Do we have a plan to deal with NOTREACHED on the platform side?

[Ian Barkley-Yeung]

My understanding is that the Chromium side is going to eventually make NOTREACHED fatal with the following steps:

1. Make NOTREACHED do dump-without-crashing
2. Clean up the major dump-without-crashing reports
3. Do a finch experiment kNotReachedIsFatal to see if NOTREACHED is causing any further problems
4. Change NOTREACHED into a synonym for CHECK(false)

This is great for the Chromium side, but what is our plan on the ChromiumOS Platform side? 

1. As far as I understand, the DumpWithoutCrashing() in src/platform/libchrome/base/debug/dump_without_crashing.h is a no-op in Platform code because we don't link in crashpad and never call SetDumpWithoutCrashingFunction(). (The libchrome_internal::DumpWithoutCrashing() function in src/platform/libchrome/dump_without_crashing.h might work but that's not the version that's being called by NOTREACHED)
2. The code that checks the experiment status will always consider the kNotReachedIsFatal experiment to be off because base::FeatureList::GetInstance() will always be nullptr in Platform code.

So it sounds like we will get no warning that a NOTREACHED is being hit until step 4, at which point there may be a whole bunch of things crashing. 

Is this already something that people have considered? 

[Hidehiko Abe]

Looping in the libchrome team.

2024年4月16日(火) 13:16 Ian Barkley-Yeung <i...@chromium.org>:

--
--
Chromium OS Developers mailing list: chromiu...@chromium.org
View archives, change email options, or unsubscribe:
https://groups.google.com/a/chromium.org/group/chromium-os-dev