Commit graph

6304 commits

Author SHA1 Message Date
Andreas Hansson
87f4c956c4 mem: Add DRAM power states to the controller
This patch adds power states to the controller. These states and the
transitions can be used together with the Micron power model. As a
more elaborate use-case, the transitions can be used to drive the
DRAMPower tool.

At the moment, the power-down modes are not used, and this patch
simply serves to capture the idle, auto refresh and active modes. The
patch adds a third state machine that interacts with the refresh state
machine.
2014-05-09 18:58:48 -04:00
Andreas Hansson
babf072c1c mem: Ensure DRAM refresh respects timings
This patch adds a state machine for the refresh scheduling to
ensure that no accesses are allowed while the refresh is in progress,
and that all banks are propely precharged.

As part of this change, the precharging of banks of broken out into a
method of its own, making is similar to how activations are dealt
with. The idle accounting is also updated to ensure that the refresh
duration is not added to the time that the DRAM is in the idle state
with all banks precharged.
2014-05-09 18:58:48 -04:00
Andreas Hansson
5c2c3f598e mem: Make DRAM read/write switching less conservative
This patch changes the read/write event loop to use a single event
(nextReqEvent), along with a state variable, thus joining the two
control flows. This change makes it easier to follow the state
transitions, and control what happens when.

With the new loop we modify the overly conservative switching times
such that the write-to-read switch allows bank preparation to happen
in parallel with the bus turn around. Similarly, the read-to-write
switch uses the introduced tRTW constraint.
2014-05-09 18:58:48 -04:00
Ali Saidi
dbaf43394b arm: Make sure UndefinedInstructions are properly initialized 2014-04-17 16:56:09 -05:00
Ali Saidi
a00b44ebe8 arm: allow DC instructions by default so SE mode works 2014-04-17 16:55:54 -05:00
Ali Saidi
c4a2f76fea sim, arm: implement more of the at variety syscalls
Needed for new AArch64 binaries
2014-04-17 16:55:05 -05:00
Andrew Bardsley
f5c3f60601 cpu: Useful getters for ActivityRecorder
Add some useful getters to ActivityRecorder
2014-05-09 18:58:48 -04:00
Andrew Bardsley
bf78299f04 cpu: Add flag name printing to StaticInst
This patch adds a the member function StaticInst::printFlags to allow all
of an instruction's flags to be printed without using the individual
is... member functions or resorting to exposing the 'flags' vector

It also replaces the enum definition StaticInst::Flags with a
Python-generated enumeration and adds to the enum generation mechanism
in src/python/m5/params.py to allow Enums to be placed in namespaces
other than Enums or, alternatively, in wrapper structs allowing them to
be inherited by other classes (so populating that class's name-space
with the enumeration element names).
2014-05-09 18:58:47 -04:00
Andrew Bardsley
8087d2622d cpu: Timebuf const accessors
Add const accessors for timebuf elements.
2014-05-09 18:58:47 -04:00
Andrew Bardsley
f7d80348fa arm: Add branch flags onto macroops
Mark branch flags onto macroops to allow branch prediction before
microop decomposition
2014-05-09 18:58:47 -04:00
Andrew Bardsley
eab00f4966 cpu: Allow setWhen on trace objects
Allow setting of 'when' in trace records.  This allows later times
than the arbitrary record creation point to be used as inst. times
2014-05-09 18:58:47 -04:00
Curtis Dunham
af39ab297f arm: add preliminary ISA splits for ARM arch 2014-05-09 18:58:47 -04:00
Curtis Dunham
fe27f937aa arch: teach ISA parser how to split code across files
This patch encompasses several interrelated and interdependent changes
to the ISA generation step.  The end goal is to reduce the size of the
generated compilation units for instruction execution and decoding so
that batch compilation can proceed with all CPUs active without
exhausting physical memory.

The ISA parser (src/arch/isa_parser.py) has been improved so that it can
accept 'split [output_type];' directives at the top level of the grammar
and 'split(output_type)' python calls within 'exec {{ ... }}' blocks.
This has the effect of "splitting" the files into smaller compilation
units.  I use air-quotes around "splitting" because the files themselves
are not split, but preprocessing directives are inserted to have the same
effect.

Architecturally, the ISA parser has had some changes in how it works.
In general, it emits code sooner.  It doesn't generate per-CPU files,
and instead defers to the C preprocessor to create the duplicate copies
for each CPU type.  Likewise there are more files emitted and the C
preprocessor does more substitution that used to be done by the ISA parser.

Finally, the build system (SCons) needs to be able to cope with a
dynamic list of source files coming out of the ISA parser. The changes
to the SCons{cript,truct} files support this. In broad strokes, the
targets requested on the command line are hidden from SCons until all
the build dependencies are determined, otherwise it would try, realize
it can't reach the goal, and terminate in failure. Since build steps
(i.e. running the ISA parser) must be taken to determine the file list,
several new build stages have been inserted at the very start of the
build. First, the build dependencies from the ISA parser will be emitted
to arch/$ISA/generated/inc.d, which is then read by a new SCons builder
to finalize the dependencies. (Once inc.d exists, the ISA parser will not
need to be run to complete this step.) Once the dependencies are known,
the 'Environments' are made by the makeEnv() function. This function used
to be called before the build began but now happens during the build.
It is easy to see that this step is quite slow; this is a known issue
and it's important to realize that it was already slow, but there was
no obvious cause to attribute it to since nothing was displayed to the
terminal. Since new steps that used to be performed serially are now in a
potentially-parallel build phase, the pathname handling in the SCons scripts
has been tightened up to deal with chdir() race conditions. In general,
pathnames are computed earlier and more likely to be stored, passed around,
and processed as absolute paths rather than relative paths.  In the end,
some of these issues had to be fixed by inserting serializing dependencies
in the build.

Minor note:
For the null ISA, we just provide a dummy inc.d so SCons is never
compelled to try to generate it. While it seems slightly wrong to have
anything in src/arch/*/generated (i.e. a non-generated 'generated' file),
it's by far the simplest solution.
2014-05-09 18:58:47 -04:00
Geoffrey Blake
0c1913336a config: Avoid generating a reference to myself for Parent.any
The unproxy code for Parent.any can generate a circular reference
in certain situations with classes hierarchies like those in ClockDomain.py.
This patch solves this by marking ouself as visited to make sure the
search does not resolve to a self-reference.
2014-05-09 18:58:47 -04:00
Geoffrey Blake
85940fd537 arch, arm: Preserve TLB bootUncacheability when switching CPUs
The ARM TLBs have a bootUncacheability flag used to make some loads
and stores become uncacheable when booting in FS mode. Later the
flag is cleared to let those loads and stores operate as normal.  When
doing a takeOverFrom(), this flag's state is not preserved and is
momentarily reset until the CPSR is touched. On single core runs this
is a non-issue. On multi-core runs this can lead to crashes on the O3
CPU model from the following series of events:
 1) takeOverFrom executed to switch from Atomic -> O3
 2) All bootUncacheability flags are reset to true
 3) Core2 tries to execute a load covered by bootUncacheability, it
    is flagged as uncacheable
 4) Core2's load needs to replay due to a pipeline flush
 3) Core1 core does an action on CPSR
 4) The handling code for CPSR then checks all other cores
    to determine if bootUncacheability can be set to false
 5) Asynchronously set bootUncacheability on all cores to false
 6) Core2 replays load previously set as uncacheable and notices
    it is now flagged as cacheable, leads to a panic.
This patch implements takeOverFrom() functionality for the ARM TLBs
to preserve flag values when switching from atomic -> detailed.
2014-05-09 18:58:47 -04:00
Curtis Dunham
1028c03320 cpu: add more instruction mix statistics
For the o3, add instruction mix (OpClass) histogram at commit (stats
also already collected at issue). For the simple CPUs we add a
histogram of executed instructions
2014-05-09 18:58:47 -04:00
Mitch Hayenga
a15b713cba mem: Squash prefetch requests from downstream caches
This patch squashes prefetch requests from downstream caches,
so that they do not steal cachelines away from caches closer
to the cpu.  It was originally coded by Mitch Hayenga and
modified by Aasheesh Kolli.
2014-05-09 18:58:46 -04:00
Stephan Diestelhorst
b9e6c260a0 stats: Method stats source
This source for stats binds an object and a method / function from the object
to a stats object. This allows pulling out stats from object methods without
needing to go through a global, or static shim.

