diff options
author | Matt Caswell <matt@openssl.org> | 2016-11-14 11:55:13 +0000 |
---|---|---|
committer | Matt Caswell <matt@openssl.org> | 2016-11-14 11:55:13 +0000 |
commit | 1fda5bc435ada1c70f2d3342bb9db98ac5840dc9 (patch) | |
tree | 30e1d5e9314ad99c2d2445169a61ef53e4561310 | |
parent | e72040c1dcd61d6669762a60924b8fa3a48c37fc (diff) | |
download | openssl-1fda5bc435ada1c70f2d3342bb9db98ac5840dc9.tar.gz |
Revert "Fixed deadlock in CRYPTO_THREAD_run_once for Windows"
This reverts commit 349d1cfddcfa33d352240582a3803f2eba39d9a0.
The proposed fix is incorrect. It marks the "run_once" code as having
finished before it has. The intended semantics of run_once is that no
threads should proceed until the code has run exactly once. With this
change the "second" thread will think the run_once code has already been
run and will continue, even though it is still in progress. This could
result in a crash or other incorrect behaviour.
Reviewed-by: Tim Hudson <tjh@openssl.org>
-rw-r--r-- | crypto/threads_win.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/crypto/threads_win.c b/crypto/threads_win.c index 5347c9e46b..4e0de908ee 100644 --- a/crypto/threads_win.c +++ b/crypto/threads_win.c @@ -78,8 +78,8 @@ int CRYPTO_THREAD_run_once(CRYPTO_ONCE *once, void (*init)(void)) do { result = InterlockedCompareExchange(lock, ONCE_ININIT, ONCE_UNINITED); if (result == ONCE_UNINITED) { - *lock = ONCE_DONE; init(); + *lock = ONCE_DONE; return 1; } } while (result == ONCE_ININIT); |