289 lines
8.8 KiB
Plaintext
289 lines
8.8 KiB
Plaintext
Important Notice - this is not the sign-off regression
|
|
list for IVL - that is a separate list that Steve Williams
|
|
currently has. It is a subset of these tests.
|
|
|
|
21 April 2007
|
|
|
|
I've added to CVS regression_report-devel.txt that is the
|
|
expected regression test results for the snapshot at the
|
|
time of the checkin. I expect to tag this file with the
|
|
makes of snapshots that are made, so that the expected
|
|
result for a given snapshot can be found.
|
|
|
|
11/3/2000
|
|
|
|
Added format.v, and gold/format.gold and test4.v
|
|
|
|
|
|
10/6/2000
|
|
|
|
Fixed bug in function3.11D.v - now runs on verilog XL okay.
|
|
Added #1 before big if clause in drive_strength.v
|
|
|
|
8/13/2000
|
|
|
|
Went thru older test submissions and found some missing. Added
|
|
stask_parm1.v, assign_mem1, task_inpad, task_omemw2
|
|
|
|
6/12/2000
|
|
|
|
Re-release with defparam race condition fixed and mail
|
|
headers removed on param_string.
|
|
|
|
6/11/2000
|
|
|
|
Added several tests (most sent from Steve Williams as bug
|
|
reports.) Added CRASH detection for CE class tests.
|
|
Also added Tom Coonan's pic processor to the suite.
|
|
Reworked the sregress.pl script to allow addition of
|
|
diff/gold AND module choices.
|
|
|
|
5/08/2000
|
|
|
|
Modified the script to support log file comparison against gold files and
|
|
against created files against gold files. Added Steven Tell's new
|
|
fdisplay, and fopen tests. Added div16.v to the contrib list now that
|
|
we can handle $random (this from Tom Coonan)
|
|
|
|
5/04/2000
|
|
|
|
Wrote some simple preprocessor tests - didn't find anything -didn't
|
|
think I would.. ;-) ifdef1-ifdef4, else1-else3, include1-include3, and
|
|
define1. Also now using iverilog.
|
|
|
|
|
|
4/30/2000
|
|
|
|
Received 4 tests from Peter Monta via Steve Williams. I'm including
|
|
three of them (reworked two to be self-checking..) The new tests are
|
|
assign_nb1, wireland, and param_concat and function_exp.
|
|
|
|
Also added an aliasing mechanism so that we can recover with tests
|
|
that barf. Now you simply type ./regress and it'll run ./sregress.pl.
|
|
The difference is that this script sets the coredumpsize to zero via
|
|
csh. This seems to help ALOT. No arguments are passed - if you want
|
|
to use sregress.pl directly - run it under csh with this set!!
|
|
|
|
4/29/2000
|
|
|
|
Received scope1 from Steve Williams.
|
|
|
|
4/28/2000
|
|
|
|
Added three tests from Steve Williams and commented out wiresr/wiresl
|
|
since they're crashing the test bench right now at run time.
|
|
Note - We've got some contributed code to compare .vcd results. The
|
|
problem is it's in C and everything in this suite todate has been
|
|
source files or a perl program -still cogitating how to handle it.
|
|
|
|
4/27/2000
|
|
|
|
Several new tests - a couple by myself and a bunch by Steve Williams.
|
|
I added a test for Procedural continuous assignment (pca1) and used
|
|
this to create another test for Wildcard sensitivity ( always @ *)
|
|
from the verilog 2000 spec.
|
|
|
|
4/16/2000
|
|
|
|
Added nblkorder to validate case where same variable is assigned
|
|
sequentially in a non-blocking always block. This was suggested from
|
|
a question on a the comp.lang.verilog usenet group. IVL does the
|
|
right thing according to IEEE1364-Draft, page 5-3, section 5.4.1.
|
|
|
|
This test still needs to be refereed by JML though.
|
|
|
|
4/10/2000
|
|
|
|
Added posedge.v and event_list.v sent by S Williams.
