23 Commits

Author SHA1 Message Date
David Horstmann
51fc6cf378 Explore sanitizers for memory poisoning
Consider MSan, ASan and Valgrind as options for implementing memory
poisoning tests. Come to the altered conclusion that Valgrind is the
best option.

Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-20 18:40:15 +01:00
David Horstmann
17b3716c5a Tweak compiler optimization evaluation section
* Remove references to the platform - this is unlikely to affect whether
copies are optimized.
* Note that the evaluation should test extreme optimisation settings.

Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-20 18:39:14 +01:00
David Horstmann
4e54abf182 Add section on possible use of Valgrind tracing
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 17:59:45 +01:00
David Horstmann
05ca3d9a1b Expand design for validation of careful access
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 16:45:37 +01:00
David Horstmann
a72b4ca734 Modify optimize-testing instructions
Mention -flto and whole-program optimization as this is the most
important aspect.

Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 15:22:15 +01:00
David Horstmann
3f7e42a750 Move implementation by module table earlier
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 15:14:50 +01:00
David Horstmann
dae0ad439f Add more detail in design of memory poisoning
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 15:12:34 +01:00
David Horstmann
0bd87f5959 Change unsigned int to uint8_t
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-19 13:45:21 +01:00
David Horstmann
23661cc232 Detailed design of memory protection strategy
Signed-off-by: David Horstmann <david.horstmann@arm.com>
2023-10-18 18:20:33 +01:00
Gilles Peskine
8ebeb9c180 Test for read-read inconsistency with mprotect and ptrace/gdb
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-16 18:37:02 +02:00
Gilles Peskine
87889ebe86 Fix editorial error with semantic consequences
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-16 15:40:02 +02:00
Gilles Peskine
a3ce6437bf Typos
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-16 15:39:37 +02:00
Gilles Peskine
1f2802c403 Suggest validating copy by memory poisoning
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 21:49:17 +02:00
Gilles Peskine
6998721c69 Add a section skeleton for copy bypass
It's something we're likely to want to do at some point.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 20:05:32 +02:00
Gilles Peskine
7bc1bb65e9 Short explanations of what is expected in the design sections
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 20:05:25 +02:00
Gilles Peskine
35de1f7a7d Distinguish whole-message signature from other asymmetric cryptography
Whole-message signature may process the message multiple times (EdDSA
signature does it).

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 20:04:16 +02:00
Gilles Peskine
9cad3b3a70 Design change for cipher/AEAD
There are many reasons why a driver might violate the security requirements
for plaintext or ciphertext buffers, so mandate copying.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 20:03:18 +02:00
Gilles Peskine
2859267a27 Clarify terminology: built-in driver
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 20:02:00 +02:00
Gilles Peskine
db00543b3a Add a section on write-read feedback
It's a security violation, although it's not clear whether it really needs
to influence the design.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 19:57:53 +02:00
Gilles Peskine
352095ca86 Simplify the relaxed output-output rule
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 19:56:22 +02:00
Gilles Peskine
60c453ee72 Expand explanations of the vulnerabilities
Add a few more examples.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 19:07:56 +02:00
Gilles Peskine
8daedaeac9 Fix typos and copypasta
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-13 18:47:29 +02:00
Gilles Peskine
f7806ca782 Analyze requirements for protection of arguments in shared memory
Propose a dual-approach strategy where some buffers are copied and others
can remain shared.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2023-10-12 16:00:11 +02:00