Previously printing an mshr would trigger an assertion if the MSHR was
not in service or if the targets list was empty. This patch changes
the print function to bypasses the accessor functions for
postInvalidate and postDowngrade and avoid the relevant assertions. It
also checks if the targets list is empty before calling print on it.
Change-Id: Ic18bee6cb088f63976112eba40e89501237cfe62
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Secure and non-secure data can coexist in the cache and therefore the
snoop filter should treat differently packets with secure and non
secure accesses. This patch uses the lower bits of the line address to
keep track of whether the packet is addressing secure memory or not.
Change-Id: I54a5e614dad566a5083582bede86c86896f2c2c1
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Stephan Diestelhorst <stephan.diestelhorst@arm.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
This patch changes the default behaviour of the SystemXBar, adding a
snoop filter. With the recent updates to the snoop filter allocation
behaviour this change no longer causes problems for the regressions
without caches.
Change-Id: Ibe0cd437b71b2ede9002384126553679acc69cc1
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
This patch improves the snoop filter allocation decisions by not only
looking at whether a port is snooping or not, but also if the packet
actually came from a cache. The issue with only looking at isSnooping
is that the CPU ports, for example, are snooping, but not actually
caching. Previously we ended up incorrectly allocating entries in
systems without caches (such as the atomic and timing quick
regressions). Eventually these misguided allocations caused the snoop
filter to panic due to an excessive size.
On the request path we now include the fromCache check on the packet
itself, and for responses we check if we actually have a snoop-filter
entry.
Change-Id: Idd2dbc4f00c7e07d331e9a02658aee30d0350d7e
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Stephan Diestelhorst <stephan.diestelhorst@arm.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
This patch takes yet another step in maintaining the clusivity, in
that it allows a mostly-inclusive cache to hold on to blocks even when
responding to a ReadExReq or UpgradeReq. Previously the cache simply
invalidated these blocks, but there is no strict need to do so.
The most important part of this patch is that we simply mark the block
clean when satisfying the upstream request where the cache is allowed
to keep the block. The only tricky part of the patch is in the memory
management of deferred snoops, where we need to distinguish the cases
where only the packet was copied (we expected to respond), and the
cases where we created an entirely new packet and request (we kept it
only to replay later).
The code in satisfyRequest is definitely ready for some refactoring
after this.
Change-Id: I201ddc7b2582eaa46fb8cff0c7ad09e02d64b0fc
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
This patch changes how the mostly exclusive policy is enforced to
ensure that we drop blocks when we should. As part of this change, the
actual invalidation due to the clusivity enforcement is moved outside
the hit handling, to a separate method maintainClusivity. For the
timing mode that means we can deal with all MSHR targets before taking
any action and possibly dropping the block. The method
satisfyCpuSideRequest is also renamed satisfyRequest as part of this
change (since we only ever see requests from the cpu-side port).
Change-Id: If6f3d1e0c3e7be9a67b72a55e4fc2ec4a90fd3d2
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
This patch adds a FromCache attribute to the packet, and updates a
number of the existing request commands to reflect that the request
originates from a cache. The attribute simplifies checking if a
requests came from a cache or not, and this is used by both the cache
and snoop filter in follow-on patches.
Change-Id: Ib0a7a080bbe4d6036ddd84b46fd45bc7eb41cd8f
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Tony Gutierrez <anthony.gutierrez@amd.com>
Reviewed-by: Steve Reinhardt <stever@gmail.com>
There are cases where we want to put boot ROMs on the PIO bus. Ruby
currently doesn't support functional accesses to such memories since
functional accesses are always assumed to go to physical memory. Add
the required support for routing functional accesses to the PIO bus.
Change-Id: Ia5b0fcbe87b9642bfd6ff98a55f71909d1a804e3
Signed-off-by: Andreas Sandberg <andreas.sandberg@arm.com>
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Brad Beckmann <brad.beckmann@amd.com>
Reviewed-by: Michael LeBeane <michael.lebeane@amd.com>
Currently garnet will not run due to double statistic registration of new
stats in ClockedObject. This occurs because a temporary array named 'cls'
is being added as a child to garnet internal and external link SimObjects.
This patch simply renames the temporary array which prevents it from
being added as a child object and avoids the assertion that a statistic
was already registered.
Committed by Jason Lowe-Power <jason@lowepower.com>
Sync DRAMPower to external tool
This patch syncs the DRAMPower library of gem5 to the external
one on github (https://github.com/ravenrd/DRAMPower) of which
I am a maintainer.
