So that the dcache could in future be used by an MMU, this moves
logic to do with data formatting, rA updates for update-form
instructions, and handling of unaligned loads and stores out of
dcache and into loadstore1. For now, dcache connects only to
loadstore1, and loadstore1 now has the connection to writeback.
Dcache generates a stall signal to loadstore1 which indicates that
the request presented in the current cycle was not accepted and
should be presented again. However, loadstore1 doesn't currently
use it because we know that we can never hit the circumstances
where it might be set.
For unaligned transfers, loadstore1 generates two requests to
dcache back-to-back, and then waits to see two acks back from
dcache (cycles where d_in.valid is true).
Loadstore1 now has a FSM for tracking how many acks we are
expecting from dcache and for doing the rA update cycles when
necessary. Handling for reservations and conditional stores is
still in dcache.
Loadstore1 now generates its own stall signal back to decode2,
so we no longer need the logic in execute1 that generated the stall
for the first two cycles.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
Currently we don't get the result from a load that hits in the dcache
until the fourth cycle after the instruction was presented to
loadstore1. This trims this back to 3 cycles by taking the low order
bits of the address generated in loadstore1 into dcache directly (not
via the output register of loadstore1) and using them to address the
read port of the dcache data RAM. We use the lower 12 address bits
here in the expectation that any reasonable data cache design will
have a set size of 4kB or less in order to avoid the aliasing problems
that can arise with a virtually-indexed physically-tagged cache if
the set size is greater than the smallest page size provided by the
MMU.
With this we can get rid of r2 and drive the signals going to
writeback from r1, since the load hit data is now available one
cycle earlier. We need a multiplexer on the read address of the
data cache RAM in order to handle the second doubleword of an
unaligned access.
One small complication is that we now need an extra cycle in the case
of an unaligned load which misses in the data cache and which reads
the 2nd-last and last doublewords of a cache line. This is the reason
for the PRE_NEXT_DWORD state; if we just go straight to NEXT_DWORD
then we end up having the write of the last doubleword of the cache
line and the read of that same doubleword occurring in the same
cycle, which means we read stale data rather than the just-fetched
data.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This involves plumbing the (existing) 'reserve' and 'rc' bits in
the decode tables down to dcache, and 'rc' and 'store_done' bits
from dcache to writeback.
It turns out that we had 'RC' set in the 'rc' column for several
ordinary stores and for the attn instruction. This corrects them
to 'NONE', and sets the 'rc' column to 'ONE' for the conditional
stores.
In writeback we now have logic to set CR0 when the input from dcache
has rc = 1.
In dcache we have the reservation itself, which has a valid bit
and the address down to cache line granularity. We don't currently
store the reservation length. For a store conditional which fails,
we set a 'cancel_store' signal which inhibits the write to the
cache and prevents the state machine from starting a bus cycle or
going to the STORE_WAIT_ACK state. Instead we set r1.stcx_fail
which causes the instruction to complete in the next cycle with
rc=1 and store_done=0.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This allows us to use the bypass at the input of execute1 for the
address and data operands for loadstore1.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This enables back-to-back execution of integer instructions where
the first instruction writes a GPR and the second reads the same
GPR. This is done with a set of multiplexers at the start of
execute1 which enable any of the three input operands to be taken
from the output of execute1 (i.e. r.e.write_data) rather than the
input from decode2 (i.e. e_in.read_data[123]).
This also requires changes to the hazard detection and handling.
Decode2 generates a signal indicating that the GPR being written
is available for bypass, which is true for instructions that are
executed in execute1 (rather than loadstore1/dcache). The
gpr_hazard module stores this "bypassable" bit, and if the same
GPR needs to be read by a subsequent instruction, it outputs a
"use_bypass" signal rather than generating a stall. The
use_bypass signal is then latched at the output of decode2 and
passed down to execute1 to control the input multiplexer.
