[Pal/Linux-SGX] Add sgx.protected_mr{enclave,signer}_files manifest options
#2484
Conversation
There was a problem hiding this comment.
Reviewed 1 of 10 files at r1.
Reviewable status: 1 of 10 files reviewed, 3 unresolved discussions, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @dimakuv)
Documentation/manifest-syntax.rst, line 493 at r1 (raw file):
sgx.protected_files.[identifier] = "[URI]" sgx.protected_mrenclave_files.[identifier] = "[URI]" sgx.protected_mrsigner_files.[identifier] = "[URI]"
Could we use the proper syntax for this? (i.e. TOML arrays, not dictionaries)
We'll be changing this for all other file types soon, so I don't like adding this it this way and then rewriting 3 weeks later :)
Documentation/manifest-syntax.rst, line 521 at r1 (raw file):
enclaves from the same signer
I'd say "enclaves signed with the same key", it's less convoluted (i.e. it's IMO not obvious what "signer" means precisely).
LibOS/shim/test/regression/test_libos.py, line 728 at r1 (raw file):
self.assertIn(f'{node}/hugepages/hugepages-1048576kB/nr_hugepages: file', lines) def test_060_protected_file(self):
shouldn't this be named "sealed_file"?
There was a problem hiding this comment.
Reviewable status: 0 of 10 files reviewed, 3 unresolved discussions, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: Intel), "fixup! " found in commit messages' one-liners (waiting on @mkow)
Documentation/manifest-syntax.rst, line 493 at r1 (raw file):
Previously, mkow (Michał Kowalczyk) wrote…
Could we use the proper syntax for this? (i.e. TOML arrays, not dictionaries)
We'll be changing this for all other file types soon, so I don't like adding this it this way and then rewriting 3 weeks later :)
The old sgx.protected_files and the new sgx.protected_{mrenclave,mrsigner}_files are both handled by the same (old) code. This old code uses TOML dictionaries. So it will not be straight-forward to use TOML dicts only for the new options.
Anyway, it looks like high time to change from TOML dicts -> TOML arrays in our SGX files. I will submit a separate PR on this... And then I guess we'll rebase this PR on top of that other one.
Documentation/manifest-syntax.rst, line 521 at r1 (raw file):
Previously, mkow (Michał Kowalczyk) wrote…
enclaves from the same signerI'd say "enclaves signed with the same key", it's less convoluted (i.e. it's IMO not obvious what "signer" means precisely).
Done.
LibOS/shim/test/regression/test_libos.py, line 728 at r1 (raw file):
Previously, mkow (Michał Kowalczyk) wrote…
shouldn't this be named "sealed_file"?
Do you mean to rename just this test? Or do you want to rename sgx.protected_mrenclave_files/sgx.protected_mrsigner_files?
Aynway, I currently changed the name of this test and the name of the generated test file.
… options Previously, only `sgx.protected_files` were available in the manifest. This kind of protected files needs a provisioned master (wrap) key. But sometimes it is enough to seal files on the same platform for later usage by the same enclave or by enclaves of the same signer: this is the SGX sealing feature. This commit adds two more options to support SGX sealing: `sgx.protected_mrenclave_files` and `sgx.protected_mrsigner_files`. Similarly to `sgx.protected_files`, these new options specify lists of files that are encrypted by the SGX key generated via SGX instruction `EGETKEY(SEAL_KEY)`, bound to the MRENCLAVE/MRSIGNER enclave measurement (so that only instances of the same enclave/only enclaves with the same signer may decrypt protected files). A corresponding LibOS test is added and documentation is updated to reflect this. Signed-off-by: Dmitrii Kuvaiskii <[email protected]>
dimakuv
force-pushed
the
dimakuv/add-seal-key-to-protected-files-2
branch
from
1bcadc4 to
2dd6c64
Compare
There was a problem hiding this comment.
Reviewable status: 0 of 11 files reviewed, 3 unresolved discussions, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @mkow)
Documentation/manifest-syntax.rst, line 493 at r1 (raw file):
Previously, dimakuv (Dmitrii Kuvaiskii) wrote…
The old
sgx.protected_filesand the newsgx.protected_{mrenclave,mrsigner}_filesare both handled by the same (old) code. This old code uses TOML dictionaries. So it will not be straight-forward to use TOML dicts only for the new options.Anyway, it looks like high time to change from TOML dicts -> TOML arrays in our SGX files. I will submit a separate PR on this... And then I guess we'll rebase this PR on top of that other one.
Done. Rebased to the latest master. @mkow Please re-review.
|
Recreated this PR in gramineproject/gramine#101. Closing here. |
Description of the changes
Previously, only
sgx.protected_fileswere available in the manifest. This kind of protected files needs a provisioned master (wrap) key. But sometimes it is enough to seal files on the same platform for later usage by the same enclave or by enclaves of the same signer: this is the SGX sealing feature.This PR adds two more options to support SGX sealing:
sgx.protected_mrenclave_filesandsgx.protected_mrsigner_files.Similarly to
sgx.protected_files, these new options specify lists of files that are encrypted by the SGX key generated via SGX instructionEGETKEY(SEAL_KEY), bound to the MRENCLAVE/MRSIGNER enclave measurement (so that only instances of the same enclave/only enclaves with the same signer may decrypt protected files).This is a reincarnation of #2328 but with a different user interface (and more flexible -- now we have three keys at the same time).
Closes #2328.
How to test this PR?
A corresponding LibOS test is added and documentation is updated.
This change is