Syntax is somewhat unpleasant, but the templates and method pointer type
specification were quite tricky.  Interface is very clean though; and similar
to .functor
2014-05-09 18:58:46 -04:00
Akash Bagdia
2b1a01ee6c cpu, arm: Allow the specification of a socket field
Allow the specification of a socket ID for every core that is reflected in the
MPIDR field in ARM systems.  This allows studying multi-socket / cluster
systems with ARM CPUs.
2014-05-09 18:58:46 -04:00
Sascha Bischoff
e940bac278 mem: Auto-generate CommMonitor trace file names
Splits the CommMonitor trace_file parameter into three parameters. Previously,
the trace was only enabled if the trace_file parameter was set, and would be
written to this file. This patch adds in a trace_enable and trace_compress
parameter to the CommMonitor.

No trace is generated if trace_enable is set to False. If it is set to True, the
trace is written to a file based on the name of the SimObject in the simulation
hierarchy. For example, system.cluster.il1_commmonitor.trc. This filename can be
overridden by additionally specifying a file name to the trace_file parameter
(more on this later).

The trace_compress parameter will append .gz to any filename if set to True.
This enables compression of the generated traces. If the file name already ends
in .gz, then no changes are made.

The trace_file parameter will override the name set by the trace_enable
parameter. In the case that the specified name does not end in .gz but
trace_compress is set to true, .gz is appended to the supplied file name.
2014-05-09 18:58:46 -04:00
Geoffrey Blake
29601eada7 arm: Panics in miscreg read functions can be tripped by O3 model
Unimplemented miscregs for the generic timer were guarded by panics
in arm/isa.cc which can be tripped by the O3 model if it speculatively
executes a wrong path containing a mrs instruction with a bad miscreg
index. These registers were flagged as implemented and accessible.
This patch changes the miscreg info bit vector to flag them as
unimplemented and inaccessible. In this case, and UndefinedInst
fault will be generated if the register access is not trapped
by a hypervisor.
2014-05-09 18:58:46 -04:00
Chris Emmons
a3306d0d5e dev: Set HDLCD default pixel clock for 1080p @ 60Hz
This patch changes the default pixel clock to effectively generate
1080p resolution at 60 frames per second. It is dependent upon the
kernel device tree file using the specified resolution / display
string in the comments.
2014-05-09 18:58:46 -04:00
Matt Evans
73dc89e542 arm: quick hack to allow a greater number of CPUs to a guest OS
This is a quick hack to communicate a greater number of CPUs to a guest OS via
the ARM A9 SCU config register. Some OSes (Linux) just look at the bottom field
to count CPUs and with a small change can look at bits [3:0] to learn about up
to 16 CPUs.

Very much unsupported (and contains warning messages as such) but useful for
running 8 core sims without hardwiring CPU count in the guest OS.
2014-05-09 18:58:46 -04:00
Curtis Dunham
7f1603d207 arch: remove inline specifiers on all inst constrs, all ISAs
With (upcoming) separate compilation, they are useless.  Only
link-time optimization could re-inline them, but ideally
feedback-directed optimization would choose to do so only for
profitable (i.e. common) instructions.
2014-05-09 18:58:46 -04:00
Curtis Dunham
eb61f0123b arm: cleanup ARM ISA definition 2014-05-09 18:58:46 -04:00
Curtis Dunham
ad019c5c58 scons: Require SWIG >= 2.0.4 and remove vector typemaps
SWIG commit fd666c1 (*) made it unnecessary for gem5 to have these
typemaps to handle Vector types.

* fd666c1440
2014-05-09 18:58:46 -04:00
Curtis Dunham
ecf774bc56 arm: Correctly display disassembly of vldmia/vstmia
The MicroMemOp class generates the disassembly for both integer
and floating point instructions, but it would always print its
first operand as an integer register without considering that the
op may be a floating instruction in which case a float register
should be displayed instead.
2014-04-23 05:18:30 -04:00
Andreas Hansson
6cd82c1116 sim: Use correct unit for abort message
This patch fixes the unit in the abort printout.
2014-04-23 05:18:27 -04:00
Mitchell Hayenga
bf25c53a7d cpu: Fix setTranslateLatency() bug for squashed instructions
setTranslateLatency could sometimes improperly access a deleted request
packet after an instruction was squashed.
2014-04-23 05:18:26 -04:00
Sascha Bischoff
2031c03c09 misc: Proper type check and import for PortRef
Rewriting the type checking around PortRef, which was interacting strangely
with other Python scripts.

Tested-by: stephan.diestelhorst@arm.com
2014-04-23 05:18:25 -04:00
Mitch Hayenga
e4086878f6 cpu: Fix case where o3 lsq could print out uninitialized data
In the O3 LSQ, data read/written is printed out in DPRINTFs.  However,
the data field is treated as a character string with a null terminated.
However the data field is not encoded this way.  This patch removes
that possibility by removing the data part of the print.
2014-04-01 14:22:06 -05:00
Mitch Hayenga
a0d30f36a6 mem: Don't print out the data of a cache block
This never actually worked since it was printing out only a word
of the cache block and not the entire thing and doubly didn't work
csprintf overrides the %#x specifier and assumes a char* array is
actually a string.
2014-04-01 14:24:36 -05:00
Mitchell Hayenga
0fad0c7f7d arm: Don't use a stack allocated mnemonic
FailUnimplemented passed a stack created mnemonic as a const char * which
causes some grief when the stack goes away.
2014-04-23 05:18:20 -04:00
Dam Sunwoo
84f8fe637c cpu: Add O3 CPU width checks
O3CPU has a compile-time maximum width set in o3/impl.hh, but checking
the configuration against this limit was not implemented anywhere
except for fetch. Configuring a wider pipe than the limit can silently
cause various issues during the simulation. This patch adds the proper
checking in the constructor of the various pipeline stages.
2014-04-23 05:18:18 -04:00
Curtis Dunham
c9071ff95e base: explicitly suggest potential use of 'All' debug flags
Without this declaration, new clangs will complain about this value
being unused. It has no explicit use in the codebase, but it can be
useful to turn on all debugging flags while in a debugger to greatly
increase simulator verbosity.
2014-04-23 05:17:59 -04:00
Curtis Dunham
e651188f75 arch: remove 'null update' check in isa-parser
SCons already does this for all build steps.
2014-04-23 05:17:57 -04:00
Curtis Dunham
fa4a262204 stats: better error message for uninitialized statistic
As suggested by Nathan Binkert in 2008:
http://permalink.gmane.org/gmane.comp.emulators.m5.users/2676
2014-02-10 18:24:20 -06:00
Nilay Vaish
4ceeda20aa ruby: slicc: remove old documentation
Has not been maintained at all.  Since there is alternate documentation
available on gem5.org, no need to have this separately.
2014-04-19 09:00:31 -05:00
Nilay Vaish
183100b8cb ruby: slicc: slight change to rule for transitions
It had an unnecessary pairs token which is being removed.
2014-04-19 09:00:31 -05:00
Faissal Sleiman
a1570f544f o3: Fix occupancy checks for SMT
A number of calls to isEmpty() and numFreeEntries()
should be thread-specific.

