RFR: JDK-8319704: LogTagSet::set_output_level() should not accept NULL as LogOutput

[Johan Sjölen]

On Wed, 8 Nov 2023 11:00:35 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:

> Trivial change to add an assert.
> 
> See [JDK-8319104](https://bugs.openjdk.org/browse/JDK-8319104)
> 
> `LogTagSet::set_output_level()` and `LogOutputList::set_output_level` should not accept NULL as LogOutput* argument.
> 
> They do, currently, which leads to delayed crashes later since the NULL output is registered in the output list.
> 
> That triggered [JDK-8319104](https://bugs.openjdk.org/browse/JDK-8319104), where - due to a fluke in C++ initialization order and a mislabeling of LogTagSet tests as "TEST", without "_VM" - `LogTagSet::set_output_level()` was called with `LogConfiguration::StdoutLog` which was still uninitialized (null). 
> 
> An assert would have saved investigation time.

LGTM.

Oops, apparently this is a draft, sorry about that.

src/hotspot/share/logging/logOutputList.cpp line 55:

> 53: 
> 54: void LogOutputList::set_output_level(LogOutput* output, LogLevelType level) {
> 55:   assert(output != nullptr, "LogOutput is NULL");

HotSpot comment and logging style is for `nullptr` to be printed as "null"

-------------

Marked as reviewed by jsjolen (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/16555#pullrequestreview-1720127182
PR Comment: https://git.openjdk.org/jdk/pull/16555#issuecomment-1801720376
PR Review Comment: https://git.openjdk.org/jdk/pull/16555#discussion_r1386487244