| Age | Commit message (Collapse) | Author |
|
|
|
properly yet but run_power.native is connected to it
|
|
|
|
testing register tracking/tainting
|
|
instead of internal actions
|
|
There are problems that warrant discussion about handling special registers that hold records or data structures previously.
|
|
then hits an error. So still in progress
Note: moved barrier kind changes to interp_interface, along with other memory kinds
|
|
Also allows register writing functions to be on the left hand side of an assignment in the same way.
The last parameter to a writing function is the value to be written, and should appear on the right hand side of an assignment expression.
|
|
constraint solving (instead of hardcoding 64 as the default).
|
|
This is now possible because we handle coercions better.
|
|
- remember mode (run, step or next) between instructions
- display continuation by default in step mode
- start in step mode by default
- incompatible change: the shorthand for stack is now bt (=backtrace),
since s becomes the shorthand for step
- incompatible change: pressing enter now repeats the current mode,
instead of "step"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
- Move FDE loop to the OCaml side of the Power model (avoid leaking
memory due to lack of TCO in interpreter)
- Display cycle count
- Check the value of CIA at the end of each cycle and stop if it is
equal to the initial value of LR, returning the value in GPR3.
|
|
In fact, increasing order is big-endian, not little-endian.
This is good news for us, because it means we do not need any
switch for the POWER spec.
|
|
Many limitations and bugs currently, but loads binary and decodes
opcodes correctly (using endianness hacks). Disabled in the default
test suite (but still compiled), run "make test_power" to try it.
|
|
Roughly 110KB/s (~5 seconds to load "hello" test).
|
|
|