Found the problem. This happens because the NSGigE Ethernet controller's
packet filter relies on the overloaded == operator for comparing two
EthAddr objects. This operator returns a memcmp() of the left and right
hand sides, but memcmp returns 0 if each byte of the blocks being compared
are equal. So packets kept getting incorrectly dropped exactly during the
opposite case of when they should.

Because data wasn't being completely set before I guess we were getting
lucky and the bytes had garbage that didn't match up, which would return a
non-zero value and caused the comparison to return the correct bool by
sheer luck.

returning !memcmp() allows that test to pass now. I'm running the complete
regressions to make sure everything still works. I was sure I had run the
regressions before, but I guess not.

Anthony Gutierrez
http://web.eecs.umich.edu/~atgutier


On Mon, Jul 7, 2014 at 9:25 AM, Anthony Gutierrez <atgut...@umich.edu>
wrote:

> I killed the jobs. I'll look into this.
>
>
> Anthony Gutierrez
> http://web.eecs.umich.edu/~atgutier
>
>
> On Mon, Jul 7, 2014 at 8:30 AM, Andreas Hansson via gem5-dev <
> gem5-dev@gem5.org> wrote:
>
>> Hi all,
>>
>> It seems that the regression
>> build/ALPHA/tests/opt/quick/fs/80.netperf-stream/alpha/linux/twosys-tsunami-simple-atomic
>> livelocks with the latest changeset:
>>
>> changeset:   10251:878f2f30b12d
>> tag:         tip
>> user:        Anthony Gutierrez <atgut...@umich.edu>
>> date:        Wed Jul 02 13:19:13 2014 -0400
>> summary:     base: fix some bugs in EthAddr
>>
>> Tony, could you have a look and see what is going wrong?
>>
>> This also explains why there have been no regression e-mails the last few
>> days. Perhaps someone could kill the runs?
>>
>> Thanks,
>>
>> Andreas
>>
>>
>> -- IMPORTANT NOTICE: The contents of this email and any attachments are
>> confidential and may also be privileged. If you are not the intended
>> recipient, please notify the sender immediately and do not disclose the
>> contents to any other person, use it for any purpose, or store or copy the
>> information in any medium. Thank you.
>>
>> ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ,
>> Registered in England & Wales, Company No: 2557590
>> ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ,
>> Registered in England & Wales, Company No: 2548782
>> _______________________________________________
>> gem5-dev mailing list
>> gem5-dev@gem5.org
>> http://m5sim.org/mailman/listinfo/gem5-dev
>>
>
>
_______________________________________________
gem5-dev mailing list
gem5-dev@gem5.org
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to