In cpu.cc, the fact that tid is /*commented*/ out is a bug. Say the rob
has instructions from thread 0 (isEmpty() returns false), and none from
thread 1. If we are trying to squash all of thread 1, then
readTailInst(thread 1) will be called because rob->isEmpty() returns
false. The result is end_it is not in the list and the while
statement loops indefinitely back over the cpu's instList.

In iew_impl.hh, all threads are told they have the entire remaining IQ, when
each thread actually has a certain allocation. The result is extra stalls at
the iew dispatch stage which the rename stage usually takes care of.

In commit_impl.hh, rob->readHeadInst(thread 1) can be called if the rob only
contains instructions from thread 0. This returns a dummyInst (which may work
since we are trying to squash all instructions, but hardly seems like the right
way to do it).

In rob_impl.hh this fix skips the rest of the function more frequently and is
more efficient.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-04-19 09:00:30 -05:00
Marco Elver
d9fa950396 ruby: recorder: Fix (de-)serializing with different cache block-sizes
Upon aggregating records, serialize system's cache-block size, as the
cache-block size can be different when restoring from a checkpoint. This way,
we can correctly read all records when restoring from a checkpoints, even if
the cache-block size is different.

Note, that it is only possible to restore from a checkpoint if the
desired cache-block size is smaller or equal to the cache-block size
when the checkpoint was taken; we can split one larger request into
multiple small ones, but it is not reliable to do the opposite.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-04-19 09:00:30 -05:00
Andreas Sandberg
02b51afb7e kvm, x86: Add initial support for multicore simulation
Simulating a SMP or multicore requires devices to be shared between
multiple KVM vCPUs. This means that locking is required when accessing
devices. This changeset adds the necessary locking to allow devices to
execute correctly. It is implemented by temporarily migrating the KVM
CPU to the VM's (and devices) event queue when handling
MMIO. Similarly, the VM migrates to the interrupt controller's event
queue when delivering an interrupt.

The support for fast-forwarding of multicore simulations added by this
changeset assumes that all devices in a system are simulated in the
same thread and each vCPU has its own thread. Special care must be
taken to ensure that devices living under the CPU in the object
hierarchy (e.g., the interrupt controller) do not inherit the parent
CPUs thread and are assigned to device thread. The KvmVM object is
assumed to live in the same thread as the other devices in the system.
2014-04-09 16:01:58 +02:00
Andreas Sandberg
221f4f232a dev: Protect PollEvent processing when running in parallel mode
The calling thread is undefined when the PollQueue services events.
This implies that PollEvents need to handle the case where they are
processed from a different thread than the thread that created the
event. This changeset adds temporary event queue migrations to the VNC
server, the ethernet tap device, and the terminal to protect them from
inter-thread calls.
2014-04-09 16:01:43 +02:00
Nilay Vaish
d805e42b81 ruby: slicc: change enqueue statement
As of now, the enqueue statement can take in any number of 'pairs' as
argument.  But we only use the pair in which latency is the key.  This
latency is allowed to be either a fixed integer or a member variable of
controller in which the expression appears.  This patch drops the use of pairs
in an enqueue statement.  Instead, an expression is allowed which will be
interpreted to be the latency of the enqueue.  This expression can anything
allowed by slicc including a constant integer or a member variable.
2014-04-08 13:26:30 -05:00
Nilay Vaish
e689c00b16 ruby: coherence protocols: drop the phrase IntraChip
The phrase is no longer valid since we do not distinguish between
inter and intra chip communication.
2014-04-08 13:26:29 -05:00
Andreas Sandberg
838bcd3b19 sim: Add the ability to lock and migrate between event queues
We need the ability to lock event queues to enable device accesses
across threads. The serviceOne() method now takes a service lock prior
to handling a new event. By locking an event queue, a different
thread/eq can effectively execute in the context of the locked event
queue. To simplify temporary event queue migrations, this changeset
introduces the EventQueue::ScopedMigration class that unlocks the
current event queue, locks a new event queue, and updates the current
event queue variable.

In order to prevent deadlocks, event queues need to be released when
waiting on barriers. This is implemented using the
EventQueue::ScopedRelease class. An instance of this class is, for
example, used in the BaseGlobalEvent class to release the event queue
when waiting on the synchronization barrier.

The intended use for this functionality is when devices need to be
accessed across thread boundaries. For example, when fast-forwarding,
it might be useful to run devices and CPUs in separate threads. In
such a case, the CPU locks the device queue whenever it needs to
perform IO.  This functionality is primarily intended for KVM.

Note: Migrating between event queues can lead to non-deterministic
timing. Use with extreme care!

--HG--
extra : rebase_source : 23e3a741a1fd73861d1339782dbbe1bc76285315
2014-04-03 11:22:49 +02:00
Marco Elver
b884fcf412 cpu: o3: lsq: Fix TSO implementation
This patch fixes violation of TSO in the O3CPU, as all loads must be
ordered with all other loads. In the LQ, if a snoop is observed, all
subsequent loads need to be squashed if the system is TSO.

Prior to this patch, the following case could be violated:

 P0         | P1          ;
 MOV [x],mail=/usr/spool/mail/nilay | MOV EAX,[y] ;
 MOV [y],mail=/usr/spool/mail/nilay | MOV EBX,[x] ;

exists (1:EAX=1 /\ 1:EBX=0) [is a violation]

The problem was found using litmus [http://diy.inria.fr].

Committed by: Nilay Vaish <nilay@cs.wisc.edu
2014-03-25 13:15:04 -05:00
Andreas Hansson
a00383a40a mem: Track DRAM read/write switching and add hysteresis
This patch adds stats for tracking the number of reads/writes per bus
turn around, and also adds hysteresis to the write-to-read switching
to ensure that the queue does not oscilate around the low threshold.
2014-03-23 11:12:14 -04:00
Andreas Hansson
7c18691db1 mem: Rename SimpleDRAM to a more suitable DRAMCtrl
This patch renames the not-so-simple SimpleDRAM to a more suitable
DRAMCtrl. The name change is intended to ensure that we do not send
the wrong message (although the "simple" in SimpleDRAM was originally
intended as in cleverly simple, or elegant).

As the DRAM controller modelling work is being presented at ISPASS'14
our hope is that a broader audience will use the model in the future.