The version used is the commit:
902a00a1797c48a9df97ec88868f20e847680ae6
from 07. May. 2016.
Committed by Jason Lowe-Power <jason@lowepower.com>
In this new hmc configuration we have used the existing components in gem5
mainly [SerialLink] [NoncoherentXbar]& [DRAMCtrl] to define 3 different
architecture for HMC.
Highlights
1- It explores 3 different HMC architectures
2- It creates 4-HMC crossbars and attaches 16 vault controllers with it.
This will connect vaults to serial links
3- From the previous version, HMCController with round robin funtionality
is being removed and all the serial links are being accessible directly
from user ports
4- Latency incorporated by HMCController (in previous version) is being
added to SerialLink
Committed by Jason Lowe-Power <jason@lowepower.com>
The snoop filter handles requests in two steps which preceed and
follow the call to send the packet downstream. An address mapper could
possibly change the address of the packet when it is sent downstream
breaking the snoop filter assumption that the address is unchanged
Change-Id: Ib2db755e9ebef4f2f7c0169a46b1b11185ffbe79
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
Fixing an issue with regStats not calling the parent class method
for most SimObjects in Gem5. This causes issues if one adds new
stats in the base class (since they are never initialized properly!).
Change-Id: Iebc5aa66f58816ef4295dc8e48a357558d76a77c
Reviewed-by: Andreas Sandberg <andreas.sandberg@arm.com>
We want to extend the stats of objects hierarchically and thus it is necessary
to register the statistics of the base-class(es), as well. For now, these are
empty, but generic stats will be added there.
Patch originally provided by Akash Bagdia at ARM Ltd.
This implements SwapReq for Ruby memory.
A SwapReq should be treated like a write, except that the response
packet contains the overwritten data.
Note that, in particular, the conditional checking for isStore/isLoad
needs to be reversed, as a SwapReq is both.
This patch fixes a memory leak where deferred snoop packets never got
deallocated. On the call to MSHR::handleSnoop these snoops were
treated as if a response will be sent, as the MSHR was
pendingModified. Consequently, a copy of the packet was created and
added to the MSHR targets. However, an preceeding target to the same
MSHR, originally from a CPU, was serviced before the snoop, and caused
the block to be invalidated. This happens for ReadExReq and
UpgradeReq.
Note that the original snoop will receive a response, just not from
the cache in question, but instead from the cache upstream that issued
the ReadExReq or UpgradeReq.
Change-Id: I4ac012fbc8a46cf693ca390fe9476105d444e6f4
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
This patch changes the flow control for HSHR::handleSnoop to ensure
that we only set cacheResponding on the snoop packet if we are
actually responding. This avoids situations where a responder is
stalling indefinitely on a response that never arrives.
Change-Id: I691dd01755b614b30203581aa74fc743b350eacc
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
This patch fixes the type of the unique_ptr instances, to ensure that
the data that is allocated with new[] is also deleted with
delete[]. The issue was highlighted by ASAN.
Change-Id: I2c5510424959d862a9954d83e728d901bb18d309
Reviewed-by: Nikos Nikoleris <nikos.nikoleris@arm.com>
Reviewed-by: Curtis Dunham <curtis.dunham@arm.com>
Reviewed-by: Stephan Diestelhorst <stephan.diestelhorst@arm.com>
Allow usage of packet class in ruby for convenience purposes. This may be
used to access members of the packet/request class (e.g., via helper
functions) and/or push protocol specific information to the packets
SenderState without needing to modify SLICC types and protocols in multiple
locations.
This patch removes the write-queue entry tracking previously used for
uncacheable writes. The write-queue entry is now deallocated as soon
as the packet is sent. As a result we also forego the stats for
uncacheable writes. Additionally, there is no longer a need to attach
the write-queue entry to the packet.
Ruby's controller block_on behavior aimed to block MessageBuffer requests into
SLICC controllers when a Locked_RMW was in flight. Unfortunately, this
functionality only partially works: When non-Locked_RMW memory accesses are
issued to the sequencer to an address with an in-flight Locked_RMW, the
sequencer may pass those accesses through to the controller. At the controller,
a number of incorrect activities can occur depending on the protocol. In
MOESI_hammer, for example, an intermediate IFETCH will cause an L1D to L2
transfer, which cannot be serviced, because the block_on functionality blocks
the trigger queue, resulting in a deadlock. Further, if an intermediate store
arrives (e.g. from a separate SMT thread), the sequencer allows the request
through to the controller, and the atomicity of the Locked_RMW may be broken.