At the moment there is no bypass on the inputs to loadstore1, but that
is OK because all load and store instructions are marked as
single-issue.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This moves the sign extension done by the extsb, extsh and extsw
instructions back into execute1. This means that we no longer need
any data formatting in writeback for results coming from execute1,
so this modifies writeback so the data formatter inputs come
directly from the loadstore unit output. The condition code
updates for RC=1 form instructions are now done on the value from
execute1 rather than the output of the data formatter, which should
help timing.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
For multiply and divide operations, execute1 now records the
destination GPR number, RC and OE from the instruction, and the
XER value. This means that the multiply and divide units don't
need to record those values and then send them back to execute1.
This makes the interface to those units a bit simpler. They
simply report an overflow signal along with the result value, and
execute1 takes care of updating XER if necessary.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
With this, the divider is a unit that execute1 sends operands to and
which sends its results back to execute1, which then send them to
writeback. Execute1 now sends a stall signal when it gets a divide
or modulus instruction until it gets a valid signal back from the
divider. Divide and modulus instructions are no longer marked as
single-issue.
The data formatting step that used to be done in decode2 for div
and mod instructions is now done in execute1. We also do the
absolute value operation in that same cycle instead of taking an
extra cycle inside the divider for signed operations with a
negative operand.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
With this, the multiplier isn't a separate pipe that decode2 issues
instructions to, but rather is a unit that execute1 sends operands
to and which sends the result back to execute1, which then sends it
to writeback. Execute1 now sends a stall signal when it gets a
multiply instruction until it gets a valid signal back from the
multiplier.
This all means that we no longer need to mark the multiply
instructions as single-issue.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This stores the most common SPRs in the register file.
This includes CTR and LR and a not yet final list of others.
The register file is set to 64 entries for now. Specific types
are defined that can represent a GPR index (gpr_index_t) or
a GPR/SPR index (gspr_index_t) along with conversion functions
between the two.
On order to deal with some forms of branch updating both LR and
CTR, we introduced a delayed update of LR after a branch link.
Note: We currently stall the pipeline on such a delayed branch,
but we could avoid stalling fetch in that specific case as we
know we have a branch delay. We could also limit that to the
specific case where we need to update both CTR and LR.
This allows us to make bcreg, mtspr and mfspr pipelined. decode1
will automatically force the single issue flag on mfspr/mtspr to
a "slow" SPR.
[paulus@ozlabs.org - fix direction of decode2.stall_in]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
We will want to store some SPRs in the register file using
a set of "extra" registers. This provides a function for
doing the translation along with some SPR definitions.
This isn't used yet
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
The carry is currently internal to execute1. We don't handle any of
the other XER fields.
This creates type called "xer_common_t" that contains the commonly
used XER bits (CA, CA32, SO, OV, OV32).
The value is stored in the CR file (though it could be a separate
module). The rest of the bits will be implemented as a separate
SPR and the two parts reconciled in mfspr/mtspr in latter commits.
We always read XER in decode2 (there is little point not to)
and send it down all pipeline branches as it will be needed in
writeback for all type of instructions when CR0:SO needs to be
updated (such forms exist for all pipeline branches even if we don't
yet implement them).
To avoid having to track XER hazards, we forward it back in EX1. This
assumes that other pipeline branches that can modify it (mult and div)
are running single issue for now.
One additional hazard to beware of is an XER:SO modifying instruction
in EX1 followed immediately by a store conditional. Due to our writeback
latency, the store will go down the LSU with the previous XER value,
thus the stcx. will set CR0:SO using an obsolete SO value.
I doubt there exist any code relying on this behaviour being correct
but we should account for it regardless, possibly by ensuring that
stcx. remain single issue initially, or later by adding some minimal
tracking or moving the LSU into the same pipeline as execute.
Missing some obscure XER affecting instructions like addex or mcrxrx.
[paulus@ozlabs.org - fix CA32 and OV32 for OP_ADD, fix order of
arguments to set_ov]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
Use a function to obtain the integer number and use constants
with the architected numbers. Replace std_match with a case
statement.
This also has the side effect of returning 0 instead of some
random previous result on mfspr of an unknown SPR.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This replaces loadstore2 with a dcache
The dcache unit is losely based on the icache one (same basic cache
layout), but has some significant logic additions to deal with stores,
loads with update, non-cachable accesses and other differences due to
operating in the execution part of the pipeline rather than the fetch
part.
