On 02/11/2023 21:29, Rémy Maucherat wrote:
On Thu, Nov 2, 2023 at 7:37 PM <ma...@apache.org> wrote:
This is an automated email from the ASF dual-hosted git repository.
markt pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tomcat.git
The following commit(s) were added to refs/heads/main by this push:
new 55e8e599b1 Fix BZ 67929 - TLS config reload can trigger JVM crash
55e8e599b1 is described below
commit 55e8e599b17f512ebdab6860237bed6f73096321
Author: Mark Thomas <ma...@apache.org>
AuthorDate: Thu Nov 2 18:26:01 2023 +0000
Fix BZ 67929 - TLS config reload can trigger JVM crash
---
java/org/apache/tomcat/util/net/SSLHostConfig.java | 4 ++--
java/org/apache/tomcat/util/net/SSLHostConfigCertificate.java | 2 +-
webapps/docs/changelog.xml | 4 ++++
3 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/java/org/apache/tomcat/util/net/SSLHostConfig.java
b/java/org/apache/tomcat/util/net/SSLHostConfig.java
index a447c7fec0..9917fe6673 100644
--- a/java/org/apache/tomcat/util/net/SSLHostConfig.java
+++ b/java/org/apache/tomcat/util/net/SSLHostConfig.java
@@ -73,11 +73,11 @@ public class SSLHostConfig implements Serializable {
private String hostName = DEFAULT_SSL_HOST_NAME;
- private transient Long openSslConfContext = Long.valueOf(0);
+ private transient volatile Long openSslConfContext = Long.valueOf(0);
// OpenSSL can handle multiple certs in a single config so the reference
to
// the context is here at the virtual host level. JSSE can't so the
// reference is held on the certificate.
- private transient Long openSslContext = Long.valueOf(0);
+ private transient volatile Long openSslContext = Long.valueOf(0);
I don't understand the purpose of this (in trunk).
This is only used by the manager servlet (as a pseudo boolean) to
display various certificate info, but then the default (when it is 0)
should work fine anyway.
Am I missing anything ?
At the time I ported this from 9.0.x to 10.1.x and 11.0.x I was only
looking at which parts of the 9.0.x would not apply due to the removal
of APR. I wasn't looking at how the code was used in 10.1.x and 11.0.x.
Your are right that neither of the above changes are going to impact the
stability of the Connector during reload as they play no part in TLS
connections.
Having thought about it some more this morning, the only argument I have
for keeping these changes is that it keeps the code aligned between
versions. That isn't a great argument as we could always add a comment
in 9.0.x and 8.5.x to ensure that volatile was retained.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
