA, rbx, addr);
+}
+
+static inline int __ewb(struct sgx_pageinfo *pginfo, void __iomem *addr,
+			void __iomem *va)
+{
+	return __encls_ret_3(SGX_EWB, pginfo, addr, va);
+}
+
+static inline int __eaug(struct sgx_pageinfo *pginfo, void __iomem *addr)
+{
+	return __encls_2(SGX_EAUG, pginfo, addr);
+}
+
+static inline int __emodpr(struct sgx_secinfo *secinfo, void __iomem *addr)
+{
+	return __encls_ret_2(SGX_EMODPR, secinfo, addr);
+}
+
+static inline int __emodt(struct sgx_secinfo *secinfo, void __iomem *addr)
+{
+	return __encls_ret_2(SGX_EMODT, secinfo, addr);
+}
+
 #endif /* _ASM_X86_SGX_H */
Signature · 1 line
2.17.1
reviewer
#56 lore raw

PATCH v14 11/19 x86/sgx: Add wrappers for ENCLS leaf functions

Quoted 16 lines · Jarkko Sakkinen

Add wrappers for Intel(R) SGX ENCLS opcode leaf functions except ENCLS(EINIT). ENCLS invokes privileged functions for managing (creation, initialization and swapping) and debugging enclaves.

---
 arch/x86/include/asm/sgx.h | 244 +++++++++++++++++++++++++++++++++++++
 1 file changed, 244 insertions(+)
arch/x86/include/asm/sgx.h +0 -0
index f4f82f0453a9..e66e2572011e 100644

arch/x86/include/asm/sgx.h line 12

Quoted 3 lines
arch/x86/include/asm/sgx.h +0 -0
@@ -10,4 +10,248 @@
 extern bool sgx_enabled;
 extern bool sgx_lc_enabled;
 

Hi,
Please don't use "/**" to begin comment blocks that are not kernel-doc notation, like the 3 below. Note: I might have seen one (or some) of these "fixed" in other patches in this series (not sure).

Quoted 20 lines
arch/x86/include/asm/sgx.h +20 -0
+/**
+ * ENCLS_FAULT_FLAG - flag signifying an ENCLS return code is a trapnr
+ *
+ * ENCLS has its own (positive value) error codes and also generates
+ * ENCLS specific #GP and #PF faults.  And the ENCLS values get munged
+ * with system error codes as everything percolates back up the stack.
+ * Unfortunately (for us), we need to precisely identify each unique
+ * error code, e.g. the action taken if EWB fails varies based on the
+ * type of fault and on the exact SGX error code, i.e. we can't simply
+ * convert all faults to -EFAULT.
+ *
+ * To make all three error types coexist, we set bit 30 to identify an
+ * ENCLS fault.  Bit 31 (technically bits N:31) is used to differentiate
+ * between positive (faults and SGX error codes) and negative (system
+ * error codes) values.
+ */
+#define ENCLS_FAULT_FLAG 0x40000000
+
+/**
+ * Check for a fault by looking for a postive value with the fault

arch/x86/include/asm/sgx.h line 42

Quoted 10 lines
arch/x86/include/asm/sgx.h +10 -0
+ * flag set.  The postive value check is needed to filter out system
+ * error codes since negative values will have all higher order bits
+ * set, including ENCLS_FAULT_FLAG.
+ */
+#define IS_ENCLS_FAULT(r) ((int)(r) > 0 && ((r) & ENCLS_FAULT_FLAG))
+
+/**
+ * Retrieve the encoded trapnr from the specified return code.
+ */
+#define ENCLS_TRAPNR(r) ((r) & ~ENCLS_FAULT_FLAG)

thanks,

Signature · 1 line
~Randy
#57 lore raw

Re: [PATCH v14 11/19] x86/sgx: Add wrappers for ENCLS leaf functions

Quoted 4 lines · Randy Dunlap

Hi,
Please don't use "/**" to begin comment blocks that are not kernel-doc notation, like the 3 below. Note: I might have seen one (or some) of these "fixed" in other patches in this series (not sure).

Thanks Randy.

/Jarkko

#58 lore raw

PATCH v14 12/19 x86/sgx: Add data structures for tracking the EPC pages

In reply to Jarkko Sakkinen

Add data structures to track Enclave Page Cache (EPC) pages. EPC is divided into multiple banks (1-N) of which addresses and sizes can be enumerated with CPUID by the OS.

On NUMA systems a node can have at most bank. A bank can be at most part of two nodes. SGX supports both nodes with a single memory controller and also sub-cluster nodes with severals memory controllers on a single die.

---
 arch/x86/include/asm/sgx.h      |  55 +++++++++++++++
 arch/x86/kernel/cpu/intel_sgx.c | 115 ++++++++++++++++++++++++++++++++
 2 files changed, 170 insertions(+)
arch/x86/include/asm/sgx.h +55 -0
index e66e2572011e..468e609147cd 100644
@@ -5,10 +5,65 @@
 #ifndef _ASM_X86_SGX_H
 #define _ASM_X86_SGX_H
 
+#include <linux/bitops.h>
+#include <linux/err.h>
+#include <linux/rwsem.h>
 #include <linux/types.h>
+#include <asm/sgx_arch.h>
+#include <asm/asm.h>
+
+struct sgx_epc_page {
+	unsigned long desc;
+	struct list_head list;
+};
+
+/**
+ * struct sgx_epc_section
+ *
+ * The firmware can define multiple chunks of EPC to the different areas of the
+ * physical memory e.g. for memory areas of the each node. This structure is
+ * used to store EPC pages for one EPC section and virtual memory area where
+ * the pages have been mapped.
+ */
+struct sgx_epc_section {
+	unsigned long pa;
+	void __iomem *va;
+	struct sgx_epc_page **pages;
+	unsigned long free_cnt;
+	spinlock_t lock;
+};
 
 extern bool sgx_enabled;
 extern bool sgx_lc_enabled;
+extern struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
+
+/*
+ * enum sgx_epc_page_desc - bits and masks for an EPC page's descriptor
+ * %SGX_EPC_SECTION_MASK:	SGX allows to have multiple EPC sections in the
+ *				physical memory. The existing and near-future
+ *				hardware defines at most eight sections, hence
+ *				three bits to hold a section.
+ * %SGX_EPC_PAGE_RECLAIMABLE:	The page page is reclaimable. Used when freeing
+ *				a page to know that we also need to remove the
+ *				page from the list of reclaimable pages.
+ */
+enum sgx_epc_page_desc {
+	SGX_EPC_SECTION_MASK			= GENMASK_ULL(3, 0),
+	SGX_EPC_PAGE_RECLAIMABLE		= BIT(4),
+	/* bits 12-63 are reserved for the physical page address of the page */
+};
+
+static inline struct sgx_epc_section *sgx_epc_section(struct sgx_epc_page *page)
+{
+	return &sgx_epc_sections[page->desc & SGX_EPC_SECTION_MASK];
+}
+
+static inline void __iomem *sgx_epc_addr(struct sgx_epc_page *page)
+{
+	struct sgx_epc_section *section = sgx_epc_section(page);
+
+	return section->va + (page->desc & PAGE_MASK) - section->pa;
+}
 
 /**
  * ENCLS_FAULT_FLAG - flag signifying an ENCLS return code is a trapnr
arch/x86/kernel/cpu/intel_sgx.c +115 -0

131 lines not on this page · raw message

Signature · 1 line
2.17.1
reviewer
#59 lore raw

Re: [PATCH v14 12/19] x86/sgx: Add data structures for tracking the EPC pages

Quoted 13 lines · Jarkko Sakkinen

Add data structures to track Enclave Page Cache (EPC) pages. EPC is divided into multiple banks (1-N) of which addresses and sizes can be enumerated with CPUID by the OS.

