[openssl/openssl] a09a68: Add one more trace message to the torture_rcu_high...

0 views
Skip to first unread message

Bernd Edlinger

unread,
Sep 10, 2025, 4:38:37 PM (2 days ago) Sep 10
to openssl...@openssl.org
Branch: refs/heads/openssl-3.6
Home: https://github.com/openssl/openssl
Commit: a09a68cef79bcf4f91da1b3b28410a4d317bfb94
https://github.com/openssl/openssl/commit/a09a68cef79bcf4f91da1b3b28410a4d317bfb94
Author: Bernd Edlinger <bernd.e...@hotmail.de>
Date: 2025-09-10 (Wed, 10 Sep 2025)

Changed paths:
M test/threadstest.c

Log Message:
-----------
Add one more trace message to the torture_rcu_high test

It is interesting that in the very rare cases, where this
test failure has been observed so far, the rcu torture value
went always backwards to 0. This could be either due to
ossl_rcu_deref(&writer_ptr) returning NULL, or the initial
value of "new = CRYPTO_zalloc(sizeof(uint64_t), NULL, 0)"
still visible despite ossl_rcu_assign_ptr(&writer_ptr, &new)
immediatley after the "*new = global_ctr++" statement.
Add one additional trace message to find out what exactly
happens here, when it happens again.
Additionally, we do no longer initialize the new value to
zero but something else, so it can also be detected.

Related to #27267

Reviewed-by: Neil Horman <nho...@openssl.org>
Reviewed-by: Tomas Mraz <to...@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28487)

(cherry picked from commit 17d5c9297ef6f6a4d72788082d6772deb7a18f24)



To unsubscribe from these emails, change your notification settings at https://github.com/openssl/openssl/settings/notifications
Reply all
Reply to author
Forward
0 new messages