* [libre-riscv-dev] [Bug 353] formal proof of soc.regfile classes RegFile and RegFileArray needed
From: bugzilla-daemon @ 2020-06-01 17:51 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-353-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=353
--- Comment #31 from Michael Nolan <mtnolan2640@gmail.com> ---
(In reply to Michael Nolan from comment #30)
> I'm taking a look at regfile.py as well as the proof. Part of the reason you
> see everything failing is that the truth table in the proof isn't right; the
> register file doesn't have a defined output when read enable is low. I have
> an idea as to how to prove that register writes work, would you like me to
> write it or leave it up to you?
Actually, it turned out that the register file wasn't generating read and write
ports.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 353] formal proof of soc.regfile classes RegFile and RegFileArray needed
From: bugzilla-daemon @ 2020-06-01 17:06 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-353-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=353
--- Comment #30 from Michael Nolan <mtnolan2640@gmail.com> ---
(In reply to Cole Poirier from comment #26)
> (In reply to Luke Kenneth Casson Leighton from comment #25)
> > (In reply to Cole Poirier from comment #24)
> >
> > > Luke, is the use of Settle() warranted in the synchronous part of this proof?
> >
> > not at all.
> >
> > l.
>
> Thanks, this helps clarify a little bit more for me the use and
> functionality of Settle().
>
> I think the proof as it stands may need a re-org... I tried filling in the
> quadrants according the to 4-way truth table comment draft that I did, which
> I saw you fixed. But nothing passes but the same Assert you added when you
> fixed the truth table a few days ago (this is actually the else condition in
> the code, but modified here to If for clarity):
>
> ```
> with m.If(not Past(wp.wen)):
> # if wen not set, reg should not change
> comb += Assert(reg == Past(reg))
> ```
>
> This leads me to believe that either the proof is wrong and needs to be
> reorganized to conform to how Register is implemented in regfile/regfile.py,
> or that Regfile itself doesn't properly implement what the proof says it
> should... I'm thinking I should try re-organizing the proof first to see if
> that improves things? What do you think?
I'm taking a look at regfile.py as well as the proof. Part of the reason you
see everything failing is that the truth table in the proof isn't right; the
register file doesn't have a defined output when read enable is low. I have an
idea as to how to prove that register writes work, would you like me to write
it or leave it up to you?
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] daily kan-ban update 01jun2020
From: Tobias Platen @ 2020-06-01 16:20 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <CAPweEDxaWfprCC=Aj61mpBdDS4QLyRRm+HzeHJaecJSaWDc=yQ@mail.gmail.com>
today:
working on formal verification of the datamerger class.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 339] create POWER9 ROTATE (SHIFTROT) pipeline
From: bugzilla-daemon @ 2020-06-01 12:40 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-339-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=339
Luke Kenneth Casson Leighton <lkcl@lkcl.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|create POWER9 ROTATE |create POWER9 ROTATE
|pipeline |(SHIFTROT) pipeline
--- Comment #3 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
ha, hilarious.
took a while to notice: M-Form is this:
rlwnm RA,RS,RB,MB,ME (Rc=0)
where XO-Form is this:
divdeu RT,RA,RB
however, underneath, the 2nd argument is *consistently a constant*. this needs
to be swapped in the CSV file. see issue #360
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 360] New: move RS to 1st or 2nd operand in CSV files
From: bugzilla-daemon @ 2020-06-01 12:39 UTC (permalink / raw)
To: libre-riscv-dev
https://bugs.libre-soc.org/show_bug.cgi?id=360
Bug ID: 360
Summary: move RS to 1st or 2nd operand in CSV files
Product: Libre-SOC's first SoC
Version: unspecified
Hardware: PC
OS: Mac OS
Status: CONFIRMED
Severity: enhancement
Priority: ---
Component: Source Code
Assignee: lkcl@lkcl.net
Reporter: lkcl@lkcl.net
CC: libre-riscv-dev@lists.libre-riscv.org
NLnet milestone: ---
we are doing this - a lot - i.e. in nearly every single Function Unit:
# RA (or RC)
reg3_ok = yield dec2.e.read_reg3.ok
reg1_ok = yield dec2.e.read_reg1.ok
assert reg3_ok != reg1_ok
if reg3_ok:
data1 = yield dec2.e.read_reg3.data
res['a'] = sim.gpr(data1).value
elif reg1_ok:
data1 = yield dec2.e.read_reg1.data
res['a'] = sim.gpr(data1).value
if instead the CSV files *specified* that RS was in position (read_reg1)
*already*, this would no longer be necessary.
