Add TPM support

This patch adds support for an Atmel TPM chip.

Background: TPMs are rather complex chips, supporting many commands and implementing complex crypto protocols like Direct Anonymous Attestation (DAA). Therefore, this patch does not directly implement a TPM chip, but instead utilizes the TPM emulator project (http://tpm-emulator.berlios.de/).
The TPM emulator can be run as a daemon, communicating through a unix domain socket.

This patch adds a "-tpm path" parameter to qemu, where "path" is the unix domain socket of the TPM emulator. If the parameter is given, the chip is registered in the emulated system. Otherwise, behaviour is unchanged.

The interface presented inside qemu is that of an Atmel TPM chip, simply because there is a Linux driver for this chip and the interface is very simple. I do not own any TPM chip, therefore the interface was written purely by looking at the Linux driver.

Use case: This patch makes it possible to experiment with software like IBMs Integrity Measurement Architecture (IMA), without having an actual TPM (this patch was developed for a demonstration involving IMA, among other things). It should also be possible to use Microsofts BitLocker technology, although this hasn't been tested yet.

---
 Makefile.target |    3 +
 hw/tpm.c        |  219 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 vl.c            |   12 +++
 vl.h            |    4 +
 4 files changed, 238 insertions(+), 0 deletions(-)
 create mode 100644 hw/tpm.c

Makefile.target +3 -0

index 33dc285..3665e82 100644
@@ -460,6 +460,9 @@ VL_OBJS += ne2000.o
 VL_OBJS += pcnet.o
 VL_OBJS += rtl8139.o
 
+# TPM device
+VL_OBJS += tpm.o
+
 ifeq ($(TARGET_BASE_ARCH), i386)
 # Hardware support
 VL_OBJS+= ide.o pckbd.o ps2.o vga.o $(SOUND_HW) dma.o $(AUDIODRV)

hw/tpm.c new file +5 -0

new file mode 100644
index 0000000..eaebd46
@@ -0,0 +1,219 @@
+/*
+ * TPM emulation
+ * Written by Thomas Bleher <[email protected]>.
+ *
+ * This driver emulates a TPM chip. TPM chips are quite complex, and a TPM

Quoted 13 lines · Thomas Bleher

hw/tpm.c new file +10 -0

@@ -0,0 +1,219 @@
+/*
+ * TPM emulation
+ * Written by Thomas Bleher <[email protected]>.
+ *
+ * This driver emulates a TPM chip. TPM chips are quite complex, and a TPM
+ * emulator already exists, therefore this driver just connects to this
+ * emulator and forwards all the data. For the TPM emulator project, see
+ * http://tpm-emulator.berlios.de/
+ *
+ * The author does not own any TPM chip himself, so the Linux Kernel driver for

hw/tpm.c line 15

Quoted 5 lines

hw/tpm.c new file +5 -0

+ * Atmel TPM chips was taken as a reference. The code works fine with the Linux
+ * driver, but no tests have been done on other operating systems.
+ *
+ * Some structures are copied from the Linux Kernel source code.
+ */

So the License of this file is "GPL, Version 2"? The license should be mentioned in the comment.

This patch adds support for an Atmel TPM chip.

Background: TPMs are rather complex chips, supporting many commands and implementing complex crypto protocols like Direct Anonymous Attestation (DAA). Therefore, this patch does not directly implement a TPM chip, but instead utilizes the TPM emulator project (http://tpm-emulator.berlios.de/).
The TPM emulator can be run as a daemon, communicating through a unix domain socket.

This patch adds a "-tpm path" parameter to qemu, where "path" is the unix domain socket of the TPM emulator. If the parameter is given, the chip is registered in the emulated system. Otherwise, behaviour is unchanged.

The interface presented inside qemu is that of an Atmel TPM chip, simply because there is a Linux driver for this chip and the interface is very simple. I do not own any TPM chip, therefore the interface was written purely by looking at the Linux driver.