To avoid these problems, disallow the Sequencer from passing any memory
accesses to the controller besides Locked_RMW_Write when a Locked_RMW is in-
flight.
A few warnings (and thus errors) pop up after being added to -Wall:
1. -Wmisleading-indentation
In the auto-generated code there were instances of if/else blocks that
were not indented to gcc's liking. This is addressed by adding braces.
2. -Wshift-negative-value
gcc is clever enougn to consider ~0 a negative constant, and
rightfully complains. This is addressed by using mask() which
explicitly casts to unsigned before shifting.
That is all. Porting done.
Queued prefetcher entries now count with a priority field. The idea is to
add packets ordered by priority and then by age.
For the existing algorithms in which priority doesn't make sense, it is set
to 0 for all deferred packets in the queue.
Some common functionality added to the base prefetcher, mainly dealing with
extracting the block address, page address, block index inside the page and
some other information that can be inferred from the block address. This is
used for some prefetching algorithms, and having the methods in the base,
as well as the block size and other information is the sensible way.
In general, the ThreadID parameter is unnecessary in the memory system
as the ContextID is what is used for the purposes of locks/wakeups.
Since we allocate sequential ContextIDs for each thread on MT-enabled
CPUs, ThreadID is unnecessary as the CPUs can identify the requesting
thread through sideband info (SenderState / LSQ entries) or ContextID
offset from the base ContextID for a cpu.
This is a re-spin of 20264eb after the revert (bd1c6789) and includes
some fixes of that commit.
The following patches had unexpected interactions with the current
upstream code and have been reverted for now:
e07fd01651f3: power: Add support for power models
831c7f2f9e39: power: Low-power idle power state for idle CPUs
4f749e00b667: power: Add power states to ClockedObject
Signed-off-by: Andreas Sandberg <andreas.sandberg@arm.com>
--HG--
extra : amend_source : 0b6fb073c6bbc24be533ec431eb51fbf1b269508
In general, the ThreadID parameter is unnecessary in the memory system
as the ContextID is what is used for the purposes of locks/wakeups.
Since we allocate sequential ContextIDs for each thread on MT-enabled
CPUs, ThreadID is unnecessary as the CPUs can identify the requesting
thread through sideband info (SenderState / LSQ entries) or ContextID
offset from the base ContextID for a cpu.
Add 4 power states to the ClockedObject, provides necessary access functions
to check and update the power state. Default power state is UNDEFINED, it is
responsibility of the respective simulation model to provide the startup state
and any other logic for state change.
Add number of transition stat.
Add distribution of time spent in clock gated state.
Add power state residency stat.
Add dump call back function to allow stats update of distribution and residency
stats.
The cache queue reserve is there as an overflow to give us enough
headroom based on when we block the cache, and how many transactions
we may already have accepted before actually blocking. The previous
values were probably chosen to be "big enough", when we actually know
that we check the MSHRs after every single allocation, and for the
write buffers we know that we implicitly may need one entry for every
outstanding MSHR.
* * *
mem: Adjust cache queue reserve to more conservative values
The cache queue reserve is there as an overflow to give us enough
headroom based on when we block the cache, and how many transactions
we may already have accepted before actually blocking. The previous
values were probably chosen to be "big enough", when we actually know
that we check the MSHRs after every single allocation, and for the
write buffers we know that we implicitly may need one entry for every
outstanding MSHR.
This patch breaks out the cache write buffer into a separate class,
without affecting any stats. The goal of the patch is to avoid
encumbering the much-simpler write queue with the complex MSHR
handling. In a follow on patch this simplification allows us to
implement write combining.
The WriteQueue gets its own class, but shares a common ancestor, the
generic Queue, with the MSHRQueue.
This patch adds assertions that enforce that only invalidating snoops
will ever reach into the logic that tracks in-order load completion and
also invalidation of LL/SC (and MONITOR / MWAIT) monitors. Also adds
some comments to MSHR::replaceUpgrades().
This patch fixes an issue where an InvalidationReq only traversed one
level of the cache hierarchy, and was subsequently turned into a
ReadExReq due to it needing writable, and the command not being
checked for explicitly.
