[
https://issues.apache.org/jira/browse/AMQCPP-759?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Arjun Ray updated AMQCPP-759:
-----------------------------
Description:
The code has a defective test for overflow (at line 221)
{color:#cccccc} {color}{color:#c586c0}if{color}{color:#cccccc}
({color}{color:#9cdcfe}max{color}{color:#cccccc}
{color}{color:#d4d4d4}>{color}{color:#cccccc}
{color}{color:#9cdcfe}result{color}{color:#cccccc}) {{color}
{color:#cccccc} {color}{color:#c586c0}throw{color}{color:#cccccc}
{color}{color:#4ec9b0}exceptions{color}{color:#cccccc}::{color}{color:#4ec9b0}NumberFormatException{color}{color:#cccccc}({color}{color:#569cd6}{_}{{_}}FILE{{_}}{_}{color}{color:#cccccc},
{color}{color:#569cd6}{_}{{_}}LINE{{_}}{_}{color}{color:#cccccc},{color}
{color:#cccccc} {color}{color:#ce9178}"Long::parseLong - Parsed
value greater than max for radix."{color}{color:#cccccc});{color}
{color:#cccccc} }{color}
The bug shows up when the code base is compiled with optimization turned on
(typically, with -O2). With that the compiler (gcc) assumes that overflow will
not occur and therefore does not even check for it. As a result, an exception
is not thrown, which causes a unit test (which was expecting the exception for
invalid input) to fail. Without optimization, the overflow causes undefined
behavior (according to the C++ standards), which in this particular case
results in the unit test passing accidentally.
This bug occurs only with unusual inputs, but in theory should not be present
at all with a correct implementation of overflow detection. This issue has
been discussed in some depth at
[https://stackoverflow.com/questions/3944505/detecting-signed-overflow-in-c-c]
was:
The code involves a defective test for overflow.
{color:#cccccc} {color}{color:#c586c0}if{color}{color:#cccccc}
({color}{color:#9cdcfe}max{color}{color:#cccccc}
{color}{color:#d4d4d4}>{color}{color:#cccccc}
{color}{color:#9cdcfe}result{color}{color:#cccccc}) {{color}
{color:#cccccc} {color}{color:#c586c0}throw{color}{color:#cccccc}
{color}{color:#4ec9b0}exceptions{color}{color:#cccccc}::{color}{color:#4ec9b0}NumberFormatException{color}{color:#cccccc}({color}{color:#569cd6}_{_}FILE{_}_{color}{color:#cccccc},
{color}{color:#569cd6}_{_}LINE{_}_{color}{color:#cccccc},{color}
{color:#cccccc} {color}{color:#ce9178}"Long::parseLong - Parsed
value greater than max for radix."{color}{color:#cccccc});{color}
{color:#cccccc} }{color}
The bug shows up when the code base is compiled with optimization turned on
(typically, with -O2). With that the compiler (gcc) assumes that overflow will
not occur and therefore does not even check for it. As a result, an exception
is not thrown, which causes a unit test (which was expecting the exception for
invalid input) to fail. Without optimization, the overflow causes undefined
behavior (according to the C++ standards), which in this particular case
results in the unit test passing accidentally.
This bug occurs only with unusual inputs, but in theory should not be present
at all with a correct implementation of overflow detection. This issue has
been discussed in some depth at
https://stackoverflow.com/questions/3944505/detecting-signed-overflow-in-c-c
> src/main/decaf/lang/Long.cpp does not detect overflow properly
> --------------------------------------------------------------
>
> Key: AMQCPP-759
> URL: https://issues.apache.org/jira/browse/AMQCPP-759
> Project: ActiveMQ C++ Client
> Issue Type: Bug
> Components: Decaf
> Affects Versions: 3.9.0, 3.9.1, 3.9.2, 3.9.3, 3.9.4, 3.9.5
> Reporter: Arjun Ray
> Assignee: Timothy A. Bish
> Priority: Major
>
> The code has a defective test for overflow (at line 221)
> {color:#cccccc} {color}{color:#c586c0}if{color}{color:#cccccc}
> ({color}{color:#9cdcfe}max{color}{color:#cccccc}
> {color}{color:#d4d4d4}>{color}{color:#cccccc}
> {color}{color:#9cdcfe}result{color}{color:#cccccc}) {{color}
> {color:#cccccc} {color}{color:#c586c0}throw{color}{color:#cccccc}
> {color}{color:#4ec9b0}exceptions{color}{color:#cccccc}::{color}{color:#4ec9b0}NumberFormatException{color}{color:#cccccc}({color}{color:#569cd6}{_}{{_}}FILE{{_}}{_}{color}{color:#cccccc},
> {color}{color:#569cd6}{_}{{_}}LINE{{_}}{_}{color}{color:#cccccc},{color}
> {color:#cccccc} {color}{color:#ce9178}"Long::parseLong -
> Parsed value greater than max for radix."{color}{color:#cccccc});{color}
> {color:#cccccc} }{color}
> The bug shows up when the code base is compiled with optimization turned on
> (typically, with -O2). With that the compiler (gcc) assumes that overflow
> will not occur and therefore does not even check for it. As a result, an
> exception is not thrown, which causes a unit test (which was expecting the
> exception for invalid input) to fail. Without optimization, the overflow
> causes undefined behavior (according to the C++ standards), which in this
> particular case results in the unit test passing accidentally.
> This bug occurs only with unusual inputs, but in theory should not be present
> at all with a correct implementation of overflow detection. This issue has
> been discussed in some depth at
> [https://stackoverflow.com/questions/3944505/detecting-signed-overflow-in-c-c]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information, visit: https://activemq.apache.org/contact