Use case: This patch makes it possible to experiment with software like IBMs Integrity Measurement Architecture (IMA), without having an actual TPM (this patch was developed for a demonstration involving IMA, among other things). It should also be possible to use Microsofts BitLocker technology, although this hasn't been tested yet.

---

Quoted 14 lines · Thiemo Seufer

Quoted 13 lines · Thomas Bleher

hw/tpm.c new file +10 -0

@@ -0,0 +1,219 @@
+/*
+ * TPM emulation
+ * Written by Thomas Bleher <[email protected]>.
+ *
+ * This driver emulates a TPM chip. TPM chips are quite complex, and a TPM
+ * emulator already exists, therefore this driver just connects to this
+ * emulator and forwards all the data. For the TPM emulator project, see
+ * http://tpm-emulator.berlios.de/
+ *
+ * The author does not own any TPM chip himself, so the Linux Kernel driver for

Quoted 8 lines

hw/tpm.c line 16

Quoted 5 lines

hw/tpm.c new file +5 -0

+ * Atmel TPM chips was taken as a reference. The code works fine with the Linux
+ * driver, but no tests have been done on other operating systems.
+ *
+ * Some structures are copied from the Linux Kernel source code.
+ */

So the License of this file is "GPL, Version 2"? The license should be mentioned in the comment.

I think that the parts I copied are not copyrightable, as I only copied the two enums (I didn't copy any structures, the comment was wrong) and, modulo naming, I see no other way to implement this.

So I would be willing to license this under a more liberal license, but to be on the safe side, GNU GPLv2 is the best choice.

Quoted 24 lines · Thomas Bleher

Quoted 14 lines · Thiemo Seufer

Quoted 13 lines · Thomas Bleher

hw/tpm.c new file +10 -0

@@ -0,0 +1,219 @@
+/*
+ * TPM emulation
+ * Written by Thomas Bleher <[email protected]>.
+ *
+ * This driver emulates a TPM chip. TPM chips are quite complex, and a TPM
+ * emulator already exists, therefore this driver just connects to this
+ * emulator and forwards all the data. For the TPM emulator project, see
+ * http://tpm-emulator.berlios.de/
+ *
+ * The author does not own any TPM chip himself, so the Linux Kernel driver for

Quoted 8 lines

hw/tpm.c line 16

Quoted 5 lines

hw/tpm.c new file +5 -0

+ * Atmel TPM chips was taken as a reference. The code works fine with the Linux
+ * driver, but no tests have been done on other operating systems.
+ *
+ * Some structures are copied from the Linux Kernel source code.
+ */

So the License of this file is "GPL, Version 2"? The license should be mentioned in the comment.

Quoted 3 lines

I think that the parts I copied are not copyrightable, as I only copied the two enums (I didn't copy any structures, the comment was wrong) and, modulo naming, I see no other way to implement this.

Ok, so the Kernel license isn't relevant here.

Quoted 2 lines

So I would be willing to license this under a more liberal license, but to be on the safe side, GNU GPLv2 is the best choice.

I didn't intend to enforce GPL licensing, I just concluded from the description that the patch would include substantial parts of kernel source code. Since this isn't the case, feel free to choose your preferred license for it.

Quoted 31 lines · Thiemo Seufer

Quoted 24 lines · Thomas Bleher

Quoted 14 lines · Thiemo Seufer

Quoted 13 lines · Thomas Bleher

hw/tpm.c new file +10 -0

@@ -0,0 +1,219 @@
+/*
+ * TPM emulation
+ * Written by Thomas Bleher <[email protected]>.
+ *
+ * This driver emulates a TPM chip. TPM chips are quite complex, and a TPM
+ * emulator already exists, therefore this driver just connects to this
+ * emulator and forwards all the data. For the TPM emulator project, see
+ * http://tpm-emulator.berlios.de/
+ *
+ * The author does not own any TPM chip himself, so the Linux Kernel driver for

Quoted 8 lines

hw/tpm.c line 16

Quoted 5 lines

hw/tpm.c new file +5 -0

+ * Atmel TPM chips was taken as a reference. The code works fine with the Linux
+ * driver, but no tests have been done on other operating systems.
+ *
+ * Some structures are copied from the Linux Kernel source code.
+ */

So the License of this file is "GPL, Version 2"? The license should be mentioned in the comment.

