Commit graph

7689 commits

Author SHA1 Message Date
Steve Reinhardt
13a15c55a4 stats: update stats for previous cset
Coherence protocol change basically got rid
of UpgradeReqs in L2 caches, other minor
related cache stat changes.
2010-09-21 23:07:35 -07:00
Steve Reinhardt
e918536380 cache: improve coherence handling of writebacks
If we write back an exclusive copy, we now mark it
as such, so the cache receiving the writeback can
mark its copy as exclusive.  This avoids some
unnecessary upgrade requests when a cache later
tries to re-acquire exclusive access to the block.
2010-09-21 23:07:34 -07:00
Steve Reinhardt
0bd9cea340 diff-out: clean up options
Make diff-out sort stats changes by percentage
by default, with '-a' to use current alpha sort
(instead of requiring '-p' to sort by percentage).
Other minor options cleanup too.
2010-09-21 23:07:34 -07:00
Steve Reinhardt
db2f226834 tests: print if output files match
Add '-s' flag to diff command generating outdiff
file so we have positive confirmation when
outputs match.
2010-09-21 23:07:34 -07:00
Gabe Black
ab8d7eee76 CPU: Fix O3 and possible InOrder segfaults in FS. 2010-09-20 02:46:42 -07:00
Steve Reinhardt
3f9f4bf3d6 devices: undo cset 017baf09599f that added timer drain functions.
It's not the right fix for the checkpoint deadlock problem
Brad was having, and creates another bug where the system can
deadlock on restore.  Brad can't reproduce the original bug
right now, so we'll wait until it arises again and then try
to fix it the right way then.
2010-09-16 20:24:05 -07:00
Gabe Black
2dd9f4fcf0 X86: Make the halt microop non-speculative.
Executing this microop makes the CPU halt even if it was misspeculated.
2010-09-14 12:31:37 -07:00
Gabe Black
0bbd88eb40 X86: Make unrecognized instructions behave better in x86. 2010-09-14 12:27:30 -07:00
Gabe Black
0dd1f7f01a CPU: Trim unnecessary includes from some common files.
This reduces the scope of those includes and makes it less likely for there to
be a dependency loop. This also moves the hashing functions associated with
ExtMachInst objects to be with the ExtMachInst definitions and out of
utility.hh.
2010-09-14 00:29:38 -07:00
Gabe Black
8f3fbd2d13 CPU: Get rid of the now unnecessary getInst/setInst family of functions.
This code is no longer needed because of the preceeding change which adds a
StaticInstPtr parameter to the fault's invoke method, obviating the only use
for this pair of functions.
2010-09-13 21:58:34 -07:00
Gabe Black
6833ca7eed Faults: Pass the StaticInst involved, if any, to a Fault's invoke method.
Also move the "Fault" reference counted pointer type into a separate file,
sim/fault.hh. It would be better to name this less similarly to sim/faults.hh
to reduce confusion, but fault.hh matches the name of the type. We could change
Fault to FaultPtr to match other pointer types, and then changing the name of
the file would make more sense.
2010-09-13 19:26:03 -07:00
Nathan Binkert
2edfcbbaee swig: make all generated files go into the m5.internal package
This is necessary because versions of swig older than 1.3.39 fail to
do the right thing and try to do relative imports for everything (even
with the package= option to %module).  Instead of putting params in
the m5.internal.params package, put params in the m5.internal package
and make all param modules start with param_.  Same thing for
m5.internal.enums.

Also, stop importing all generated params into m5.objects.  They are
not necessary and now with everything using relative imports we wound
up with pollution of the namespace (where builtin-range got overridden).

