| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
If the CPL changes mid macroop, the end of the instruction might not be
priveleged enough to execute the beginning.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
not in the GDT.
|
|
|
|
|
|
the fake TLB miss faults.
|
|
the timing simple CPU to use it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Just use the stuff directly and things ought to be more clear
|
|
Basically, this means renaming several things called data to info, which
is information about the statistics. Things that are named data now are
actual data stored for the statistic.
|
|
|
|
|
|
This is mainly to allow the unit test to run without requiring the standard
M5 stats from being initialized (e.g. sim_seconds, sim_ticks, host_seconds)
|
|
|
|
accordingly.
|
|
comments and O3 extensions InOrder Thread Context
|
|
Also, while we're at it, remember that priorities are in the Event class
and add a disable method to disable tracing.
|
|
We previously used the actual wall time for the base timestamps,
making etherdumps non-deterministic. This fixes that problem and
gets rid of the "malformed packet" at the front that we needed to
provide the right base timestamp to wireshark/tcpdump.
|
|
been tested on Alpha, compiles for the rest but not tested. I don't see why it wouldn't work though.
|
|
Apparently we broke it with the cache rewrite and never noticed.
Thanks to Bao Yungang <baoyungang@gmail.com> for a significant part
of these changes (and for inspiring me to work on the rest).
Some other overdue cleanup on the prefetch code too.
|
|
|
|
|
|
|
