Subject: DirectC: C layer revised
From: Andrzej Litwiniuk (Andrzej.Litwiniuk@synopsys.com)
Date: Fri Jan 10 2003 - 07:50:45 PST
Team,
Enclosed is the complete revised version of the C layer of SV DirectC Interface.
Regards,
Andrzej
------------------------ DirectC C Layer --------------------------------------
Overview of C layer
===================
Representation of the data
--------------------------
The following is assumed about the representation of SV data:
a) representation of packed types is implementation dependent
b) basic integer and real data types are represented as defined in LRM 3.3,
3.4.2, see also "Data type mapping" below.
c) layout of unpacked structures is same as used by C compiler (LRM 3.7)
d) layout of sized unpacked arrays is same as used by C compiler;
this includes arrays embedded in structures and the standalone arrays
(i.e. not embedded in any structure)
Note that this is a restriction imposed on the SV side of the interface!
Depending on the implementation, a particular array may or may be not
accepted as an actual argument for the formal argument which is a sized
array (it will be always accepted for unsized array).
e) layout of the unsized (aka open) standalone unpacked arrays
is implementation dependent with the following restriction:
an element of an array must have the same representation as
individual a value of the same type, with the exception of scalars
(bit or logic) and packed arrays as a type of an element.
Hence array's elements other than scalars or packed arrays can be
accessed via pointers similarly to individual values.
Note that d) actually does not impose any restrictions on how unpacked arrays
are implemented; it says only that an array that does not satisfy d)
may not be passed as an actual argument for the formal argument which is
a sized array; it may be passed, however, for unsized array.
Therefore, the correctness of an actual argument may be implementation
dependent. Nevertheless an open array provides implementation independent
solution. This seems to be a reasonable trade-off.
Data type mapping
-----------------
The basic SV data types are represented as follows:
char -> char
shortint -> shortint
int -> int
longint -> long
byte -> char
real -> double
shortreal -> float
'opaque pointer'-> void*
'string' -> char*
Representation of SV specific data types like packed bit and logic vectors is
implementation dependent and generally transparent to the user.
Nevertheless, for the sake of performance, applications may be tuned for a
specific implementation and make use of the actual representation used by that
implementation; such applications will not be binary compatible, however.
Access to data
--------------
DirectC interface defines the canonical API representation of packed 2-state
and 4-state arrays. (Actually based on PLI's avalue/bvalue for 4-state.)
Library functions will provide the translation between the representation
used in a simulator and the canonical API representation.
Abstract access is provided only for the unsized (open) packed and unpacked
arrays. All other objects are directlly accessible.
C compatible data types are accessed either directly or through pointers.
Packed arrays are accessible via canonical representation; interface provides
functions for moving data between implementation representation and canonical
representation, any necessary convertion is performed on the fly.
There are also functions for bit selects and limited (up to 32 bits) partial bit
selects.
Array slices are not supported.
For unsized (open) arrays the original SV ranges are used in indexing.
For sized arrays normalized ranges are assumed.
For example, if 'logic [2:3][1:3][2:0] b [1:10]' is used in SV type,
it will have to be defined in C as if it were declared in SV in the following
normalized form: 'logic [17:0] b [0:9]'.
That practically means that the packed part must be linearized and the ranges
must be normalized.
By and large the actual arguments will be passed by reference, with a few
exceptions (e.g. small input arguments).
There will be no copying of arguments (other than resulting from coercing)
Please recall item 10 from "17 items"!
[10a: "xf must not modify the values of its input args'
10b: "The initial values of formal output args are unspecified and may be
implementation dependent."]
The actual arguments passed by reference by and large will be passed as they are, without changing their representation from the one used by a simulator.
(Again, there are some exceptions, mainly for the 'open' alias 'unsized' arrays.)
Therefore there will be no overhead on argument passing because no copying or
translation between different representations will be required.
C data types will be directly accessible.
SV specific types will be accessible both via the interface library functions,
what grants the compatibility, and directly thru pointers, what allows for
the simulator-specific tuning of the application.
'Open' alias 'unsized' arrays will be accessible through the abstract access
mode, i.e. via the interface library functions.
Depending on the data types used for the external (or exported) functions,
either binary level or C source level compatibility is granted.
The binary level compatibility is granted for all data types that do not mix
C types with SV specific types, though either category separately is fine.
For example, if a formal argument type is a C structure or a packed array
of bit or logic, then binary level compatibility is granted.
On the other hand, if a packed structure is embedded in an unpacked data
type, then only a source level compatibility is granted.
