Only a small quantity of prefetches were issued, as the positive
feedback mechanism was not implemented. This commit adds a new
action po_observeHit, which notifies the RubyPrefetcher of
successful prefetches and resets the prefetch flag.
When a cache line was replaced by a prefetch, the wrong queue could
be stalled. This commit adds a new event PF_L1_Replacement, which
stalls the correct queue.
The behavior when receiving a prefetch or instruction fetch while
in PF_IS_I (prefetch caused GETs, but got invalidated before the
response was received) was undefined. This was changed to drop the
prefetch request or change the state to non-prefetch, respectively.
This behavior is analogous to IS_I (non-prefetch caused GETs, but
got invalidated before the response was received) and the data case,
respectively.
In my local branch a major (20+%) performance increase can be
observed in SPEC2006 gobmk and leslie3d when enabling the
prefetcher. Some other benchmarks like bwaves, GemsFDTD, sphinx and
wrf show smaller (~10%) performance increases. Unfortunately, the
performance in most other SPEC benchmarks is still poor, most likely
as the prefetcher does not detect strides fast/often enough. In
order to push the change timely (most benchmarks have runtimes in
the order of days on my machine even with the smallest parameters)
after checkout, I have only run gobmk with the base repository
+ this commit. The results match those of my local branch.
Change-Id: I9903a2fcd02060ea5e619b409f31f7d6fac47ae8
Reviewed-on: https://gem5-review.googlesource.com/8801
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Swapnil Haria <swapnilster@gmail.com>
Maintainer: Jason Lowe-Power <jason@lowepower.com>
This patch introduces the TLB IPA-Based invalidating instructions in
aarch32. In the entry selection policy the level of translation is not
taken into account.
This means that no difference stands between (e.g.) TLBIIPAS2 and
TLBIPAS2L.
Change-Id: Ieeb54665480874d2041056f356d86448c45043cb
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8822
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
In the pool of TLB Invalidate system register a category of instruction
was missing: the ones operating on entries added to the TLB during the
last level only of a table walk. (E.g. TLBIVMAL). This patch is not
considering this matching criteria when invalidating the entries and it
is rather performing the invalidation on all levels.
Change-Id: I5f2186cfdd73793e76c90b260f7128be187903fe
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8821
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
g++ seems to think there are some missing brackets when initializing
the sparc fault information. Passify it by adding extra brackets.
Change-Id: I826995f88b8ac8a21721c949a244dec480831b80
Reviewed-on: https://gem5-review.googlesource.com/8763
Maintainer: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Gabe Black <gabeblack@google.com>
The fault status code generated by a Prefetch/Data Fault was containing
a wrong value when the fault was triggered in aarch32 but handled in
aarch64. This because the encoding differs between the two ISAs and the
encoder was just checking the starting ISA rather than the the ending
one. In this case the getFsr must be called after we know which is the
ending ISA, which happens only after ArmFault::invoke gets called. The
fsc update hence happens before writing into the Syndrome register.
Change-Id: I725f12b6dcc0178f608233bd3d15e466d1cd1ffc
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8362
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
There is a set of internal variables in ArmFault thats get updated once
the fault is invoked (ArmFault::invoke). Sometimes we rely on those even
if the fault is generated but not invoked (e.g. when checking if a
memory access is producing a fault). This patch is moving the update
functionalities inside a public method so that a client can make use of
it even when not invoking the fault.
Change-Id: I3ac5b6835023f28ec569fe25487dffa356e1b2fd
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8361
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
invalidate was defined as a separate function in the base associative
and fully-associative tags classes although both functions should
implement identical functionality. This patch moves the invalidate
function in the base tags class.
Change-Id: I206ee969b00ab9e05873c6d87531474fcd712907
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8286
Reviewed-by: Daniel Carvalho <odanrc@yahoo.com.br>
Maintainer: Nikos Nikoleris <nikos.nikoleris@arm.com>
Starting with version 3, scons imposes using the print function instead
of the print statement in code it processes. To get things building
again, this change moves all python code within gem5 to use the
function version. Another change by another author separately made this
same change to the site_tools and site_init.py files.
Change-Id: I2de7dc3b1be756baad6f60574c47c8b7e80ea3b0
Reviewed-on: https://gem5-review.googlesource.com/8761
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Maintainer: Gabe Black <gabeblack@google.com>
If the switch block inside the RFDR case selects the non-default case
and breaks out, it will fall through into the BRAR case, which seems
incorrect. Put in a break to ensure that it breaks out of the parent
switch block as well.
Change-Id: Ie4cedf66954b7e8f4b884ad9e3a653968bbfaef7
Signed-off-by: Siddhesh Poyarekar <siddhesh.poyarekar@gmail.com>
Reviewed-on: https://gem5-review.googlesource.com/8563
Reviewed-by: Gabe Black <gabeblack@google.com>
Maintainer: Gabe Black <gabeblack@google.com>
Writebacks write data to either an existing block or a newly allocated
block. In either case we need to populate the whenReady field of the
block which will determine when the new value can be used.
Change-Id: I5788fad0b8086a1be96714639bf6a9470b334926
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Curtis Dunham <curtis.dunham@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8285
Maintainer: Nikos Nikoleris <nikos.nikoleris@arm.com>
The secure bit should be set when we fill a block with data from a
secure location, as indicated by the packet that triggers the fill.
This patch fixes a bug in which the cache wouldn't populate the secure
bit when filling the temp block.