--HG--
rename : src/mem/SimpleDRAM.py => src/mem/DRAMCtrl.py
rename : src/mem/simple_dram.cc => src/mem/dram_ctrl.cc
rename : src/mem/simple_dram.hh => src/mem/dram_ctrl.hh
2014-03-23 11:12:12 -04:00
Andreas Hansson
3dd1587afc mem: Change memory defaults to be more representative
Make the default memory type DDR3-1600 x64, and use the open-adaptive
page policy. This change is aiming to ensure that users by default are
using a realistic memory system.
2014-03-23 11:12:10 -04:00
Wendy Elsasser
bbbae677ed mem: Add close adaptive paging policy to DRAM controller model
This patch adds a second adaptive page policy to the DRAM controller,
closing the page unless there are already queued accesses to the open
page.
2014-03-23 11:12:08 -04:00
Andreas Hansson
03a1aed803 mem: DRAM controller tidying up
Minor tidying up and removing of redundant code, including the
printing of queue state every million accesses.
2014-03-23 11:12:06 -04:00
Andreas Hansson
bc83eb2197 mem: Fix bug in DRAM bytes per activate
This patch ensures that we do not sample the bytes per activate when
the row has already been closed.
2014-03-23 11:12:05 -04:00
Andreas Hansson
116985d661 mem: Limit the accesses to a page before forcing a precharge
This patch adds a basic starvation-prevention mechanism where a DRAM
page is forced to close after a certain number of accesses. The limit
is combined with the open and open-adaptive page policy and if reached
causes an auto-precharge.
2014-03-23 11:12:03 -04:00
Andreas Hansson
6557741311 mem: Make DRAM write queue draining more aggressive
This patch changes the triggering condition for the write draining
such that we grab the opportunity to issue writes if there are no
reads waiting (as opposed to waiting for the writes to reach the high
threshold). As a result, we potentially drain some of the writes in read
idle periods (if any).

A low threshold is added to be able to control how many write bursts
are kept in the memory controller queue (acting as on-chip storage).

The high and low thresholds are updated to sensible values for a 32/64
size write buffer. Note that the thresholds should be adjusted along
with the queue sizes.

This patch also adds some basic initialisation sanity checks and moves
part of the initialisation to the constructor.
2014-03-23 11:12:01 -04:00
Neha Agarwal
364a51181e cpu: DRAM Traffic Generator
This patch enables a new 'DRAM' mode to the existing traffic
generator, catered to generate specific requests to DRAM based on
required hit length (stride size) and bank utilization. It is an add on
to the Random mode.

The basic idea is to control how many successive packets target the
same page, and how many banks are being used in parallel. This gives a
two-dimensional space that stresses different aspects of the DRAM
timing.

The configuration file needed to use this patch has to be changed as
follow: (reference to Random Mode, LPDDR3 memory type)

'STATE 0 10000000000 RANDOM 50 0 134217728 64 3004 5002 0'
-> 'STATE 0 10000000000 DRAM 50 0 134217728 32 3004 5002 0 96 1024 8 6 1'

The last 4 parameters to be added are:
<stride size (bytes), page size(bytes), number of banks available in DRAM,
    number of banks to be utilized, address mapping scheme>

The address mapping information is used to get the stride address
stream of the specified size and to know where to find the bank
bits. The configuration file has a parameter where '0'-> RoCoRaBaCh,
'1'-> RoRaBaCoCh/RoRaBaChCo address-mapping schemes. Note that the
generator currently assumes a single channel and a single rank. This
is to avoid overwhelming the traffic generator with information about
the memory organisation.
2014-03-23 11:11:58 -04:00
Neha Agarwal
43abaf518f mem: DDR3 config for comparing with DRAMSim2
This patch adds a new DDR3 configuration to match with the parameters
that are specified in one of the DDR3 configs used in DRAMSim2.
2014-03-23 11:11:56 -04:00
Andreas Hansson
7e7b67472a mem: More descriptive address-mapping scheme names
This patch adds the row bits to the name of the address mapping
schemes to make it more clear that all the current schemes places the
row bits as the most significant bits.
2014-03-23 11:11:53 -04:00
Stan Czerniawski
4f77bc230a misc: Fix -q (quiet) flag
Check the right flag.
2014-03-23 11:11:49 -04:00
Andreas Hansson
9ac4f781ec ruby: Move Ruby debug flags to ruby dir and remove stale options
This patch moves the Ruby-related debug flags to the ruby
sub-directory, and also removes the state SConsopts that add the
no-longer-used NO_VECTOR_BOUNDS_CHECK.
2014-03-23 11:11:48 -04:00
Andreas Hansson
9f018d2f5a mem: Include the DRAMSim2 wrapper in NULL build
This patch makes sure DRAMSim2 is included in a build of the NULL ISA.
2014-03-23 11:11:44 -04:00
Sascha Bischoff
548d47ea2c mem: CommMonitor trace warn on non-timing mode
Add a warning to the CommMonitor which will alert the user if they try
and record a trace when the system is not in timing mode.
2014-03-23 11:11:40 -04:00
Stan Czerniawski
e18d0e04a2 cpu: Add basic check to TrafficGen initial state
Prevent incomplete configuration of TrafficGen class from causing
segmentation faults. If an 'INIT' line is not present in the
configuration file then the currState variable will remain
uninitialized which may result in a crash.
2014-03-23 11:11:39 -04:00
Andrew Bardsley
0c001e729a dev: Fix IsaFake's cxx_header setting
cxx_header was set incorrectly on IsaFake
2014-03-23 11:11:37 -04:00
Eric Van Hensbergen
7630168a75 arm: m5ops readfile64 args broken, offset coming through garbage
There were several sections of the m5ops code which were
essentially copy/pasted versions of the 32-bit code. The
problem is that some of these didn't account fo4 64-bit
registers leading to arguments being in the wrong registers.
This patch addresses the args for readfile64, writefile64,
and addsymbol64 -- all of which seemed to suffer from a
similar set of problems when moving to 64-bit.
2014-03-23 11:11:34 -04:00
Andreas Hansson
5093e58dc2 base: Fix error message time unit (cycle -> tick)
This patch fixes the unit used in all error messages.
2014-03-23 11:11:32 -04:00
Nilay Vaish
52a83c1d0e ruby: consumer: avoid accessing wakeup times when waking up
Each consumer object maintains a set of tick values when the object is supposed
to wakeup and do some processing.  As of now, the object accesses this set both
when scheduling a wakeup event and when the object actually wakes up.  The set
is accessed during wakeup to remove the current tick value from the set.  This
functionality is now being moved to the scheduling function where ticks are
removed at a later time.
2014-03-20 09:14:14 -05:00
Nilay Vaish
4b67ada89e ruby: garnet: convert network interfaces into clocked objects
This helps in configuring the network interfaces from the python script and
these objects no longer rely on the network object for the timing information.
2014-03-20 09:14:14 -05:00
Nilay Vaish
4f7ef51efb ruby: slicc: code refactor 2014-03-20 09:14:14 -05:00
Nilay Vaish
9b3418d163 ruby: no piobus in se mode
Piobus was recently added to se scripts for ruby so that the interrupt
controller can be connected to something (required since the interrupt
controller sends address range messages).  This patch removes the piobus
and instead, the pio port of ruby port will now ignore the range change
messages in se mode.
2014-03-20 08:03:09 -05:00
Nilay Vaish
f7e7fa6d90 ruby: remove some of the unnecessary code 2014-03-17 17:40:14 -05:00
Andreas Sandberg
11ffa379ab kvm: Clean up signal handling
KVM used to use two signals, one for instruction count exits and one
for timer exits. There is really no need to distinguish between the
two since they only trigger exits from KVM. This changeset unifies and
renames the signals and adds a method, kick(), that can be used to
raise the control signal in the vCPU thread. It also removes the early
timer warning since we do not normally see if the signal was
delivered.