|
|
|
|
4/5/2000
|
|
|
|
Incorporated 3 tests sent by Steve Williams (dff1, idiv, con_tri) and
|
|
wrote fork3.19A, and fork3.19B.
|
|
|
|
3/13/2000
|
|
|
|
Well - been a month - geewhiz. defparam needs main added in
|
|
regress.list and had to change the summary error print message
|
|
to REPORT file in sregress.pl
|
|
|
|
2/13/2000
|
|
|
|
Added readmemb and readmemh tests.
|
|
|
|
2/11/20000
|
|
|
|
Been awhile! Anyway, got a bug report from George Gallant concerning
|
|
functions with either case or if statements. Thus sdw_function4 and
|
|
sdw_function5 both which fail. Also need to add readmemh and readmemb!
|
|
|
|
1/23/2000
|
|
|
|
Added drive_strength.v and pullupdown.v to test suite. Contributed!
|
|
|
|
01/13/2000
|
|
|
|
Wrote mult1.v to test reg and wire multiply case, and modulus.v
|
|
to test % operator. Fixed the race in them. Fixed contrib8.5
|
|
in regress.list - it needed to NOT call the top level mod.
|
|
Added xnor_test.v sent to me from Steve Williams(via Guy
|
|
Hutchinson)
|
|
|
|
01/10/2000
|
|
|
|
Added inout.v and modparam.v tests. inout does some sneaky
|
|
stuff with modname(a,a); inout a; This is legal to short
|
|
two signals together. modparam passes paramters down from
|
|
a top level module and overides the loever level. Added udp_bufg
|
|
back to list. It doesn't work, but doesn't core dump or hang
|
|
anymore.
|
|
|
|
01/02/2000
|
|
|
|
Fixed the error message from fifo.v to get correctly detected.
|
|
(It should say "Failed.") Added gencrc and onehot to the regress.list.
|
|
|
|
01/01/2000
|
|
|
|
Added a set of unary operator tests. unary_and, unary_nand,
|
|
unary_or,unary_nor, unary_xor, unary_xnor1, and unary_xnor2.
|
|
There are three elaboration problems in this last bunch. Added
|
|
a scan for "Unable" which seems to qualify elaboration
|
|
errors to the sregress.pl script.
|
|
|
|
12/27/99
|
|
|
|
This is a BIG release for one simple fact. The entire suite has
|
|
been tested against Cadence Verilog XL....and we found a bug in
|
|
XL! See time2.v for the failing testcase.
|
|
|
|
Did some basic maintenance/QA on test suite. Changed sregress.pl
|
|
to use the -l option for XL, as oppossed to redirection. Works
|
|
more reliably on Sun(well - this FIXED a problem with all the
|
|
verilog runs going into backround with the >& syntax?)
|
|
|
|
Fixed always3.1.9A.v, Fixed always3.1.9B.v. Added tests always3.1.9C,D.v
|
|
Fixed wiremod1.v, syntax error in ptest006-008 in $display. Howcome
|
|
IVL likes these? Removed sdw_template.v. Added "PASSED" printout
|
|
to z1,z2. This should make them compatible with XL.
|
|
Fixed assign3.2D.v syntax error, moved syntax error into assign3.2E.v
|
|
to test for it!
|
|
Fixed contrib8.4.v so that it prints PASSED.
|
|
|
|
12/11/99
|
|
|
|
Added contrib directory and fifo.v in same donated by Tom Coonan.
|
|
He's GPL this and other work for inclusion within the test suite!
|
|
THANKS TOM!!!!!
|
|
|
|
Note that two tests always3.1.11B and udp_bufg are commented
|
|
out because they hang the testsuite currently.
|
|
|
|
12/1/99
|
|
|
|
Added qmark6.v contributed by Dan Nelsen. Muxtest.v validates
|
|
X on sel, when inputs are both 1 or 0.