Change-Id: I95c706146449804ff42b205b25dd79750f3e882a
Reviewed-by: Curtis Dunham <curtis.dunham@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8284
Maintainer: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Daniel Carvalho <odanrc@yahoo.com.br>
The temperature used for the power calculations was fixed at 0
degrees, unless a thermal model was setup. This commit allows
the user to set the temperature that needs to be used by the
power calculation during gem5 configuration. This value will be
overwritten if there are thermal models present.
Change-Id: I7ca8fa6766bdcba9d362c12fc75d1e1f74385f35
Reviewed-by: Sascha Bischoff <sascha.bischoff@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8602
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
Proposed changes to SPARC FS simulation, testing indicates that checkpoints are now loaded correctly with the following command: build/SPARC/gem5.opt configs/example/fs.py -r 1
Change-Id: Icd44f01a74c41a78828ef6fd7b661e584bdb6966
Reviewed-on: https://gem5-review.googlesource.com/8581
Reviewed-by: Gabe Black <gabeblack@google.com>
Maintainer: Gabe Black <gabeblack@google.com>
The HLT instruction is used to trap into semihosting. The semihosting
code can change the contents of memory behind the back of the CPU,
which requires instructions triggering semihosting to be
non-speculative and memory barriers.
Change-Id: I735166251aa194120ad49c08082d4ac65fe96524
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8373
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
A new class of Semihosting constructor templates has been added. Their
main purpose is to check if the Exception Generation Instructions (HLT,
SVC) are actually a semihosting command. If that is the case, the
IsMemBarrier flag is raised, so that in the O3 model we perform a
coherent memory access during the semihosting operation.
Change-Id: Ib87fdeb70ee7a930659563230a80cce0e1372c32
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8370
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
This patch fixes the disassembly of AArch64 Exception Generating
instructions, which were not printing the encoded immediate field. This
has been accomplished by changing their underlying type to a newly
defined one.
Change-Id: If58ae3e620d2baa260e12ecdc850225adfcf1ee5
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8368
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
There are cases where the IEW adds a non-speculative instruction to
the IQ twice. This can happen if an instruction is flagged as
IsMemBarrier and IsNonSpeculative. Avoid adding non-speculative
instructions in the IEW to the IQ by checking if it has been added
already.
Change-Id: Ifcff676a451b57b2406ce00ed8dae19ed399515f
Signed-off-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Javier Setoain <javier.setoain@arm.com>
Reviewed-by: Giacomo Gabrielli <giacomo.gabrielli@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8374
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Since b8b13206c8, the '.fast' build has failed to compile with an error
caused by a variable and an assert.
As a reminder, assert macros are optimized out of the build for '.fast'.
If an assert check requires a variable that is unused anywhere else in
the code, the compiler complains that the variable is unused and the
scons build fails. The solution is to add a M5_VAR_USED specifier to
tell the compiler to ignore the variable.
Change-Id: I38f6bbed1e4c0506c5bbc1206c21f1f7e3d8dfe6
Reviewed-on: https://gem5-review.googlesource.com/8462
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Anthony Gutierrez <anthony.gutierrez@amd.com>
Maintainer: Jason Lowe-Power <jason@lowepower.com>
Maintainer: Anthony Gutierrez <anthony.gutierrez@amd.com>
There is a bug in RISC-V's compressed branch instructions where the
offsets are not stored in ImmOp's immediate field, causing incorrect
branchTarget() return values. This patch adds a new compressed branch
op format, CBOp, which correctly stores the offset.
Change-Id: Iac6e9b091d63f3dce4717ee5a9ec31a7cbd6c377
Reviewed-on: https://gem5-review.googlesource.com/8441
Reviewed-by: Tuan Ta <qtt2@cornell.edu>
Maintainer: Alec Roelke <ar4jc@virginia.edu>
The current physical port proxy doesn't know how to tag memory
accesses as secure. Refactor the class slightly to create a set of
methods (readBlobPhys, writeBlobPhys, memsetBlobPhys) that always
access physical memory and take a set of Request::Flags as an
argument. The new port proxy, SecurePortProxy, uses this interface to
issue secure physical accesses.
Change-Id: I8232a4b35025be04ec8f91a00f0580266bacb338
Signed-off-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8364
Maintainer: Nikos Nikoleris <nikos.nikoleris@arm.com>
A new pseudo register has been added to the Misc pool. It is the
implementation defined register. This kinds of registers are covered by
the architecture and must be treated differently than UNIMPLEMENTED
registers: their access can be trapped to EL2 (See HCR.TIDCP bit in the
arm arm).
Some previously undecoded registers in c9,c10,c11 have now this register
type.
Change-Id: Ibfc35982470b9dea0ecf39aaa6b1012a21852f53
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/7922
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
A new identifier has been introduced: NUM_PHYS_MISCREGS, which is used
as a boundary for the number of physical (real) Misc registers in the
system. Pseudo registers (like CP15_UNIMPL) have been moved after the
NUM_PHYS_MISCREGS identifier, so that their enum number is
(NUM_PHYS_MISCREGS < number < NUM_MISCREGS). Moving away those
registers has created some free slots that can be used for future Misc
register implementation.
SERIALIZE and UNSERIALIZE now only save/restore PHYSICAL Misc Registers.
This allows us to define as many pseudo registers as we want without
being concerned about checkpoint compatibility.
Change-Id: I7e297b814eeaa4bee640e81bee625fb66710af45
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/7921
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
do{Long,L1,L2}Descriptor was not able to load descriptors correctly
for big-endian situations, causing recognised Descriptors. Added
big-endian related data conversions to correct them.
Change-Id: I0fdfbbdf56f94bbed19172acae1b6e4a0382b5a0
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/8144
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
