recognized as a valid command when file locking has been disabled as
a compile-time option. The command then generates an error on
"locking enable" but simply ignores the command "locking disable".
gates when the gate length minimum may be larger than the gate
width minimum, and to implement an "angles" option on the "width"
rule that can implement rules where a layer width must be wider
as measured from angled edges than from straight edges.
the "width" rule is assumed to be symmetric, and not checked in all
four directions, as that would be redundant. But non-Manhattan tiles
are not symmetric and must be checked all four directions. Implemented
in a way that does not increase the DRC processing time.
in github issue #149. This causes magic to no longer write log files
for "make" and "make install" but will properly exit with a non-zero
return code on any error during compile.
Also: Corrected the command "tech drc surround <type1> <type2>" so
that it now returns the correct value when <type1> and <type2> are in
the same plane. Added new command "tech drc directional <type1> <type2>"
which works the same way as "tech drc surround" except for directional
surround rules. Used this to generate vias from "def write" with the
correct metal surround amounts included in the via definiton. The
route analysis then ignores tile slivers that make up the surrounding
material around contacts. Also implemented a method that handles
routes that are made of multiple thin tiles due to the maximum horizontal
stripes rule. Now magic handles "def write" well except for not dealing
with non-minimum-width routes unless they're specifically called out as
"special" nets.
Most of this had to do with the incorrect use of the parent's substrate
name in extHierSubstrate(). After the correction, there still remains
an issue that is caused when a labeled isolated substrate region overlaps
an extraction tile boundary. I believe that this particular error has
existed for some time and is not new, so I am committing these changes.
The extSubtree() routine cuts a layout into squares and extracts
each separately, checking for subcell interactions. In each
square it parses all labels looking for unconnected ones. This
section of code not only parses all labels M x N times, but it
then marks interaction areas where there may be none, forcing
additional unnecessary processing. This commit makes the first
quick optimization, which is to change the return value of
DRCFindInteractions() from boolean to integer, allowing it to
return a value indicating that there are no subcells in the
area. This prevents the loop through labels from happening in
cases where there can never be interactions. More to come.
CmdRS.c:1269:22: warning: & has lower precedence than ==; == will be evaluated first [-Wparentheses]
DRCtech.c:2573:16: warning: & has lower precedence than !=; != will be evaluated first [-Wparentheses]
account for the fact that non-Manhattan tiles are processed twice
in the search, with the side mask bit 1 and 0 for each call. The
DRC check, like the Manhattan checks, only needs to check one of
these.
to copy up errors from non-interacting subcells. The routine was
only copying up TT_ERROR_P type errors, but for deep hierarchies,
TT_ERROR_S type errors may have to be propagated up as well.
"|"), pointed out by Jim Everitt. The error is pretty major, but
because the section of code it affects is just eliminating
unnecessary DRC rules, I believe that the only effect is that the
DRC ruleset ends up using more memory than it needs to. But, good
to have fixed.
the layers2->layers1 swapped case from being implemented; the
surround_ok rule type is by definition asymmetric and the two
layer sets cannot just be swapped.
geometry (which had gone unnoticed due to the lack of use of
"cifspacing" in any rule decks). The rule was not checking for
all synthetic edges, because the tile type was expected to match
the rule type when the function is called, but with a non-Manhattan
tile, that may or may not be true and needs to be checked.
checks on non-Manhattan tiles were made only on the straight edges;
this was sufficient for most checks. However, it can miss the case
of facing non-Manhattan edges. This check does not do triggered
rules because there is no non-Manhattan maxwidth algorithm implemented,
and because the triggering clipping area is a triangle and needs an
extension to support it.
area + halo but then failing to limit overlap checks to that clip
area, resulting in bizarre errors whenever an array is made. Not
sure why the error didn't show up more often.
one is an "off_grid" DRC type, which can be used to check geometry
that is below the manufacturing grid. Normally magic prevents the
grid from being subdivided below the manufacturing grid, but this
limit can be removed and replaced by DRC checks to check for such
errors in a GDS file of unknown origin. The second version looks
for interactions between subcells that end up with intersections
of non-manhattan geometry landing on points that are not on the
database internal grid. Such errors cannot be seen by magic's DRC
engine by definition, and so must be detected while flattening
geometry for the DRC checks.
to stop the search whenever a cell is not found. Used this to implement
a new option for GDS writes, "gds undefined allow|disallow" (default
"disallow") controls whether or not GDS with undefined references will
be allowed to be written. Similarly affects CIF and LEF writes, extraction,
and DRC (when running "drc check" from the top).
hierarchical processing from CIFGenSubcells() and CIFGenArrays(), and to
avoid certain operators that are useless and harmful when applied
hierarchically; namely squares, slots, bbox, boundary, and net.
the handling of subcell instances generally. Previously it would check
the interaction between neighboring cells in an array without regard to
any material in the parent cell which might remove those errors;
consequently, the array would have to be DRC clean by itself in order for
the parent cell to show as DRC clean. The array check has been moved
inside the DRCInteractionCheck() routine, so that it runs only where
arrayed instances do not interact with anything else. Within interaction
areas, the area is flattened and checked, so the array check is not
needed.
from child cells, which was incorrectly descending all the way down
into the hierarchy; not only can this produce the incorrect result
but it also wastes time searching cells that don't need to be
searched.
are interpreted in the dimensions used in the cifoutput section rather
than always in centimicrons (otherwise, rules at, say, 5nm cannot be
represented in the DRC section).
the bottom in non-interacting areas means that any change in DRC
to a subcell must be handled before checking DRC in the parent.