On NUMA systems a node can have at most bank. A bank can be at most part of two nodes. SGX supports both nodes with a single memory controller and also sub-cluster nodes with severals memory controllers on a single die.

Quoted 4 lines
---
 arch/x86/include/asm/sgx.h      |  55 +++++++++++++++
 arch/x86/kernel/cpu/intel_sgx.c | 115 ++++++++++++++++++++++++++++++++
 2 files changed, 170 insertions(+)

Hi,

Several comment blocks begin with "/**" but they are either not intended to be kernel-doc notation or they are incomplete kernel-doc notation.

Quoted 19 lines
arch/x86/include/asm/sgx.h +9 -0
index e66e2572011e..468e609147cd 100644
@@ -5,10 +5,65 @@
 #ifndef _ASM_X86_SGX_H
 #define _ASM_X86_SGX_H
 
+#include <linux/bitops.h>
+#include <linux/err.h>
+#include <linux/rwsem.h>
 #include <linux/types.h>
+#include <asm/sgx_arch.h>
+#include <asm/asm.h>
+
+struct sgx_epc_page {
+	unsigned long desc;
+	struct list_head list;

arch/x86/include/asm/sgx.h line 27

Quoted 10 lines
arch/x86/include/asm/sgx.h +10 -0
+};
+
+/**
+ * struct sgx_epc_section
+ *
+ * The firmware can define multiple chunks of EPC to the different areas of the
+ * physical memory e.g. for memory areas of the each node. This structure is
+ * used to store EPC pages for one EPC section and virtual memory area where
+ * the pages have been mapped.
+ */

Incomplete...

Quoted 12 lines
arch/x86/include/asm/sgx.h +9 -0
+struct sgx_epc_section {
+	unsigned long pa;
+	void __iomem *va;
+	struct sgx_epc_page **pages;
+	unsigned long free_cnt;
+	spinlock_t lock;
+};
 
 extern bool sgx_enabled;
 extern bool sgx_lc_enabled;
+extern struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
+

arch/x86/include/asm/sgx.h line 49

Quoted 10 lines
arch/x86/include/asm/sgx.h +10 -0
+/*
+ * enum sgx_epc_page_desc - bits and masks for an EPC page's descriptor
+ * %SGX_EPC_SECTION_MASK:	SGX allows to have multiple EPC sections in the
+ *				physical memory. The existing and near-future
+ *				hardware defines at most eight sections, hence
+ *				three bits to hold a section.
+ * %SGX_EPC_PAGE_RECLAIMABLE:	The page page is reclaimable. Used when freeing
+ *				a page to know that we also need to remove the
+ *				page from the list of reclaimable pages.
+ */

OTOH, this one is close to ready for kernel-doc but it does not begin with "/**". But that's OK, your choice.

Quoted 80 lines
arch/x86/include/asm/sgx.h +17 -0
+enum sgx_epc_page_desc {
+	SGX_EPC_SECTION_MASK			= GENMASK_ULL(3, 0),
+	SGX_EPC_PAGE_RECLAIMABLE		= BIT(4),
+	/* bits 12-63 are reserved for the physical page address of the page */
+};
+
+static inline struct sgx_epc_section *sgx_epc_section(struct sgx_epc_page *page)
+{
+	return &sgx_epc_sections[page->desc & SGX_EPC_SECTION_MASK];
+}
+
+static inline void __iomem *sgx_epc_addr(struct sgx_epc_page *page)
+{
+	struct sgx_epc_section *section = sgx_epc_section(page);
+
+	return section->va + (page->desc & PAGE_MASK) - section->pa;
+}
 
 /**
  * ENCLS_FAULT_FLAG - flag signifying an ENCLS return code is a trapnr
arch/x86/kernel/cpu/intel_sgx.c +52 -0
index 138af9b9a39a..b24d6287442d 100644
@@ -14,10 +14,121 @@ bool sgx_enabled __ro_after_init;
 EXPORT_SYMBOL_GPL(sgx_enabled);
 bool sgx_lc_enabled __ro_after_init;
 EXPORT_SYMBOL_GPL(sgx_lc_enabled);
+struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
+EXPORT_SYMBOL_GPL(sgx_epc_sections);
+
+static int sgx_nr_epc_sections;
+
+static __init void sgx_free_epc_section(struct sgx_epc_section *section)
+{
+	int i;
+
+	for (i = 0; i < section->free_cnt && section->pages[i]; i++)
+		kfree(section->pages[i]);
+	kfree(section->pages);
+	iounmap(section->va);
+}
+
+static __init int sgx_init_epc_section(u64 addr, u64 size, unsigned long index,
+				       struct sgx_epc_section *section)
+{
+	unsigned long nr_pages = size >> PAGE_SHIFT;
+	unsigned long i;
+
+	section->va = ioremap_cache(addr, size);
+	if (!section->va)
+		return -ENOMEM;
+
+	section->pa = addr;
+	section->free_cnt = nr_pages;
+	spin_lock_init(&section->lock);
+
+	section->pages = kcalloc(nr_pages, sizeof(struct sgx_epc_page *),
+				 GFP_KERNEL);
+	if (!section->pages)
+		goto out;
+
+	for (i = 0; i < nr_pages; i++) {
+		section->pages[i] = kzalloc(sizeof(struct sgx_epc_page),
+					    GFP_KERNEL);
+		if (!section->pages[i])
+			goto out;
+
+		section->pages[i]->desc = (addr + (i << PAGE_SHIFT)) | index;
+	}
+
+	return 0;
+out:
+	sgx_free_epc_section(section);
+	return -ENOMEM;
+}
+
+static __init void sgx_page_cache_teardown(void)
+{
+	int i;

arch/x86/kernel/cpu/intel_sgx.c bool sgx_enabled __ro_after_init; · line 78

Quoted 10 lines
arch/x86/kernel/cpu/intel_sgx.c +10 -0
+
+	for (i = 0; i < sgx_nr_epc_sections; i++)
+		sgx_free_epc_section(&sgx_epc_sections[i]);
+}
+
+/**
+ * A section metric is concatenated in a way that @low bits 12-31 define the
+ * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
+ * metric.
+ */

not kernel-doc notation.

arch/x86/kernel/cpu/intel_sgx.c bool sgx_enabled __ro_after_init; · line 83

Quoted 5 lines
arch/x86/kernel/cpu/intel_sgx.c +5 -0
+static inline u64 sgx_calc_section_metric(u64 low, u64 high)
+{
+	return (low & GENMASK_ULL(31, 12)) +
+	       ((high & GENMASK_ULL(19, 0)) << 32);
+}

thanks,

Signature · 1 line
~Randy
#60 lore raw

Re: [PATCH v14 12/19] x86/sgx: Add data structures for tracking the EPC pages

Quoted 19 lines · Randy Dunlap
Quoted 13 lines · Jarkko Sakkinen

Add data structures to track Enclave Page Cache (EPC) pages. EPC is divided into multiple banks (1-N) of which addresses and sizes can be enumerated with CPUID by the OS.

