You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
microwatt/fetch1.vhdl

134 lines
3.8 KiB
VHDL

library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;
library work;
use work.common.all;
entity fetch1 is
generic(
RESET_ADDRESS : std_logic_vector(63 downto 0) := (others => '0')
);
port(
clk : in std_ulogic;
rst : in std_ulogic;
-- Control inputs:
stall_in : in std_ulogic;
flush_in : in std_ulogic;
stop_in : in std_ulogic;
-- redirect from execution unit
e_in : in Execute1ToFetch1Type;
-- Request to icache
i_out : out Fetch1ToIcacheType
);
end entity fetch1;
architecture behaviour of fetch1 is
type stop_state_t is (RUNNING, STOPPED, RESTARTING);
type reg_internal_t is record
stop_state: stop_state_t;
end record;
signal r, r_next : Fetch1ToIcacheType;
signal r_int, r_next_int : reg_internal_t;
begin
regs : process(clk)
begin
if rising_edge(clk) then
if r /= r_next then
report "fetch1 rst:" & std_ulogic'image(rst) &
Add TLB to icache This adds a direct-mapped TLB to the icache, with 64 entries by default. Execute1 now sends a "virt_mode" signal from MSR[IR] to fetch1 along with redirects to indicate whether instruction addresses should be translated through the TLB, and fetch1 sends that on to icache. Similarly a "priv_mode" signal is sent to indicate the privilege mode for instruction fetches. This means that changes to MSR[IR] or MSR[PR] don't take effect until the next redirect, meaning an isync, rfid, branch, etc. The icache uses a hash of the effective address (i.e. next instruction address) to index the TLB. The hash is an XOR of three fields of the address; with a 64-entry TLB, the fields are bits 12--17, 18--23 and 24--29 of the address. TLB invalidations simply invalidate the indexed TLB entry without checking the contents. If the icache detects a TLB miss with virt_mode=1, it will send a fetch_failed indication through fetch2 to decode1, which will turn it into a special OP_FETCH_FAILED opcode with unit=LDST. That will get sent down to loadstore1 which will currently just raise a Instruction Storage Interrupt (0x400) exception. One bit in the PTE obtained from the TLB is used to check whether an instruction access is allowed -- the privilege bit (bit 3). If bit 3 is 1 and priv_mode=0, then a fetch_failed indication is sent down to fetch2 and to decode1, which generates an OP_FETCH_FAILED. Any PTEs with PTE bit 0 (EAA[3]) clear or bit 8 (R) clear should not be put into the iTLB since such PTEs would not allow execution by any context. Tlbie operations get sent from mmu to icache over a new connection. Unfortunately the privileged instruction tests are broken for now. Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
5 years ago
" IR:" & std_ulogic'image(e_in.virt_mode) &
" P:" & std_ulogic'image(e_in.priv_mode) &
" R:" & std_ulogic'image(e_in.redirect) &
" S:" & std_ulogic'image(stall_in) &
" T:" & std_ulogic'image(stop_in) &
" nia:" & to_hstring(r_next.nia) &
" SM:" & std_ulogic'image(r_next.stop_mark);
end if;
r <= r_next;
r_int <= r_next_int;
end if;
end process;
comb : process(all)
variable v : Fetch1ToIcacheType;
variable v_int : reg_internal_t;
variable increment : boolean;
begin
v := r;
v_int := r_int;
if rst = '1' then
v.nia := RESET_ADDRESS;
Add TLB to icache This adds a direct-mapped TLB to the icache, with 64 entries by default. Execute1 now sends a "virt_mode" signal from MSR[IR] to fetch1 along with redirects to indicate whether instruction addresses should be translated through the TLB, and fetch1 sends that on to icache. Similarly a "priv_mode" signal is sent to indicate the privilege mode for instruction fetches. This means that changes to MSR[IR] or MSR[PR] don't take effect until the next redirect, meaning an isync, rfid, branch, etc. The icache uses a hash of the effective address (i.e. next instruction address) to index the TLB. The hash is an XOR of three fields of the address; with a 64-entry TLB, the fields are bits 12--17, 18--23 and 24--29 of the address. TLB invalidations simply invalidate the indexed TLB entry without checking the contents. If the icache detects a TLB miss with virt_mode=1, it will send a fetch_failed indication through fetch2 to decode1, which will turn it into a special OP_FETCH_FAILED opcode with unit=LDST. That will get sent down to loadstore1 which will currently just raise a Instruction Storage Interrupt (0x400) exception. One bit in the PTE obtained from the TLB is used to check whether an instruction access is allowed -- the privilege bit (bit 3). If bit 3 is 1 and priv_mode=0, then a fetch_failed indication is sent down to fetch2 and to decode1, which generates an OP_FETCH_FAILED. Any PTEs with PTE bit 0 (EAA[3]) clear or bit 8 (R) clear should not be put into the iTLB since such PTEs would not allow execution by any context. Tlbie operations get sent from mmu to icache over a new connection. Unfortunately the privileged instruction tests are broken for now. Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
