Tomáš Mráz
Tomáš Mráz
This should be merged to all the active branches.
I'd be OK with this backport. Could you please retarget this against the openssl-3.5 branch?
Merged to the 3.5, 3.4, 3.3, 3.2 and 3.0 branches. Thank you for your contribution.
This needs a rebase due to conflicts.
This looks like a valid approach, except for one thing - what about the performance of using the sparse array as the backend storage. Would it make sense to keep...
There is one more possible radical way how to avoid per-context thread locals (and all the thread locking stuff as well!). Basically have a boolean in an OSSL_LIB_CTX indicating the...
The reasoning for the above - with it you can have basically two application models: 1. shared libctxes, new libctx is used only for either some special operation like fips-enabled...
> I think the approach of thread-friendly libctx and single-thread libctx is like adding yet another knob user/application can use to hurt its knee. If user/vendor is absolutely sure not...
I am afraid that what we overdesigned is the inclusion of locking and thread-locals and whatnot in the libcrypto core. As that is what we see on performance results. In...
> Is there any button that I can click on which will re-run all checks? Nope, that can be done only by a maintainer. Once your first PR is merged...