On NUMA systems a node can have at most bank. A bank can be at most part of two nodes. SGX supports both nodes with a single memory controller and also sub-cluster nodes with severals memory controllers on a single die.

Quoted 4 lines
---
 arch/x86/include/asm/sgx.h      |  55 +++++++++++++++
 arch/x86/kernel/cpu/intel_sgx.c | 115 ++++++++++++++++++++++++++++++++
 2 files changed, 170 insertions(+)
Quoted 4 lines

Hi,

Several comment blocks begin with "/**" but they are either not intended to be kernel-doc notation or they are incomplete kernel-doc notation.

I'll enumerate them through.

Thank you for the feedback.

/Jarkko

#61 lore raw

PATCH v14 13/19 x86/sgx: Enclave Page Cache (EPC) memory manager

In reply to Jarkko Sakkinen

Add a Enclave Page Cache (EPC) memory manager that can be used to allocate and free EPC pages.

---
 arch/x86/include/asm/sgx.h      |  7 ++-
 arch/x86/kernel/cpu/intel_sgx.c | 86 +++++++++++++++++++++++++++++++++
 2 files changed, 92 insertions(+), 1 deletion(-)
arch/x86/include/asm/sgx.h +6 -1
index 468e609147cd..6ea5a6dbd36d 100644
@@ -14,6 +14,7 @@
 
 struct sgx_epc_page {
 	unsigned long desc;
+	void *owner;
 	struct list_head list;
 };
 
@@ -65,7 +66,7 @@ static inline void __iomem *sgx_epc_addr(struct sgx_epc_page *page)
 	return section->va + (page->desc & PAGE_MASK) - section->pa;
 }
 
-/**
+/*
  * ENCLS_FAULT_FLAG - flag signifying an ENCLS return code is a trapnr
  *
  * ENCLS has its own (positive value) error codes and also generates
@@ -309,4 +310,8 @@ static inline int __emodt(struct sgx_secinfo *secinfo, void __iomem *addr)
 	return __encls_ret_2(SGX_EMODT, secinfo, addr);
 }
 
+struct sgx_epc_page *sgx_alloc_page(void *owner, bool reclaim);
+int __sgx_free_page(struct sgx_epc_page *page);
+void sgx_free_page(struct sgx_epc_page *page);
+
 #endif /* _ASM_X86_SGX_H */
arch/x86/kernel/cpu/intel_sgx.c +86 -0

92 lines not on this page · raw message

Signature · 1 line
2.17.1
#62 lore raw

PATCH v14 14/19 x86/sgx: Add sgx_einit() for initializing enclaves

In reply to Jarkko Sakkinen

From: Sean Christopherson <[email protected]>

Add a function to perform ENCLS(EINIT), which initializes an enclave, which can be used by a driver for running enclaves and VMMs.

Writing the LE hash MSRs is extraordinarily expensive, e.g. 3-4x slower than normal MSRs, so we use a per-cpu cache to track the last known value of the MSRs to avoid unnecessarily writing the MSRs with the current value.

---
 arch/x86/include/asm/sgx.h      |  2 ++
 arch/x86/kernel/cpu/intel_sgx.c | 53 +++++++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+)
arch/x86/include/asm/sgx.h +2 -0
index 6ea5a6dbd36d..d5c535db094c 100644
@@ -313,5 +313,7 @@ static inline int __emodt(struct sgx_secinfo *secinfo, void __iomem *addr)
 struct sgx_epc_page *sgx_alloc_page(void *owner, bool reclaim);
 int __sgx_free_page(struct sgx_epc_page *page);
 void sgx_free_page(struct sgx_epc_page *page);
+int sgx_einit(struct sgx_sigstruct *sigstruct, struct sgx_einittoken *token,
+	      struct sgx_epc_page *secs, u64 *lepubkeyhash);
 
 #endif /* _ASM_X86_SGX_H */
arch/x86/kernel/cpu/intel_sgx.c +53 -0

65 lines not on this page · raw message

Signature · 1 line
2.17.1
#63 lore raw

PATCH v14 15/19 platform/x86: Intel SGX driver

In reply to Jarkko Sakkinen

Intel Software Guard eXtensions (SGX) is a set of CPU instructions that can be used by applications to set aside private regions of code and data. The code outside the enclave is disallowed to access the memory inside the enclave by the CPU access control.

SGX driver provides a ioctl API for loading and initializing enclaves. Address range for enclaves is reserved with mmap() and they are destroyed with munmap(). Enclave construction, measurement and initialization is done with the provided the ioctl API.

