> On April 7, 2013, 9:55 p.m., Nilay Vaish wrote:
> > I am not sure if this is required. Which revision are you working with?
I only witnessed the bug that this patch fixes and thoroughly tested the patch
with revision 9478.
The assertion that was failing when I ran this was MessageBuffer.cc:212:
assert(m_clockobj_ptr->curCycle() >= msg_ptr->getLastEnqueueTime() &&
"ensure we aren't dequeued early");
I just spent some time in the repo site and found this assertion has been
updated in changeset 9603:bf5e46a02a38 (March 22), which addresses the same
checkpoint/restore bug that this patch does. My apologies for not catching that
here (this patch has been in our queue since March 9). I will drop this
request.
- Joel
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviews.gem5.org/r/1817/#review4205
-----------------------------------------------------------
On April 7, 2013, 8:58 p.m., Joel Hestness wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviews.gem5.org/r/1817/
> -----------------------------------------------------------
>
> (Updated April 7, 2013, 8:58 p.m.)
>
>
> Review request for Default.
>
>
> Description
> -------
>
> Changeset 9634:4a87fbab0988
> ---------------------------
> Sequencers: Reset clocks after cache warmup/cooldown
>
> Sequencers can be accessed by CPU cores after cache warm-up or cool-down but
> before their clocks have been reset in normal simulation, which can cause
> MessageBuffer assertion failures that the enqueue time of a message is later
> than the send time. This change ensures that the Sequencer clocks are reset
> after warm-up and cool-down to avoid this problem.
>
>
> Diffs
> -----
>
> src/mem/ruby/recorder/CacheRecorder.hh fa31189e1fb5
> src/mem/ruby/recorder/CacheRecorder.cc fa31189e1fb5
> src/mem/ruby/system/Sequencer.hh fa31189e1fb5
> src/mem/ruby/system/System.cc fa31189e1fb5
>
> Diff: http://reviews.gem5.org/r/1817/diff/
>
>
> Testing
> -------
>
> Numerous checkpoint restore tests with x86 + Ruby
>
>
> Thanks,
>
> Joel Hestness
>
>
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