(Binary level compatibility means that an application compiled for a given
platform shall work with every SV simulator on that platform.
Source level compatibility means that an application will have to be re-compiled for each SV simulator and that the implementation specific definitions
will be required for the compilation.)
Argument passing
=================
Actual arguments generally are passed by reference or by value and will
be directly accessible in C code.
In some cases, defined separately, an argument will be passed by handle and will
be accessible via library functions (abstract acess mode).
For the arguments passed by reference, their original simulator-specific
representation will be used and a reference to the original data object
will be passed.
If an argument of type 'T' is passed by reference, than the formal argument
shall be of the type 'T *'.
Output and inout arguments are passed by reference.
Input arguments are passed by value or by reference, depending on the size.
'Small' values of formal input arguments are passed by value.
The following data types are considered 'small':
- char, byte, int, real, shortreal (have I omitted something?)
- pointer, string
- bit (i.e. 2-state) vectors up to 32-bit; canonical representation will
be used, similarly for function result
Input arguments of other types are passed by reference.
Include files
=================
(All names are provisional and subject to discussion and improvment.)
The C layer of SV DirectC interface defines two include files corresponding to
the two levels of compatibility (binary level and source code level):
- "svc_bin.h"
- "svc_src.h"
"svc_bin.h" is fully defined by the interface and is implementation independent.
It defines the canonical API representation of 2-state (bit) and 4-state (logic)
values, defines the types used for passing references to SV data objects,
provides function headers and defines a number of helper macros and constants.
"svc_src.h" provides implementation dependent definitions.
The contens of this file, i.e. what symbols are defined (constants, macros,
typedefs), is defined by the interface.
The actual definitions of the symbols, however, are implementation specific and
will be provided by the vendors.
"svc_src.h" defines data structures for implementation specific representation
of 2-state and 4-state vectors, as well as a number of helper macros and
constants.
User's applications that require "svc_src.h" file will be only source-level
compatible, i.e. they will have to be compiled with the version of "svc_src.h"
provided for a particular implementation of SV.
The applications that use only "svc_bin.h" will be binary compatible with
all SV simulators.
The values of C-like types may be directly accessed via the corresponding
C type definitions.
The values of SV specific types, like packed arrays of bit or logic, may
be accessed via interface functions using the canonical representation of
2-state and 4-state vectors.
They also may be directly accessed using the implementation representation.
The former mode will assure binary level compatibility, the later one
will allow for tool-specific performance oriented tuning of an application.
There will be no confusion whether passing a particular data type
is binary compatible or only source level compatible.
The rule is straightforward: if a correcponding type definition can be
written in C without the need to include "svc_src.h" file, then the binary
compatibility is granted. Everything that requires "svc_src.h" is not binary
compatible and needs recompilation for each simulator of choice.
Applications that pass solely C compatible data types or standalone packed
arrays (both 2-st and 4-st) will require only "svc_bin.h" and therefore will be
binary compatible will all simulators.
Applications that pass complex data types that contain at the same time
packed arrays and C-compatible types, will require also "svc_src.h" file and
therefore will not be binary compatible will all simulators. The source level
compatibility is, however, granted.
"svc_bin.h"
=================
This file contains the following definitions:
/* canonical API representation */
#define sv_0 0
#define sv_1 1
#define sv_z 2 /* representation of 4-st scalar z */
#define sv_x 3 /* representation of 4-st scalar x */
/* common type for 'bit' and 'logic' scalars. */
typedef unsigned char svScalar;
typedef svScalar svBit; /* scalar */
typedef svScalar svLogic; /* scalar */
/* 2-state and 4-state vectors, modelled upon PLI's avalue/bvalue */
#define VEC32_NEEDED(WIDTH) (((WIDTH)+31)>>5)
typedef unsigned int
svBitVec32; /* (a chunk of) packed bit array */
typedef struct { unsigned int c; unsigned int d;} /* as in VCS */
svLogicVec32; /* (a chunk of) packed logic array */
/* reference to a standalone packed array */
typedef void* svBitPackedArr;
typedef void* svLogicPackedArr;
/* a handle to a generic object (actually, unsized array) */
typedef void* svHandle;
/* functions for translation between simulator's and canonical representations */
/* s=source, d=destination, w=width */
/* actual <-- canonical */
void svPutBitVec32 (svBitPackedArr d, svBitVec32* s, int w);
void svPutLogicVec32 (svLogicPackedArr d, svLogicVec32* s, int w);
/* canonical <-- actual */
void svGetBitVec32 (svBitVec32* d, svBitPackedArr s, int w);
void svGetLogicVec32 (svLogicVec32* d, svLogicPackedArr s, int w);
The above functions copy the whole array in either direction.