---
 arch/x86/include/uapi/asm/sgx.h            |  60 ++
 drivers/platform/x86/Kconfig               |   2 +
 drivers/platform/x86/Makefile              |   1 +
 drivers/platform/x86/intel_sgx/Kconfig     |  20 +
 drivers/platform/x86/intel_sgx/Makefile    |  12 +
 drivers/platform/x86/intel_sgx/sgx.h       | 180 +++++
 drivers/platform/x86/intel_sgx/sgx_encl.c  | 826 +++++++++++++++++++++
 drivers/platform/x86/intel_sgx/sgx_ioctl.c | 234 ++++++
 drivers/platform/x86/intel_sgx/sgx_main.c  | 267 +++++++
 drivers/platform/x86/intel_sgx/sgx_util.c  |  85 +++
 drivers/platform/x86/intel_sgx/sgx_vma.c   |  43 ++
 11 files changed, 1730 insertions(+)
 create mode 100644 arch/x86/include/uapi/asm/sgx.h
 create mode 100644 drivers/platform/x86/intel_sgx/Kconfig
 create mode 100644 drivers/platform/x86/intel_sgx/Makefile
 create mode 100644 drivers/platform/x86/intel_sgx/sgx.h
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_encl.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_ioctl.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_main.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_util.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_vma.c
arch/x86/include/uapi/asm/sgx.h new file +60 -0
new file mode 100644
index 000000000000..58bd5d5c3c9d
@@ -0,0 +1,60 @@
+
+/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */
+/**
+ * Copyright(c) 2016-18 Intel Corporation.
+ */
+#ifndef _UAPI_ASM_X86_SGX_H
+#define _UAPI_ASM_X86_SGX_H
+
+#include <linux/types.h>
+#include <linux/ioctl.h>
+
+#define SGX_MAGIC 0xA4
+
+#define SGX_IOC_ENCLAVE_CREATE \
+	_IOW(SGX_MAGIC, 0x00, struct sgx_enclave_create)
+#define SGX_IOC_ENCLAVE_ADD_PAGE \
+	_IOW(SGX_MAGIC, 0x01, struct sgx_enclave_add_page)
+#define SGX_IOC_ENCLAVE_INIT \
+	_IOW(SGX_MAGIC, 0x02, struct sgx_enclave_init)
+
+/* IOCTL return values */
+#define SGX_POWER_LOST_ENCLAVE		0x40000000
+
+/**
+ * struct sgx_enclave_create - parameter structure for the
+ *                             %SGX_IOC_ENCLAVE_CREATE ioctl
+ * @src:	address for the SECS page data
+ */
+struct sgx_enclave_create  {
+	__u64	src;
+};
+
+/**
+ * struct sgx_enclave_add_page - parameter structure for the
+ *                               %SGX_IOC_ENCLAVE_ADD_PAGE ioctl
+ * @addr:	address within the ELRANGE
+ * @src:	address for the page data
+ * @secinfo:	address for the SECINFO data
+ * @mrmask:	bitmask for the measured 256 byte chunks
+ */
+struct sgx_enclave_add_page {
+	__u64	addr;
+	__u64	src;
+	__u64	secinfo;
+	__u16	mrmask;
+} __attribute__((__packed__));
+
+
+/**
+ * struct sgx_enclave_init - parameter structure for the
+ *                           %SGX_IOC_ENCLAVE_INIT ioctl
+ * @addr:	address within the ELRANGE
+ * @sigstruct:	address for the SIGSTRUCT data
+ */
+struct sgx_enclave_init {
+	__u64	addr;
+	__u64	sigstruct;
+};
+
+#endif /* _UAPI_ASM_X86_SGX_H */
drivers/platform/x86/Kconfig +2 -0
index 0c1aa6c314f5..66a1ab6235e2 100644
@@ -1229,6 +1229,8 @@ config I2C_MULTI_INSTANTIATE
 	  To compile this driver as a module, choose M here: the module
 	  will be called i2c-multi-instantiate.
 
+source "drivers/platform/x86/intel_sgx/Kconfig"
+
 endif # X86_PLATFORM_DEVICES
 
 config PMC_ATOM
drivers/platform/x86/Makefile +1 -0
index e6d1becf81ce..d0db8c66ed3c 100644
@@ -92,3 +92,4 @@ obj-$(CONFIG_MLX_PLATFORM)	+= mlx-platform.o
 obj-$(CONFIG_INTEL_TURBO_MAX_3) += intel_turbo_max_3.o
 obj-$(CONFIG_INTEL_CHTDC_TI_PWRBTN)	+= intel_chtdc_ti_pwrbtn.o
 obj-$(CONFIG_I2C_MULTI_INSTANTIATE)	+= i2c-multi-instantiate.o
+obj-$(CONFIG_INTEL_SGX) += intel_sgx/
drivers/platform/x86/intel_sgx/Kconfig new file +20 -0
new file mode 100644
index 000000000000..7d22d44acce9
@@ -0,0 +1,20 @@
+#
+# Intel SGX
+#
+
+config INTEL_SGX
+	tristate "Intel(R) SGX Driver"
+	depends on X86_64 && CPU_SUP_INTEL && INTEL_SGX_CORE
+	select MMU_NOTIFIER
+	select CRYPTO
+	select CRYPTO_SHA256
+	help
+	Intel(R) SGX is a set of CPU instructions that can be used by
+	applications to set aside private regions of code and data.  The code
+	outside the enclave is disallowed to access the memory inside the
+	enclave by the CPU access control.
+
+	The firmware uses PRMRR registers to reserve an area of physical memory
+	called Enclave Page Cache (EPC). There is a hardware unit in the
+	processor called Memory Encryption Engine. The MEE encrypts and decrypts
+	the EPC pages as they enter and leave the processor package.
drivers/platform/x86/intel_sgx/Makefile new file +12 -0
new file mode 100644
index 000000000000..117e97effeff
@@ -0,0 +1,12 @@
+#
+# Intel SGX
+#
+
+obj-$(CONFIG_INTEL_SGX) += intel_sgx.o
+
+intel_sgx-$(CONFIG_INTEL_SGX) += \
+	sgx_encl.o \
+	sgx_ioctl.o \
+	sgx_main.o \
+	sgx_util.o \
+	sgx_vma.o \
drivers/platform/x86/intel_sgx/sgx.h new file +180 -0

180 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_encl.c new file +826 -0

826 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +234 -0

234 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_main.c new file +267 -0

267 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_util.c new file +85 -0

85 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_vma.c new file +43 -0

43 lines not on this page · raw message

Signature · 1 line
2.17.1
#64 lore raw

Re: [PATCH v14 15/19] platform/x86: Intel SGX driver

Quoted 9 lines · Jarkko Sakkinen

Intel Software Guard eXtensions (SGX) is a set of CPU instructions that can be used by applications to set aside private regions of code and data. The code outside the enclave is disallowed to access the memory inside the enclave by the CPU access control.

SGX driver provides a ioctl API for loading and initializing enclaves. Address range for enclaves is reserved with mmap() and they are destroyed with munmap(). Enclave construction, measurement and initialization is done with the provided the ioctl API.

...

Quoted 30 lines
drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +30 -0
+/**
+ * sgx_ioc_enclave_init - handler for %SGX_IOC_ENCLAVE_INIT
+ *
+ * @filep:	open file to /dev/sgx
+ * @cmd:	the command value
+ * @arg:	pointer to an &sgx_enclave_init instance
+ *
+ * Flushes the remaining enqueued EADD operations and performs EINIT. Does not
+ * allow the EINITTOKENKEY attribute for an enclave.
+ *
+ * Return:
+ *   0 on success,
+ *   SGX error code on EINIT failure,
+ *   -errno otherwise
+ */
+static long sgx_ioc_enclave_init(struct file *filep, unsigned int cmd,
+				 unsigned long arg)
+{
+	struct sgx_enclave_init *initp = (struct sgx_enclave_init *)arg;
+	struct sgx_sigstruct *sigstruct;
+	struct sgx_einittoken *einittoken;
+	struct sgx_encl *encl;
+	struct page *initp_page;
+	int ret;
+
+	initp_page = alloc_page(GFP_HIGHUSER);
+	if (!initp_page)
+		return -ENOMEM;
+
+	sigstruct = kmap(initp_page);

drivers/platform/x86/intel_sgx/sgx_ioctl.c line 182

Quoted 10 lines
drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +10 -0
+	einittoken = (struct sgx_einittoken *)
+		((unsigned long)sigstruct + PAGE_SIZE / 2);
+	memset(einittoken, 0, sizeof(*einittoken));
+
+	ret = copy_from_user(sigstruct, (void __user *)initp->sigstruct,
+			     sizeof(*sigstruct));
+	if (ret)
+		goto out;
+	if (sigstruct->attributes & SGX_ATTR_EINITTOKENKEY) {
+		ret = EINVAL;

This should be "ret = -EINVAL".