This patch essentially rolls back 10518:30e3715c9405 to make RubyPort the
parent class of DMASequencer. It removes redundant code and restores some
features which were lost when directly inheriting from MemObject. For
example,
DMASequencer can now communicate to other devices using PIO, which is useful
for memmory-mapped communication between multiple DMADevices.
Avoid being overly conservative in clearing load locks in the cache,
and allow writes to the line if they are from the same context. This
is in line with ALPHA and ARM.
This patch introduces the ability of making the coherent crossbar the
point of coherency. If so, the crossbar does not forward packets where
a cache with ownership has already committed to responding, and also
does not forward any coherency-related packets that are not intended
for a downstream memory controller. Thus, invalidations and upgrades
are turned around in the crossbar, and the memory controller only sees
normal reads and writes.
In addition this patch moves the express snoop promotion of a packet
to the crossbar, thus allowing the downstream cache to check the
express snoop flag (as it should) for bypassing any blocking, rather
than relying on whether a cache is responding or not.
Adopt the same flow as in timing mode, where the caches on the path to
memory get to keep the line (if present), and we use the
responderHadWritable flag to determine if we need to forward the
(invalidating) packet or not.
This patch unifies the snoop handling in case of hitting writebacks
with how we handle snoops hitting in the tags. As a result, we end up
using the same optimisation as the normal snoops, where we inform the
downstream cache if we encounter a line in Modified (writable and
dirty) state, which enables us to avoid sending out express snoops to
invalidate any Shared copies of the line. A few regressions
consequently change, as some transactions are sunk higher up in the
cache hierarchy.
This patch changes how the cache determines if snoops should be
forwarded from the memory side to the CPU side. Instead of having a
parameter, the cache now looks at the port connected on the CPU side,
and if it is a snooping port, then snoops are forwarded. Less error
prone, and less parameters to worry about.
The patch also tidies up the CPU classes to ensure that their I-side
port is not snooping by removing overrides to the snoop request
handler, such that snoop requests will panic via the default
MasterPort implement
Result of running 'hg m5style --skip-all --fix-control -a' to get
rid of '== true' comparisons, plus trivial manual edits to get
rid of '== false'/'== False' comparisons.
Left a couple of explicit comparisons in where they didn't seem
unreasonable:
invalid boolean comparison in src/arch/mips/interrupts.cc:155
>> DPRINTF(Interrupt, "Interrupts OnCpuTimerINterrupt(tc) == true\n");<<
invalid boolean comparison in src/unittest/unittest.hh:110
>> "EXPECT_FALSE(" #expr ")", (expr) == false)<<
mem: support for gpu-style RMWs in ruby
This patch adds support for GPU-style read-modify-write (RMW) operations in
ruby. Such atomic operations are traditionally executed at the memory controller
(instead of through an L1 cache using cache-line locking).
Currently, this patch works by propogating operation functors through the memory
system.
The new Packet::setRaw() method incorrectly still contained
an htog() conversion. As a result, calls to the old set()
method (now defined as setRaw(htog(v))) underwent two htog
conversions, which breaks things when htog() is not a no-op.
Interestingly the only test that caught this was a SPARC
boot test, where an IsaFake device with a non-zero return
value was getting swapped twice resulting in a register
getting loaded with 0x100000000000000 instead of 1.
(Good reason for keeping SPARC around, perhaps?)
Some of the DPRINTFs added to the classic cache in cset 45df88079f04,
while useful to those unfamiliar with the cache code, end up being
noise when you're familiar with the code but are trying to debug tricky
protocol issues. (Particularly getting two messages from each cache
as it receives a snoop request then declares that there was no match.)
This patch introduces a CacheVerbose debug flag, and moves a subset of
the added DPRINTFs into that category, so that Cache by itself returns
to being a more succinct summary of cache activity.
Also added a CacheAll compound flag to turn on all the cache-related
debug flags (other than CacheTags, which you *really* have to want badly
to turn it on, IMO).
This patch removes the NeedsWritable flag for all responses, as it is
really only the request that needs a writable response. The response,
on the other hand, should in these cases always provide the line in a
writable state, as indicated by the hasSharers flag not being set.
When we send requests that has NeedsWritable set, the response will
always have the hasSharers flag not set. Additionally, there are cases
where the request did not have NeedsWritable set, and we still get a
writable response with the hasSharers flag not set. This never happens
on snoops, but is used by downstream caches to pass ownership
upstream.
