[Webkit-unassigned] [Bug 163125] [GTK] Use libgcrypt instead of GnuTLS for CryptoDigest and SubtleCrypto HMAC implementation

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=163125

--- Comment #11 from Carlos Garcia Campos <cgarcia at igalia.com> ---
Comment on attachment 290961
  --> https://bugs.webkit.org/attachment.cgi?id=290961
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=290961&action=review

> Source/WebCore/crypto/gcrypt/CryptoAlgorithmHMACGCrypt.cpp:40
> +static int getGCryptDigestAlgorithm(CryptoAlgorithmIdentifier hashFunction)

Don't use "get" here. You could use cryptoAlgorithmIdentifierToGCryptDigestAlgorithm or something similar, or gcryptDigestAlgorithmForCryptoAlgorithmIdentifier

> Source/WebCore/crypto/gcrypt/CryptoAlgorithmHMACGCrypt.cpp:66
> +    gcry_mac_open(&hd, algorithm, 0, 0);

0, nullptr

> Source/WebCore/crypto/gcrypt/CryptoAlgorithmHMACGCrypt.cpp:69
> +    gcry_mac_read(hd, result.data(), &digestLength);

Don't you need to handle the return value of digestLength? Can it really be different to result.size()? I guess it can't be bigger than value returned by gcry_mac_get_algo_maclen, but can it be smaller? If it should be the same size I would add an assert, otherwise you would need to resize the vector here.

> Source/WebCore/crypto/gcrypt/CryptoAlgorithmHMACGCrypt.cpp:72
> +    return result;

Most of the gcry functions used above return a gcry_error_t that is not handled at all. We should call the failureCallback in platformSign if any of those fail.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20161010/3ef4d9a4/attachment.html>