User is responsible for providing the correct width and for allocating an array
in the canonical representation.
Although the put/get functionality provided for 'bit' and 'logic'
packed arrays is sufficient yet basic, it requires unnecessary copying of
the whole packed array when perhaps only some bits are needed.
For the sake of the convenience and improved performance, the bit selects and
limited (up to 32 bits) part-selects are also supported.
Functions for part-select allow to access (read/write) only a narrow subranges
of up to 32 bits. A canonical representation will be used for such narrow
vectors.
For the sake of symmetry a single chunk of an array in the canonical
representation is used both for 'logic' and 'bit'.
One may argue, however, that in the case of 'bit' such small vector
is actually a single int, so perhaps a more intuitive signature could be
used.
/* functions for bit-select and limited width part-select */
/* Packed arrays are assumed to be indexed n-1:0,
where 0 is the index of least significant bit */
/* functions for bit-select */
/* s=source, d=destination, i=bit-index */
svScalar svGetSelectBit(svBitPackedArr s, int i);
svScalar svGetSelectLogic(svLogicPackedArr s, int i);
void svPutSelectBit(svBitPackedArr d, int i, scalar s);
void svPutSelectLogic(svLogicPackedArr d, int i, scalar s);
/*
* functions for part-select
*
* a narrow (<=32 bits) part select is copied between
* the implementation representation and a single chunk of
* canonical representation
*
* s=source, d=destination, i=starting bit index, w=width
* like for variable part selects; limitations: w <= 32
*/
Please note that for the sake of symmetry a canonical representation
(i.e. an array) is used both for 'bit' and 'logic', though a simpler
int could be used for 'bit' part selects <= 32-bit.
void svGetPartSelectBit(svBitVec32* d, svBitPackedArr s, int i, int w);
void svPartGetSelectLogic(svLogicVec32* d, svLogicPackedArr s, int i, int w);
/*
* for 'bit' type a part select <= 32-bit is really an int
* so a function could be used:
* int svGetPartSelectBit(svBitPackedArr s, int i, int w);
*/
void svPutPartSelectBit(svBitPackedArr d, svBitPackedArr s, int i, int w);
void svPutPartSelectLogic(svLogicPackedArr d, svLogicPackedArr s, int i, int w);
/*
* for 'bit' type a part select <= 32-bit is really an int
* so simpler arg could be used:
* svPutPartSelectBit(svBitPackedArr d, int up_to_32-bits, int i, int w);
*/
More functions may be added, for example, for converting packed arrays into
char* strings (for printing) or other way round (for reading), like in VCS
(cf. VCS DirectC.h).
"svc_src.h"
===========
This file provides implementation specific definitions.
In particular, it defines the macros for specifying variables representing
SV packed arrays:
#define BIT_PACKED_ARRAY(WIDTH,NAME) ...
#define LOGIC_PACKED_ARRAY(WIDTH,NAME) ...
(For example, VCS might defined the later macro as follows:
#define LOGIC_PACKED_ARRAY(WIDTH,NAME) vec32 NAME [ VEC32_NEEDED(WIDTH) ]
)
Example 1 - binary compatible application
=========================================
SV:
typedef struct {int a; int b;} pair;
extern void foo(input int i1, pair i2, output logic [63:0] o3);
C:
#include "svc_bin.h"
typedef struct {int a; int b;} pair;
void foo(int i1, pair *i2, svLogicPackedArr o3)
{
svLogicVec32 arr[VEC32_NEEDED(64)]; /* 2 chunks needed */
printf("%d\n", i1);
arr[1].c = i2->a;
arr[1].d = 0;
arr[2].c = i2->b;
arr[2].d = 0;
svPutLogicVec32 (o3, arr, 64);
}
Example 2 - source level compatible application
===============================================
SV:
typedef struct {int a; bit [6:1][1:8] b [65:2]; int c;} triple;
// troublesome mix od C types and packed arrays
extern void foo(input triple i);
C:
#include "svc_bin.h"
#include "svc_src.h"
typedef struct {
int a;
BIT_PACKED_ARRAY(6*8, b) [64];
int c;
} triple;
/* Note that 'b' is defined as for 'bit [6*8-1:0] b [63:0]' */
void foo(triple *i)
{
int j;
svBitVec32 arr[VEC32_NEEDED(6*8)]; /* 6*8 packed bits */
printf("%d %d\n", i->a, i->c);
for (j=0; j<64; j++) {
svGetBitVec32(arr, (svBitPackedArr)&(i->b[j]), 6*8);
...
