Wipe output on failure: key generation #586
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Current code is hard to follow and has complicated management for key generation contexts. However, this is not essential complexity. The core insight here is that while the context is shared, exporting private and public key are two independent operations. Therefore we can try exporting both keys, free the context, and only then deal with the results. It then becomes really natural to introduce a helper function to combine two results into one. While we're here, make sure that all local variables are initialized. This too makes error handling path easier to write and understand.
Now, let's make sure that we don't leak any information if key generation fails. For example, if the private key has been extracted just fine but extracting public key fails then the private key has already been written out into the user buffer. This is not nice. We can easily insert soter_wipe() calls to remove results from output buffers if key generation is not succesful. Note that we *do not* wipe the buffer if it is reported to be too small. In this case the length is now equal to the desired buffer length, not actual one provided by the caller. We'd probably overrun the buffer if we touch it. Instead, we can simply not wipe it because in this case we have not written anything into it yet. Symmetric key generation simply calls soter_rand() which wipes after itself like an adult. We leave a comment just in case the code gets more complex in the future, so that the developers don't forget to add a soter_wipe() call if necessary. This covers Themis code for key generation API: - themis_gen_sym_key - themis_gen_rsa_key_pair - themis_gen_ec_key_pair
In OpenSSL EVP_PKEY_CTX_ctrl() penultimate argument "p1" has "int" type. We use this function to set RSA key length in bits, and we incorrectly use "unsigned" type when interfacing with EVP_PKEY_CTX_ctrl(). Update the type to just "int" to avoid warnings by clang-tidy about narrowing casts. We can use "int" in rsa_key_length() because it's wide enough for current maximum length of 4096 bits.
ilammy
added
core
ilammy
requested review from
ignatk,
Lagovas,
storojs72 and
vixentael
as code owners
vixentael
approved these changes
Jan 31, 2020
Which parts, exactly? I can elaborate if you wish.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
More wiping 🧹
Refactor high-level EC/RSA key generation functions in Themis so that they are easier to follow and it is easier to insert additional actions into the error path. Then insert soter_wipe() for already exported public key if private key export fails or vice versa.
While we're here, I've noticed and fixed clang-tidy complaints about mismatched integer types in OpenSSL code for RSA key generation. (BoringSSL code is fine, it already uses correct types.)
Checklist