logSelection.cpp : too high tag values show up in some jtreg tests

[Johan Sjolen]

Hi,

Stefan's analysis seems correct to me. I have opened a ticket for this issue and assigned it to me. Please see JDK-8340923.

All the best,
Johan Sjölén

________________________________________
From: hotspot-runtime-dev <hotspot-runtime-dev-retn at openjdk.org> on behalf of Stefan Karlsson <stefan.karlsson at oracle.com>
Sent: Wednesday, September 25, 2024 15:25
To: Baesken, Matthias; hotspot-runtime-dev at openjdk.java.net
Subject: Re: logSelection.cpp : too high tag values show up in some jtreg tests

Isn't the problem that the LogSelection constructor doesn't initialize all elements (Only elements that are __NO_TAG gets initialized):
```
LogSelection::LogSelection(const LogTagType tags[LogTag::MaxTags], bool wildcard, LogLevelType level)
    : _ntags(0), _wildcard(wildcard), _level(level), _tag_sets_selected(0) {
  while (_ntags < LogTag::MaxTags && tags[_ntags] != LogTag::__NO_TAG) {
    _tags[_ntags] = tags[_ntags];
    _ntags++;
  }
```

And then later when the code copies LogTagType arrays, it hits this ubsan warning when it reads the uninitialized value.. It looks like this could be fixed by always fully initializing _tags memory.

StefanK

On 2024-09-25 13:23, Baesken, Matthias wrote:
Hi,  it has been observed (with ubsan,  and also without when adding a little check + guarantee)  that in logSelection.cpp
  we run into very high / messed up  tag values :

[JDK-8333090] ubsan: logSelection.cpp:41:49: runtime error: load of value 32597, which is not a valid value for type 'type' - Java Bug System (openjdk.org)<https://bugs.openjdk.org/browse/JDK-8333090>

When running   jtreg tests  cds/appcds/LotsOfClasses  or  cds/appcds/cacheObject/ArchivedIntegerCacheTest  we see those high tag values .
Is there someone more familiar with  UL  who has an idea  why the bad tag values occur ?

Or should / can we ignore those  because even bad/uninitialized values work “somehow”  ?

Best regards, Matthias