}
}
Note that 'a', 'b', 'c' are directly accessed as fields in a structure.
In the case of 'b', which represents unpacked array of packed arrays,
individual element is accessed via library function svGetBitVec32(),
by passing its address to the function.
Open (Unsized) Arrays and Abstract Access - Overview
====================================================
All open arrays, packed (i.e. vectors) and unpacked (i.e. arrays per se)
will be passed by handle and accessed mainly via accessory functions (abstract
access).
Abstract access for open arrays will allow inquires about the dimensions
and the original boundaries of SV actual argument and will allow to access
the elements of an open array using the same range of indices as in SV.
The programmer will always have a choice, whether to specify a formal
argument as a sized array or as an open (unsized) array.
In the former case all indices will be normalized on the C side (i.e. 0 and up)
and the programmer is assumed to know the size of an array.
Programmer is also assumed to be capable of figuring out how the ranges of
the actual argument will map onto C-style ranges.
Hint: programmers may decide to stick to [n:0]name[0:k] style ranges in SV.
In the later case, i.e. open array, the abstract access mode will be used,
what would facilitate SV-style of indexing with the original boundaries of
the actual argument, all this for the price of some overhead.
Note that this provides some degree of flexibility and allows programmer
to control the trade-off of performance vs. convenience.
If a formal argument is specified as a sized array, then it will be passed by reference, with no overhead, and will be directly accessible as a normalized
array.
If a formal argument is specified as a open (unsized) array, then it will be
passed by handle, with some overhead, and will be accessible mostly indirectly,
again with some overhead, although with the original boundaries.
Unsized formal arguments
========================
For input, output and inout arguments that are declared as open array,
the corresponding formal argument in C will be of type svHandle.
Type svHandle denotes an opaque pointer to a descriptor generated by SV
compiler for an actual argument. Descriptor will provide comprehensive
information about an actual argument. The internal structure of a descriptor
is implementation dependent and fully irrelevant to the user.
Limitations
===========
The unpacked part of an open array may be multidimensional.
The packed part, however, is restricted to a single dimension.
Note that any packed data type in SV is eventually equivalent to
a one-dimensional packed array. Hence the above limitation is practically
not restrictive.
Open array querying functions
=============================
These functions are modelled upon SV array querying functions, with
the same semantics (cf. LRM 16.3):
/* h= handle to open array, d=dimension */
/* For unpacked part */
int svUnpackedLeft(svHandle h, int d);
int svUnpackedRight(svHandle h, int d);
int svUnpackedLow(svHandle h, int d);
int svUnpackedHigh(svHandle h, int d);
int svUnpackedIncrement(svHandle h, int d);
int svUnpackedLength(svHandle h, int d);
int svUnpackedDimensions(svHandle h);
/* For 1-dimensional packed part */
int svPackedLeft(svHandle h);
int svPackedRight(svHandle h);
int svPackedLow(svHandle h);
int svPackedHigh(svHandle h);
int svPackedIncrement(svHandle h);
int svPackedLength(svHandle h);
Open array access functions
=============================
Similarly to sized arrays, there are functions for copying data between
the simulator representation and the canonical representation.
It will be also possible to get the actual address of SV data object
or of an individual element of an unpacked array.
This may be useful for the simulator-specific tuning of the application.
Depending on the type of an element of an unpacked array, different access
methods are used:
- packed arrays ('bit' or 'logic') are accessed via copying to or from
the canonical representation
- scalars (1-bit value of type 'bit' or 'logic') are accessed directly
- other types of values (e.g. structures) are accessed via generic pointers;
a library function calculates an address and the user should provide
the appropriate casting
- all types but scalars may be accessed via pointers, as described above
SV allows arbitrary dimensions and hence an arbitrary number of indices.
To facilitate this, a variable argument list functions will be used.
For the sake of performance the specialized versions of all indexing functions
are provided for 1, 2 and 3 indices.
Access via canonical representation
-----------------------------------
This group of functions is meant for accessing elements which are packed
arrays ('bit' or 'logic').
The following functions will copy a single vector from a canonical
representation to an element of an open array or other way round.
Element of an array is identified by indices bound by the ranges of the
actual argument. In other words, original SV values are used for indexing.
