Make instruction counter in simulators 64-bit (issue 22302).

Make instruction counter in simulators 64-bit (issue 22302).
This requires supporting uint64_t flag values.

R=zra@google.com

Committed: https://code.google.com/p/dart/source/detail?r=43660

[regis, 2015-02-07 00:08:42]

regis@google.com changed reviewers:
+ johnmccutchan@google.com, zra@google.com

[regis, 2015-02-07 00:08:42]

One potential issue: is ULLONG_MAX defined on all platforms (Mac, Windows, MIPS,
ARM, ...)? The buildbots will tell us.

John, for now, I skip the uint64_t flags when writing to JSON. I hardly imagine
requiring the observatory when stopping at instructions in the simulator. But
this new flag type could be used for other things in the future. I am not sure
where to add support to read the value back.

Thanks,
Regis

[Cutch, 2015-02-09 04:08:41]

On 2015/02/07 00:08:42, regis wrote:
> One potential issue: is ULLONG_MAX defined on all platforms (Mac, Windows,
MIPS,
> ARM, ...)? The buildbots will tell us.
> 
> John, for now, I skip the uint64_t flags when writing to JSON. I hardly
imagine
> requiring the observatory when stopping at instructions in the simulator. But
> this new flag type could be used for other things in the future. I am not sure
> where to add support to read the value back.
> 
> Thanks,
> Regis

Regis, please just print the value as a uint64_t. Depending on the client, we
can load it on the other end.

[Florian Schneider, 2015-02-09 11:21:16]

fschneider@google.com changed reviewers:
+ fschneider@google.com

[Florian Schneider, 2015-02-09 11:21:17]

dbc: would it be enough to just use int64_t internally in the simulators, but
leave the flag as is? Just to avoid the wrapping around after reaching the
32-bit limit.

[regis, 2015-02-09 21:05:17]

On 2015/02/09 04:08:41, Cutch wrote:
> On 2015/02/07 00:08:42, regis wrote:
> > One potential issue: is ULLONG_MAX defined on all platforms (Mac, Windows,
> MIPS,
> > ARM, ...)? The buildbots will tell us.
> > 
> > John, for now, I skip the uint64_t flags when writing to JSON. I hardly
> imagine
> > requiring the observatory when stopping at instructions in the simulator.
But
> > this new flag type could be used for other things in the future. I am not
sure
> > where to add support to read the value back.
> > 
> > Thanks,
> > Regis
> 
> Regis, please just print the value as a uint64_t. Depending on the client, we
> can load it on the other end.

Done. Thanks.

[regis, 2015-02-09 21:07:02]

On 2015/02/09 11:21:17, Florian Schneider wrote:
> dbc: would it be enough to just use int64_t internally in the simulators, but
> leave the flag as is? Just to avoid the wrapping around after reaching the
> 32-bit limit.

Yes, a 32-bit flag value could be copied to a 64-bit counter as we start the
simulator. But it is useful to be able to provide a 64-bit address as the flag
value.

[regis, 2015-02-09 21:07:14]

PTAL

[zra, 2015-02-10 16:56:06]

lgtm

[regis, 2015-02-10 17:40:15]

New patchsets have been uploaded after l-g-t-m from zra@google.com

[regis, 2015-02-10 17:40:20]

Thanks!

I'll keep an eye on the buildbots to see if ULLONG_MAX is declared on all
platforms.

[regis, 2015-02-10 17:41:18]

Committed patchset #3 (id:40001) manually as r43660 (presubmit successful).