Re: Open parser issues

Ron:

Commenting out the [Pin Numbers] entries is one way to get Segmentation faults
on a Sun Unix system. The same file is attached, but with pin 5 commented
out.

Bob Ross,
Interconnectix, Inc.

> Hello Bob,

> Thanks for the feedback.

> A couple of open issues:

> [1] Fixed bug when Number_of_Pins < actual Pin_Numbers list count
> (mtx.c:323:assertion failed). However, you also mentioned a
> segmentation fault not associated with an assertion. I have so
> far been unable to locate (still looking, Bounds Checker is not as
> slick as Purify). If you have an ibs file that demos this
> failure, I would appreciate a copy of the relevant section.

> [2] "Banded_matrix last rows must contain fewer entries" - Please let me
> know if you get consensus on a change to this area.

> [3] "Pin" to "Pin_Numbers" correlation - Again, please let me know if
> consensus is achieved on a change.

> thanx, ron

|***************************************************************
|
[IBIS Ver] 2.1
[File Name] pkg.ibs
[File Rev] 0.1
[Date] May 25, 1995
[Source] Interconnectix Inc.
[Disclaimer] Provided for modeling purposes only.
                  Results are not guaranteed.
|
|***************************************************************
|
[Component] PACKAGE-TEST
[Manufacturer] Test File
[Package]
| variable typ min max
R_pkg 0.0m NA NA
L_pkg 0.0nH NA NA
C_pkg 0.0pF NA NA
|
|***************************************************************
|
[Pin] signal_name model_name R_pin L_pin C_pin
A1 P_1 NC
2 P_2 NC
5 P_3 NC
A4 P_4 NC | There is nothing stated in the
| 3 P_5 NC | Spec regarding correlating
                                          | [Pin] entries and [Pin Numbers]
                                          | entries by number of pins or by
                                          | actual pin number correlation.
                                          | NO error is reported.
                                          | I assume the order can differ.
[Package Model] test

[Define Package Model] test
[Manufacturer] anyone
[OEM] this file shows
[Description] some failed test cases

[Number of Pins] 5 | get CRASH if changed to 1

[Pin Numbers]
A1
2
3
4
| 5 | Commenting out [Pin Numbers]
                                           | entries can produce Segmentation
                                           | faults

[Model Data]
[Inductance Matrix] Banded_matrix | Success!
[Bandwidth] 2
[Row] A1
.01 .01 .01
[Row] 2
.01 .01 .01
[Row] 3
.01 .01 .01
[Row] 4
.01 .01 | I did not expect this triangle matrix for the last
[Row] 5 | terms, but this is correct according to the Spec.
.01 | Otherwise, parser will give an error.
                       |
                       | However, I believe the Spec is wrong. There is no
                       | way, for example, to describe the coupling of a
                       | 100 pin Quad Flat Pack with from pins 99 and 100 to
                       | pin 1 (and visa-versa) when using a bandwidth of 2
                       | and pin 1 is centered on one edge.

[Capacitance Matrix] Full_matrix | Success!
[Row] A1
.01 .01 .01 .01 .01
[Row] 2
.01 .01 .01 .01
[Row] 3
.01 .01 .01
[Row] 4
.01 .01
[Row] 5
.01

[Resistance Matrix] Sparse_matrix
[Row] A1
A1 .01 | Get Failure here, but if A1 changed to 1, get
2 .01 | Incorrect Success.
[Row] 2
3 .01
[Row] 3
5 .01
3 .01
6 .01 | Error Message text is not correct here.

[End Model Data]
[End Package Model]
|
[End]
Received on Tue May 30 09:19:01 1995