--HG--
extra : rebase_source : cd0e45ca90894c3d6f6aa115b9b06a1d8f0fda4d
2014-03-16 17:40:58 +01:00
Andreas Sandberg
5db547bca4 kvm: x86: Adjust PC to remove the CS segment base address
gem5 seems to store the PC as RIP+CS_BASE. This is not what KVM
expects, so we need to subtract CS_BASE prior to transferring the PC
into KVM. This changeset adds the necessary PC manipulation and
refactors thread context updates slightly to avoid reading registers
multiple times from KVM.

--HG--
extra : rebase_source : 3f0569dca06a1fcd8694925f75c8918d954ada44
2014-03-16 17:30:24 +01:00
Andreas Sandberg
f791e7b313 kvm: x86: Add support for x86 INIT and STARTUP handling
This changeset adds support for INIT and STARTUP IPI handling. We
currently handle both of these interrupts in gem5 and transfer the
state to KVM. Since we do not have a BIOS loaded, we pretend that the
INIT interrupt suspends the CPU after reset.

--HG--
extra : rebase_source : 7f3b25f3801d68f668b6cd91eaf50d6f48ee2a6a
2014-03-16 17:28:23 +01:00
Paul Rosenfeld
32bf74cb8e alpha: Small removal of dead comments/code from alpha ISA
Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-03-12 07:03:22 -05:00
Andreas Hansson
62fe81e9c1 cpu: Make CPU and ThreadContext getters const
This patch merely tidies up the CPU and ThreadContext getters by
making them const where appropriate.
2014-03-07 15:56:23 -05:00
Geoffrey Blake
c4a8e5c36c arm: Handle functional TLB walks properly
The table walker code currently accounts for two types of walks,
Atomic and Timing, and treats them differently. Atomic walks keep a
single instance of WalkerState around for all walks to use in
currState. Timing mode keeps a queue of in-flight WalkerStates and
maintains currState as NULL between walks.

If a functional walk is done during Timing mode, it is treated as an
atomic walk and either creates a persistent WalkerState if in between
Timing walks, or stomps an existing currState for an in-progress
Timing walk.

This patch distinguishes functional walks as being able to exist at
any time and sets up a temporary WalkerState for its exclusive use and
then cleans up when finished, leaving any in progress Atomic or Timing
walks undisturbed.
2014-03-07 15:56:23 -05:00
Prakash Ramrakhyani
e88cffb30a mem: Fix incorrect assert failure in the Cache
This patch fixes an assert condition that is not true at all
times. There are valid situations that arise in dual-core
dual-workload runs where the assert condition is false. The function
call following the assert however needs to be called only when the
condition is true (a block cannot be invalidated in the tags structure
if has not been allocated in the structure, and the tempBlock is never
allocated). Hence the 'assert' has been replaced with an 'if'.
2014-03-07 15:56:23 -05:00
Radhika Jagtap
c446dc40bd mem: Edit proto Packet and enhance the python script
This patch changes the decode script to output the optional fields of
the proto message Packet, namely id and flags. The flags field is set
by the communication monitor.

The id field is useful for CPU trace experiments, e.g. linking the
fetch side to decode side. It had to be renamed because it clashes
with a built in python function id() for getting the "identity" of an
object.

This patch also takes a few common function definitions out from the
multiple scripts and adds them to a protolib python module.
2014-03-07 15:56:23 -05:00
Stephan Diestelhorst
45677ffa97 misc: Add panic_if / fatal_if / chatty_assert
This snippet can be used to replace if + {panics, fatals, asserts} constructs.
The idea is to have both the condition checking and a verbose printout in a single statement.  The interface is as follows:

panic_if(foo != bar, "These should be equal: foo %i bar %i", foo, bar);
fatal_if(foo != bar, "These should be equal: foo %i bar %i", foo, bar);
chatty_assert(foo == bar, "These should be equal: foo %i bar %i", foo, bar);
2014-03-07 15:56:23 -05:00
Mitch Hayenga
b9a9d99b22 scons: Fixes uninitialized warnings issued by clang
Small fixes to appease recent clang versions.
2014-03-07 15:56:23 -05:00
Stephan Diestelhorst
bef2086f5b arm: Fix uninitialised warning with gcc 4.8
Small fix for a warning that prevents compilation with gcc 4.8.1 due
to detecting that a variable might be uninitialised. The fix is to
assign a safe default.
2014-03-07 15:56:23 -05:00
Ali Saidi
bf39a475fe mem: Wakeup sleeping CPUs without caches on LLSC
For systems without caches, the LLSC code does not get snoops for
wake-ups. We add the LLSC code in the abstract memory to do the job
for us.
2014-03-07 15:56:23 -05:00
Andreas Sandberg
f4a897d8e3 sim: Schedule the global sync event at curTick() + simQuantum
The global synchronization event used to be scheduled at
simQuantum. This prevented repeated entries into gem5 from Python as
it can be scheduled in the past. This changeset ensures that the first
global synchronization happens at curTick() + simQuantum instead.
2014-03-06 15:59:53 +01:00
Andreas Sandberg
be246cef62 x86: Setup correct TSL/TR segment attributes on INIT
The TSL/LDT & TR/TSS segments didn't contain valid attributes. This
caused problems when transfering the state into KVM where invalid
state is a no-go. Fixup the attributes with values from AMD's
architecture programmer's manual.
2014-03-03 14:44:57 +01:00
Andreas Sandberg
e7d230ede0 kvm: x86: Always assume segments to be usable
When transferring segment registers into kvm, we need to find the
value of the unusable bit. We used to assume that this could be
inferred from the selector since segments are generally unusable if
their selector is 0. This assumption breaks in some weird corner
cases.  Instead, we just assume that segments are always usable. This
is what qemu does so it should work.
2014-03-03 14:34:33 +01:00
Andreas Sandberg
739cc0128b kvm: Initialize signal handlers from startupThread()
Signal handlers in KVM are controlled per thread and should be
initialized from the thread that is going to execute the CPU. This
changeset moves the initialization call from startup() to
startupThread().
2014-03-03 14:31:39 +01:00
Nilay Vaish
5cd9dd29bd ruby: message buffer: changes related to tracking push/pop times
The last pop operation is now tracked as a Tick instead of in Cycles.
This helps in avoiding use of the receiver's clock during the enqueue
operation.
2014-03-01 23:59:58 -06:00
Nilay Vaish
67cd04b6fe ruby: make the max_size variable of the MessageBuffer unsigned 2014-03-01 23:59:57 -06:00
Christopher Torng
919baa603d cpu: Enable fast-forwarding for MIPS InOrderCPU and O3CPU
A copyRegs() function is added to MIPS utilities
to copy architectural state from the old CPU to
the new CPU during fast-forwarding. This
addition alone enables fast-forwarding for the
o3 cpu model running MIPS.

The patch also adds takeOverFrom() and
drainResume() functions to the InOrderCPU to
enable it to take over from another CPU. This
change enables fast-forwarding for the inorder
cpu model running MIPS, but not for Alpha.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-03-01 23:35:23 -06:00
Nilay Vaish
a533f3f983 ruby: profiler: statically allocate stats variable
Couple of users observed segmentation fault when the simulator tries to
register the statistical variable m_IncompleteTimes.  It seems that there
is some problem with the initialization of these variables when allocated
in the constructor.
2014-03-01 23:35:21 -06:00
Nilay Vaish
7e27860ef4 ruby: route all packets through ruby port
Currently, the interrupt controller in x86 is connected to the io bus
directly.  Therefore the packets between the io devices and the interrupt
controller do not go through ruby.  This patch changes ruby port so that
these packets arrive at the ruby port first, which then routes them to their
destination.  Note that the patch does not make these packets go through the
ruby network.  That would happen in a subsequent patch.
2014-02-23 19:16:16 -06:00
Andreas Hansson
5755fff998 ruby: Simplify RubyPort flow control and routing
This patch simplfies the retry logic in the RubyPort, avoiding
redundant attributes, and enforcing more stringent checks on the
interactions with the normal ports. The patch also simplifies the
routing done by the RubyPort, using the port identifiers instead of a
heavy-weight sender state.

