Commit graph

8729 commits

Author SHA1 Message Date
Nilay Vaish
f19b3f30b4 X86 Regressions: Update stats due to introduction of TSO 2012-01-28 19:09:17 -06:00
Nilay Vaish
5c2fc35e02 O3 CPU LSQ: Implement TSO
This patch makes O3's LSQ maintain total order between stores. Essentially
only the store at the head of the store buffer is allowed to be in flight.
Only after that store completes, the next store is issued to the memory
system. By default, the x86 architecture will have TSO.
2012-01-28 19:09:04 -06:00
Andreas Hansson
4acca8a053 ns_gige: Fix a missing curly brace in if-statement
This patch adds a missing curly brace when clearing and setting the
appropriate bits in the ns_gige.cc code.

This commit is not based on any runtime bug experienced, but rather
inspection of the code.
2012-01-27 12:54:11 -05:00
Ronald Dreslinski
38ee552798 configs: actually add ARMv7a-like cpu/cache file 2012-01-26 16:44:43 -05:00
Ronald Dreslinski
fc7cf40de6 configs: A more realistic configuration of an ARM-like processor 2012-01-26 14:53:48 -05:00
Andreas Hansson
53c130bf2f MEM: Fix fs.py by specifying the range size rather than end
This patch fixes the currently broken fs.py by specifying the size of
the bridge range rather than the end address. This effectively
subtracts one when determining the address range for the IO bridge
(from IO bus to membus), and thus avoids the overlapping ranges.
2012-01-25 12:44:43 -05:00
Mitchell Hayenga
698408bce2 Fix memory corruption issue with CopyStringOut()
CopyStringOut() improperly indexed setting the null
character, would result in zeroing a random byte
of memory after(out of bounds) the character array.
2012-01-12 15:27:20 -06:00
Ali Saidi
a17dbdf883 stats: Update stats for final tick and memory bandwidth patches 2012-01-25 17:19:50 +00:00
Ali Saidi
bd55c9e2af sim: display final value of curTick in stats
Different from sim_ticks in that this value is restored from checkpoints and is never reset.
Useful for aligning with framebuffer output ticks
2012-01-25 17:18:25 +00:00
Ali Saidi
e1c48dfce5 Mem: Add simple bandwidth stats to PhysicalMemory 2012-01-25 17:18:25 +00:00
Nilay Vaish
24c2300998 Config: Enable using O3 CPU and Ruby in SE mode 2012-01-23 11:33:52 -06:00
Nilay Vaish
63563c9df2 O3, Ruby: Forward invalidations from Ruby to O3 CPU
This patch implements the functionality for forwarding invalidations and
replacements from the L1 cache of the Ruby memory system to the O3 CPU. The
implementation adds a list of ports to RubyPort. Whenever a replacement or an
invalidation is performed, the L1 cache forwards this to all the ports, which
is the LSQ in case of the O3 CPU.
2012-01-23 11:07:14 -06:00
Nilay Vaish
9481d05b8a MemCmd: Add a command for invalidation requests to LSQ
This command will be sent from the memory system (Ruby) to the LSQ of
an O3 CPU so that the LSQ, if it needs to, invalidates the address in
the request packet.
2012-01-23 11:07:11 -06:00
Andreas Hansson
acd289b7ef MEM: Make the bus default port yet another port
This patch removes the idiosyncratic nature of the default bus port
and makes it yet another port in the list of interfaces. Rather than
having a specific pointer to the default port we merely track the
identifier of this port. This change makes future port diversification
easier and overall cleans up the bus code.
2012-01-17 12:55:09 -06:00
Andreas Hansson
55cf3f4ac1 MEM: Removing the default port peer from Python ports
In preparation for the introduction of Master and Slave ports, this
patch removes the default port parameter in the Python port and thus
forces the argument list of the Port to contain only the
description. The drawback at this point is that the config port and
dma port of PCI and DMA devices have to be connected explicitly. This
is key for future diversification as the pio and config port are
slaves, but the dma port is a master.
2012-01-17 12:55:09 -06:00
Andreas Hansson
2208ea049f MEM: Make the bus bridge unidirectional and fixed address range
This patch makes the bus bridge uni-directional and specialises the
bus ports to be a master port and a slave port. This greatly
simplifies the assumptions on both sides as either port only has to
deal with requests or responses. The following patches introduce the
notion of master and slave ports, and would not be possible without
this split of responsibilities.

