conversion of and operation having a function from c# to tcl - tcl

Question

how to do the and operation given as one line statement in tcl in tcl where pcieDeviceControlRegister is a function given as in the code:
code:
pcieDeviceControlRegister = cfgSpace.pcieDeviceControlRegister & (~((uint)0xF));
Reference for pcieDeviceControlRegister function is :
public uint pcieDeviceControlRegister
{
get
{
if (pcieCapabilityOffset != 0)
return (ReadDW((int)(pcieCapabilityOffset + 8) / 4, 0xF)) & 0xFFFF;
else
return 0;
}
set
{
if (pcieCapabilityOffset != 0)
{
uint val = ReadDW((int)(pcieCapabilityOffset + 8) / 4, 0xF)& 0xFFFF0000;
val |= value;
// write should be done with byte enables !!!
WriteDW((int)(pcieCapabilityOffset + 8) / 4, val, 0xF);
}
}
}

Answer 1

You'll have to arrange for the mapping of ReadDW and WriteDW into Tcl, probably by writing a little C or C++ code that makes commands (with the same names) that do those operations. I'm assuming that you've already done that. (SWIG can generate the glue code if you need it.)
Then, we define a command like this:
proc pcieDeviceControlRegister {{newValue ""}} {
global pcieCapabilityOffset
# Filter the bogus setup case early; if this is really an error case though,
# it is better to actually throw an error instead of struggling on badly.
if {$pcieCapabilityOffset == 0} {
return 0
# error "PCIE capability offset is zero"
}
set offset [expr {($pcieCapabilityOffset + 8) / 4}]
if {$newValue eq ""} {
# This is a read operation
return [expr {[ReadDW $offset 0xF] & 0xFFFF}]
} else {
# This is a write operation
set val [expr {[ReadDW $offset 0xF] & 0xFFFF0000}]
# Note that we do the bit filtering HERE
set val [expr {$val | ($newValue & 0xFFFF)}]
WriteDW $offset $val 0xF
return
}
}
With that, which you should be able to see is a pretty simple translation of the C# property code (with a bit of minor refactoring), you can then write your calling code like this:
pcieDeviceControlRegister [expr {[pcieDeviceControlRegister] & ~0xF}]
With Tcl, you don't write casts to different types of integers: Tcl just has numbers (which are theoretically of infinite width) so instead you need to do a few more bit masks in key places.
The conversion of the above code to a method on an object is left as an exercise. It doesn't change very much…