Previously the order of checks was reversed, moving parent cells
to the beginning of the check list. This prevents the error cited
in the previous commit which was showing up as a delayed DRC check
when creating parameterized cells.
processes "interaction areas". This should eliminate weirdnesses
where errors will fail to show up in a subcell that does not
interact with paint or other subcells in the top level edit cell.
These errors cannot be reported directly in the top level cell,
but a new error message has been created to direct the user to
check the subcell for errors. Also: Modified the toolkit procedures
to force DRC to be run on newly created or modified parameterized
cell layouts. There is some oddity about the process that causes
DRC errors to be delayed unless a print statement is put before the
DRC check; I would like to investigate this further.
the rule is a normal database rule or a CIF-DRC rule. For the latter,
the flag is used when substituting for escape strings in the "why"
rule explanation to produce the correct value in microns.
errors in the periphery of where a change has been made. For some
reason this was not apparent before, but seems to be from a change
dating back to 2008---which seems unlikely. The fact that it has
not been seen before may have something to do with the size of the
DRC halo compared to the DRC step size in the SkyWater PDK, where
it has suddenly become apparent. Jury is still out on this one.
Before this modification "directed" surround rule only worked for layers1 and layers2 belonging to the same plane.
After this modification the "directed" rule works also for different planes, even if layer1 is a contact type.
templayers. This permits some useful interactions like growing to
the size of a bounding box, or abutment box, as well as many other
possibilities. Also: Corrected the use of "cif see" for the boundary
(abutment box) layer, which was not working because the "cif see"
command uses a flattened CellDef that does not have the boundary
property of the cell it was flattened from.
where the function drcFindOtherCells() was missing, which causes
interaction areas to be missed and messes up the DRC checks between
parent and child cells.
Conflicts:
VERSION
Merged recent changes from master back into bplane, as the efficiency of
bplane for doing extraction on large layouts is unquestionably better.
Fixed the implementation of DBMoveCell() for bplane. Corrected an error
in the bplane version of dbScaleCell() that enumerates cell uses but
does not free the list.
Conflicts:
VERSION
calma/Depend
cif/Depend
cmwind/Depend
commands/Depend
database/Depend
dbwind/Depend
debug/Depend
drc/Depend
ext2sim/Depend
ext2spice/Depend
extflat/Depend
extract/Depend
garouter/Depend
gcr/Depend
graphics/Depend
grouter/Depend
irouter/Depend
lef/Depend
lisp/Depend
mzrouter/Depend
netmenu/Depend
plot/Depend
plow/Depend
resis/Depend
router/Depend
select/Depend
sim/Depend
tcltk/Depend
textio/Depend
tiles/Depend
utils/Depend
windows/Depend
wiring/Depend
Merged recent changes from master branch into bplane branch. Testing the
bplane implementation which has about a 5x improvement in extraction times
for large layouts, which is significant enough to move ahead with the bplane
implementation; however, the bplane implementation has not been thoroughly
vetted yet, so it will remain a branch until such time that it has been
validated.
Conflicts:
VERSION
database/DBcellsrch.c
database/DBconnect.c
extract/ExtInter.c
lef/Depend
utils/Depend
Updated bplane branch with all changes to master since the bplane branch
was last modified.
DRC records contain an index into a string array instead of containing
a copy of a string. This is preliminary to changing the way the DRC
error plane is painted, so that the types painted will mark the error
type. This will (1) allow "drc why" to simply scan the DRC error
plane rather than running the DRC engine, (2) allow DRC errors to be
counted by area rather than by tile, and (3) let the DRC count be the
same whether done by "drc listall why" or "drc count".
which had been changed a few months back to remove the individual
cell count and only list the top level cell. The behavior has been
changed a bit so that "list" returns values for the top level cell
only, but "listall" returns a complete list. "drc list count total"
gives the DRC count for the top cell, but "drc listall count total"
gives the DRC count for everything (probably not very useful).
