Hi Meng,
There are some more exotic instructions that we don't
print the correct dissambly for or get the data a bit wrong. The way the
data works is pretty much the last register that is written to becomes
the final_value of the instruction and that is what is printed.
You
can see some beautiful (j/k) generated code that looks something like
this:
uint32_t final_val = Dest;
((dest == PCReg) ? setNextPC(xc,
Dest) :
xc->setIntRegOperand(this, 0, Dest))
;
if (traceData) {
traceData->setData(final_val); }
}
{
uint32_t final_val = Dest2;
((dest2 == PCReg) ? setNextPC(xc, Dest2) :
xc->setIntRegOperand(this,
1, Dest2))
;
if (traceData) { traceData->setData(final_val); }
In
the particular case I picked the data printed is always what was in
Dest2.
Ali
On 26.06.2013 13:37, Meng Wang wrote:
> Hi all,
> "ldrd
Rm, Rn, <addr>" is an ARM instruction for loading two continuous words
at <addr> into two registers Rm, Rn. It can execute correctly but its
dumping information seems wrong. I wrote an simple assembly code to
test:
>
> .section .text, "ax"
> global _start
> _start:
> mov r0,
#0xff // r0 = 0xff
> mov r1, #0x11 // r1 = 0x11
> str r0, [sp], #4 //
store 0x11 to <addr>, sp +=4
> str r1, [sp], #4 // store 0x11 to
<addr+4>, sp +=4
> ldrD r2, r3, [sp, #-8] // r2 = value in <addr>, r3 =
value in <addr+4>
> mov r4, r2 // r4 = r2
> mov r5, r3 // r5 = r3
> #
following is exiting.
> exit:
> mov r0, #0
> mov r7, #1
> svc 0
>
.end
>
> Run it on SE mode and dump it with Exec flag:
>
> 1. 0x8054 :
mov r0, #255 : IntAlu : D=0x00000000000000ff
> 2. 0x8058 : mov r1, #17 :
IntAlu : D=0x0000000000000011
> 3. 0x805c.0 : str r0, [sp] #4 : MemWrite
: D=0x00000000000000ff A=0xbeffff20
> 4. 0x805c.1 : addi_uop sp, sp, #4
: IntAlu : D=0x00000000beffff24
> 5. 0x8060.0 : str r1, [sp] #4 :
MemWrite : D=0x0000000000000011 A=0xbeffff24
> 6. 0x8060.1 : addi_uop
sp, sp, #4 : IntAlu : D=0x00000000beffff28
> 7. 0x8064 : ldrd r2, r3,
[sp, #-8] : MemRead : D=0x0000000000000011 A=0xbeffff20
> 8. 0x8068 :
mov r4, r2 : IntAlu : D=0x00000000000000ff
> 9. 0x806c : mov r5, r3 :
IntAlu : D=0x0000000000000011
> 10. 0x8070 : mov r0, #0 : IntAlu :
D=0x0000000000000000
> 11. 0x8074 : mov r7, #1 : IntAlu :
D=0x0000000000000001
> 12. 0x8078 : svc : IntAlu :
>
> In line 7 of
dumped trace, should trace value "D=0x0000000000000011" be
"D=0x00000011000000ff"? (Assuming the configuration is little endian).
>
> Thanks,
> Meng
> _______________________________________________
>
gem5-dev mailing list
> [email protected]
>
http://m5sim.org/mailman/listinfo/gem5-dev [1]
Links:
------
[1]
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