|
|
|
|
11/12/99
|
|
|
|
Added several ga_* tests that use a primitive gate vector. The ga_nand.v test
|
|
of this sequence fails. Went thru the ptestxxx.v series and replaced all
|
|
!= or == constructs with !==/===.
|
|
|
|
11/8/99
|
|
|
|
Fixed stuff added on 11/6 - it now says PASSED!
|
|
Added wirele.v, wirege.v, wireeq.v tests. These with the
|
|
CVS 11/7/99 cut that I used to test.
|
|
|
|
11/6/99
|
|
|
|
Added tests wireadd1.v wiremod1.v wiresl.v wiresr.v wiresub1.v wirexor1.v
|
|
Several of the tests in regress.list had the .v extension so they
|
|
weren't running correctly. That's fixed.
|
|
|
|
10/14/99
|
|
|
|
This is mainly a debug release. See below.
|
|
|
|
lh_varindx added (tests left hand variable index of a vector.)
|
|
Rewrote always3.1.6A-C COMPLETELY. Changed time1.v to use
|
|
always block and #5 d = a; Added time3.v that looks at same
|
|
structure for non-blocking (#5 d <= a).
|
|
|
|
Added an always3.1.6D.v ( a case test from Steve Williams.)
|
|
|
|
There was a race condition in the always3.1.6A-C that was
|
|
due in part to the nature of always case (x) ...default #1; whoops!
|
|
|
|
Removed qmark2.v - it became redundant(besides it was broken sematically)
|
|
|
|
10/5/99
|
|
|
|
Changed location of #0 to first item in the initial
|
|
block for always3.1.6A-C.v This also counts as a
|
|
test for proper operation of the semantics for #0
|
|
which is used to defeat what would otherwise be
|
|
a race condition. Researched the operation of #0 in
|
|
Moorby 4th edition. This seems to be kosher (though
|
|
not GOOD programming practice.)
|
|
|
|
Added several contributed tests by many folks.
|
|
|
|
Note that the regression_report.txt is the results of
|
|
running with the 9/28 IVL release.
|
|
|
|
9/27/99
|
|
|
|
Added a #0 to take care of a race in always3.1.6A-C.
|
|
Added the ability to put an optional main module name as
|
|
a 4th arguement within the regress.list line.
|
|
|
|
9/20/99
|
|
|
|
This release collects the tests together along with a
|
|
regression script(sregress.pl) which is a modified
|
|
version of a script contributed by Guy Hutchison.
|
|
(Thanks Guy!!!)
|
|
|
|
To run the testsuite - ensure that you have the
|
|
appropriate IVL environment variables set - the
|
|
script doesn't check for THAT yet - maybe next
|
|
time;-) You need the appropriate stuff in your
|
|
search path for IVL and set VPI_MODULE_PATH and
|
|
LD_LIBRARY_PATH as necessary. See the IVL docs
|
|
for info.
|
|
|
|
Now - simply move into the testsuite directory
|
|
and type: ./sregress.pl <cr>
|
|
|
|
This will run the regression script, and create a
|
|
series of log files in the log directory. The test
|
|
runs are summarized in regression_report.txt. The
|
|
tests run are obtained from regress.list.
|
|
|
|
9/9/99
|
|
|
|
Completed thru 3.14F(tasks). So Added from Section
|
|
3.11-3.14. Updated ivl_test.html file. Also have included
|
|
some of the "contributed" tests. Haven't worked them all
|
|
into the test document yet.
|
|
|
|
9/6/99
|
|
|
|
Completed all of the always block tests. Added tests
|
|
for sections 3.2-3.10. Reworked the 3.1.5 tests that
|
|
had bugs. Updated the ivl_tests.html file to
|
|
reflect work to date.
|
|
|
|
Note that runsh doesn't use the "verilog" script yet.
|
|
|
|
|
|
8/18/99
|
|
|
|
New tests added 3.1.5C - 3.1.7B
|
|
|
|
8/12/99
|
|
|
|
Initial release to Steve Williams.
|