Linux-man Archive on
 help / color / Atom feed
* [PATCH] sgx.7: New page with overview of Software Guard eXtensions (SGX)
@ 2020-12-02  7:15 Jarkko Sakkinen
  2020-12-02 10:37 ` Michael Kerrisk (man-pages)
  2020-12-02 11:50 ` Alejandro Colomar (mailing lists; readonly)
  0 siblings, 2 replies; 13+ messages in thread
From: Jarkko Sakkinen @ 2020-12-02  7:15 UTC (permalink / raw)
  To: mtk.manpages; +Cc: linux-man, linux-sgx, dave.hansen, x86, Jarkko Sakkinen

Signed-off-by: Jarkko Sakkinen <>
 man7/sgx.7 | 198 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 198 insertions(+)
 create mode 100644 man7/sgx.7

diff --git a/man7/sgx.7 b/man7/sgx.7
new file mode 100644
index 000000000..429c9b64d
--- /dev/null
+++ b/man7/sgx.7
@@ -0,0 +1,198 @@
+.\" Copyright (C) 2020 Intel Corporation
+.\" Permission is granted to make and distribute verbatim copies of this
+.\" manual provided the copyright notice and this permission notice are
+.\" preserved on all copies.
+.\" Permission is granted to copy and distribute modified versions of this
+.\" manual under the conditions for verbatim copying, provided that the
+.\" entire resulting derived work is distributed under the terms of a
+.\" permission notice identical to this one.
+.\" Since the Linux kernel and libraries are constantly changing, this
+.\" manual page may be incorrect or out-of-date.  The author(s) assume no
+.\" responsibility for errors or omissions, or for damages resulting from
+.\" the use of the information contained herein.  The author(s) may not
+.\" have taken the same level of care in the production of this manual,
+.\" which is licensed free of charge, as they might when working
+.\" professionally.
+.\" Formatted or processed versions of this manual, if unaccompanied by
+.\" the source, must acknowledge the copyright and authors of this work.
+.TH SGX 7 2020-12-02 "Linux" "Linux Programmer's Manual"
+sgx - overview of Software Guard eXtensions
+Intel Software Guard eXtensions (SGX) allow user space applications to
+set aside private memory regions of code and data.
+These memory regions are called as enclaves.
+SGX must be enabled by the BIOS.
+If SGX appears to be unsupported on a system having the hardware
+support, ensure that SGX is enabled in the BIOS.
+If a BIOS presents a choice between \[lq]Enabled\[rq] and \[lq]Software
+Enabled\[rq] modes for SGX, choose \[lq]Enabled\[rq].
+An enclave can be only entered at a fixed set of entry points.
+Each of them can hold a single hardware thread at a time.
+While the enclave is loaded from a regular binary file, only the threads
+inside the enclave can access its memory.
+Although carved out of normal DRAM, enclave memory is marked in the
+system memory map as reserved and is not managed by the Linux memory
+There may be several regions spread across the system.
+Each contiguous region is called an Enclave Page Cache (EPC) section.
+EPC sections are enumerated via CPUID.
+These regions are encrypted when they leave the LLC.
+SGX is available only if the kernel was configured and built with the
+\f[B]CONFIG_X86_SGX\f[R] option.
+The hardware support for SGX can be observed from
+\f[I]/proc/cpuinfo\f[R] with the \[lq]flags\[rq] field containing
+.SS Enclave management
+Enclave\[cq]s life-cycle starts by opening \f[I]/dev/sgx_enclave\f[R],
+and ends once all the references to the opened file have been closed.
+After opening the enclave, its contents must be populated with the ioctl
+interface provided by \f[I]<asm/sgx.h>\f[R].
+ENCLS is a privileged (ring-0) instruction with the functionality for
+managing enclave memory, and the ioctl inteface provides a wrapper for
+Enclave construction starts by calling \f[B]SGX_IOC_ENCLAVE_CREATE\f[R],
+which takes in the initial version of SGX Enclave Control Structure
+SGX Enclave Control Structure (SECS) contains the description of the
+The ioctl calls ENCLS[ECREATE] function, which will copy SECS to the SGX
+reserved memory.
+SECS is never mapped to the process address space, and thus cannot be
+directly referenced.
+Among other things, SECS contains the base address and size of the
+enclave, meaning that the addresss space must be carved out before the
+There is also a hardware constrain that the size must be a power of two,
+and the base address must be also naturally aligned relative to the
+Some of the SECS fields must be initialized to zero because their values
+are determined dynamically after the enclave has been created.
+Most importantly SECS contains two SHA256 hashes: MRENCLAVE and
+Each enclave invocation, during the enclave construction, hashes its
+defining parameters MRENCLAVE.
+Arbitrary data can be also hashed into MRENCLAVE via the ENCLS[EEXTEND]
+MRSIGNER contains the hash of the enclave signer key, which is used to
+sign the SIGSTRUCT structure passed to the ENCLS[EINIT] function.
+A running enclave can use both of these fields as material for keys
+acquired with the ENCLU[EGETKEY]unction.
+After the enclave has been created, a series of
+\f[B]SGX_IOC_ENCLAVE_ADD_PAGES\f[R]calls are issued.
+This ioctl copies the provided data to the enclave memory by invoking
+the ENCLS[EADD] function, and optionally hashes it with help of the
+ENCLS[EEXTEND] function.
+Hashing is optional because it is a slow operation and sometimes not
+For instance, one might only care to measure the code pages.
+Finally, when all the data has been copied to the enclave memory,
+\f[B]SGX_IOC_ENCLAVE_INIT\f[R] is called, which invokes the ENCLS[EINIT]
+This function inspects that the accumulated MRENCLAVE matches the
+MRENCLAVE inside the caller provided SIGSTRUCT, andlocks down the
+enclave from further build operations.
+After this, the enclave can be invoked.
+.SS Enclave memory mapping
+The processor tracks EPC pages in a hardware metadata structure called
+the \f[I]Enclave Page Cache Map (EPCM)\f[R].
+The EPCM contains an entry for each EPC page which describes the owning
+enclave, access rights and page type among the other things.
+EPCM permissions are separate from the normal page tables.
+This prevents the kernel from, for instance, allowing writes to data
+which an enclave wishes to remain read-only.
+EPCM permissions may only impose additional restrictions on top of
+normal x86 page permissions.
+For all intents and purposes, the SGX architecture allows the processor
+to invalidate all EPCM entries at will.
+This requires that software be prepared to handle an EPCM fault at any
+In practice, this can happen on events like power transitions when the
+ephemeral key that encrypts enclave memory is lost.
+Kernel records EPCM permissions when \f[B]SGX_IOC_ENCLAVE_ADD_PAGES\f[R]
+is called.
+When the pages are mapped to memory via mmap() or mprotect(), the EPCM
+permissions are compared against the declared permissions.
+If the declared permissions have bits set that are not part of the EPCM
+permissions, the operation fails with \f[B]-EACCES\f[R].
+.SS Enclave invocation
+Enclaves encounter exceptions for lots of reasons: everything from
+enclave page faults to NULL pointer de-references, to system calls that
+must be called by a delegate from outside the enclave.
+Also, the enclave memory can be invalidated at any point of time when
+system state changes.
+For instance, when system wakes up from a sleep state, all EPC regions
+get invalidated.
+An SGX run-time must be prepared to all of this, and must be ready to
+even rebuild the enclave when the memory is invalidated.
+This type of exception handling has been traditionally been handled in
+SIGSEGV handlers, registered by the library.
+But, being process-wide, shared state, signal handling and shared
+libraries do not mix well.
+In order to assist the run-time, kernel provides a vDSO,
+\f[B]vsgx_enter_enclave\f[R], which wraps enclave entry functions
+ENCLU[EENTER] and EENTER[ERESUME], and returns exceptions to the caller
+at the point they happen.
+The exception information is filled in RDI, RSI and RDX.
+The kernel-provided user space portion of the vDSO handler will place
+this information in a user-provided buffer, or optionally trigger a
+user-provided callback at the time of the exception.
+The vDSO function calling convention uses the standard RDI RSI, RDX,
+RCX, R8 and R9 registers.
+This makes it possible to declare the vDSO as a C prototype, but other
+than that there is no specific support for SystemV ABI.
+Things like storing XSAVE are the responsibility of the enclave and the
+.SS Page reclaimer
+Just like normal RAM, there is a limited amount of enclave memory
+available and over-committing it is a very valuable tool to reduce
+resource use.
+Introduce a simple reclaim mechanism for enclave pages.
+In contrast to normal page reclaim, the kernel cannot directly access
+enclave memory.
+To get around this, the SGX architecture provides a set of functions to
+Among other things, these functions copy enclave memory to and from
+normal memory, encrypting it and protecting its integrity in the
+Kernel provides a page reclaimer by using these functions.
+It picks victim pages in LRU fashion from all the enclaves running in
+the system.
+A new kernel thread (ksgxd) reclaims pages in the background based on
+watermarks, similar to normal kswapd.
+All enclave pages can be reclaimed, architecturally.
+But, there are some limits to this, such as the special SECS metadata
+page which must be reclaimed last.
+The page version array (used to mitigate replaying old reclaimed pages)
+is also architecturally reclaimable, but not yet implemented.
+The end result is that the vast majority of enclave pages are currently

