Until now, the number of L1 client channels was set statically in the
configuration. This static configuration also assumed the same number of
cached and uncached channels per tile. As we plan to move towards
heterogenous multicore systems, this restriction should be removed.
This commit changes the generator so that number of channels per tile
can be independently set (using cde.Parameters.alterPartial).
The OuterMemorySystem will dynamically compute the number of cached and
uncached channels by summing the number of each kind of channel per core.
This ensures that assertion failures, which currently print a message to
the console but return a successful exit code, now will cause non-zero
exit code. This is meant to help automated tools like travis and
buildbot do a better job at catching assertions.
This impacts the various run-* targets in the simulation
directories.
Until recently, we were assuming that the data channel in AXI was always
right-justified. However, for narrow writes, the data must actually be
aligned within the byte lanes. This commit changes some of the
converters in order to fix this issue.
There was a bug in the L2 cache in which a merged get request was
causing the tracker to read the old data from the data array,
overwriting the updated data acquired from outer memory. Changed it so
that pending_reads is no longer set if the data in the buffer is already
valid.
There was a bug in the PortedTileLinkCrossbar. The new GrantFromSrc and
FinishToDst types used client_id for routing to managers. This caused
bits to get cut off, which meant the Finish messages could not be routed
correctly. Changed to use manager_id instead.
We are planning on switching to a TileLink interconnect throughout and
convert to AXI only on the very edge. Therefore, we need to get rid of
all the existing AXI masters other than the TileLink to AXI converter.
* Get rid of DMA engine for now
* Connect RTC to TileLink interconnect instead of AXI interconnect
This allows users to specify if they want to build RocketChip against
Chisel 2 or 3. Since Chisel 3 is now open source we can add these
submodule pointers directly to avoid a fork of upstream.
It turns out the Chisel C++ backend can't emit correct initialization
code for a 128 bit wide NastiROM. Rather than trying to fix Chisel, I'm
just going to hack up the backup memory port Verilog harness a bit more
to make it work.
Note that the backup memory port Verilog already couldn't take arbitrary
parameters for MIF_*, so it's not like we're losing any flexibility
here.
The backup memory port doesn't work on multi-channel configurations, it
just screws up the Nasti tag bits. This patch always instantiates a
single-channel backup memory port, which relies on the memory channel
selector to only enable a single memory channel when the backup memory
port is enabled. There are some assertions to make sure this happens,
as otherwise memory gets silently corrupted.
While this is a bit of a hack, the backup memory port will be going away
soon so I don't want to spend a whole lot of time fixing it. The
generated hardware is actually very similar: we used to elaborate a
Nasti arbiter inside the backup memory support, now there's one outside
of it instead.
We need this to work for our chip, and it's not been tested in a long
time in upstream -- it didn't even used to build since the Nasti
conversion. This makes a few changes:
* Rather than calling the backup memory port parameters MEM_*, it calls
them MIF_* (to match the MIT* paramater objects). A new name was
necessary because the Nasti stuff is now dumped as MEM_*, which has
similar names but incompatible values.
* p(MIFDataBits) was changed back to 128, as otherwise the backup
memory port doesn't work (it only send half a TileLink transaction).
64 also causes readmemh to bail out, but changing the elf2hex parameters
works around that.
* A configuration was added that enabled the backup memory port in the
tester. While this is kind of an awkward way to do it, I want to
make sure I can start testing this regularly and this makes it easy to
integrate.
We're building a chip with 8 memory channels. Since this will require a
complicated test setup we want to also be able to bring up the chip with fewer
memory channels. This commit adds a SCR that controls the number of active
memory channels on a chip. Toggling this SCR will scramble memory and drop
Nasti messages, so it's only possible to change while the chip is booting.
By default this just adds a 1-bit SCR, which essentially no extra logic.
When multiple memory channel configurations are enabled at elaboration time, a
NastiMemoryInterconnect is generated for each channel configuration. The
number of outstanding misses is increased to coorespond to the maximum number
of banks per memory channel (added as a parameter), which I believe is
necessary to avoid deadlock in the memory system.
A configuration is added that supports 8 memory channels but has only 1 enabled
by default.
This uses the new SCRFile changes to generate a header file containing a list
of all the SCRs in a core to remove the magic constant "63" (the HTIF clock
divider control register) and replace it with a generated number (which is
still 63).
* DMA frontend-backend communication tunneled over TileLink/AXI
* Split MMIO and Mem requests in l1tol2net instead of in AXI interconnect
* Don't make NIOMSHRs configurable. Fixed at 1.
* Connect accelerator-internal CSRs into the CSR file
* Make mtvec register configurable and writeable
In addition to the fix, there are several additions to the
RegressionTest module. The set of regressions is now parameterized and
split into ones for the cache and ones for the broadcast hub.