--HG--
rename : src/python/m5/internal/enums/__init__.py => src/python/m5/internal/enums.py
rename : src/python/m5/internal/params/__init__.py => src/python/m5/internal/params.py
2010-09-12 15:41:34 -07:00
Nathan Binkert
afafaf1dcb style: fix sorting of includes and whitespace in some files 2010-09-10 14:58:04 -07:00
Nathan Binkert
47ef97b9ca scons: Stop building the big monolithic swigged params module
kill params.i and create a separate .i for each object (param, enums, etc.)
2010-09-09 14:26:29 -07:00
Nathan Binkert
e6ee56c657 init: don't build files that centralize python and swig code
Instead of putting all object files into m5/object/__init__.py, interrogate
the importer to find out what should be imported.
Instead of creating a single file that lists all of the embedded python
modules, use static object construction to put those objects onto a list.
Do something similar for embedded swig (C++) code.
2010-09-09 14:15:42 -07:00
Nathan Binkert
710ed8f492 scons: use code_formatter wherever we can in the build system 2010-09-09 14:15:41 -07:00
Nathan Binkert
c514ad9b09 code_formatter: make it easier to insert whitespace
a newline by just doing "code()". indent() and dedent() now take a
"count" parameter to indent/dedent multiple levels.
2010-09-09 14:15:41 -07:00
Nathan Binkert
18ef1bcfa2 swig: don't override SWIG_name anymore
It doesn't appear to be necessary and it is somewhat odd.  I'm pretty
sure that the package parameter to %module does whatever this might
have been before.  It's necessary in future revisions anyway.
2010-09-09 14:15:40 -07:00
Steve Reinhardt
9e45ada171 stats: update stats for preceding coherence changes
Because the handling of the E state for multilevel caches
has changed, stats are affected for any non-ruby config
with caches, even uniprocessor simple CPU.
2010-09-09 14:40:19 -04:00
Steve Reinhardt
1249728494 cache: fail SC when invalidated while waiting for bus
Corrects an oversight in cset f97b62be544f.  The fix there only
failed queued SCUpgradeReq packets that encountered an
invalidation, which meant that the upgrade had to reach the L2
cache.  To handle pending requests in the L1 we must similarly
fail StoreCondReq packets too.
2010-09-09 14:40:19 -04:00
Steve Reinhardt
6dc599ea9b mem: fix functional accesses to deal with coherence change
We can't just obliviously return the first valid cache block
we find any more... see comments for details.
2010-09-09 14:40:19 -04:00
Steve Reinhardt
71aca6d29e cache: coherence protocol enhancements & bug fixes
Allow lower-level caches (e.g., L2 or L3) to pass exclusive
copies to higher levels (e.g., L1).  This eliminates a lot
of unnecessary upgrade transactions on read-write sequences
to non-shared data.

Also some cleanup of MSHR coherence handling and multiple
bug fixes.
2010-09-09 14:40:18 -04:00
Gabe Black
7c4dc4491a ARM: Get rid of the checkFpEnableFault function in ARM. 2010-08-31 09:50:49 -07:00
Gabe Black
ebf5c5b91b Alpha: Alpha's mt.hh was including mips header files. 2010-08-31 09:48:05 -07:00
Gabe Black
c9d01c6557 CPU: Get rid of the unused ev5_trap function on the simple and checker CPUs. 2010-08-31 09:47:29 -07:00
Brad Beckmann
8f1f83d13d config: fixed numa high bit setting bug 2010-08-30 15:26:51 -07:00
Brad Beckmann
08b360e0e0 config: minor fix to probe filter index setting 2010-08-30 12:07:21 -07:00
Nathan Binkert
6d207bf40d ruby: None, not none 2010-08-29 07:02:42 -07:00
Gabe Black
794ca517f2 X86: Change the copyright holder to AMD.
I accidentally left myself as a placeholder copyright holder on this file when
I checked it in. Copyright should be assigned to AMD.
2010-08-27 15:35:36 -07:00
Steve Reinhardt
3ffc4505f7 mem: fix m5.fast compile bug in previous cset 2010-08-26 08:03:20 -07:00
Steve Reinhardt
1bf944be62 cache: fix a bug in atomic multilevel snoops 2010-08-25 21:55:55 -07:00
Steve Reinhardt
ee6a92863a memtest: fix/cleanup functional access testing
Don't assert that the response packet is marked as a response
since it won't always be so for functional accesses.