the same thing applies to ShiftRot - except this time RS and RB should
be actually *swapped over*.
this is quite a big change as it affects every single Function Unit,
unit test, and formal proof.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] microwatt decoder tables: M-Form and X-Form switched RS and RB
From: Luke Kenneth Casson Leighton @ 2020-06-01 12:20 UTC (permalink / raw)
To: openpower-hdl-cores; +Cc: Libre-RISCV General Development
anton, paul, and team, hi,
we noticed that in microwatt decode1.vhdl tables, M-Form has:
* op1 is RA
* op2 is a constant or RB, which is in the *3rd* field (M-Form) position
* op3 is RS, which is in the *second* Form position
where in X-Form it is:
* op1 is RA
* op2 is a constant or RB, where RB is from the *2nd* field
* op3 is RS (or RC), which is in the 3rd position.
so decode1.vhdl has this, for rldcl:
0b1000,SHIFT_ROT,OP_RLCL,NONE,RB,RS,RA
we're going to flip that to:
0b1000,SHIFT_ROT,OP_RLCL,NONE,RS,RB,RA
this will put the shift-rot pipeline "in line" with the arithmetic
(and logical) ones.
it *might* be the case that if you did the same thing, you would be
able to cut at least one pair of 64-bit MUXes from microwatt, because
RS and RB (2nd and 3rd operands) would no longer be
special-case-swapped for any Shift-Rot operations. i mention it
because microwatt is intended to be tiny.
best,
l.
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-06-01 12:02 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #45 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Cesar Strauss from comment #43)
> So rdmask, as an input, must be driven and held high, for all the time that
> busy_o is active?
yes - for now. i will look into whether (and how, and if) it should be
moved into CompXXXOpSubsets (into oper_i).
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-06-01 12:01 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #44 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
(In reply to Cesar Strauss from comment #41)
> I updated the unit test to deal with rdmaskn.
>
> For how long rdmaskn must be held valid?
the entire time (the entire busy cycle)
> If I assert it for only one cycle,
> together with issue_i, it doesn't work. rel still rises on the next cycle.
>
> Shouldn't it be latched, as with zero_a and imm_ok?
hmmm hmmm that's a really good point. actually... it should be generated
from the Decoder2 information (Decode2Execute1 in power_decoder2.py),
and brought in via the CompXXXOpSubsets.
if it came in via those, it would be part of oper_i and would automatically
be latched.
it's *different* information depending on which Function Unit
is used, based on the regspecs which are collected together in
soc/fu/compunits/compunits.py (from soc/fu/*/pipe_data.py)
i'll need to think about it - can you set it manually for now and hold
it set for the duration of busy_o == True?
btw it should not be necessary to set the entire mask to all 1s, because
there does not exist an FU which takes zero operands.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-06-01 11:55 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #43 from Cesar Strauss <cestrauss@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #42)
> rdmask disables (entirely disables) bits of rd.rel, and consequently
> rd.go *should* never be raised. rd.rel being the output (which is
> masked out), rd.go being the input, it's a "protocol violation" if
> raised.
So rdmask, as an input, must be driven and held high, for all the time that
busy_o is active?
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-06-01 11:46 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #42 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
Cesar: this is kiiinda true... indirectly :) i'm mulling over the words...
rdmask disables (entirely disables) bits of rd.rel, and consequently
rd.go *should* never be raised. rd.rel being the output (which is
masked out), rd.go being the input, it's a "protocol violation" if
raised.
commit f930c1814f19c41f082c819494cf246edd09082d
Author: Cesar Strauss <cestrauss@gmail.com>
Date: Mon Jun 1 07:56:12 2020 -0300
Add rdmaskn parameter and assert it along issue_i
Like zero_a and imm_ok, rdmaskn disallows the activation of go.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-06-01 11:09 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #41 from Cesar Strauss <cestrauss@gmail.com> ---
I updated the unit test to deal with rdmaskn.