The patch also fixes a bug in the sending of responses from PIO
ports. Previously these responses bypassed the queue in the queued
port, and ignored the return value, potentially leading to response
packets being lost.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-02-23 19:16:16 -06:00
Nilay Vaish
7572ab71b5 ruby: message buffer: refactor code
Code in two of the functions was exactly the same.  This patch moves
this code to a new function which is called from the two functions
mentioned initially.
2014-02-23 19:16:15 -06:00
Nilay Vaish
cde20fd476 ruby: remove few not required #includes 2014-02-23 19:16:15 -06:00
Nilay Vaish
82378f7301 ruby: slicc: remove unused COPY_HEAD functionality 2014-02-23 19:16:15 -06:00
Nilay Vaish
13ad07601b ruby: protocols: remove unused action z_stall 2014-02-23 19:16:15 -06:00
Nilay Vaish
cd33f9bc42 ruby: network: move message buffers to base network class. 2014-02-21 08:02:05 -06:00
Nilay Vaish
bd8f954526 ruby: network: garnet: fixed: removes net_ptr from links 2014-02-21 08:02:04 -06:00
Nilay Vaish
307f53e164 ruby: cache: remove not required variable m_cache_name 2014-02-21 08:02:02 -06:00
Nilay Vaish
f8f8b7e5c2 ruby: network: garnet: fixed: removes next cycle functions
At several places, there are functions that take a cycle value as input
and performs some computation.  Along with each such function, another
function was being defined that simply added one more cycle to input and
computed the same function.  This patch removes this second copy of the
function.  Places where these functions were being called have been updated
to use the original function with argument being current cycle + 1.
2014-02-20 17:28:01 -06:00
Nilay Vaish
896654746a ruby: controller: slight code refactoring 2014-02-20 17:27:45 -06:00
Nilay Vaish
0ce8c25919 ruby: mesi three level: rename incorrectly named files
Two files had been incorrectly named with a .cache suffix.

--HG--
rename : src/mem/protocol/MESI_Three_Level-L0.cache => src/mem/protocol/MESI_Three_Level-L0cache.sm
rename : src/mem/protocol/MESI_Three_Level-L1.cache => src/mem/protocol/MESI_Three_Level-L1cache.sm
2014-02-20 17:27:17 -06:00
Nilay Vaish
db5b3d37fe ruby: network: removes unused code. 2014-02-20 17:27:07 -06:00
Nilay Vaish
dd5c72e5a7 ruby: slicc: slight code refactoring 2014-02-20 17:26:49 -06:00
Nilay Vaish
b312a41f21 ruby: message buffer: removes some unecessary functions. 2014-02-20 17:26:41 -06:00
Andreas Sandberg
0d6009e8dc kvm: Add support for multi-system simulation
The introduction of parallel event queues added most of the support
needed to run multiple VMs (systems) within the same gem5
instance. This changeset fixes up signal delivery so that KVM's
control signals are delivered to the thread that executes the CPU's
event queue. Specifically:

  * Timers and counters are now initialized from a separate method
    (startupThread) that is scheduled as the first event in the
    thread-specific event queue. This ensures that they are
    initialized from the thread that is going to execute the CPUs
    event queue and enables signal delivery to the right thread when
    exiting from KVM.

  * The POSIX-timer-based KVM timer (used to force exits from KVM) has
    been updated to deliver signals to the thread that's executing KVM
    instead of the process (thread is undefined in that case). This
    assumes that the timer is instantiated from the thread that is
    going to execute the KVM vCPU.

  * Signal masking is now done using pthread_sigmask instead of
    sigprocmask. The behavior of the latter is undefined in threaded
    applications.

  * Since signal masks can be inherited, make sure to actively unmask
    the control signals when setting up the KVM signal mask.

There are currently no facilities to multiplex between multiple KVM
CPUs in the same event queue, we are therefore limited to
configurations where there is only one KVM CPU per event queue. In
practice, this means that multi-system configurations can be
simulated, but not multiple CPUs in a shared-memory configuration.
2014-02-20 15:43:53 +01:00
Andreas Hansson
4b81585c49 mem: Fix bug in PhysicalMemory use of mmap and munmap
This patch fixes a bug in how physical memory used to be mapped and
unmapped. Previously we unmapped and re-mapped if restoring from a
checkpoint. However, we never checked that the new mapping was
actually the same, it was just magically working as the OS seems to
fairly reliably give us the same chunk back. This patch fixes this
issue by relying entirely on the mmap call in the constructor.
2014-02-18 05:51:01 -05:00
Andreas Hansson
f0ea79c41f dev: Include basic devices in NULL ISA build
This patch enbles use of the basic PIO devices as part of the NULL
build. Although it might seem counter intuitive to have a PIO device
without being able to execute a driver, this change enables us to
break a device class hierarchy into an ISA-agnostic part, and an
ISA-specific part, without requiring multiple-inheritance. The
ISA-agnostic base class is a PIO device, but does not make use of the
port.
2014-02-18 05:50:59 -05:00
Andreas Hansson
969b436243 mem: Filter cache snoops based on address ranges
This patch adds a filter to the cache to drop snoop requests that are
not for a range covered by the cache. This fixes an issue observed
when multiple caches are placed in parallel, covering different
address ranges. Without this patch, all the caches will forward the
snoop upwards, when only one should do so.
2014-02-18 05:50:58 -05:00
Andreas Hansson
bf2f178f85 mem: Add a wrapped DRAMSim2 memory controller
This patch adds DRAMSim2 as a memory controller by wrapping the
external library and creating a sublass of AbstractMemory that bridges
between the semantics of gem5 and the DRAMSim2 interface.

The DRAMSim2 wrapper extracts the clock period from the config
file. There is no way of extracting this information from DRAMSim2
itself, so we simply read the same config file and get it from there.

To properly model the response queue, the wrapper keeps track of how
many transactions are in the actual controller, and how many are
stacking up waiting to be sent back as responses (in the wrapper). The
latter requires us to move away from the queued port and manage the
packets ourselves. This is due to DRAMSim2 not having any flow control
on the response path.

DRAMSim2 assumes that the transactions it is given are matching the
burst size of the choosen memory. The wrapper checks to ensure the
cache line size of the system matches the burst size of DRAMSim2 as
there are currently no provisions to split the system requests. In
theory we could allow a cache line size smaller than the burst size,
but that would lead to inefficient use of the DRAM, so for not we
fatal also in this case.
2014-02-18 05:50:53 -05:00
Andreas Hansson
c9cb492e1c mem: Fix input to DPRINTF in CommMonitor
Minor fix of the debug message parameters.
2014-02-18 05:50:51 -05:00
Andreas Sandberg
c52190a695 cpu: simple: Add support for using branch predictors
This changesets adds branch predictor support to the
BaseSimpleCPU. The simple CPUs normally don't need a branch predictor,
however, there are at least two cases where it can be desirable:

  1) A simple CPU can be used to warm the branch predictor of an O3
     CPU before switching to the slower O3 model.