The cache is store-through, though a hit with an existing line will
update the line rather than invalidate it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Since the condition setting got moved to writeback, execute2 does
nothing aside from wasting a cycle. This removes it.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This makes the exts[bhw] instructions do the sign extension in the
writeback stage using the sign-extension logic there instead of
having unique sign extension logic in execute1. This requires
passing the data length and sign extend flag from decode2 down
through execute1 and execute2 and into writeback. As a side bonus
we reduce the number of values in insn_type_t by two.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This adds code to writeback to format data and test the result
against zero for the purpose of setting CR0. The data formatter
is able to shift and mask by bytes and do byte reversal and sign
extension. It can also put together bytes from two input
doublewords to support unaligned loads (including unaligned
byte-reversed loads).
The data formatter starts with an 8:1 multiplexer that is able
to direct any byte of the input to any byte of the output. This
lets us rotate the data and simultaneously byte-reverse it.
The rotated/reversed data goes to a register for the unaligned
cases that overlap two doublewords. Then there is per-byte logic
that does trimming, sign extension, and splicing together bytes
from a previous input doubleword (stored in data_latched) and the
current doubleword. Finally the 64-bit result is tested to set
CR0 if rc = 1.
This removes the RC logic from the execute2, multiply and divide
units, and the shift/mask/byte-reverse/sign-extend logic from
loadstore2.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
Consolidate and/andc/nand, or/orc/nor and xor/eqv, using a common
invert on the input and output. This saves us about 200 LUTs.
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
The goal is to have the icache fit in BRAM by latching the output
into a register. In order to avoid timing issues , we need to give
the BRAM a full cycle on reads, and thus we souce the BRAM address
directly from fetch1 latched NIA.
(Note: This will be problematic if/when we want to hash the address,
we'll probably be better off having fetch1 latch a fully hashed address
along with the normal one, so the icache can use the former to address
the BRAM and pass the latter along)
One difficulty is that we cannot really stall the icache without adding
more combo logic that would break the "one full cycle" BRAM model. This
means that on stalls from decode, by the time we stall fetch1, it has
already gone to the next address, which the icache is already latching.
We work around this by having a "stash" buffer in fetch2 that will stash
away the icache output on a stall, and override the output of the icache
with the content of the stash buffer when unstalling.
This requires a rewrite of the stop/step debug logic as well. We now
do most of the hard work in fetch1 which makes more sense.
Note: Vivado is still not inferring an built-in output register for the
BRAMs. I don't want to add another cycle... I don't fully understand why
it wouldn't be able to treat current_row as such but clearly it won't. At
least the timing seems good enough now for 100Mhz, possibly more.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This changes the names of the mul_32bit and mul_signed fields of
decode_rom_t to is_32bit and is_signed, so they can be used with
other types of operations besides multiplies.
This plumbs the is_32bit and is_signed flags down into execute1,
though they are not used at this point.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This aims to simplify the logic between the instruction image and
the register file read address ports and reduce the size of the decode
tables. With this patch, the input_reg_a column of the decode tables
can only select RA or zeroes, the input_reg_b column can only select
RB or a constant (0, -1, or an immediate value from the instruction),
and the input_reg_c columns can only select RS or zeroes.
That means that the rotate/shift/logical ops now have their first
input coming in via the input_reg_c column. That means we need to
add a read_data3 field to the Decode2ToExecuteType record, but that
will go away again when we split out the rotate/mask/logical ops to
their own unit.
As a related but not tightly connected change, this patch also sets
the read1_enable signal to the register file be 0 when RA=0 and the
input_reg_a for the instruction is RA_OR_ZERO (previously it was 1).
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
All of the PPC add and subtract instructions, including carrying
and extended versions, do much the same arithmetic operation:
result = (I xor A) + B + C
where A is the value from RA, I provides a logical inversion of A
(i.e. I is 0 or -1), B is either from RB or is a constant 0 or -1,
and C is 0, 1 or the carry bit from XER (CA).
