Brave-browser: Unable to open Brave, crashing on start-up

[Nuc1eoN]

Brave suddenly stopped working on my system. Every time I launch it, the window opens but is frozen and crashes after a few seconds.

I am not yet sure, but it appears this happened after updating Mesa v19 to v20.

Console output:

[2920:2920:0330/141930.824149:ERROR:gl_implementation.cc(277)] Failed to load /usr/lib/brave-bin/swiftshader/libGLESv2.so: /usr/lib/brave-bin/swiftshader/libGLESv2.so: cannot open shared object file: No such file or directory
[2920:2920:0330/141930.825051:ERROR:viz_main_impl.cc(161)] Exiting GPU process due to errors during initialization
[2962:2962:0330/141932.364283:ERROR:gpu_channel_manager.cc(459)] ContextResult::kFatalFailure: Failed to create shared context for virtualization.
[2898:3003:0330/141932.923474:ERROR:dat_file_util.cc(22)] GetDATFileData: the dat file is not found or corrupted /home/nuc/.config/BraveSoftware/Brave-Browser/jmomcjcilfpbaaklkifaijjcnancamde/1.0.529/rs-E71426E7-E898-401C-A195-177945415F38.dat
[2898:2898:0330/141932.943153:ERROR:ad_block_base_service.cc(225)] Could not obtain ad block data
[2898:3121:0330/141934.088623:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: PRAGMA auto_vacuum
[2898:3121:0330/141934.088690:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: PRAGMA journal_mode=TRUNCATE
[2898:3121:0330/141934.088770:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: SELECT 1 FROM sqlite_master WHERE type=? AND name=?
[2898:3121:0330/141934.088886:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: SELECT 1 FROM sqlite_master WHERE type=? AND name=?
[2898:3121:0330/141934.088936:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: SELECT 1 FROM sqlite_master WHERE type=? AND name=?
[2898:3121:0330/141934.088977:ERROR:database.cc(1584)] publisher_info_db sqlite error 5, errno 0: database is locked, sql: CREATE TABLE activity_info (publisher_id LONGVARCHAR NOT NULL,duration INTEGER DEFAULT 0 NOT NULL,score DOUBLE DEFAULT 0 NOT NULL,percent INTEGER DEFAULT 0 NOT NULL,weight DOUBLE DEFAULT 0 NOT NULL,category INTEGER NOT NULL,month INTEGER NOT NULL,year INTEGER NOT NULL,CONSTRAINT fk_activity_info_publisher_id    FOREIGN KEY (publisher_id)    REFERENCES publisher_info (publisher_id)    ON DELETE CASCADE)
[2898:3121:0330/141934.089002:ERROR:publisher_info_database.cc(1036)] DB: Error with MigrateV0toV1
/usr/bin/brave: line 27:  2898 Segmentation fault      (core dumped) /usr/lib/brave-bin/brave "$@" $SANDBOX_FLAG $PEPPER_FLASH_FLAG $USER_FLAGS

Archlinux
Brave 1.5.115
AMD RX470, open radeon drivers

[Nuc1eoN]

After testing, it appears downgrading mesa does not help. So I don't know which change caused the bug to appear.

However right now when testing, Brave browser stopped segfaulting. However animations are very stuttery and performance is bad. Also Brave leaves a trailing process in the background when closes it's window over the "x" button.

[bsclifton]

@Nuc1eoN I suspect you're having an out of control Bat Ledger Service process. Can you please try to go to Hamburger menu (top right, 3 lines) => More Tools => Task Manager?

Once there, you can see what is eating up the CPU. My guess is that it's Bat Ledger Service - which we have tracked with https://github.com/brave/brave-browser/issues/8846

[Nuc1eoN]

@bsclifton No, it does not appear to be the case for my bug:

[Nuc1eoN]

Wait a second, it is currently working as normal again. The stuttering stopped and it closes in a clean manner (no trailing process). Also instead of ~14% CPU (see screenshot above) it is at 1-2% now.

I have no idea what fixed it. But it's kinda scary.

EDIT: Gonna keep this open for some time, until I can verify it went away.

[bbugh]

There's another thread about crashing on start, #7914. Maybe there's an answer for you there?

[bsclifton]

@Nuc1eoN how are things looking now? Any better?

[Nuc1eoN]

@bsclifton Yeah the issue is not there anymore 🤔 I wonder what caused it though.

Feel free to close.

[bsclifton]

@iefremov will be looking into crash reports soon - we've had a bunch of similar reports on macOS too. Hoping we can find it soon and have a fix 😄