I've attached the log output. Also, I obtained this version using the scripts from https://github.com/scheib/chromium-latest-linux.

And the current build (of 62.0.3181.0, I presume) is 492966.

Deleted: chrome-crash.log 23.3 KB

chrome-crash.log 23.3 KB View Download

It appears the function that's crashing is retrieving static resources included with chrome: https://chromium.googlesource.com/chromium/src/+/5a52527e6a2e2/content/child/blink_platform_impl.cc#606

If it's randomly failing to gunzip that would imply to me that you may have some sort of hardware issue like bad ram. 
(It could also be a corrupted file in your installation, but I'd expect that to reproduce reliably, not randomly.)

Are you able to reproduce this on more than one machine?

I've been able to reproduce it on at least three different linux machines, one of which has 32GB of RAM.

It also looks like that line of code was introduced in commit 60f0d3f.

https://chromium.googlesource.com/chromium/src/+/60f0d3fdcd32ace98f0baafab27ee420637d3760

I can't figure out if that changeset has been put into the dev release channel yet, but I tried the latest dev version (62.0.3178.0), and--on the same machines--I don't have this issue. It's only with the nightly build.

Thank you for providing more feedback. Adding requester "mattm@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

That change has been in since M59: https://storage.googleapis.com/chromium-find-releases-static/60f.html#60f0d3fdcd32ace98f0baafab27ee420637d3760

I was able to repro with https://www.googleapis.com/download/storage/v1/b/chromium-browser-snapshots/o/Linux_x64%2F493142%2Fchrome-linux.zip?alt=media

It appears that the crash occurs immediately after printing [0809/150532.266271:VERBOSE1:network_delegate.cc(31)] NetworkDelegate::NotifyBeforeURLRequest: https://www.youtube.com/yts/jsbin/player-vflIVpVc9/en_US/remote.js

On the runs that don't crash, that NotifyBeforeURLRequest is not in the output. So the randomness is probably due to either some variability in what content the site serves, or perhaps headless mode decides to exit sooner sometimes than others, before all the subresources have loaded?

Indeed, adding --repl argument makes the CHECK happen reliably since it stops chrome exiting automatically.

However, I can't seem to repro on a local debug build (@493145), even when it's loading /remote.js.

I'll see if I can bisect-build it.

bisect-builds says:
You are probably looking for a change made after 492297 (known good), but no later than 492299 (first known bad).
CHANGELOG URL: https://chromium.googlesource.com/chromium/src/+log/067d2c7f8fc65efd7debca8673fc11225cce9430..ac52dfae0506a18df0c1ccaa6cda19fcc502e658

Only two changes in that range, and this one does change blink resources, so looks seems quite likely:
https://chromium.googlesource.com/chromium/src/+/68e47c3e05ed056586f167cb3d5da4ab75fd3238

I looked at the Media Controls CSS files that were changed in that commit and they are present and gzipped in the PAK file.

I could not reproduce this locally with the following GN args:

  4 is_component_build = false
  5 is_debug = false
  6 strip_absolute_paths_from_debug_symbols = true
  7 use_goma = true
  8 headless_use_embedded_resources = true

But I could download the snapshot and reproduce it.

Are the snapshots built with different flags?

We did some gzip work on the media controls and this is now fixed in latest.

Hello, we're getting this crash continuously on our non-headless chromium build. Can you please point to a commit so we can consider taking this as a pre-release patch? Thanks!

Found it, I think: https://chromium.googlesource.com/chromium/src/+/8ba04f2a94c6c70164b35cd603b286f512246caa%5E%21/content/child/blink_platform_impl.cc