For how long rdmaskn must be held valid? If I assert it for only one cycle,
together with issue_i, it doesn't work. rel still rises on the next cycle.
Shouldn't it be latched, as with zero_a and imm_ok?
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] daily kan-ban update 01jun2020
From: Luke Kenneth Casson Leighton @ 2020-06-01 10:40 UTC (permalink / raw)
To: Libre-RISCV General Development
yesterday:
* got MultiCompUnit functional when interacting through the pipeline
API with the ALU that it manages. (MultiCompUnit now qualifies as the
2nd or 3rd most complex piece of software engineering i've ever
tackled, in 30 years. yes i keep a list. with one exception, all of
the others involve SQL)
* got the ALU MultiCompUnit test working (found several bugs)
* got the Logical MultiCompUnit test working (found one bug)
* got the Condition Register MultiCompUnit test working (found two bugs)
* got the Branch Register MultiCompUnit test written
this morning:
* answered Marketnext about the upcoming Hackathon. they got it
(which is great) - namely that the process we follow also applies to
Marketnext entrepreneurs just as much.
* got the Branch Register MultiCompUnit test working (swapped the
order of SPR1/SPR2)
today:
* Shift-Rot MCU (should be about 15 minutes)
* start looking at connecting MCUs to regfiles.
upcoming tasks (soon to be blockers) and blockers:
* TRAP, SPR and SYS pipelines (we need volunteers to work on these)
* MUL and DIV pipelines (Jacob)
* Interfaces for the 180nm ASIC.
l.
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] Named Records in nMigen
From: Luke Kenneth Casson Leighton @ 2020-06-01 10:15 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <E02C3F23-49C5-4C65-8B05-AD085772FDD1@gatech.edu>
On Mon, Jun 1, 2020 at 7:29 AM Yehowshua <yimmanuel3@gatech.edu> wrote:
>
> Does anybody know if its possible to have a named records
Record. give each bit (or bit range) a name and it is possible to
either assign (eq) to each bit (or bit range) *or* to the entire
thing. bear in mind caveat (1) below.
> or subsignals in nMigen?
as Jock says: python slice, and use an Enum or a defined (named) constant.
bear in mind the following:
* Record is treated as a sequence of bits... *not* a sequence of
fields. several people objected to this, and whitequark to her credit
finally listened. however the replacement for Record is still under
development.
* RecordObject was created to allow for an over-ride of the behaviour
of Record, providing us with the ability to actually have something
more like python OO behaviour.
* you can use Cat on *BOTH* sides of the equation. *PAY ATTENTION* to
the LSB-MSB ordering. Cat is in *LSB* sequential ordering starting
from the 1st argument, where numbers, in the ENGLISH LANGUAGE, start
from the *MSB*.
m.d.comb += Cat(mysig[0], mysig[5], mysig[6]).eq(Const(0b110, 3)
is equivalent to:
m.d.comb += mysig[0].eq(0)
m.d.comb += mysig[5].eq(1)
m.d.comb += mysig[6].eq(1)
* to cope with both POWER9 LE/BE and also to help with ISA field
decoding, we created something that allows for "remapping" of fields.
it's called SignalBitRange. this allows you to *remap* a sequence of
bits to completely different ranges.
l.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] Named Records in nMigen
From: Jock Tanner @ 2020-06-01 7:17 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <0C1DDF53-BB1C-4191-A850-31410BA4734B@gatech.edu>
Yehowshua,
In case you or anybody else come back to this question eventually: it
is possible to define a subsignal using built-in `slice()`. For
example, if the bits 4-7 of your instruction defines a source register,
you can do something similar to
INSTR_SRC = slice(4, 8)
# ...
m.d.comb += self.src_reg.eq(instr[INSTR_SRC])
... or even wrap slices in standard `enum.Enum`:
class InstrFormat(Enum):
dst = slice(4)
src = slice(4, 8)
# ...
m.d.comb += self.src_reg.eq(instr[InstrFormat.src])
Hope it helps.
On Mon, 2020-06-01 at 02:33 -0400, Yehowshua wrote:
> Actually - scratch that - I got it working fine.
