cpu: fix problem with forwarding and locked load

If a (regular) store is followed closely enough by a locked load that
overlaps, the LSQ will forward the store's data to the locked load and
never tell the cache about the locked load.  As a result, the cache will
not lock the address and all future store-conditional requests on that
address will fail.  This patch fixes that by preventing forwarding if
the memory request is a locked load and adding another case to the LSQ
forwarding logic that delays the locked load request if a store in the
LSQ contains all or part of the data that is requested.

[Merge second and last if blocks because their bodies are the same.]

Change-Id: I895cc2b9570035267bdf6ae3fdc8a09049969841
Reviewed-on: https://gem5-review.googlesource.com/2400
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
Maintainer: Jason Lowe-Power <jason@lowepower.com>
This commit is contained in:
Alec Roelke
2017-04-09 15:23:23 -04:00
parent 8ced1bd0b0
commit c8c406bc30

View File

@@ -671,8 +671,9 @@ LSQUnit<Impl>::read(Request *req, Request *sreqLow, Request *sreqHigh,
(req->getVaddr() + req->getSize()) >
storeQueue[store_idx].inst->effAddr;
// If the store's data has all of the data needed, we can forward.
if ((store_has_lower_limit && store_has_upper_limit)) {
// If the store's data has all of the data needed and the load isn't
// LLSC, we can forward.
if (store_has_lower_limit && store_has_upper_limit && !req->isLLSC()) {
// Get shift amount for offset into the store's data.
int shift_amt = req->getVaddr() - storeQueue[store_idx].inst->effAddr;
@@ -707,11 +708,18 @@ LSQUnit<Impl>::read(Request *req, Request *sreqLow, Request *sreqHigh,
++lsqForwLoads;
return NoFault;
} else if ((store_has_lower_limit && lower_load_has_store_part) ||
(store_has_upper_limit && upper_load_has_store_part) ||
(lower_load_has_store_part && upper_load_has_store_part)) {
} else if (
(!req->isLLSC() &&
((store_has_lower_limit && lower_load_has_store_part) ||
(store_has_upper_limit && upper_load_has_store_part) ||
(lower_load_has_store_part && upper_load_has_store_part))) ||
(req->isLLSC() &&
((store_has_lower_limit || upper_load_has_store_part) &&
(store_has_upper_limit || lower_load_has_store_part)))) {
// This is the partial store-load forwarding case where a store
// has only part of the load's data.
// has only part of the load's data and the load isn't LLSC or
// the load is LLSC and the store has all or part of the load's
// data
// If it's already been written back, then don't worry about
// stalling on it.