As part of this patch, the affected response types are updated, and
the snoop filter is similarly modified to check only the hasSharers
flag (as it should). A sanity check is also added to the packet class,
asserting that we never look at the NeedsWritable flag for responses.
No regressions are affected.
This patch looks at the request and response command to determine if
either actually has any data payload, and if not, we do not allocate
any space for packet data.
The only tricky case is where the command type is changed as part of
the MSHR functionality. In these cases where the original packet had
no data, but the new packet does, we need to explicitly call
allocate().
This patch ensures we do not respond with a Modified (dirty and
writable) line if the request is uncacheable, and that the cache
responding retains the line without modifying the state (even if
responding).
This patch changes the name of a bunch of packet flags and MSHR member
functions and variables to make the coherency protocol easier to
understand. In addition the patch adds and updates lots of
descriptions, explicitly spelling out assumptions.
The following name changes are made:
* the packet memInhibit flag is renamed to cacheResponding
* the packet sharedAsserted flag is renamed to hasSharers
* the packet NeedsExclusive attribute is renamed to NeedsWritable
* the packet isSupplyExclusive is renamed responderHadWritable
* the MSHR pendingDirty is renamed to pendingModified
The cache states, Modified, Owned, Exclusive, Shared are also called
out in the cache and MSHR code to make it easier to understand.
This patch is imported from reviewboard patch 2551 by Nilay.
This patch moves from a dynamically defined MachineType to a statically
defined one. The need for this patch was felt since a dynamically defined
type prevents us from having types for which no machine definition may
exist.
The following changes have been made:
i. each machine definition now uses a type from the MachineType enumeration
instead of any random identifier. This required changing the grammar and the
*.sm files.
ii. MachineType enumeration defined statically in RubySlicc_Exports.sm.
* * *
normal protocol fixes for nilay's parser machine type fix
This patch is imported from reviewboard patch 2550 by Nilay.
It was possible to specify multiple machine types with a single state machine.
This seems unnecessary and is being removed.
Add a sanity check to make it explicit that we currently do not allow
an I/O coherent agent to directly issue writes into the coherent part
of the memory system (it has to go via a cache, and get transformed
into a read ex, upgrade or invalidation).
This patch changes how the cache tracks which snoops are forwarded,
and which ones are created locally. Previously the identification was
based on an empty sender state of a specific class, but this method
fails to distinguish which cache actually attached the sender
state. Instead we use the same mechanism as the crossbar, and keep
track of the requests that have outstanding snoops.
This patch addresses a bug in how the cache attached the MSHR as a
sender state. Rather than overwriting any existing sender state it now
pushes a new one. The handling of upward snoops is also clarified.
This patch fixes a corner case in the deferred snoop handling, where
requests ended up being used by multiple packets with different
lifetimes, and inadvertently got deleted while they were still in use.
This patch allows the ruby random tester to use ruby ports that may only
support instr or data requests. This patch is similar to a previous changeset
(8932:1b2c17565ac8) that was unfortunately broken by subsequent changesets.
This current patch implements the support in a more straight-forward way.
Since retries are now tested when running the ruby random tester, this patch
splits up the retry and drain check behavior so that RubyPort children, such
as the GPUCoalescer, can perform those operations correctly without having to
duplicate code. Finally, the patch also includes better DPRINTFs for
debugging the tester.
This patch adds support to optionally capture the virtual address and asid
for load/store instructions in the elastic traces. If they are present in
the traces, Trace CPU will set those fields of the request during replay.
the sanity check, while generally useful for exposing memory system bugs,
may be spurious with respect to GPU workloads, which may generate many more
requests than typical CPU workloads. the large number of requests generated
by the GPU may cause the req/resp queues to back up, thus queueing more than
100 packets.
This patch adds the necessary commands and cache functionality to
allow clean writebacks. This functionality is crucial, especially when
having exclusive (victim) caches. For example, if read-only L1
instruction caches are not sending clean writebacks, there will never
be any spills from the L1 to the L2. At the moment the cache model
defaults to not sending clean writebacks, and this should possibly be
re-evaluated.
The implementation of clean writebacks relies on a new packet command
WritebackClean, which acts much like a Writeback (renamed
WritebackDirty), and also much like a CleanEvict. On eviction of a
clean block the cache either sends a clean evict, or a clean
writeback, and if any copies are still cached upstream the clean
evict/writeback is dropped. Similarly, if a clean evict/writeback
reaches a cache where there are outstanding MSHRs for the block, the
packet is dropped. In the typical case though, the clean writeback
allocates a block in the downstream cache, and marks it writable if
the evicted block was writable.
