Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 15:28:12 +0000 (16:28 +0100)]
propagate missing parameters from div
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 15:27:01 +0000 (16:27 +0100)]
code comments
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 15:16:12 +0000 (16:16 +0100)]
do not set div result if overflow occurs
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 15:05:23 +0000 (16:05 +0100)]
re-enable div random tests and other regressions
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 15:04:08 +0000 (16:04 +0100)]
check for div_overflow equal to None rather than == 1
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 13:30:28 +0000 (14:30 +0100)]
re-add rc/oe back into LDST input record
this for later use with st*cx because it writes CR and OV (and SO)
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 13:21:48 +0000 (14:21 +0100)]
whew panic over, missed a bigendian argument in test_compunit.py
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 13:12:50 +0000 (14:12 +0100)]
add test7 div regression
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 09:40:15 +0000 (10:40 +0100)]
add more debug output for #425
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 09:16:28 +0000 (10:16 +0100)]
add debugging chain for #425
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 09:15:52 +0000 (10:15 +0100)]
update submodule
Luke Kenneth Casson Leighton [Fri, 10 Jul 2020 08:44:27 +0000 (09:44 +0100)]
cut/paste error writing to wrong vcd file
Jacob Lifshay [Fri, 10 Jul 2020 00:21:37 +0000 (17:21 -0700)]
switch to using Signal.width instead of Signal.shape()[0] since Shape isn't a tuple anymore
see https://github.com/nmigen/nmigen/pull/422
Jacob Lifshay [Fri, 10 Jul 2020 00:21:19 +0000 (17:21 -0700)]
format file
Jacob Lifshay [Fri, 10 Jul 2020 00:19:29 +0000 (17:19 -0700)]
update libreriscv submodule
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 23:18:53 +0000 (00:18 +0100)]
add regression test for div overflow case
see https://bugs.libre-soc.org/show_bug.cgi?id=425
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 21:20:41 +0000 (22:20 +0100)]
slightly different so handling in common output stage
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 21:08:55 +0000 (22:08 +0100)]
also set so only if OE requires it
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 21:08:15 +0000 (22:08 +0100)]
debug information related to 32/64 bit mode
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 21:07:06 +0000 (22:07 +0100)]
bug #424 - 32/64 bit is a *global* flag not a per-op one
when it comes to setting CR0
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 20:10:37 +0000 (21:10 +0100)]
test top bit 31 in 32-bit mode for CR0 creation
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:54:09 +0000 (20:54 +0100)]
ha ha very funny. pipelines being pipelines, you have to wait for them
just as with MUL, it was necessary to set the "valid" signal for
only one cycle otherwise spurious output is created
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:44:34 +0000 (20:44 +0100)]
whoops test gets copied 4 times on the If.
create intermediate signal
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:37:14 +0000 (20:37 +0100)]
ALU output stage, change logic slightly
test for oe/ok then set xer/ov data/ok if true
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:28:54 +0000 (20:28 +0100)]
set xer_ov.ok = 1
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:18:04 +0000 (20:18 +0100)]
remove unneeded xer.ca in MulOutputData
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:14:29 +0000 (20:14 +0100)]
something weird going on with div. interaction between tests
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:00:29 +0000 (20:00 +0100)]
simplify setting of mul overflow into xer_ov
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 19:00:07 +0000 (20:00 +0100)]
clarifying comments on setting xer_ov/so
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 18:59:45 +0000 (19:59 +0100)]
DIV overflow needs to be copied into both bits of XER.ov
(OV, OV32)
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 18:58:13 +0000 (19:58 +0100)]
add debug output of DIV results
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 18:57:34 +0000 (19:57 +0100)]
check result first then CR second
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 17:07:25 +0000 (18:07 +0100)]
munge alu_fsm Shifter into looking like CompALU API compliant
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 12:34:07 +0000 (13:34 +0100)]
resolving issues with div tests (turned out to be nmutil.divmod)
adding more tests to track down a CR0 issue
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 10:49:51 +0000 (11:49 +0100)]
remove xer_ca from DIV pipeline (took a bit of messing about)
Cesar Strauss [Thu, 9 Jul 2020 09:50:47 +0000 (06:50 -0300)]
Define ports for a simple sequential Shifter
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 09:55:52 +0000 (10:55 +0100)]
irony comment on how one line creates a massive array of gates
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 09:52:46 +0000 (10:52 +0100)]
add new stages etc. to get multiply working without xer_ca
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 09:32:57 +0000 (10:32 +0100)]
create new DivMulOutputData which does not have CA/CA32
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 08:32:55 +0000 (09:32 +0100)]
make carry output handling optional in common output stage
Luke Kenneth Casson Leighton [Thu, 9 Jul 2020 00:07:24 +0000 (01:07 +0100)]
identifying locations where big/little endian is in place, adding args
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 19:49:32 +0000 (20:49 +0100)]
resolving bigendian/littleendian modes in qemu sim
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 19:24:45 +0000 (20:24 +0100)]
resolving old and new behaviour for lookup of SPRs
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 19:23:39 +0000 (20:23 +0100)]
resolving old and new behaviour for lookup of SPRs
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 19:15:23 +0000 (20:15 +0100)]
adding in ALU test back in, debugging SPR setup
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 19:09:48 +0000 (20:09 +0100)]
sorting out setting of XER
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 18:45:27 +0000 (19:45 +0100)]
add spr to fast reg converter
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 18:37:14 +0000 (19:37 +0100)]
got test_issuer operational on one unit test
however needs further investigation
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 17:26:16 +0000 (18:26 +0100)]
switch assembler to little-endian
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:41:48 +0000 (17:41 +0100)]
stashing current state of investigation whilst looking for regression bug
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:22:52 +0000 (17:22 +0100)]
add test trap simulator unit test
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:22:35 +0000 (17:22 +0100)]
allow qemu to stop at specified end point
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:18:51 +0000 (17:18 +0100)]
add mtspr and bcctrl instructions to helloworld test
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:18:29 +0000 (17:18 +0100)]
add option to qemu to break at known alternate address
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 16:05:16 +0000 (17:05 +0100)]
add to/from spr test (mtspr, mfspr)
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 15:09:18 +0000 (16:09 +0100)]
add code-fragment from microwatt helloworld
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 15:05:18 +0000 (16:05 +0100)]
add a simple addis test (regression)
Luke Kenneth Casson Leighton [Wed, 8 Jul 2020 15:05:04 +0000 (16:05 +0100)]
copy binary loaded from disk into data memory as well
Cesar Strauss [Wed, 8 Jul 2020 09:42:07 +0000 (06:42 -0300)]
Start the FSM-based ALU example.
