concurrent rr record sessions on the same binary
wadehennessey at gmail.com
Mon Nov 16 07:23:22 UTC 2015
I guess you are saying Mozilla doesn't have rr record failure test farms
running. With the record -c option your scheduler should have less
nondeterminism. . I've run -c 2 to try to reproduce. problems. This finds
my problem , but takes over a day to reproduce. That's too long for
practical debugging since the replay takes a few days, and so far I haven't
gotten checkpoints to work.
Sorry for all this trouble. It's clear that Firefox doesn't have the level
concurrency that I'm trying to debug, and that's your main goal. I still
love the idea. It just may not work for my situation at the moment. -wade
On Sun, Nov 15, 2015 at 10:19 PM, Robert O'Callahan <robert at ocallahan.org>
> On Mon, Nov 16, 2015 at 4:58 PM, Wade Hennessey <wadehennessey at gmail.com>
>> There is one other approach I've been thinking of, but I don't see any
>> mention of it. Has Mozilla come up with a framework for automating rr
>> record sessions with variations in record options and allowed cpu run time?
>> The idea is to run lots of concurrent record sessions, possibly over an
>> indefinite period just trying to find a lucky run that exposes a problem.
>> It doesn't really matter what the problem is, and whole server farms of
>> machines could possibly just run rr record sessions searching for test runs
>> that rarely fail. Maybe Mozilla already does this. It doesn't seem like a
>> hard problem to address, it's just pretty brute force.-wade
> Doing more test runs should always help, if the scheduler admits a
> sufficient amount of nondeterminism. But we're not nearly there yet.
> lbir ye,ea yer.tnietoehr rdn rdsme,anea lurpr edna e hnysnenh hhe uresyf
> selthor stor edna siewaoeodm or v sstvr esBa kbvted,t
> o l euetiuruewFa kbn e hnystoivateweh uresyf tulsa rehr rdm or rnea
> .a war hsrer holsa rodvted,t nenh hneireseoouot.tniesiewaoeivatewt sstvr
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rr-dev