RFR: 8332506: SIGFPE In ObjectSynchronizer::is_async_deflation_needed() [v4]
Fredrik Bredberg
fbredberg at openjdk.org
Wed Jan 8 09:52:48 UTC 2025
On Tue, 7 Jan 2025 12:46:09 GMT, Fredrik Bredberg <fbredberg at openjdk.org> wrote:
>> This PR solves a division by zero problem, that according to the bug report happened in `ObjectSynchronizer::is_async_deflation_needed()`. As it turns out it really happened in the inlined `monitors_used_above_threshold()` function. The problematic line looked like this:
>>
>> `size_t monitor_usage = (monitors_used * 100LL) / ceiling;`
>>
>> Unfortunately the `ceiling` value was increased every time there where too many deflations without any progress. This whould eventually lead to an overflow in the `ceiling` value, and in unlucky circumstances, it would become zero. Thus causing a division by zero crash.
>>
>> This PR makes sure not to increase the `ceiling` value if `monitor_usage` is below the `MonitorUsedDeflationThreshold`.
>> It also makes sure the ceiling value is never zero, and does not wrap around.
>>
>> Tested okay in tier1-5 and `test/hotspot/jtreg/runtime/Monitor/MonitorUsedDeflationThresholdTest.java`
>
> Fredrik Bredberg has updated the pull request incrementally with one additional commit since the last revision:
>
> Update three after review
Thanks for the review.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/22815#issuecomment-2577243653
More information about the hotspot-runtime-dev
mailing list