Hi all,
I am using m5 to implement a DataFlow-like architecture.
I am having a weird Segmentation Fault when I insert an instruction in a
wire. I can send a bunch of instructions before this one with no problem,
and after a little bit I get a Segmentation Fault.
This is part of the code that I am using, and the function that generates
the error (RefCounted::decref()).
In my design I store a bunch of instructions (usually a Basic Block) in a
wire and move them to the next pipeline stage. I advance the Queue when the
next stage receives and stores the instructions in its own local buffer, is
there anything else I need to do?
Thanks!
-Andrea
Code that causes Segmentation Fault (fetch_impl.hh):
//tg_units[index_tg_unit]->getFetchQueue()->insts[toDecode->size++] =
instruction;
//toDecode->insts[toDecode->size++] = instruction;
typename TimeBuffer<FetchStruct>::wire toTGUnitDecode =
tg_units[index_tg_unit]->getOutputWire();
toTGUnitDecode->insts[toTGUnitDecode->size++] = instruction; // <- This
function breaks
-----------------------------------------------------------
Output:
...
FullO3CPU: Ticking main, FullO3CPU.
301000: system.cpu.fetch: Running stage.
301000: system.cpu.fetch: Attempting to fetch from [tid:0]
301000: system.cpu.fetch: [tid:0]: Adding instructions to queue to decode.
301000: system.cpu.fetch: [tid:0]: Instruction PC 0x409a73 (0) created
[sn:95].
301000: system.cpu.fetch: [tid:0]: Instruction is: ADD_R_I : limm t1,
0x8
301000: system.cpu.fetch: [tid:0]: Instruction PC 0x409a73 (1) created
[sn:96].
301000: system.cpu.fetch: [tid:0]: Instruction is: ADD_R_I : add rdi,
rdi, t1
Segmentation fault
-----------------------------------------------------------
Function that breaks:
28 RefCounted::decref() refcnt.hh:51 0x000000000040a8cc
Code:
void decref() { if (--count <= 0) delete this; }
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