Also cleanup code to refer to functional accesses rather
than "probes" (old terminology), and mention in the
DPRINTF which type of access we're doing.
2010-08-25 21:55:44 -07:00
Steve Reinhardt
6d9fc4175e memtest: scale associativity and mshrs according to config
Use the actual fanouts in the tree specification to scale
cache associativity and mshrs instead of dumb constants.
2010-08-25 21:55:42 -07:00
Ali Saidi
546eaa6109 CPU: Print out traces for faluting inst when the flag ExecFaulting is set 2010-08-25 19:10:43 -05:00
Min Kyu Jeong
dee8f3d500 ARM: Support unaligned memory access.
Without this flag set, page-crossing requests were not split into two mem
request.

Depending on the alignment bit in the SCTLR, misaligned access could
raise a fault. However it seems unnecessary to implement that.
2010-08-25 19:10:43 -05:00
Gene WU
b52fed4747 ARM: Seperate the queues of L1 and L2 walker states. 2010-08-25 19:10:43 -05:00
Min Kyu Jeong
c23e8c31eb ARM: Adding a bogus fault that does nothing.
This fault can used to flush the pipe, not including the faulting instruction.

The particular case I needed this was for a self-modifying code. It needed to
drain the store queue and force the following instruction to refetch from
icache. DCCMVAC cp15 mcr instruction is modified to raise this fault.
2010-08-25 19:10:43 -05:00
William Wang
8376f7bca3 ARM: Remove ALPHA KSeg functions.
These were erronously copied years ago into the ARM directory.
2010-08-25 19:10:43 -05:00
Ali Saidi
c0b54f579c ARM: Limited implementation of dprintk.
Does not work with vfp arguments or arguments passed on the stack.
2010-08-25 19:10:43 -05:00
Min Kyu Jeong
e1168e72ca ARM: Fixed register flattening logic (FP_Base_DepTag was set too low)
When decoding a srs instruction, invalid mode encoding returns invalid instruction.
This can happen when garbage instructions are fetched from mispredicted path
2010-08-25 19:10:43 -05:00
Ali Saidi
edca5f7da6 ARM: Make VMSR, RFE PC/LR etc non speculative, and serializing 2010-08-25 19:10:43 -05:00
Ali Saidi
e6d3fe8a0c ARM: Update regression tests for ldr/str microcode changes. 2010-08-25 19:10:42 -05:00
Gene WU
4d8f4db8d1 ARM: Use fewer micro-ops for register update loads if possible.
Allow some loads that update the base register to use just two micro-ops. three
micro-ops are only used if the destination register matches the offset register
or the PC is the destination regsiter. If the PC is updated it needs to be
the last micro-op otherwise O3 will mispredict.
2010-08-25 19:10:42 -05:00
Ali Saidi
c2d5d2b53d ARM: Set the high bits in the part number so it's considered new by some code. 2010-08-25 19:10:42 -05:00
Ali Saidi
99fafb72b8 ARM: Fix VFP enabled checks for mem instructions 2010-08-25 19:10:42 -05:00
Gabe Black
63464d950e ARM: Seperate out the renamable bits in the FPSCR. 2010-08-25 19:10:42 -05:00
Gabe Black
93ce7238bf ARM: Eliminate some unused enums. 2010-08-25 19:10:42 -05:00
Gabe Black
0efe2f6769 ARM: Fix type comparison warnings in Neon. 2010-08-25 19:10:42 -05:00
Gabe Black
54a919f225 ARM: Implement CPACR register and return Undefined Instruction when FP access is disabled. 2010-08-25 19:10:42 -05:00
Gabe Black
6368edb281 ARM: Implement all ARM SIMD instructions. 2010-08-25 19:10:42 -05:00