summaryrefslogtreecommitdiff
path: root/src/dev
diff options
context:
space:
mode:
authorAli Saidi <Ali.Saidi@ARM.com>2011-03-17 19:20:19 -0500
committerAli Saidi <Ali.Saidi@ARM.com>2011-03-17 19:20:19 -0500
commita432d8e0851de8d090676697e29ca6ed4be64fb7 (patch)
tree50546c6a98e9d766ad53aa05eb75381ce6d7f298 /src/dev
parent2f40b3b8ae4fddcdd167fc86469254f40736c888 (diff)
downloadgem5-a432d8e0851de8d090676697e29ca6ed4be64fb7.tar.xz
Mem: Fix issue with dirty block being lost when entire block transferred to non-cache.
This change fixes the problem for all the cases we actively use. If you want to try more creative I/O device attachments (E.g. sharing an L2), this won't work. You would need another level of caching between the I/O device and the cache (which you actually need anyway with our current code to make sure writes propagate). This is required so that you can mark the cache in between as top level and it won't try to send ownership of a block to the I/O device. Asserts have been added that should catch any issues.
Diffstat (limited to 'src/dev')
-rw-r--r--src/dev/io_device.cc3
1 files changed, 3 insertions, 0 deletions
diff --git a/src/dev/io_device.cc b/src/dev/io_device.cc
index be97bc4ad..ffe8fdf06 100644
--- a/src/dev/io_device.cc
+++ b/src/dev/io_device.cc
@@ -139,6 +139,9 @@ DmaPort::recvTiming(PacketPtr pkt)
assert(pendingCount >= 0);
assert(state);
+ // We shouldn't ever get a block in ownership state
+ assert(!(pkt->memInhibitAsserted() && !pkt->sharedAsserted()));
+
state->numBytes += pkt->req->getSize();
assert(state->totBytes >= state->numBytes);
if (state->totBytes == state->numBytes) {