Also: Implemented a behavior by request to automatically removed
the (UNNAMED) cell whenever a new cell is loaded and the (UNNAMED)
cell has not been modified.
to be investigated here. I am no longer sure why I removed the
cell search from DRC count, but it appears that the cell search is
non-functional, and it should be determined why. There is no
particular reason not to have a DRC count search. It could be
implemented such that "list" vs. "listall" counts the top cell vs.
all cells. First it must be determined why there are no subcell
counts.
flag definition had been put in database.h instead of database.h.in,
and so was deleted on "make clean". Also, corrected a problem that
causes the cif output style to be lost when running DRC-CIF checks
if the first output style is the DRC style.
only at the time of running the command "load". But cells are generally
loaded only on an as-needed basis, so the dereferencing option must be
saved as a flag in the cell and honored whenever its subcells are expanded
or otherwise read at a later time.
categorized by error type and scrolled through conveniently. However,
it needs work dealing with finding the actual error bounds. The
"DRC count" counts tiles, which is tile-plane-geometry-specific, and
"DRC listall why" fractures errors both over tiles and over the square
areas that the interactive DRC splits the layout into, for performance.
The DRC error plane needs to be changed to hold different types for
each error class, so that errors can be scanned by boundary instead of
by tile (work to be done).
rid of redundant port entries in subcircuits. There is still an outstanding
issue as to whether nodes and connections need to be recursively iterated
to the hierarchy bottom. The current fix corrected the test case. Also,
added a "-dereference" option to the "load" command to revert to the
original behavior of using only search paths from "addpath" when searching
for files to load.
(such as widespacing or directional surround) that will cause the
rule to be triggered without a cause due to a failure to reset the
error count from a previous triggered rule (the condition of failure
is much more rare than this explanation makes it sound, which is why
it went undiscovered for so long).
error whenever there are no DRC-CIF rules in the techfile. The
same error will be raised anyway when reading the techfile if
DRC-CIF rules are declared without a style being specified.
reduction of memory and startup time, which was to maintain only
one CIF style in memory. The new method is just to read in and
keep the DRC CIF style separately from the output CIF style.
Because the CIF sections of the techfile are read before the DRC
sections, and the CIF DRC style is declared in the DRC section,
the CIF DRC style is read in on the fly during the first DRC
checking.
the compiler. Some are obscure functions (plot verstatec hasn't
been used in years) but others (like SPICE distributed junctions)
are potentially significant sources of unexpected crashes on
systems that don't zero uninitialized memory.
DRC rules. The substitutions are specified by "%d" for the main
rule distance, "%c" for the corner rule distance (sometimes
interpreted differently; e.g., as width in the widespacing rule),
and "%a" for rule area (e.g., maxarea rule). In addition to
simplifying the process of writing rule violation strings, the
benefits are twofold: (1) The output is in meaningful physical
units, but in the case of SCMOS technology, will scale properly
depending on the selected GDS output style, and in the case of
all technologies, will scale properly with internal grid division;
and (2) when using lambda, but where rules are given in vendor
minimum dimensions, the rules will be based on the lambda rule
approximation (that is, distances will be rounded to the nearest
lambda but reported in microns). Behavior is unchanged from
previous versions for "why" strings not using the defined
substitution sequences.
behaviors: (1) An additional syntax for "widespacing" that allows
both the triggering metal's width AND run-length, which is typical
of rules in 65nm-and-below processes; and (2) a new "option"
statement for the DRC section, with (for now) one possible flag
"wide-width-noninclusive", indicating that the metal width given
for "widespacing" rules means that a violation is only triggered
for material with a width greater than the given rule width (as
opposed to the default interpretation of a width greater than
or equal to the given rule width).
to be not quite right, and all sets of rules need to be updated
to match the latest DRC rule parsing. Also, plowing does not
handle contact types correctly.
enabling the "plow" command. At this time, however, not all
rules are being implemented correctly given the numerous changes
to the DRC handling since the last time "plow" was working. Fixes
are ongoing.
errors from the area checked must use the same area as is used
by the interaction checker to recompute errors. In earlier code,
both these areas were much larger than necessary. In the last
code update, the area to make revised checks was reduced, but
the area cleared of errors was not, resulting in errors that
would disappear from the layout as paint changes were made, until
a full "drc check" was run again. This code commit makes the
areas the same again so that no errors are lost.
much larger area than necessary to the interaction check, and
the other which failed to scale the tech halo distance down
after scaling all the rules. Both of these led to huge areas
of a layout being unnecessarily checked when even a tiny part
of the layout was modified. Corrected behavior matches
expectations for interactive DRC response.
to have come from an unfinished (or at least unchecked) development
effort on improving the checks for DRC interaction errors (which
also affects extraction).
avoid the problem where subcells violate DRC themselves but do
not violate DRC in the context of the parent (i.e., the parent
adds metal to avoid minimum area error), DRC errors in subcells
are neither counted nor displayed, but they remain present in
the subcell database. The DRC check enumerates all subcell
errors as mandatory check areas in DRCFindInteractions(), so
that those areas are always counted with the interaction areas.
Previously, if a subcell violated a DRC rule by itself, that
error would be counted and displayed in the parent even if the
parent had paint causing the error to no longer exist.
development had been halted since it was first created back in April.
Version 8.2 is now the official development version, with the first
development push to create a Cairo graphics interface.
Tim Edwards
Anton Blanchard
Brad Smith
fabian-l-c