Quoted 16 lines
drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +16 -0
+		goto out;
+	}
+
+	ret = sgx_encl_get(initp->addr, &encl);
+	if (ret)
+		goto out;
+
+	ret = sgx_encl_init(encl, sigstruct, einittoken);
+
+	kref_put(&encl->refcount, sgx_encl_release);
+
+out:
+	kunmap(initp_page);
+	__free_page(initp_page);
+	return ret;
+}
#65 lore raw

Re: [PATCH v14 15/19] platform/x86: Intel SGX driver

Quoted 47 lines · Sean Christopherson
Quoted 9 lines · Jarkko Sakkinen

Intel Software Guard eXtensions (SGX) is a set of CPU instructions that can be used by applications to set aside private regions of code and data. The code outside the enclave is disallowed to access the memory inside the enclave by the CPU access control.

SGX driver provides a ioctl API for loading and initializing enclaves. Address range for enclaves is reserved with mmap() and they are destroyed with munmap(). Enclave construction, measurement and initialization is done with the provided the ioctl API.

...

Quoted 34 lines
drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +34 -0
+/**
+ * sgx_ioc_enclave_init - handler for %SGX_IOC_ENCLAVE_INIT
+ *
+ * @filep:	open file to /dev/sgx
+ * @cmd:	the command value
+ * @arg:	pointer to an &sgx_enclave_init instance
+ *
+ * Flushes the remaining enqueued EADD operations and performs EINIT. Does not
+ * allow the EINITTOKENKEY attribute for an enclave.
+ *
+ * Return:
+ *   0 on success,
+ *   SGX error code on EINIT failure,
+ *   -errno otherwise
+ */
+static long sgx_ioc_enclave_init(struct file *filep, unsigned int cmd,
+				 unsigned long arg)
+{
+	struct sgx_enclave_init *initp = (struct sgx_enclave_init *)arg;
+	struct sgx_sigstruct *sigstruct;
+	struct sgx_einittoken *einittoken;
+	struct sgx_encl *encl;
+	struct page *initp_page;
+	int ret;
+
+	initp_page = alloc_page(GFP_HIGHUSER);
+	if (!initp_page)
+		return -ENOMEM;
+
+	sigstruct = kmap(initp_page);
+	einittoken = (struct sgx_einittoken *)
+		((unsigned long)sigstruct + PAGE_SIZE / 2);
+	memset(einittoken, 0, sizeof(*einittoken));
+
Quoted 8 lines

drivers/platform/x86/intel_sgx/sgx_ioctl.c line 182

Quoted 6 lines
drivers/platform/x86/intel_sgx/sgx_ioctl.c new file +6 -0
+	ret = copy_from_user(sigstruct, (void __user *)initp->sigstruct,
+			     sizeof(*sigstruct));
+	if (ret)
+		goto out;
+	if (sigstruct->attributes & SGX_ATTR_EINITTOKENKEY) {
+		ret = EINVAL;

This should be "ret = -EINVAL".

Ouch :-( Thank you for spotting this out.

/Jarkko

#66 lore raw

PATCH v14 16/19 platform/x86: Add swapping functionality to the Intel SGX driver

In reply to Jarkko Sakkinen

Add functions to swap EPC pages to the driver:

* sgx_encl_page_get() pins and sgx_encl_page_put() unpins an enclave page.
* sgx_encl_page_reclaim() tries to mark page as being reclaimed. The page is considered reclaimable if it hasn't been accessed recently and it isn't reserved by the driver for other use.
* sgx_encl_page_block() and sgx_encl_page_track() implement EBLOCK and ETRACK operations.
* sgx_encl_page_write() writes the page to the regular memory with EWB.

This commit also contains the #PF handler for loading the swapped pages back.

---
 drivers/platform/x86/intel_sgx/Makefile       |   2 +
 drivers/platform/x86/intel_sgx/sgx.h          |  32 ++++
 drivers/platform/x86/intel_sgx/sgx_encl.c     | 129 ++++++++++++-
 .../platform/x86/intel_sgx/sgx_encl_page.c    | 177 ++++++++++++++++++
 drivers/platform/x86/intel_sgx/sgx_fault.c    | 107 +++++++++++
 drivers/platform/x86/intel_sgx/sgx_util.c     |  71 +++++++
 drivers/platform/x86/intel_sgx/sgx_vma.c      |  15 ++
 7 files changed, 532 insertions(+), 1 deletion(-)
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_encl_page.c
 create mode 100644 drivers/platform/x86/intel_sgx/sgx_fault.c
drivers/platform/x86/intel_sgx/Makefile +2 -0
index 117e97effeff..5023c5ebd57d 100644
@@ -6,6 +6,8 @@ obj-$(CONFIG_INTEL_SGX) += intel_sgx.o
 
 intel_sgx-$(CONFIG_INTEL_SGX) += \
 	sgx_encl.o \
+	sgx_encl_page.o \
+	sgx_fault.o \
 	sgx_ioctl.o \
 	sgx_main.o \
 	sgx_util.o \
drivers/platform/x86/intel_sgx/sgx.h +32 -0
index 67bd8ea1d53d..f707701e422d 100644
@@ -36,16 +36,32 @@
 #define SGX_EINIT_SLEEP_COUNT	50
 #define SGX_EINIT_SLEEP_TIME	20
 
+#define SGX_VA_SLOT_COUNT 512
+
+struct sgx_va_page {
+	struct sgx_epc_page *epc_page;
+	DECLARE_BITMAP(slots, SGX_VA_SLOT_COUNT);
+	struct list_head list;
+};
+
 /**
  * enum sgx_encl_page_desc - defines bits for an enclave page's descriptor
  * %SGX_ENCL_PAGE_TCS:			The page is a TCS page.
  * %SGX_ENCL_PAGE_LOADED:		The page is not swapped.
+ * %SGX_ENCL_PAGE_RESERVED:		The page cannot be reclaimed.
+ * %SGX_ENCL_PAGE_RECLAIMED:		The page is in the process of being
+ *					reclaimed.
+ * %SGX_ENCL_PAGE_VA_OFFSET_MASK:	Holds the offset in the Version Array
+ *					(VA) page for a swapped page.
  * %SGX_ENCL_PAGE_ADDR_MASK:		Holds the virtual address of the page.
  */
 enum sgx_encl_page_desc {
 	SGX_ENCL_PAGE_TCS		= BIT(0),
 	SGX_ENCL_PAGE_LOADED		= BIT(1),
 	/* Bits 11:3 are available when the page is not swapped. */
+	SGX_ENCL_PAGE_RESERVED		= BIT(3),
+	SGX_ENCL_PAGE_RECLAIMED		= BIT(4),
+	SGX_ENCL_PAGE_VA_OFFSET_MASK	= GENMASK_ULL(11, 3),
 	SGX_ENCL_PAGE_ADDR_MASK		= PAGE_MASK,
 };
 