>
> Yehowshua
>
> _______________________________________________
> libre-riscv-dev mailing list
> libre-riscv-dev@lists.libre-riscv.org
> http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] Named Records in nMigen
From: Yehowshua @ 2020-06-01 6:33 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <E02C3F23-49C5-4C65-8B05-AD085772FDD1@gatech.edu>
Actually - scratch that - I got it working fine.
Yehowshua
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] Named Records in nMigen
From: Yehowshua @ 2020-06-01 6:29 UTC (permalink / raw)
To: Libre-RISCV General Development
Does anybody know if its possible to have a named records or subsignals in nMigen?
Consider mysig
I’d like to do something to the effect of
```
>>> mysig.byte1
mysig[0:4]
```
Yehowshua
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] more pipeline instructions needed
From: Luke Kenneth Casson Leighton @ 2020-06-01 2:03 UTC (permalink / raw)
To: Libre-RISCV General Development
https://git.libre-soc.org/?p=soc.git;a=blob;f=src/soc/fu/trap/main_stage.py;h=92112e612960179fac1fdda6398ac8eeb91305be;hb=HEAD
we need SPR, TRAP and SYS pipeline instructions. does anyone want to help
with these?
l.
--
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] daily kan-ban update 31may2020
From: Luke Kenneth Casson Leighton @ 2020-05-31 22:14 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <20200531221202.34622bc40d9a893a5c5894b0@platen-software.de>
On Sunday, May 31, 2020, Tobias Platen <libre-soc@platen-software.de> wrote:
> today:
> mostly working on formal verification of the module I have written in the
> last week.
great to hear.
l.
--
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] daily kan-ban update 31may2020
From: Tobias Platen @ 2020-05-31 20:12 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <CAPweEDznJw-y=VJE=q0RbDcJ5LPGJ6wpv7ADq4Ek2hfAnXPUTA@mail.gmail.com>
today:
mostly working on formal verification of the module I have written in the last week.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 353] formal proof of soc.regfile classes RegFile and RegFileArray needed
From: bugzilla-daemon @ 2020-05-31 18:20 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-353-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=353
Cesar Strauss <cestrauss@gmail.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |cestrauss@gmail.com
--- Comment #29 from Cesar Strauss <cestrauss@gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #27)
> basically i have no idea what i am doing with Formal Proofs. i can read
> them, i can *maybe* spot omissions once they're written, and i can just about
> manage combinatorial ones. these sync/time-based ones: really no idea.
Hi Luke,
As I learned about formal verification, I found the initial slides of
http://www.clifford.at/papers/2017/smtbmc-sby/slides.pdf very enlightening for
basic concepts. See for instance slides 6 to 16.
(In reply to Luke Kenneth Casson Leighton from comment #28)
> https://zipcpu.com/blog/2018/03/10/induction-exercise.html
>
> all i can say is: argh :)
This illustrates the difficulties in induction proofs.
It is apparent, from the circuit, that the two shift-registers will be equal
for all time, so the XOR output will be always zero.
However, bounded model check can only check for a finite number of cycles. To
prove it for all time, you prove it that, whenever the assertions hold for n
cycles, it will hold for the n+1 cycle.
The problem with induction, is that it does not start with the initial state
(see slide 11). So, the "initial" state can be in the unreachable state area
(see slide 9).
In the example, an initial state in which the shift register are different, is
unreachable. But the proof checker does not know that. So, it can choose an
initial state where the last bit is equal, passing the assertion, but the
next-to-last bit is different.
Then, it holds the shift enable de-asserted for n cycles and asserts it for the
n+1 cycle, finally shifting out the differing bits. So, it passes for n cycles,
but fails in n+1.
The easy way to solve this is to assert that the shift register contents are
always equal.
Another way is limiting the time in which the shift-enable can be kept
de-asserted. This is done with an assumption.
Without an assumption, the proof checker is free to toggle any unconnected
input signal, in the most perverse way possible, to make the proof fail. In
reality, the input signals must obey some protocol. For instance, you tell the
proof checker that it must keep (assume) issue_i low whenever busy_o is high.
Otherwise it would be free to assert issue_i in the middle of a busy_o cycle,
which it's a violation of the protocol.
