caused by the new bloat-all with templayers, which used the distance
record in the bloat structure differently than the other bloat
functions, and therefore was messing up the enumeration of layers
needing to be handled by the hierarchical output. This was probably
also wrong for the existing bloat-all function, which might explain
some problems recently with the GDS output.
"lef write", which had one typo in the formatting, plus was using
a static string method for generating the formatted output that was
implementation-dependent on fprintf(). These have been fixed.
minimum manufacturing grid (normally 1000 but can be altered by the
"angstroms" flag in the cifoutput section, and by "gridlimit").
The output values then are truncated such that the floating-point
output value, when multiplied by the UNITS value, is always an
integer. e.g., "gridlimit 5" will change UNITS to 200, and values
will be minimum precision 0.005, or 5 nanometers.
probably needs revisiting, because "lef write" and "lef writeall"
need handling to generate the PROPERTYDEFINITIONS block for the
PROPERTY entries to be correct.
layers (apart from the fact that contacts are output as magic's
contact layer representation, and not as cuts; this still needs to
be handled properly).
I missed the LAYER value and the geometry was one level to high.
Previous;
```
PORT
LAYER li1 ;
RECT 1.145000 1.075000 1.690000 1.275000 ;
RECT 3.720000 1.075000 4.490000 1.275000 ;
LAYER met1 ;
RECT 1.105000 1.260000 1.395000 1.305000 ;
RECT 3.765000 1.260000 4.055000 1.305000 ;
```
After;
```
PORT
LAYER li1 ;
RECT 1.145000 1.075000 1.690000 1.275000 ;
RECT 3.720000 1.075000 4.490000 1.275000 ;
LAYER met1 ;
RECT 1.105000 1.260000 1.395000 1.305000 ;
RECT 3.765000 1.260000 4.055000 1.305000 ;
```
areas and writes ANTENNAGATEAREA and ANTENNADIFFAREA values.
(2) Determines "USE POWER" or "USE GROUND" from label names
matching Tcl variables $VDD and $GND, if the USE has not been
registered as a cell property (knowning the use allows magic
to avoid writing an ANTENNADIFFAREA for power rails, although
doing so should not be an issue).
because otherwise all pins will flag metal-to-obstruction spacing
within the cell if the cell is wide enough that the obstruction
layer satisfies the width requirement for the rule. It is too
complicated to try to find specific places where the wide spacing
might not be needed. Potentially this could be a problem for
technologies that define a number of graded wide-spacing rules,
as the largest-width rule is always used now by "lef write -hide",
and the largest-width rule could theoretically allow enough space
to route through, which would cause a short that cannot be
detected. That would be a pathological case that may not show up
in practice.
labels by expanding a zero area label rectangle, but then if "select
chunk" returns nothing, it sets the area to the zero area label
rectangle instead of the expanded one that it just created. This
is the reason that "lef write" is producing pins with no geometry
in the LEF file output.
of the pin port geometry and using those areas to create the
spacing between them and the obstruction layer. Otherwise, the
existing method used different databases (source vs. flattened) to
find the pin area, and they did not always agree on the exact
dimensions, leading to spacing errors within the LEF view.
multiple ports; also, when using the "-hide" option, the obstruction
area is computed from layer geometry, not from the bounding box.
Still left to do: Ensure minimum width on pins, and remove slivers
of obstruction that are below minimum width.
the right argument type (float, not int). Otherwise all resistances
from extresist come out zero when doing "ext2spice extresist on"
and "ext2spice hierarchy on". Also changed the format of the resistance
in the SPICE output to type float, since values are in standard units of
ohms, and rounding to the nearest ohm seems excessively coarse-grained.
by FIXED_BBOX derived from GDS and the new "boundary" cif input
rule, then the bbox property values take precedence over the
extent-of-geometry bounding box.
is placed over multiple types. This causes SelectChunk() to fail
and the pin will have no geometry output in the LEF file. To avoid
this, the area of the label is always painted into the select cell
so that if SelectChunk() fails, the label area still exists with
the label tile type.
it only complains about (i.e., issues an error message) ports with
the same index but different text, indicating a read port number
collision and a true error.
the undo mechanism in the wrong state when writing a LEF header
with no route track information. Also added code to the DEF write
routine to handle subcell arrays. Thanks to Martin Devera for
both patches.
the output to have an obstruction area over the entire cell except
for a keep-out area around each pin. Instead of marking every
part of the pin geometry, only the "chunk" (largest immediate
rectangle) surrounding the port label is output as part of port
LEF geometry. This avoids making unnecessarily complicated
abstract views, and makes it easier for other tools to read and
manage the same abstract views.
Tim Edwards
Tim 'mithro' Ansell