@@ -79,10 +95,12 @@ struct sgx_encl {
 	struct mutex lock;
 	struct mm_struct *mm;
 	struct file *backing;
+	struct file *pcmd;
 	struct kref refcount;
 	unsigned long base;
 	unsigned long size;
 	unsigned long ssaframesize;
+	struct list_head va_pages;
 	struct radix_tree_root page_tree;
 	struct list_head add_page_reqs;
 	struct work_struct add_page_work;
@@ -102,6 +120,16 @@ static inline pgoff_t sgx_encl_page_backing_index(struct sgx_encl_page *page,
 	return PFN_DOWN(page->desc - encl->base);
 }
 
+static inline int sgx_encl_page_pcmd_offset(struct sgx_encl_page *page,
+					    struct sgx_encl *encl)
+{
+	int index;
+
+	index = sgx_encl_page_backing_index(page, encl);
+	return (index & (PAGE_SIZE / sizeof(struct sgx_pcmd) - 1)) *
+	       sizeof(struct sgx_pcmd);
+}
+
 extern struct workqueue_struct *sgx_add_page_wq;
 extern u64 sgx_encl_size_max_32;
 extern u64 sgx_encl_size_max_64;
@@ -174,6 +202,10 @@ int sgx_test_and_clear_young(struct sgx_encl_page *page);
 void sgx_flush_cpus(struct sgx_encl *encl);
 void sgx_set_page_loaded(struct sgx_encl_page *encl_page,
 			 struct sgx_epc_page *epc_page);
+struct sgx_epc_page *sgx_alloc_va_page(void);
+unsigned int sgx_alloc_va_slot(struct sgx_va_page *va_page);
+void sgx_free_va_slot(struct sgx_va_page *va_page, unsigned int offset);
+bool sgx_va_page_full(struct sgx_va_page *va_page);
 struct page *sgx_get_backing(struct file *file, pgoff_t index);
 void sgx_put_backing(struct page *backing_page, bool write);
 
drivers/platform/x86/intel_sgx/sgx_encl.c +128 -1

203 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_encl_page.c new file +177 -0

177 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_fault.c new file +107 -0

107 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_util.c +71 -0

77 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_vma.c +15 -0
index 17e95a0c734c..cc0993b4fd40 100644
@@ -37,7 +37,22 @@ static void sgx_vma_close(struct vm_area_struct *vma)
 	kref_put(&encl->refcount, sgx_encl_release);
 }
 
+static int sgx_vma_fault(struct vm_fault *vmf)
+{
+	unsigned long addr = (unsigned long)vmf->address;
+	struct vm_area_struct *vma = vmf->vma;
+	struct sgx_encl_page *entry;
+
+	entry = sgx_fault_page(vma, addr, 0);
+
+	if (!IS_ERR(entry) || PTR_ERR(entry) == -EBUSY)
+		return VM_FAULT_NOPAGE;
+	else
+		return VM_FAULT_SIGBUS;
+}
+
 const struct vm_operations_struct sgx_vm_ops = {
 	.close = sgx_vma_close,
 	.open = sgx_vma_open,
+	.fault = sgx_vma_fault,
 };
Signature · 1 line
2.17.1
#67 lore raw

PATCH v14 17/19 x86/sgx: Add a simple swapper for the EPC memory manager

In reply to Jarkko Sakkinen

The swapper thread ksgxswapd reclaims pages on the event when the number of free EPC pages goes below %SGX_NR_LOW_PAGES up until it reaches %SGX_NR_HIGH_PAGES.

Pages are reclaimed in LRU fashion from a global list. The consumers take care of calling EBLOCK (block page from new accesses), ETRACK (restart counting the entering hardware threads) and EWB (write page to the regular memory) because executing these operations usually (if not always) requires to do some subsystem-internal locking operations.

---
 arch/x86/Kconfig                           |   1 +
 arch/x86/include/asm/sgx.h                 |   7 +
 arch/x86/kernel/cpu/intel_sgx.c            | 226 ++++++++++++++++++++-
 drivers/platform/x86/intel_sgx/sgx_encl.c  |   3 +-
 drivers/platform/x86/intel_sgx/sgx_fault.c |   1 +
 5 files changed, 226 insertions(+), 12 deletions(-)
arch/x86/Kconfig +1 -0
index b47e1a144409..5c329dfd0fd9 100644
@@ -1916,6 +1916,7 @@ config X86_INTEL_MEMORY_PROTECTION_KEYS
 config INTEL_SGX_CORE
 	bool "Intel SGX core functionality"
 	depends on X86_64 && CPU_SUP_INTEL
+	select INTEL_SGX
 	help
 	Intel Software Guard eXtensions (SGX) CPU feature that allows ring 3
 	applications to create enclaves: private regions of memory that are
arch/x86/include/asm/sgx.h +7 -0
index d5c535db094c..6bc4f3a38119 100644
@@ -315,5 +315,12 @@ int __sgx_free_page(struct sgx_epc_page *page);
 void sgx_free_page(struct sgx_epc_page *page);
 int sgx_einit(struct sgx_sigstruct *sigstruct, struct sgx_einittoken *token,
 	      struct sgx_epc_page *secs, u64 *lepubkeyhash);
+void sgx_page_reclaimable(struct sgx_epc_page *page);
+
+bool sgx_encl_page_get(struct sgx_epc_page *epc_page);
+void sgx_encl_page_put(struct sgx_epc_page *epc_page);
+bool sgx_encl_page_reclaim(struct sgx_epc_page *epc_page);
+void sgx_encl_page_block(struct sgx_epc_page *epc_page);
+void sgx_encl_page_write(struct sgx_epc_page *epc_page);
 
 #endif /* _ASM_X86_SGX_H */
arch/x86/kernel/cpu/intel_sgx.c +215 -11

282 lines not on this page · raw message

drivers/platform/x86/intel_sgx/sgx_encl.c +2 -1
index f3306bd58978..d49eac05a88b 100644
@@ -171,6 +171,7 @@ static bool sgx_process_add_page_req(struct sgx_add_page_req *req,
 	encl->secs_child_cnt++;
 	sgx_set_page_loaded(encl_page, epc_page);
 	sgx_test_and_clear_young(encl_page);
+	sgx_page_reclaimable(encl_page->epc_page);
 	return true;
 }
 