To consolidate all the add/subtract instructions into a single
OP_ADD, we add a column to decode_rom_t to indicate when A should
be inverted, and change the input_carry field to a 3-state selector
to select C in the equation above.
This also adds a new "CONST_M1" value for input_reg_b_t to indicate
that B is a constant -1. This allows us to implement addme and
subfme.
The addex instruction appears not to exist, so the comments referring
to it are removed.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
The register file is currently implemented as a whole pile of individual
1-bit registers instead of LUT memory which is a huge waste of FPGA
space.
This is caused by the output signal exposing the register file to the
outside world for simulation debug.
This removes that output, and moves the dumping of the register file
to the register file module itself. This saves about 8% of fpga on
the little Arty A7-35T.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The const* fields of decode_rom_t drove multiplexers in decode2 that
picked out various instruction fields and put them into the const*
fields of the Decode2ToExecute1Type record, from where they were
used in execute1. However, the code in execute1 can just as easily
use the appropriate fields of the original instruction word, since
that is now available in execute1. This therefore changes the
code to do that, resulting in smaller decode tables.
Suggested-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
Instead of doing mfctr, mflr, mftb, mtctr, mtlr as separate ops,
just pass down mfspr and mtspr ops with the spr number and let
execute1 decode which SPR we're addressing. This will help reduce
the number of instruction bits decode1 needs to look at.
In fact we now pass down the whole instruction from decode2 to
execute1. We will need more bits of the instruction in future,
and the tools should just optimize away any that we don't end
up using. Since the 'aa' bit was just a copy of an instruction
bit, we can now remove it from the record.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This moves the negation of negative operands for signed divide and
modulus operations out of the decode2 stage and into the divider.
If either of the operands for a signed divide or modulus operation
is negative, the divider now takes an extra cycle to negate the
operands that are negative.
The interface to the divider now has an 'is_signed' signal rather
than a 'neg_result' signal, and the dividend and divisor can be
negative, so divider_tb had to be updated for the new interface.
The reason for doing this is that one of the worst timing violations
on the Arty A7-100 at 100MHz involved the carry chain in the adders
that did the negation of the dividend and divisor in the decode stage.
Moving the negations to a separate cycle fixes that and also seems to
reduce the total number of slice LUTs used.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This adds a divider unit, connected to the core in much the same way
that the multiplier unit is connected. The division algorithm is
very simple-minded, taking 64 clock cycles for any division (even
32-bit division instructions).
The decoding is simplified by making use of regularities in the
instruction encoding for div* and mod* instructions. Instead of
having PPC_* encodings from the first-stage decoder for each of the
different div* and mod* instructions, we now just have PPC_DIV and
PPC_MOD, and the inputs to the divider that indicate what sort of
division operation to do are derived from instruction word bits.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
This module adds some simple core controls:
reset, stop, start, step
along with icache clear and reading the NIA and core
status bits
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org
We only need two write ports for load with update instructions.
Having two write ports just for this instruction is expensive.
For now we will force them to be the only instruction in the
pipeline, and take two cycles of writeback.
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
It simulated fine, but didn't synthesize. Fix some obvious issues
to get us going again.
Fixes: 9fbaea6f08 ("Rework CR file and add forwarding")
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
Right now we continually print all 3 possible GPRs an instruction
may be using. Add signals so we only print GPRs when they are
actually read. This should hopefully optimise away when synthesized.
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
Handle the CR as a single field with per nibble enables. Forward any
writes in the same cycle.
If this proves to be an issue for timing, we may want to revisit
this in the future. For now, it keeps things simple.
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
The decode2 stage was spaghetti code and needed cleaning up.
Create a series of functions to pull fields from a ppc instruction
and also a series of helpers to extract values for the execution
units.
As suggested by Paul, we should pass all signals to the execution
units and only set the valid signal conditionally, which should
use less resources.
Signed-off-by: Anton Blanchard <anton@linux.ibm.com>
Paul Mackerras
Anton Blanchard