This started over in JENKINS-6254, but I think it's not limited to Perforce, so I'm submitting this in Core.
Today, after upgrading to 1.354, I'm seeing cases where builds move from their job into other jobs instead. In JENKINS-6254 I note one case where two builds (412+413) from one job move to another job, and possibly cause the Perforce plugin to get very confused about where to build.
Earlier I had two jobs where 3 builds from each had switched sides and now were listed as belonging to the other job.
The switching was complete in the sense that the build directories on the Hudson server were in the wrong places. I stopped the service and moved the directories around, and they looked fine again.
One of the those jobs consistently failed like this:

FATAL: null
java.lang.NullPointerException
	at hudson.tasks.ArtifactArchiver.prebuild(ArtifactArchiver.java:147)
	at hudson.model.AbstractBuild$AbstractRunner.preBuild(AbstractBuild.java:595)
	at hudson.model.AbstractBuild$AbstractRunner.preBuild(AbstractBuild.java:590)
	at hudson.model.AbstractBuild$AbstractRunner.preBuild(AbstractBuild.java:586)
	at hudson.model.Build$RunnerImpl.doRun(Build.java:114)
	at hudson.model.AbstractBuild$AbstractRunner.run(AbstractBuild.java:416)
	at hudson.model.Run.run(Run.java:1244)
	at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:46)
	at hudson.model.ResourceController.execute(ResourceController.java:88)
	at hudson.model.Executor.run(Executor.java:122)

I'm not really keen on going through all our jobs to scan for switched builds, so I'll likely have to downgrade to 1.353.
A curious thing, though, is that all of the switched jobs (so far) took place April 7th (a week ago). We were not running 1.354 then, but I didn't see the problems that I do now, so I'm mostly inclined to believe that it's something to do with 1.354.
After downgrading, I cannot really be of much help as we don't have a "testing" server (but I'm beginning to think we should).