Jacob Lifshay [Wed, 8 Jul 2020 02:39:36 +0000 (19:39 -0700)]
Merge branch 'master' of ssh://git.libre-riscv.org:922/soc
Jacob Lifshay [Wed, 8 Jul 2020 02:39:13 +0000 (19:39 -0700)]
add WIP pipeline loop demo
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 23:41:12 +0000 (00:41 +0100)]
add hello world binary test
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 23:23:56 +0000 (00:23 +0100)]
whoops error in test of dynamic parameter
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 23:21:48 +0000 (00:21 +0100)]
sort-of got binary execution test working
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 20:05:21 +0000 (21:05 +0100)]
code-shuffle on testing to prepare loading large files into memory
Cesar Strauss [Tue, 7 Jul 2020 09:30:02 +0000 (06:30 -0300)]
Clear input data along with valid_i
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 15:20:12 +0000 (16:20 +0100)]
ordering of tests for OP_ATTN needed shuffling. seems to be working
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 15:11:40 +0000 (16:11 +0100)]
whoops got Function.NONE test wrong in PowerDecode2
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 15:05:54 +0000 (16:05 +0100)]
remove unneeded record field from logical_input_record
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 15:02:24 +0000 (16:02 +0100)]
debugging termination (OP_ATTN)
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 14:50:30 +0000 (15:50 +0100)]
update opcode map for OP_ATTN
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 14:42:52 +0000 (15:42 +0100)]
add halted condition in ISACaller, when attn instruction encountered
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 14:27:50 +0000 (15:27 +0100)]
debugging termination / OP_ATTN
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 13:15:56 +0000 (14:15 +0100)]
add ATTN unit test
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 13:14:46 +0000 (14:14 +0100)]
add core start/stop capability, and OP_ATTN support
Luke Kenneth Casson Leighton [Tue, 7 Jul 2020 12:42:20 +0000 (13:42 +0100)]
add in SPR test cases into test_issuer.py
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:28:38 +0000 (23:28 +0100)]
use ComMULOpSubset in mul pipeline
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:26:56 +0000 (23:26 +0100)]
remove alu unneeded op record data
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:23:51 +0000 (23:23 +0100)]
remove alu unneeded op record data
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:22:45 +0000 (23:22 +0100)]
remove alu unneeded op record data
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:15:55 +0000 (23:15 +0100)]
add mul unit to test_issuer
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:13:00 +0000 (23:13 +0100)]
add mul compunit
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 22:08:53 +0000 (23:08 +0100)]
whoops forgot that the mul pipeline is actually a pipeline (3 stage, first one)
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:48:26 +0000 (20:48 +0100)]
do abs slightly differently in SelectableInt
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:43:48 +0000 (20:43 +0100)]
continue mul unit test debugging
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:19:58 +0000 (20:19 +0100)]
add MULS (signed) version of multiply
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:19:48 +0000 (20:19 +0100)]
improve debug for test_sim.py
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:19:26 +0000 (20:19 +0100)]
add mullw test to qemu sim
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 19:19:06 +0000 (20:19 +0100)]
fix SelectableInt abs
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 18:56:33 +0000 (19:56 +0100)]
add first simulator mul test
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 18:49:49 +0000 (19:49 +0100)]
investigating mul pipeline
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 18:44:58 +0000 (19:44 +0100)]
SelectableInt: make __mul__ return enough space to fit the result
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 16:02:09 +0000 (17:02 +0100)]
first cut at mul test pipeline
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 15:34:31 +0000 (16:34 +0100)]
add first cut at fu mul pipeline
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 13:12:12 +0000 (14:12 +0100)]
adding mtspr tests
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 13:11:47 +0000 (14:11 +0100)]
adding OP_MTMSR test
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 13:11:09 +0000 (14:11 +0100)]
add mtmsr internal op
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 13:10:59 +0000 (14:10 +0100)]
add mtmsr internal op
Luke Kenneth Casson Leighton [Mon, 6 Jul 2020 13:08:04 +0000 (14:08 +0100)]
sort out initialisation of TstL0CacheBuffer in ldst compunit test
Cesar Strauss [Mon, 6 Jul 2020 11:12:59 +0000 (08:12 -0300)]
Assert n.ready_i at the beginning of the cycle
This simulates the common case where we are ready for the
result as soon as the ALU delivers it.
The special case for the zero-delay operation is no longer
needed.