bcwhite@: the CL that added these histograms was landed 3 months ago, which reaches the stable channel now. Could it be possible that there are some changes in the histogram infra side that is causing this crash? I just feel that if adding that histogram is the root cause, it would have a lot of crashes since M62, not starting at M64.

Hey, sorry.  I didn't see your reply back in November.

There have been no changes in the histogram infrastructure that would cause such a crash.  It's an indication of the same histogram name being defined in multiple ways.

While I haven't investigated this specific one, others were due to histograms being created one way, serialized, and deserialized only to conflict with something defined locally.

The query linked indicates I was only looking in M64 and above which was perhaps a mistake on my part.

Unfortunately, there's no differentiation in the crash reports by histogram which means looking through them by hand to find matches of a particular histogram.

Here's an updated search query:
https://crash.corp.google.com/browse?q=product.Version%20LIKE%20%27__.%25.%25.%25%27%20AND%20product.name%20LIKE%20%27%25Chrome%25%27%20AND%20expanded_custom_data.ChromeCrashProto.magic_signature_1.component%3D%27src%2Fbase%2Fmetrics%27%20AND%20expanded_custom_data.ChromeCrashProto.magic_signature_1.name%3D%27base%3A%3AHistogram%3A%3AFactory%3A%3ABuild%27#-property-selector,-processtype,+magicsignature:50,-magicsignature2:50,-stablesignature:50,-clientid,-magicsignaturesorted:50

I hope to be able to eliminate these crashes in the near future and also more reliably show what histograms are misbehaving.  But for the moment, I can't find any particular instances of this conflict.  Your choice if you wish to investigate or not.