^ permalink raw reply	[flat|nested] 13+ messages in thread

end of thread, back to index

Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-12-02  7:15 [PATCH] sgx.7: New page with overview of Software Guard eXtensions (SGX) Jarkko Sakkinen
2020-12-02 10:37 ` Michael Kerrisk (man-pages)
2020-12-02 17:17   ` Jarkko Sakkinen
2020-12-04  1:55     ` Jarkko Sakkinen
2020-12-02 11:50 ` Alejandro Colomar (mailing lists; readonly)
2020-12-02 12:05   ` Michael Kerrisk (man-pages)
2020-12-02 14:20   ` Michael Kerrisk (man-pages)
2020-12-02 14:45     ` Alejandro Colomar (man-pages)
2020-12-02 14:47       ` Michael Kerrisk (man-pages)
2020-12-18 10:24         ` Alejandro Colomar (man-pages)
2020-12-18 12:51           ` Jarkko Sakkinen
2020-12-02 17:27     ` Jarkko Sakkinen
2020-12-02 17:21   ` Jarkko Sakkinen

Linux-man Archive on

Archives are clonable:
	git clone --mirror linux-man/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-man linux-man/ \
	public-inbox-index linux-man

Example config snippet for mirrors

Newsgroup available over NNTP:

AGPL code for this site: git clone