You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Ensure that HACL* states are not read concurrently #155835
Sometimes, it's possible to read an HACL* state concurrently:
sha3 reads the live HACL state outside the object lock at six sites
blake2's digest_size getter reads the live HACL state outside the lock
Instead, we can (1) add this static information just after construction (2) compute it inside the lock sections. The problem is that HACL* update() functions update the entire state so reading static information is unsafe.
Activity
pythongh-155835: fix `digest_size` and `block_size` data races on SHA…
pythongh-155835: fix `digest_size` and `block_size` data races on SHA…
pythongh-155835: fix `digest_size` and `block_size` data races on SHA…
pythongh-155835: fix `digest_size` data race on BLAKE-2 objects
gh-155835: fix `digest_size` and `block_size` data races on SHA-3 obj…
gh-155835: fix `digest_size` data race on BLAKE-2 objects (#155839)
pythongh-155835: fix `digest_size` and `block_size` data races on SHA…
pythongh-155835: fix `digest_size` data race on BLAKE-2 objects (pyth…
[3.13] gh-155835: fix `digest_size` and `block_size` data races on SH…
[3.15] gh-155835: fix `digest_size` and `block_size` data races on SH…
[3.14] gh-155835: fix `digest_size` and `block_size` data races on SH…