The patch changes the O3_ARM_v7a L1 cache configuration and the
default L1 caches in config/common/Caches.py
This patch adds a parameter to control the cache clusivity, that is if
the cache is mostly inclusive or exclusive. At the moment there is no
intention to support strict policies, and thus the options are: 1)
mostly inclusive, or 2) mostly exclusive.
The choice of policy guides the behaviuor on a cache fill, and a new
helper function, allocOnFill, is created to encapsulate the decision
making process. For the timing mode, the decision is annotated on the
MSHR on sending out the downstream packet, and in atomic we directly
pass the decision to handleFill. We (ab)use the tempBlock in cases
where we are not allocating on fill, leaving the rest of the cache
unaffected. Simple and effective.
This patch also makes it more explicit that multiple caches are
allowed to consider a block writable (this is the case
also before this patch). That is, for a mostly inclusive cache,
multiple caches upstream may also consider the block exclusive. The
caches considering the block writable/exclusive all appear along the
same path to memory, and from a coherency protocol point of view it
works due to the fact that we always snoop upwards in zero time before
querying any downstream cache.
Note that this patch does not introduce clean writebacks. Thus, for
clean lines we are essentially removing a cache level if it is made
mostly exclusive. For example, lines from the read-only L1 instruction
cache or table-walker cache are always clean, and simply get dropped
rather than being passed to the L2. If the L2 is mostly exclusive and
does not allocate on fill it will thus never hold the line. A follow
on patch adds the clean writebacks.
The patch changes the L2 of the O3_ARM_v7a CPU configuration to be
mostly exclusive (and stats are affected accordingly).
This patch optimises the handling of writebacks and clean evictions
when using a snoop filter. Instead of snooping into the caches to
determine if the block is cached or not, simply set the status based
on the snoop-filter result.
Instead of conservatively enforcing order for all packets, which may
negatively impact the simulated-system performance, this patch updates
the packet queue such that it only applies the restriction if there
are already packets with the same address in the queue.
The basic need for the order enforcement is due to coherency
interactions where requests/responses to the same cache line must not
over-take each other. We rely on the fact that any packet that needs
order enforcement will have a block-aligned address. Thus, there is no
need for the queue to know about the cacheline size.
This patch enforces insertion order transmission of packets on the
response path in the cache. Note that the logic to enforce order is
already present in the packet queue, this patch simply turns it on for
queues in the response path.
Without this patch, there are corner cases where a request-response is
faster than a response-response forwarded through the cache. This
violation of queuing order causes problems in the snoop filter leaving
it with inaccurate information. This causes assert failures in the
snoop filter later on.
A follow on patch relaxes the order enforcement in the packet queue to
limit the performance impact.
This patch updates the I/O devices, bridge and simple memory to take
the packet header and payload delay into account in their latency
calculations. In all cases we add the header delay, i.e. the
accumulated pipeline delay of any crossbars, and the payload delay
needed for deserialisation of any payload.
Due to the additional unknown latency contribution, the packet queue
of the simple memory is changed to use insertion sorting based on the
time stamp. Moreover, since the memory hands out exclusive (non
shared) responses, we also need to ensure ordering for reads to the
same address.
This patch aligns how the memory-system slaves, i.e. the various
memory controllers and the bridge, identify and deal with sinking of
inhibited packets that are only useful within the coherent part of the
memory system.
In the future we could shift the onus to the crossbar, and add a
parameter "is_point_of_coherence" that would allow it to sink the
aforementioned packets.
This patch changes the CleanEvict command type to not be considered a
write. Initially it was made a zero-sized write to match the writeback
command, but as things developed it became clear that it causes more
problems than it solves. For example, the memory modules (and bridge)
should not consider the CleanEvict as a write, but instead discard
it. With this patch it will be neither a read, nor write, and as it
does not need a response the slave will simply sink it.
This patch unifies how we deal with delayed packet deletion, where the
receiving slave is responsible for deleting the packet, but the
sending agent (e.g. a cache) is still relying on the pointer until the
call to sendTimingReq completes. Previously we used a mix of a
deletion vector and a construct using unique_ptr. With this patch we
ensure all slaves use the latter approach.