5 years ago
v.virt_mode := '0';
v.priv_mode := '1';
v_int.stop_state := RUNNING;
elsif e_in.redirect = '1' then
v.nia := e_in.redirect_nia;
Add TLB to icache This adds a direct-mapped TLB to the icache, with 64 entries by default. Execute1 now sends a "virt_mode" signal from MSR[IR] to fetch1 along with redirects to indicate whether instruction addresses should be translated through the TLB, and fetch1 sends that on to icache. Similarly a "priv_mode" signal is sent to indicate the privilege mode for instruction fetches. This means that changes to MSR[IR] or MSR[PR] don't take effect until the next redirect, meaning an isync, rfid, branch, etc. The icache uses a hash of the effective address (i.e. next instruction address) to index the TLB. The hash is an XOR of three fields of the address; with a 64-entry TLB, the fields are bits 12--17, 18--23 and 24--29 of the address. TLB invalidations simply invalidate the indexed TLB entry without checking the contents. If the icache detects a TLB miss with virt_mode=1, it will send a fetch_failed indication through fetch2 to decode1, which will turn it into a special OP_FETCH_FAILED opcode with unit=LDST. That will get sent down to loadstore1 which will currently just raise a Instruction Storage Interrupt (0x400) exception. One bit in the PTE obtained from the TLB is used to check whether an instruction access is allowed -- the privilege bit (bit 3). If bit 3 is 1 and priv_mode=0, then a fetch_failed indication is sent down to fetch2 and to decode1, which generates an OP_FETCH_FAILED. Any PTEs with PTE bit 0 (EAA[3]) clear or bit 8 (R) clear should not be put into the iTLB since such PTEs would not allow execution by any context. Tlbie operations get sent from mmu to icache over a new connection. Unfortunately the privileged instruction tests are broken for now. Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
5 years ago
v.virt_mode := e_in.virt_mode;
v.priv_mode := e_in.priv_mode;
elsif stall_in = '0' then
-- For debug stop/step to work properly we need a little bit of
-- trickery here. If we just stop incrementing and send stop marks
-- when stop_in is set, then we'll increment on the cycle it clears
-- and end up never executing the instruction we were stopped on.
--
-- Avoid this along with the opposite issue when stepping (stop is
-- cleared for only one cycle) is handled by the state machine below
--
-- By default, increment addresses
increment := true;
case v_int.stop_state is
when RUNNING =>
-- If we are running and stop_in is set, then stop incrementing,
-- we are now stopped.
if stop_in = '1' then
increment := false;
v_int.stop_state := STOPPED;
end if;
when STOPPED =>
-- When stopped, never increment. If stop is cleared, go to state
-- "restarting" but still don't increment that cycle. stop_in is
-- now 0 so we'll send the NIA down without a stop mark.
increment := false;
if stop_in = '0' then
v_int.stop_state := RESTARTING;
end if;
when RESTARTING =>
-- We have just sent the NIA down, we can start incrementing again.
-- If stop_in is still not set, go back to running normally.
-- If stop_in is set again (that was a one-cycle "step"), go
-- back to "stopped" state which means we'll stop incrementing
-- on the next cycle. This ensures we increment the PC once after
-- sending one instruction without a stop mark. Since stop_in is
-- now set, the new PC will be sent with a stop mark and thus not
-- executed.
if stop_in = '0' then
v_int.stop_state := RUNNING;
else
v_int.stop_state := STOPPED;
end if;
end case;
if increment then
v.nia := std_logic_vector(unsigned(v.nia) + 4);
end if;
end if;
v.req := not rst;
v.stop_mark := stop_in;
r_next <= v;
r_next_int <= v_int;
-- Update outputs to the icache
i_out <= r;
end process;
end architecture behaviour;