  2) The simple CPU can be used as a quick way of evaluating/debugging
     new branch predictors since it exposes branch predictor
     statistics.

Limitations:
  * Since the simple CPU doesn't speculate, only one instruction will
    be active in the branch predictor at a time (i.e., the branch
    predictor will never see speculative branches).

  * The outcome of a branch prediction does not affect the performance
    of the simple CPU.
2014-02-09 20:49:28 +01:00
Nilay Vaish
eb73a14fe2 base: calls abort() from fatal
Currently fatal() ends the simulation in a normal fashion.  This results in
the call stack getting lost when using a debugger and it is not always
possible to debug the simulation just from the information provided by the
printed error message.  Even though the error is likely due to a user's fault,
the information available should not be thrown away.  Hence, this patch to
call abort() from fatal().
2014-02-06 16:30:13 -06:00
Nilay Vaish
bb0e9119e7 ruby: memory controller: use MemoryNode * 2014-02-06 16:30:12 -06:00
Andreas Sandberg
e76a37985f x86: Fix x87 state transfer bug
Changeset 7274310be1bb (isa: clean up register constants) increased
the value of NumFloatRegs, which triggered a bug in
X86ISA::copyRegs(). This bug is caused by the x87 stack being copied
twice since register indexes past NUM_FLOATREGS are mapped into the
x87 stack relative to the top of the stack, which is undefined when
the copy takes place.

This changeset updates the copyRegs() function to use access registers
using the non-flattening interface, which guarantees that undesirable
register folding does not happen.
2014-02-05 14:08:13 +01:00
Nikos Nikoleris
c6279f2d19 x86, kvm: Fix bug in the RFlags get and set functions
The getRFlags and setRFlags utility functions were not updated
correctly when condition registers were separated into their own
register class. This lead to incorrect state transfer in calls from
kvm into the simulator (e.g., m5 readfile ended up in an infinite
loop) and when switching CPUs. This patch makes these utility
functions use getCCReg and setCCReg instead of getIntReg and setIntReg
which read and write the integer registers.

Reviewed-by: Andreas Sandberg <andreas@sandberg.pp.se>
2014-02-02 16:37:35 +01:00
Ola Jeppsson
7f16951451 unittest: Fix build errors
Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-30 12:21:58 -06:00
Mitch Hayenga
96317d466e mem: Add additional tolerance to stride prefetcher
Forces the prefetcher to mispredict twice in a row before resetting the
confidence of prefetching.  This helps cases where a load PC strides by a
constant factor, however it may operate on different arrays at times.
Avoids the cost of retraining.  Primarily helps with small iteration loops.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-29 23:21:26 -06:00
Mitch Hayenga
771c864bf4 mem: Allowed tagged instruction prefetching in stride prefetcher
For systems with a tightly coupled L2, a stride-based prefetcher may observe
access requests from both instruction and data L1 caches.  However, the PC
address of an instruction miss gives no relevant training information to the
stride based prefetcher(there is no stride to train).  In theses cases, its
better if the L2 stride prefetcher simply reverted back to a simple N-block
ahead prefetcher.  This patch enables this option.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-29 23:21:26 -06:00
Mitch Hayenga ext:(%2C%20Amin%20Farmahini%20%3Caminfar%40gmail.com%3E)
95735e10e7 mem: prefetcher: add options, support for unaligned addresses
This patch extends the classic prefetcher to work on non-block aligned
addresses.  Because the existing prefetchers in gem5 mask off the lower
address bits of cache accesses, many predictable strides fail to be
detected.  For example, if a load were to stride by 48 bytes, with 64 byte
cachelines, the current stride based prefetcher would see an access pattern
of 0, 64, 64, 128, 192.... Thus not detecting a constant stride pattern.  This
patch fixes this, by training the prefetcher on access and not masking off the
lower address bits.

It also adds the following configuration options:
1) Training/prefetching only on cache misses,
2) Training/prefetching only on data acceses,
3) Optionally tagging prefetches with a PC address.
#3 allows prefetchers to train off of prefetch requests in systems with
multiple cache levels and PC-based prefetchers present at multiple levels.
It also effectively allows a pipelining of prefetch requests (like in POWER4)
across multiple levels of cache hierarchy.

Improves performance on my gem5 configuration by 4.3% for SPECINT and 4.7%  for SPECFP (geomean).
2014-01-29 23:21:25 -06:00
Xiangyu Dong
32cc2ea8b9 cpu: fix bug when TrafficGen deschedules event
Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-29 22:35:04 -06:00
Mitch Hayenga
b77ca57f8c arm: Enable umask syscall in SE mode
Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-28 18:00:51 -06:00
Mitch Hayenga
55a4ff5f04 base: Fix race condition in the socket listen function
gem5 makes the incorrect assumption that by binding a socket, it
effectively has allocated a port. Linux only allocates ports once you call
listen on the given socket, not when you call bind.  So even if the port was
free when bind was called, another process (gem5 instance) could race in
between the bind & listen calls and steal the port. In the current code, if
the call to bind fails due to the port being in use (EADDRINUSE), gem5 retries
for a different port.  However if listen fails, gem5 just panics. The fix is
testing the return value of listen and re-trying if it was due to EADDRINUSE.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-28 18:00:51 -06:00
Amin Farmahini
ffbdaa7cce mem: Remove redundant findVictim() input argument
The patch
(1) removes the redundant writeback argument from findVictim()
(2) fixes the description of access() function

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-28 18:00:50 -06:00
Amin Farmahini
575a73f4a1 mem: Fixes a bug in simple_dram write merging
Fixes updating the value of size in the write merge function.

Committed by: Nilay Vaish <nilay@cs.wisc.edu>
2014-01-28 18:00:49 -06:00
Nilay Vaish
bdee69d0b1 x86: use lfpimm instead of limm for fptan 2014-01-27 18:50:54 -06:00
Nilay Vaish
6a543b5134 x86: implements x87 add/sub instructions 2014-01-27 18:50:53 -06:00
Nilay Vaish
5be0b846b1 x86: implements fxch instruction. 2014-01-27 18:50:52 -06:00
Nilay Vaish
4eb3b1ed0b x86: correct error in emms instruction. 2014-01-27 18:50:51 -06:00
ARM gem5 Developers
612f8f074f arm: Add support for ARMv8 (AArch64 & AArch32)
Note: AArch64 and AArch32 interworking is not supported. If you use an AArch64
kernel you are restricted to AArch64 user-mode binaries. This will be addressed
in a later patch.

Note: Virtualization is only supported in AArch32 mode. This will also be fixed
in a later patch.

