Tomáš Mráz
Tomáš Mráz
> > You could then check if the provider of the digest is the same as your provider, if not, then do not use the acceleration. > > I can...
Merged to the master branch. Thank you for your contribution.
The fix needs to be in EVP_PKEY_copy_parameters() - it should not try to use the evp_keymgmt_dup() in this situation at all, or it should fallback to using the legacy param_copy...
> Other parts of the default method are left unchanged. Commenting out the ossl_dh_is_foreign() check fixed the problem. Maybe the solution would be to improve the check instead of assuming...
> For outl the documentation currently says that the 'number of bytes written is placed in `I`'. This is clearly not true for this case since it writes zero padding...
> The documentation says it is the number of bytes written, which is not true. Would saying that it is the number of _plaintext_ bytes written, be OK?
> Do we really expect users to write code like this > > ``` > cipher(NULL, &sz); > sz += some weird additional value; > out = malloc(sz); > ```...
Squashed and merged to all the active branches. Thank you for your contribution.,
> So the log at https://github.com/openssl/openssl/actions/runs/6705970584/job/18221487067?pr=21668 found at least: Fix in https://github.com/openssl/openssl/pull/22572
Could you please rebase this and answer my outstanding question above?