In making the bridge unidirectional, the address range mechanism of
the bridge is also changed. For the cases where communication is
taking place both ways, an additional bridge is needed. This causes
issues with the existing mechanism, as the busses cannot determine
when to stop iterating the address updates from the two bridges. To
avoid this issue, and also greatly simplify the specification, the
bridge now has a fixed set of address ranges, specified at creation
time.
2012-01-17 12:55:09 -06:00
William Wang
e731cf4c1d MEM: Remove the functional ports from the memory system
The functional ports are no longer used and this patch cleans up the
legacy that is still present in buses, memories, CPUs etc. Note that
this does not refer to the class FunctionalPort (already removed), but
rather ports with the name (and use) functional.
2012-01-17 12:55:09 -06:00
Andreas Hansson
07cf9d914b MEM: Separate queries for snooping and address ranges
This patch simplifies the address-range determination mechanism and
also unifies the naming across ports and devices. It further splits
the queries for determining if a port is snooping and what address
ranges it responds to (aiming towards a separation of
cache-maintenance ports and pure memory-mapped ports). Default
behaviours are such that most ports do not have to define isSnooping,
and master ports need not implement getAddrRanges.
2012-01-17 12:55:09 -06:00
Andreas Hansson
142380a373 MEM: Remove Port removeConn and MemObject deletePortRefs
Cleaning up and simplifying the ports and going towards a more strict
elaboration-time creation and binding of the ports.
2012-01-17 12:55:09 -06:00
Andreas Hansson
6315e5bbb5 MEM: Remove the notion of the default port
This patch removes the default port and instead relies on the peer
being set to NULL initially. The binding check (i.e. is a port
connected or not) will eventually be moved to the init function of the
modules.
2012-01-17 12:55:09 -06:00
Andreas Hansson
de34e49d15 MEM: Simplify ports by removing EventManager
This patch removes the inheritance of EventManager from the ports and
moves all responsibility for event queues to the owner. Eventually the
event manager should be the interface block, which could either be the
structural owner or a subblock like a LSQ in the O3 CPU for example.
2012-01-17 12:55:09 -06:00
Andreas Hansson
b3f930c884 CPU: Moving towards a more general port across CPU models
This patch performs minimal changes to move the instruction and data
ports from specialised subclasses to the base CPU (to the largest
degree possible). Ultimately it servers to make the CPU(s) have a
well-defined interface to the memory sub-system.
2012-01-17 12:55:08 -06:00
Andreas Hansson
f85286b3de MEM: Add port proxies instead of non-structural ports
Port proxies are used to replace non-structural ports, and thus enable
all ports in the system to correspond to a structural entity. This has
the advantage of accessing memory through the normal memory subsystem
and thus allowing any constellation of distributed memories, address
maps, etc. Most accesses are done through the "system port" that is
used for loading binaries, debugging etc. For the entities that belong
to the CPU, e.g. threads and thread contexts, they wrap the CPU data
port in a port proxy.

The following replacements are made:
FunctionalPort      > PortProxy
TranslatingPort     > SETranslatingPortProxy
VirtualPort         > FSTranslatingPortProxy