Contributors:
Giacomo Gabrielli    (TrustZone, LPAE, system-level AArch64, AArch64 NEON, validation)
Thomas Grocutt       (AArch32 Virtualization, AArch64 FP, validation)
Mbou Eyole           (AArch64 NEON, validation)
Ali Saidi            (AArch64 Linux support, code integration, validation)
Edmund Grimley-Evans (AArch64 FP)
William Wang         (AArch64 Linux support)
Rene De Jong         (AArch64 Linux support, performance opt.)
Matt Horsnell        (AArch64 MP, validation)
Matt Evans           (device models, code integration, validation)
Chris Adeniyi-Jones  (AArch64 syscall-emulation)
Prakash Ramrakhyani  (validation)
Dam Sunwoo           (validation)
Chander Sudanthi     (validation)
Stephan Diestelhorst (validation)
Andreas Hansson      (code integration, performance opt.)
Eric Van Hensbergen  (performance opt.)
Gabe Black
2014-01-24 15:29:34 -06:00
Andreas Hansson
cfc4a99982 arch: Make all register index flattening const
This patch makes all the register index flattening methods const for
all the ISAs. As part of this, readMiscRegNoEffect for ARM is also
made const.
2014-01-24 15:29:30 -06:00
Geoffrey Blake
9633282fc8 checker: CheckerCPU handling of MiscRegs was incorrect
The CheckerCPU model in pre-v8 code was not checking the
updates to miscellaneous registers due to some methods
for setting misc regs were not instrumented.  The v8 patches
exposed this by calling the instrumented misc reg update
methods and then invoking the checker before the main CPU had
updated its misc regs, leading to false positives about
register mismatches. This patch fixes the non-instrumented
misc reg update methods and places calls to the checker in
the proper places in the O3 model.
2014-01-24 15:29:30 -06:00
Ali Saidi
7d0344704a arch, cpu: Add support for flattening misc register indexes.
With ARMv8 support the same misc register id  results in accessing different
registers depending on the current mode of the processor. This patch adds
the same orthogonality to the misc register file as the others (int, float, cc).
For all the othre ISAs this is currently a null-implementation.

Additionally, a system variable is added to all the ISA objects.
2014-01-24 15:29:30 -06:00
Giacomo Gabrielli
3436de0c2a cpu: Add support for Memory+Barrier instruction types in O3 cpu. 2014-01-24 15:29:30 -06:00
Ali Saidi
90b1775a8f cpu: Add support for instructions that zero cache lines. 2014-01-24 15:29:30 -06:00
Ali Saidi
6bed6e0352 cpu: Add CPU support for generatig wake up events when LLSC adresses are snooped.
This patch add support for generating wake-up events in the CPU when an address
that is currently in the exclusive state is hit by a snoop. This mechanism is required
for ARMv8 multi-processor support.
2014-01-24 15:29:30 -06:00
Giacomo Gabrielli
d3444c6603 mem: Add flag to request if it was generated by a page table walk 2014-01-24 15:29:30 -06:00
Giacomo Gabrielli
aefe9cc624 mem: Add support for a security bit in the memory system
This patch adds the basic building blocks required to support e.g. ARM
TrustZone by discerning secure and non-secure memory accesses.
2014-01-24 15:29:30 -06:00
Chris Adeniyi-Jones
7f835a59f1 sim: Add openat/fstatat syscalls and fix mremap
This patch adds support for the openat and fstatat syscalls and
broadens the support for mremap to make it work on OS X.
2014-01-24 15:29:30 -06:00
Ali Saidi
904872a01a mem: Remove explict cast from memhelper.
Previously we were casting the result type to the the memory type which
is incorrect for things like dual-memory operations which still return a
single result.
2014-01-24 15:29:30 -06:00
Timothy M. Jones
427ceb57a9 Cache: Collect very basic stats on tag and data accesses
Adds very basic statistics on the number of tag and data accesses within the
cache, which is important for power modelling.  For the tags, simply count
the associativity of the cache each time.  For the data, this depends on
whether tags and data are accessed sequentially, which is given by a new
parameter.  In the parallel case, all data blocks are accessed each time, but
with sequential accesses, a single data block is accessed only on a hit.
2014-01-24 15:29:30 -06:00
Dam Sunwoo
85e8779de7 mem: per-thread cache occupancy and per-block ages
This patch enables tracking of cache occupancy per thread along with
ages (in buckets) per cache blocks.  Cache occupancy stats are
recalculated on each stat dump.
2014-01-24 15:29:30 -06:00
Matt Horsnell
739c6df94e base: add support for probe points and common probes
The probe patch is motivated by the desire to move analytical and trace code
away from functional code. This is achieved by the probe interface which is
essentially a glorified observer model.

What this means to users:
* add a probe point and a "notify" call at the source of an "event"
* add an isolated module, that is being used to carry out *your* analysis (e.g. generate a trace)
* register that module as a probe listener
Note: an example is given for reference in src/cpu/o3/simple_trace.[hh|cc] and src/cpu/SimpleTrace.py

What is happening under the hood:
* every SimObject maintains has a ProbeManager.
* during initialization (src/python/m5/simulate.py) first regProbePoints and
  the regProbeListeners is called on each SimObject.  this hooks up the probe
  point notify calls with the listeners.

FAQs:
Why did you develop probe points:
* to remove trace, stats gathering, analytical code out of the functional code.
* the belief that probes could be generically useful.

What is a probe point:
* a probe point is used to notify upon a given event (e.g. cpu commits an instruction)

What is a probe listener:
* a class that handles whatever the user wishes to do when they are notified
  about an event.

What can be passed on notify:
* probe points are templates, and so the user can generate probes that pass any
  type of argument (by const reference) to a listener.

What relationships can be generated (1:1, 1:N, N:M etc):
* there isn't a restriction. You can hook probe points and listeners up in a
  1:1, 1:N, N:M relationship. They become useful when a number of modules
  listen to the same probe points. The idea being that you can add a small
  number of probes into the source code and develop a larger number of useful
  analysis modules that use information passed by the probes.

Can you give examples:
* adding a probe point to the cpu's commit method allows you to build a trace
  module (outputting assembler), you could re-use this to gather instruction
  distribution (arithmetic, load/store, conditional, control flow) stats.

Why is the probe interface currently restricted to passing a const reference:
* the desire, initially at least, is to allow an interface to observe
  functionality, but not to change functionality.
* of course this can be subverted by const-casting.

What is the performance impact of adding probes:
* when nothing is actively listening to the probes they should have a
  relatively minor impact. Profiling has suggested even with a large number of
  probes (60) the impact of them (when not active) is very minimal (<1%).
2014-01-24 15:29:30 -06:00
Andreas Hansson
4de69821e6 sim: Expose the current voltage for each object as a stat 2014-01-24 15:29:30 -06:00
Andreas Hansson
1d85e914a6 sim: Expose the current clock period as a stat
This patch adds observability to the clock period of the clock domains
by including it as a stat.

As a result of adding this, the regressions will be updated in a
separate patch.
2014-01-24 15:29:30 -06:00
Matt Horsnell
ca89eba79e mem: track per-request latencies and access depths in the cache hierarchy
Add some values and methods to the request object to track the translation
and access latency for a request and which level of the cache hierarchy responded
to the request.
2014-01-24 15:29:30 -06:00
Andreas Hansson
daa781d2db config: Make the Clock a Tick parameter like Latency/Frequency
This patch makes the Clock a TickParamValue just like
Latency/Frequency. There is no longer any need to distinguish it
(originally needed to support multiplication).
2014-01-24 15:29:29 -06:00
Andreas Hansson
f2b0b551cc x86: Fix memory leak in table walker
This patch fixes a memory leak in the table walker, by ensuring that
the sender state is deleted again if the request packet cannot be
successfully sent.
2014-01-24 15:29:29 -06:00
Andreas Hansson
7db542c0dd cpu: Relax check on squashed non-speculative instructions
This patch relaxes the check performed when squashing non-speculative
instructions, as it caused problems with loads that were marked ready,
and then stalled on a blocked cache. The assertion is now allowing
memory references to be non-faulting.
2014-01-24 15:29:29 -06:00