Because MediaOps have two sizes, one for sources and one for
destinations, it does not have a single dataSize member to set the size
of its operands. This was difficult to correct at the time, so a
dataSize member was created which was fixed at 0, and the instructions
themselves would use srcSize and destSize internally to do the actual
computation.
That causes problems when tracing, since the printReg function needs to
know what size to use to print some registers properly, specifically
integer registers.
To now fix that problem, some SFINAE constructors have been added which
will either pass through a dataSize member if one exists, or pass
through a pointer to the instruction itself so that operand index
selector class can pick out the member that makes sense for it (destSize
for DestOp, srcSize for Src1Op and Src2Op).
Change-Id: I6b8259a5ab27f809b81453bcf987cc6d1be4811a
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/49203
Tested-by: kokoro <noreply+kokoro@google.com>
Maintainer: Gabe Black <gabe.black@gmail.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
These methods are all identical now. The O3 versions can all be
consolidated into a single method. For the checker CPU, they can
actually be eliminated entirely, and the result queue's "emplace()"
method can be used to add items using less text than just calling the
original helper method.
Change-Id: Ifaeb3beeea257c8bbf951ee1dd8d2d5fd8bb3964
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/49128
Reviewed-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Maintainer: Giacomo Travaglini <giacomo.travaglini@arm.com>
Tested-by: kokoro <noreply+kokoro@google.com>
These *used* to have a source register which was the syscall number.
That was not an actual property of the instruction, it was a property of
the syscall mechanism in Linux which had leaked into the implementation
of the instructions themselves.
That was corrected by removing that source register and letting the
syscall handling layer figure out which system call to invoke.
Unfortunately the SyscallInst format wasn't also updated, and when
printing the disassembly for those instructions, it would try to print
that source register. That instruction doesn't bother to set up a source
operand array since it doesn't actually have any, so that ended up
accessing random memory.
Change-Id: I20890ad69605983305a13f414ca59a5fa35d4d06
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/49183
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
The function will return the absolute path of the gem5 repo
hosting the m5 library.
One of the use of this helper is to effectively refer/import
gem5 modules from EXTRAS repositories.
If I wanted to import the Ruby module from configs/ruby I could
do that with:
from m5.util import addToPath, repoPath
configs_path = os.path.join(repoPath(), configs)
addToPath(configs_path)
from ruby import Ruby
This isn't an out of tree scripts utility only: most of our configs are
currently relying on doing relative backward imports and could be ported
to use the repoPath utility:
addToPath(../..) is quite a common pattern
This makes the dependencies difficult to read/track and a bit fragile
as it all relies on the relative position between modules.
Change-Id: I26f6ef34b44f20903cc1b6248330b6156378f40b
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/49083
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Track structured data related to different binary flavors (opt, debug,
etc), using a class instead of various lists, etc. Also use a loop to
set up SCons environments to build these binaries instead of a spelled
out loop.
Change-Id: Ie35a914ab79342190e4cdc27a945a0fecd54a476
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48130
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Daniel Carvalho <odanrc@yahoo.com.br>
Maintainer: Gabe Black <gabe.black@gmail.com>
This function used makeAtomicOpFunctor to create a unique_ptr which
pointed to an AtomicOpFunctor *, which it immediately extracted with
.get(). Then since the temporary unique_ptr went out of scope, it
deleted the AtomicOpFunctor which it just returned a pointer to.
Instead, that function should create a local unique_ptr to pass
ownership of the object off to. It will still be cleaned up when it
goes out of scope, but not before it's done being used.
Change-Id: I74a0bcbb719a78a3e9ec8cb2ea5aa15120da0456
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/49023
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Bobby R. Bruce <bbruce@ucdavis.edu>
Reviewed-by: Kyle Roarty <kyleroarty1716@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
The SourceFile.filename property dynamically calculated the str()
conversion of self.tnode. Since self.tnode shouldn't be changed, it
doesn't seem useful to calculate that value over and over, especially
since it adds some extra indirection and magic to something that's
really pretty simple.