--HG--
rename : src/mem/vport.cc => src/mem/fs_translating_port_proxy.cc
rename : src/mem/vport.hh => src/mem/fs_translating_port_proxy.hh
rename : src/mem/translating_port.cc => src/mem/se_translating_port_proxy.cc
rename : src/mem/translating_port.hh => src/mem/se_translating_port_proxy.hh
2012-01-17 12:55:08 -06:00
Andreas Hansson
06c39a154c Ruby: Change the access permissions for MOESI hammer
Regression statistics update.
2012-01-17 12:55:07 -06:00
Andreas Hansson
43a45edcf0 Ruby: Change the access permissions for MOESI hammer
This patch changes the access permission for the WB_E_W state from
Busy to Read_Write to avoid having issues in follow-on patches with
functional accesses going through Ruby. This change was made after
consultation with all involved parties and is more of a work-around
than a fix.
2012-01-17 12:55:07 -06:00
Andreas Hansson
41af57f9fb MEM: Add the system port as a central access point
The system port is used as a globally reachable access point to the
memory subsystem. The benefit of using an actual port is that the
usual infrastructure is used to resolve any access and thus makes the
overall system able to handle distributed memories in any
configuration, and also makes the accesses agnostic to the address
map. This patch only introduces the port and does not actually use it
for anything.
2012-01-17 12:55:07 -06:00
Andreas Hansson
13ef7a5647 MEM: Differentiate functional cache accesses from CPU and memory
This patch changes the functionalAccess member function in the cache
model such that it is aware of what port the access came from, i.e. if
it came from the CPU side or from the memory side. By adding this
information, it is possible to respect the 'forwardSnoops' flag for
snooping requests coming from the memory side and not forward
them. This fixes an outstanding issue with the IO bus getting accesses
that have no valid destination port and also cleans up future changes
to the bus model.
2012-01-17 12:55:07 -06:00
Ali Saidi
8c7936c40c stats: undo parser change from initparam change 2012-01-16 22:37:05 -05:00
Steve Reinhardt
7a3a37307a Alpha: warn_once about broken PAL breakpoints.
A recent changeset (aae12ce9f34c) removed support for
PAL-mode breakpoints in Alpha, since it was awkward
and likely unused.  This patch lets a user know if they
potentially run into this limitation.
2012-01-16 19:01:27 -08:00
Steve Reinhardt
1585cfb5b5 debug: fix AllFlags::disable()
Looks like copy-and-paste bug, apparently I'm the first
person to ever use this since it's plainly broken.
2012-01-16 19:00:59 -08:00
Maximilien Breughe
a7394ad680 inorder: MDU deadlock fix 2012-01-12 10:15:00 -05:00
Deyuan Guo
4a59cf00b4 mips: compatibility between MIPS_SE and cross compiler from CodeSorcery 2012-01-12 09:59:01 -05:00
Deyuan Guo
31b6941a52 mips: Fix bugs in faults.cc/hh and tlb.cc for MIPS_FS 2012-01-12 09:59:00 -05:00
Deyuan Guo
a40ec5671f mips: Fix decoder of two float-convert instructions 2012-01-12 09:58:59 -05:00
Deyuan Guo
7f782a6c79 mips: definition of MIPS64_QNAN in registers.hh 2012-01-12 09:58:58 -05:00
Nilay Vaish
0e6d6a5e25 PerfectCacheMemory: Remove references to CacheMsg
The definition for the class CacheMsg was removed long back. Some declaration
had still survived, which was recently removed. Since the PerfectCacheMemory
class relied on this particular declaration, its absence let to compilation
breaking down. Hence this patch.
2012-01-12 00:35:57 -06:00
Ali Saidi
c40ae2c3fb Packet: Put back part of the assert 2012-01-11 19:27:11 -05:00
Ali Saidi
bc1c21274e Packet: Remove meaningless assert statement 2012-01-11 19:24:13 -05:00
Nilay Vaish
c57dc3ffcb Ruby: Use map option for selecting b/w sparse and memory vector 2012-01-11 13:53:38 -06:00
Nilay Vaish
03229f2575 Config: Add support for restoring using a timing CPU
Currently there is an assumption that restoration from a checkpoint will
happen by first restoring to an atomic CPU and then switching to a timing
CPU. This patch adds support for directly restoring to a timing CPU. It
adds a new option '--restore-with-cpu' which is used to specify the type
of CPU to which the checkpoint should be restored to. It defaults to
'atomic' which was the case before.
2012-01-11 13:50:18 -06:00
Nilay Vaish
bf59a9298f Ruby: Resurrect Cache Warmup Capability
This patch resurrects ruby's cache warmup capability. It essentially
makes use of all the infrastructure that was added to the controllers,
memories and the cache recorder.
2012-01-11 13:48:48 -06:00
Nilay Vaish
3f8065290a Ruby Debug Flags: Remove one, add another
The flag RubyStoreBuffer is being removed, instead RubySystem is being added
2012-01-11 13:42:00 -06:00
Nilay Vaish
2d3cae02f5 Ruby Port: Add a list of cpu ports attached to this port 2012-01-11 13:39:58 -06:00
Nilay Vaish
17fc60ee88 Ruby EventQueue: Remove unused functions 2012-01-11 13:31:04 -06:00
Nilay Vaish
8b3ad17cc3 Ruby Sparse Memory: Add function for collating blocks
This patch adds function to the Sparse Memory so that the blocks can be
recorded in a cache trace. The blocks are added to the cache recorder
which can later write them into a file.
2012-01-11 13:29:54 -06:00
Nilay Vaish
c3109f7775 Ruby: Add infrastructure for recording cache contents
This patch changes CacheRecorder, CacheMemory, CacheControllers
so that the contents of a cache can be recorded for checkpointing
purposes.
2012-01-11 13:29:15 -06:00
Nilay Vaish
ab0347a1c6 Ruby Memory Vector: Functions for collating and populating pages
This patch adds functions to the memory vector class that can be used for
collating memory pages to raw trace and for populating pages from a raw
trace.
2012-01-11 11:46:23 -06:00
Nilay Vaish
bd739a75b9 Ruby: remove the files related to the tracer
The Ruby Tracer is out of date with the changes that are being carried
out to support checkpointing. Hence, it needs to be removed.
2012-01-10 18:35:45 -06:00
Nathan Binkert
a1b4cb58f1 hgfilesize: skip files that have been removed 2012-01-10 22:50:54 -08:00
Nilay Vaish
d272bdb1bf MOESI Hammer: Update regression test output 2012-01-10 17:28:49 -06:00