@@ -890,7 +891,7 @@ int sgx_encl_load_page(struct sgx_encl_page *encl_page,
 	ret = __eldu(&pginfo, sgx_epc_addr(epc_page),
 		     sgx_epc_addr(encl_page->va_page->epc_page) + va_offset);
 	if (ret) {
-		sgx_err(encl, "ELDU returned %d\n", ret);
+		SGX_INVD(ret, encl, "ELDU returned %d (0x%x)", ret, ret);
 		ret = encls_to_err(ret);
 	}
 
drivers/platform/x86/intel_sgx/sgx_fault.c +1 -0
index 2f459329f29c..32d9e67d7556 100644
@@ -73,6 +73,7 @@ static struct sgx_encl_page *__sgx_fault_page(struct vm_area_struct *vma,
 
 	encl->secs_child_cnt++;
 	sgx_test_and_clear_young(entry);
+	sgx_page_reclaimable(entry->epc_page);
 	if (do_reserve)
 		entry->desc |= SGX_ENCL_PAGE_RESERVED;
 
Signature · 1 line
2.17.1
#68 lore raw

PATCH v14 18/19 platform/x86: ptrace() support for the SGX driver

In reply to Jarkko Sakkinen

Add VMA callbacks for ptrace() that can be used with debug enclaves. With debug enclaves data can be read and write the memory word at a time by using ENCLS(EDBGRD) and ENCLS(EDBGWR) leaf instructions.

---
 drivers/platform/x86/intel_sgx/sgx_vma.c | 109 +++++++++++++++++++++++
 1 file changed, 109 insertions(+)
drivers/platform/x86/intel_sgx/sgx_vma.c +109 -0
index cc0993b4fd40..df604e4d0d0a 100644
@@ -51,8 +51,117 @@ static int sgx_vma_fault(struct vm_fault *vmf)
 		return VM_FAULT_SIGBUS;
 }
 
+static int sgx_edbgrd(struct sgx_encl *encl, struct sgx_encl_page *page,
+		      unsigned long addr, void *data)
+{
+	unsigned long offset;
+	int ret;
+
+	offset = addr & ~PAGE_MASK;
+
+	if ((page->desc & SGX_ENCL_PAGE_TCS) &&
+	    offset > offsetof(struct sgx_tcs, gs_limit))
+		return -ECANCELED;
+
+	ret = __edbgrd(sgx_epc_addr(page->epc_page) + offset, data);
+	if (ret) {
+		sgx_dbg(encl, "EDBGRD returned %d\n", ret);
+		return encls_to_err(ret);
+	}
+
+	return 0;
+}
+
+static int sgx_edbgwr(struct sgx_encl *encl, struct sgx_encl_page *page,
+		      unsigned long addr, void *data)
+{
+	unsigned long offset;
+	int ret;
+
+	offset = addr & ~PAGE_MASK;
+
+	/* Writing anything else than flags will cause #GP */
+	if ((page->desc & SGX_ENCL_PAGE_TCS) &&
+	    offset != offsetof(struct sgx_tcs, flags))
+		return -ECANCELED;
+
+	ret = __edbgwr(sgx_epc_addr(page->epc_page) + offset, data);
+	if (ret) {
+		sgx_dbg(encl, "EDBGWR returned %d\n", ret);
+		return encls_to_err(ret);
+	}
+
+	return 0;
+}
+
+static int sgx_vma_access(struct vm_area_struct *vma, unsigned long addr,
+			  void *buf, int len, int write)
+{
+	struct sgx_encl *encl = vma->vm_private_data;
+	struct sgx_encl_page *entry = NULL;
+	unsigned long align;
+	char data[sizeof(unsigned long)];
+	int offset;
+	int cnt;
+	int ret = 0;
+	int i;
+
+	/* If process was forked, VMA is still there but vm_private_data is set
+	 * to NULL.
+	 */
+	if (!encl)
+		return -EFAULT;
+
+	if (!(encl->flags & SGX_ENCL_DEBUG) ||
+	    !(encl->flags & SGX_ENCL_INITIALIZED) ||
+	    (encl->flags & SGX_ENCL_DEAD))
+		return -EFAULT;
+
+	for (i = 0; i < len; i += cnt) {
+		if (!entry || !((addr + i) & (PAGE_SIZE - 1))) {
+			if (entry)
+				entry->desc &= ~SGX_ENCL_PAGE_RESERVED;
+
+			entry = sgx_fault_page(vma, (addr + i) & PAGE_MASK,
+					       true);
+			if (IS_ERR(entry)) {
+				ret = PTR_ERR(entry);
+				entry = NULL;
+				break;
+			}
+		}
+
+		/* Locking is not needed because only immutable fields of the
+		 * page are accessed and page itself is reserved so that it
+		 * cannot be swapped out in the middle.
+		 */
+
+		align = ALIGN_DOWN(addr + i, sizeof(unsigned long));
+		offset = (addr + i) & (sizeof(unsigned long) - 1);
+		cnt = sizeof(unsigned long) - offset;
+		cnt = min(cnt, len - i);
+
+		ret = sgx_edbgrd(encl, entry, align, data);
+		if (ret)
+			break;
+		if (write) {
+			memcpy(data + offset, buf + i, cnt);
+			ret = sgx_edbgwr(encl, entry, align, data);
+			if (ret)
+				break;
+		} else
+			memcpy(buf + i, data + offset, cnt);
+	}
+
+	if (entry)
+		entry->desc &= ~SGX_ENCL_PAGE_RESERVED;
+
+	return ret < 0 ? ret : i;
+}
+
 const struct vm_operations_struct sgx_vm_ops = {
 	.close = sgx_vma_close,
 	.open = sgx_vma_open,
 	.fault = sgx_vma_fault,
+	.access = sgx_vma_access,
 };
Signature · 1 line
2.17.1
#69 lore raw

PATCH v14 19/19 x86/sgx: Driver documentation

In reply to Jarkko Sakkinen

Documentation of the features of the Software Guard eXtensions used by the Linux kernel and basic design choices for the core and driver and functionality.

---
 Documentation/index.rst         |   1 +
 Documentation/x86/intel_sgx.rst | 185 ++++++++++++++++++++++++++++++++
 2 files changed, 186 insertions(+)
 create mode 100644 Documentation/x86/intel_sgx.rst
Documentation/index.rst +1 -0
index 5db7e87c7cb1..1cdc139adb40 100644
@@ -104,6 +104,7 @@ implementation.
    :maxdepth: 2
 
    sh/index
+   x86/index
 
 Filesystem Documentation
 ------------------------
Documentation/x86/intel_sgx.rst new file +185 -0

185 lines not on this page · raw message

Signature · 1 line
2.17.1
maintainer
#70 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

Quoted 16 lines · On Tue, 25 Sep 2018 16:06:56 +0300 Jarkko Sakkinen…

Documentation of the features of the Software Guard eXtensions used by the Linux kernel and basic design choices for the core and driver and functionality.