Change-Id: Ia0e1e8f4b0c019a026a08b5c2730d93c66de8190
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48123
Reviewed-by: Daniel Carvalho <odanrc@yahoo.com.br>
Maintainer: Bobby R. Bruce <bbruce@ucdavis.edu>
Tested-by: kokoro <noreply+kokoro@google.com>
The build for .fast was set up to produce a stripped executable, and the
unstripped executable was instead named .fast.unstripped. I think the
assumption that a stripped executable is faster than an unstripped
executable is flawed, since the parts of the binary that are removed,
debug symbols, are not loaded into memory anyway, so while the program
is executing it shouldn't be any different or take up any additional
memory. This also made .fast a special case compared to the other build
types, like .opt, .debug, etc.
Instead, this change makes .fast unstripped like all the other binaries,
and also makes it possible to request a stripped version of *any* binary
the build can produce with a .stripped suffix.
Change-Id: I2de82e0951d9f41c30594f32fba50acdd14ed69c
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48120
Reviewed-by: Bobby R. Bruce <bbruce@ucdavis.edu>
Maintainer: Bobby R. Bruce <bbruce@ucdavis.edu>
Tested-by: kokoro <noreply+kokoro@google.com>
When remote GDB attaches to gem5, handle the initial communication
(`qSupported` and alike) right away instead of scheduling a `DataEvent`
and firing the simulation loop in hope that GDB will be quick enough to
send initial packets before instructions are dispatched.
This requires attach() to be always called at instruction boundary
to make it safe to interact with the rest of gem5.
When `--wait-gdb` is used, connect() is called from workflow startup,
therefore on an instruction boundary and therefore needs not special
handling.
To handle case the GDB connects while simulation is already running,
we arrange (new) assynchronous IncommingConnectionEvent on listening
socket that, when there's a new connection being made, *only* schedules
*synchronous* ConnectEvent that handles the rest, *including* calling
an accept() on listening socket. This way it is safe to process commands
in attach().
In order to make the code more systematic and easier to understands,
detach() is also made to be called only synchronously (that is, at
intruction boundary). Asynchronous events and event handlers are
prefixed with "incoming".
This seems to fix the race described in 44612 [1]
[1]: https://gem5-review.googlesource.com/c/public/gem5/+/44612
Change-Id: I33b2922ba017205acabd51b6a8be3e6fb2d6409a
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48182
Reviewed-by: Boris Shingarov <shingarov@labware.com>
Maintainer: Boris Shingarov <shingarov@labware.com>
Tested-by: kokoro <noreply+kokoro@google.com>
At the moment the ITLBI and DTLBI operations where implicitly
implemented as generic TLBIs
e.g:
* DTLBIASID, ITLBIASID = TLBIASID
This was possible as a single TLB was either an instruction TLB
or a data TLB and no generic/shared TLB option was available.
In other words, an ITLBI Op to an ITB was invalidating all entries
as we were sure the ITB was not containing data entries.
With shared TLBs, this doesn't hold true and we need to explicitly
implement I/DTLBIs
JIRA: https://gem5.atlassian.net/browse/GEM5-790
Change-Id: I39a4add7674f6008dacaedfd1fd90560d264048e
Signed-off-by: Giacomo Travaglini <giacomo.travaglini@arm.com>
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48148
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Maintainer: Andreas Sandberg <andreas.sandberg@arm.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Ruby assumes protocols use directory controllers as memory interface.
Thus, recvAtomic() uses the machine type of directory when it calls
mapAddressToMachine(). However, it doesn't work for CHI since
CHI does not use directory controllers as memory controller interface.
Therefore, the code was modified to check which controller type is used
for memory interface between MachineType_Directory and
MachineType_Memory, which is used for CHI.
Change-Id: If35a06a8a3772ce5e5b994df05c9d94c7770c90d
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48403
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Now all occurances of the THE_ISA macro which were being used to check
for anything other than the NULL_ISA have been eliminated. We still need
to be able to check whether the current ISA is the null ISA, but we
don't want to let any preprocessor checks back in which are based on
what the current ISA is.
This change removes the THE_ISA macro, and replaces it with IS_NULL_ISA
which evaluates to 1 if the ISA is null, and 0 if it isn't.
Jira Issue: https://gem5.atlassian.net/browse/GEM5-1060
Change-Id: Iec146b40d8cab846dae03e15191390f754f2b71b
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/48709
Reviewed-by: Hoa Nguyen <hoanguyen@ucdavis.edu>
Maintainer: Gabe Black <gabe.black@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