Hope it helps,
Cesar
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 359] New: cut down on wires between decode and function units
From: bugzilla-daemon @ 2020-05-31 17:46 UTC (permalink / raw)
To: libre-riscv-dev
https://bugs.libre-soc.org/show_bug.cgi?id=359
Bug ID: 359
Summary: cut down on wires between decode and function units
Product: Libre-SOC's first SoC
Version: unspecified
Hardware: PC
OS: Linux
Status: CONFIRMED
Severity: enhancement
Priority: ---
Component: Source Code
Assignee: lkcl@lkcl.net
Reporter: lkcl@lkcl.net
CC: libre-riscv-dev@lists.libre-riscv.org
NLnet milestone: ---
subsets of decoded instructions are passed to function units. unfortunately
the expansion of wires is vast.
a solution involves passing the instruction to each pipeline and having the
*FU* perform the decode... *at the FU*.
to do this, decode2 should take the subset of fields to be decoded, rather than
assiming that all fields should always be decoded.
this can be done dynamically.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 336] ALU CompUnit needs to recognise that RA (src1) can be zero
From: bugzilla-daemon @ 2020-05-31 16:42 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-336-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=336
--- Comment #40 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
cesar i've moved the unit tests for MultiCompUnit into their own file,
they were getting very big and justified their own module.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* Re: [libre-riscv-dev] daily kan-ban update 31may2020
From: Luke Kenneth Casson Leighton @ 2020-05-31 15:13 UTC (permalink / raw)
To: Libre-RISCV General Development
In-Reply-To: <CAPweEDznJw-y=VJE=q0RbDcJ5LPGJ6wpv7ADq4Ek2hfAnXPUTA@mail.gmail.com>
On Sun, May 31, 2020 at 2:38 PM Luke Kenneth Casson Leighton
<lkcl@lkcl.net> wrote:
> today:
>
> * finding numerous errors where register-writes are being requested by
> the ALU pipeline, where the data is *not* valid ("ok" signal set when
> it should not be)
HA! tracked down and fixed a bug in MultiCompUnit's (new) write-mask
handling, where there was a test condition looking for write-request
signals coming in that would, because they were disabled, never going
to happen.
once that was sorted, the new CompUnit-based unit test for ALU passed.
it uses the exact same unit tests as from the pipeline-only-based ALU
unit test.
with the only difference between the tests being that one adds
MultiCompUnit on *top* of the other, this helps to confirm that
MultiCompUnit is functional.
i am very happy.
i'll do the same thing for the other pipelines.
l.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] [Bug 353] formal proof of soc.regfile classes RegFile and RegFileArray needed
From: bugzilla-daemon @ 2020-05-31 14:12 UTC (permalink / raw)
To: libre-riscv-dev
In-Reply-To: <bug-353-13@https.bugs.libre-soc.org/>
https://bugs.libre-soc.org/show_bug.cgi?id=353
--- Comment #28 from Luke Kenneth Casson Leighton <lkcl@lkcl.net> ---
https://zipcpu.com/blog/2018/03/10/induction-exercise.html
all i can say is: argh :)
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
* [libre-riscv-dev] daily kan-ban update 31may2020
From: Luke Kenneth Casson Leighton @ 2020-05-31 13:38 UTC (permalink / raw)
To: Libre-RISCV General Development
yesterday:
* went through inter-relationship between MultiCompUnit and ALU,
getting MCU to conform to the pipeline API. it's a mess, however it's
functional.
* started debugging ALU Function Unit through MultiCompUnit (closest
thing to actually connecting them to a regfile)
* added "read register mask" and "write register mask" capability to
MultiCompUnit
today:
* finding numerous errors where register-writes are being requested by
the ALU pipeline, where the data is *not* valid ("ok" signal set when
it should not be)
l.
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
_______________________________________________
libre-riscv-dev mailing list
libre-riscv-dev@lists.libre-riscv.org
http://lists.libre-riscv.org/mailman/listinfo/libre-riscv-dev
^ permalink raw reply
page: next prev latest
Libre RISC-V Mailing list Archives (accessible via public-inbox)
Archives are clonable: git clone --mirror https://inbox.libre-soc.org/libre-riscv-dev
Example config snippet for mirrors
AGPL code for this site: git clone https://public-inbox.org/public-inbox.git