Seems that this change should be merged with the other one.  They seem to be interdependent.

src/cpu/base_dyn_inst.hh (Diff revision 1)

Seems that these particular functions are fine to me.  It's the ones with the switch statements that are not useful.

1. Gabe Black June 21, 2011, 12:47 p.m. (June 21, 2011, 12:47 p.m.)

   Switch statements? Which are those?
   
   Generally speaking, these changes are eliminating the read and write functions for typed memory all together and replacing them with ones that just read and write blobs of whatever size. Since these are of the former type and no longer used, they're getting nuked.

2. Nathan Binkert June 22, 2011, 2:59 a.m. (June 22, 2011, 2:59 a.m.)

   I was referring to the switch statements in the other diff, though they appear to only be for x86.

src/cpu/base_dyn_inst.hh (Diff revision 1)

Does the tracing get lost?

1. Gabe Black June 21, 2011, 12:45 p.m. (June 21, 2011, 12:45 p.m.)

   Yes. I'm going to add it back in in subsequent changes, but it's going to look different since the CPU doesn't know what the type is or how to handle endianness. It'll be displayed like a string of bytes.

2. Nathan Binkert June 22, 2011, 2:59 a.m. (June 22, 2011, 2:59 a.m.)

   The CPU doesn't know its own endianness?  That seems odd.  The ISA defines it and has to provide it to the CPU, right?  Displaying as a string of bytes that the user has to byte swap in their head would suck.