Quoted 3 lines

I think that the parts I copied are not copyrightable, as I only copied the two enums (I didn't copy any structures, the comment was wrong) and, modulo naming, I see no other way to implement this.

Ok, so the Kernel license isn't relevant here.

Quoted 7 lines

Quoted 2 lines

So I would be willing to license this under a more liberal license, but to be on the safe side, GNU GPLv2 is the best choice.

I didn't intend to enforce GPL licensing, I just concluded from the description that the patch would include substantial parts of kernel source code. Since this isn't the case, feel free to choose your preferred license for it.

Is there a preferred license for qemu? I see that the code as a whole is licensed under the GPLv2, but some code is under the LGPL or some BSD license. I'm willing to license it under whatever license suits qemu best.

Thiemo Seufer told me that GPLv2 is fine for qemu, therefore I'd like to ask that this patch be included in qemu as I posted it (the second version with the clarified GPLv2 license).

Quoted 3 lines · Thomas Bleher

Thiemo Seufer told me that GPLv2 is fine for qemu, therefore I'd like to ask that this patch be included in qemu as I posted it (the second version with the clarified GPLv2 license).

I prefer that a BSD style license is used, especially if the code just contains wrappers.

Quoted 6 lines · Fabrice Bellard

Quoted 3 lines · Thomas Bleher

Thiemo Seufer told me that GPLv2 is fine for qemu, therefore I'd like to ask that this patch be included in qemu as I posted it (the second version with the clarified GPLv2 license).

I prefer that a BSD style license is used, especially if the code just contains wrappers.

Fine with me, too.

Review

Quoted 7 lines · Thomas Bleher

Quoted 1 line · Fabrice Bellard

Thomas Bleher wrote:

(quoted diff) +5 -0

+        result = write(s->tpm_fd, s->send_data, s->send_data_index);
+        if (result < s->send_data_index) {
+            fprintf(stderr, "WARNING: Failed to write data to tpm!\n");
+            return ATML_STATUS_BUSY;
+        }

You should handle EINTR and EAGAIN.

Review

Quoted 8 lines

(quoted diff) +8 -0

+        s->send_data_index = 0;
+        s->recv_data_pos = 0;
+        s->recv_data_length = 0;
+        s->data_to_send = 0;
+        s->data_to_recv = 1;
+    }
+    if (s->data_to_recv) {
+        if (poll(&(s->tpm_poll), 1, 0) > 0) {

poll is not needed as read will block.

Review

Quoted 5 lines

(quoted diff) +5 -0

+            result = read(s->tpm_fd, s->recv_data, 2048);
+            if (result < 6) { // a minimal packet is 6 bytes long
+                fprintf(stderr, "WARNING: Not enough data from tpm!\n");
+                return ATML_STATUS_BUSY;
+            }

same comment as write().

Review

Quoted 3 lines

(quoted diff) +3 -0

+    res = connect(sock, (struct sockaddr*)&addr, sizeof(struct sockaddr_un));
+    if (res < 0)
+        return -1;

same comment for EINTR.

+/* should be called first, initializes all structures and connects to the external emulator */

(quoted diff) +2 -0

+void tpm_configure(const char* tpm_socket)
+{

Suppress this function (cf tpm_register).

Review

Quoted 2 lines

(quoted diff) +2 -0

+/* split of from tpm_configure() so the configuration can be called earlier */
+void tpm_register()

Ansi C requires void in parameters. Allocate the state and return it. Suppress the global variable tpm_state. Pass the path to this function and suppress the configure function. Add a global variable for the path and register the device if it is not NULL.

A few more remarks :

- tpm_register() must be called in hw/pc.c.

- you must also handle incomplete write and reads. Use something like unix_write() in vl.c and the equivalent for read.

- The code seems to be Unix specific so it must be disabled for win32.