---
 Documentation/index.rst         |   1 +
 Documentation/x86/intel_sgx.rst | 185 ++++++++++++++++++++++++++++++++
 2 files changed, 186 insertions(+)
 create mode 100644 Documentation/x86/intel_sgx.rst
Documentation/index.rst +0 -0
index 5db7e87c7cb1..1cdc139adb40 100644

Documentation/index.rst implementation. · line 107

Quoted 4 lines
Documentation/index.rst +1 -0
@@ -104,6 +104,7 @@ implementation.
    :maxdepth: 2
 
    sh/index
+   x86/index

So you're adding this reference to x86/index, but the actual file (Documentation/x86/index.rst) doesn't exist; that will break the docs build. I do appreciate your doing your document in RST, though!

Who is the intended audience for this document? I see a bit of UAPI stuff that might be better placed in the userspace-api manual, mixed with stuff that user-space developers are unlikely to take much interest in. Might it be worth splitting this into two pieces?

Thanks,

jon

#71 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

In reply to Jarkko Sakkinen
Quoted 11 lines · Jarkko Sakkinen

Documentation of the features of the Software Guard eXtensions used by the Linux kernel and basic design choices for the core and driver and functionality.

Documentation/x86/intel_sgx.rst new file +2 -0
@@ -0,0 +1,185 @@
+===================
+Intel(R) SGX driver

Documentation/x86/intel_sgx.rst line 12

Quoted 10 lines
Documentation/x86/intel_sgx.rst new file +10 -0
+===================
+
+Introduction
+============
+
+Intel(R) SGX is a set of CPU instructions that can be used by applications to
+set aside private regions of code and data. The code outside the enclave is
+disallowed to access the memory inside the enclave by the CPU access control.
+In a way you can think that SGX provides inverted sandbox. It protects the
+application from a malicious host.

Well, recently hardware had some problems keeping its promises. So... what about rowhammer, meltdown and spectre?

Which ones apply, which ones do not, and on what cpu generations?

Documentation/x86/intel_sgx.rst line 28

Quoted 11 lines
Documentation/x86/intel_sgx.rst new file +11 -0
+Overview of SGX
+===============
+
+SGX has a set of data structures to maintain information about the enclaves and
+their security properties. BIOS reserves a fixed size region of physical memory
+for these structures by setting Processor Reserved Memory Range Registers
+(PRMRR).
+
+This memory range is protected from outside access by the CPU and all the data
+coming in and out of the CPU package is encrypted by a key that is generated for
+each boot cycle.

Encryption, that sounds nice, but it is hard to do right. If SGX protected code changes single bit in its memory, how many bits will be changed in physical RAM?

Can we get security people to look at this and perhaps tells us what properties it has?

Thanks,
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#72 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

Quoted 9 lines · Pavel Machek

Documentation/x86/intel_sgx.rst line 12

Quoted 5 lines · Jarkko Sakkinen
Documentation/x86/intel_sgx.rst new file +5 -0
+Intel(R) SGX is a set of CPU instructions that can be used by applications to
+set aside private regions of code and data. The code outside the enclave is
+disallowed to access the memory inside the enclave by the CPU access control.
+In a way you can think that SGX provides inverted sandbox. It protects the
+application from a malicious host.

Well, recently hardware had some problems keeping its promises. So... what about rowhammer, meltdown and spectre?

Doesn't hardware always have this problem over time?

Quoted 1 line

Which ones apply, which ones do not, and on what cpu generations?

Definitely should be refined.

Meltdowns approach AFAIK does not work because reads outside the enclave will always have a predefined value (-1) but only if the page is present, which was later exploited in the Foreshadow attack.

Quoted 3 lines

Encryption, that sounds nice, but it is hard to do right. If SGX protected code changes single bit in its memory, how many bits will be changed in physical RAM?

512-bit blocks and merkle tree based mac. It is pretty well documented in https://eprint.iacr.org/2016/204.pdf. I'll take not to myself to add this to the references.

Thanks for the feedback. The ocumentation is hard to drive forward w/o it.

/Jarkko

#73 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

Quoted 11 lines · Jarkko Sakkinen
Quoted 9 lines · Pavel Machek

Documentation/x86/intel_sgx.rst line 12

Quoted 5 lines · Jarkko Sakkinen
Documentation/x86/intel_sgx.rst new file +5 -0
+Intel(R) SGX is a set of CPU instructions that can be used by applications to
+set aside private regions of code and data. The code outside the enclave is
+disallowed to access the memory inside the enclave by the CPU access control.
+In a way you can think that SGX provides inverted sandbox. It protects the
+application from a malicious host.

Well, recently hardware had some problems keeping its promises. So... what about rowhammer, meltdown and spectre?

Quoted 1 line

Doesn't hardware always have this problem over time?

No, not really.

In this case, tries to protect from hardware "attacks" done by machine owner. That job is theoretically impossible, so you have harder situation than most..

Quoted 7 lines
Quoted 1 line

Which ones apply, which ones do not, and on what cpu generations?

Definitely should be refined.

Meltdowns approach AFAIK does not work because reads outside the enclave will always have a predefined value (-1) but only if the page is present, which was later exploited in the Foreshadow attack.

What about L1tf and https://github.com/lsds/spectre-attack-sgx ?

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#74 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

Quoted 7 lines · Pavel Machek
Quoted 5 lines

Definitely should be refined.

Meltdowns approach AFAIK does not work because reads outside the enclave will always have a predefined value (-1) but only if the page is present, which was later exploited in the Foreshadow attack.

What about L1tf and https://github.com/lsds/spectre-attack-sgx ?

L1TF is the vuln and Foreshadow is the attack taking advantage of the vuln. I didn't mean to patch the documention in my response or give extensive list of the vulns if you expected that.

For kernel documentation it does make sense to give a threat model but not enumerate every possible vuln.

/Jarkko

#75 lore raw

Re: [PATCH v14 19/19] x86/sgx: Driver documentation

In reply to Pavel Machek
Quoted 7 lines · Pavel Machek

Documentation/x86/intel_sgx.rst line 12

Quoted 5 lines
Documentation/x86/intel_sgx.rst new file +5 -0
+Intel(R) SGX is a set of CPU instructions that can be used by applications to
+set aside private regions of code and data. The code outside the enclave is
+disallowed to access the memory inside the enclave by the CPU access control.
+In a way you can think that SGX provides inverted sandbox. It protects the
+application from a malicious host.

Well, recently hardware had some problems keeping its promises. So... what about rowhammer, meltdown and spectre?

There's a ton of documentation out there about what kinds of protections SGX provides. I don't think this is an appropriate place to have an exhaustive discussion about it. But, there's extensive discussion of it on Intel's security site:

https://software.intel.com/security-software-guidance/

There's documentation on how L1TF affects SGX here:

https://software.intel.com/security-software-guidance/software-guidance/l1-terminal-fault

Or Spectre v2 here:

https://software.intel.com/security-software-guidance/software-guidance/bounds-check-bypass

Quoted 1 line

Which ones apply, which ones do not, and on what cpu generations?

The CVEs list this in pretty exhaustive detail. The L1TF/SGX one, for example:

https://nvd.nist.gov/vuln/detail/CVE-2018-3615

Lists a bunch of processor models.