/* functions for translation between simulator's and canonical
representations */
/* s=source, d=destination */
/* actual <-- canonical */
void svPutBitArrElemVec32 (svHandle d, svBitVec32* s, int indx1, ...);
void svPutBitArrElem1Vec32(svHandle d, svBitVec32* s, int indx1);
void svPutBitArrElem2Vec32(svHandle d, svBitVec32* s, int indx1, int indx2);
void svPutBitArrElem3Vec32(svHandle d, svBitVec32* s,
U indx1, int indx2, int indx3);
void svPutLogicArrElemVec32 (svHandle d, svLogicVec32* s, int indx1, ...);
void svPutLogicArrElem1Vec32(svHandle d, svLogicVec32* s, int indx1);
void svPutLogicArrElem2Vec32(svHandle d, svLogicVec32* s, int indx1, int indx2);
void svPutLogicArrElem3Vec32(svHandle d, svLogicVec32* s,
U indx1, int indx2, int indx3);
/* canonical <-- actual */
void svGetBitArrElemVec32 (svBitVec32* d, svHandle s, int indx1, ...);
void svGetBitArrElem1Vec32(svBitVec32* d, svHandle s, int indx1);
void svGetBitArrElem2Vec32(svBitVec32* d, svHandle s, int indx1, int indx2);
void svGetBitArrElem3Vec32(svBitVec32* d, svHandle s,
U indx1, int indx2, int indx3);
void svGetLogicArrElemVec32 (svLogicVec32* d, svHandle s, int indx1, ...);
void svGetLogicArrElem1Vec32(svLogicVec32* d, svHandle s, int indx1);
void svGetLogicArrElem2Vec32(svLogicVec32* d, svHandle s, int indx1, int indx2);
void svGetLogicArrElem3Vec32(svLogicVec32* d, svHandle s,
U indx1,U indx2, int indx3);
The above functions copy the whole packed array in either direction.
User is responsible for providing the correct width and for allocating an array
in the canonical representation.
Access to scalars ('bit' and 'logic')
-------------------------------------
Another group of functions is needed for scalars (i.e. when an element
of an array is a simple scalar, 'bit' or 'logic':
svBit svGetBitArrElem (svHandle s, int indx1, ...);
svBit svGetBitArrElem (svHandle s, int indx1);
svBit svGetBitArrElem (svHandle s, int indx1, int indx2);
svBit svGetBitArrElem (svHandle s, int indx1,U indx2, int indx3);
svLogic svGetLogicArrElem(svHandle s, int indx1, ...);
svLogic svGetLogicArrElem(svHandle s, int indx1);
svLogic svGetLogicArrElem(svHandle s, int indx1, int indx2);
svLogic svGetLogicArrElem(svHandle s, int indx1,U indx2, int indx3);
void svPutLogicArrElem(svHandle d, svBit value, int indx1, ...);
void svPutLogicArrElem(svHandle d, svBit value, int indx1);
void svPutLogicArrElem(svHandle d, svBit value, int indx1, int indx2);
void svPutLogicArrElem(svHandle d, svBit value, int indx1,U indx2, int indx3);
void svPutBitArrElem (svHandle d, svLogic value, int indx1, ...);
void svPutBitArrElem (svHandle d, svLogic value, int indx1);
void svPutBitArrElem (svHandle d, svLogic value, int indx1, int indx2);
void svPutBitArrElem (svHandle d, svLogic value, int indx1,U indx2, int indx3);
Access to the actual representation
-----------------------------------
The following functions provide an actual address of the whole array or
of its individual element. These functions will be used for accessing
elements of the arrays of types compatible with C.
These functions will be also usefull for the vendors, because
they provide access to the actual representation for all types of arrays.
If the actual layout of the SV array passed as an argument for an open unpacked
array is different than C layout, then it will not be posssible to access such
array as a whole and therefore the address and size of such array will be
undefined (zero, to be exact).
Nonetheless the adresses of individual elements of an array will be always
supported.
Note that no specific representation of an array is assumed here, hence
all functions use a generic pointer void *.
/* a pointer to the actual representation of the whole array of any type */
/* NULL if not in C layout */
void *svGetArrayPtr(svHandle);
int svSizeOfArray(svHandle); /* total size in bytes or 0 if not in C layout */
/* Return a pointer to an element of the array
or NULL if index outside the range or null pointer */
void *svGetArrElemPtr(svHandle, int indx1, ...);
/* specialized versions for 1-, 2- and 3-dimensional arrays: */
void *svGetArrElemPtr1(svHandle, int indx1);
void *svGetArrElemPtr2(svHandle, int indx1, int indx2);
void *svGetArrElemPtr3(svHandle, int indx1, int indx2, int indx3);
Access to array elements of other types
---------------------------------------
If array's elements are of a type compatible with C, then there is no need
to use the canonical representation. In such situations the elements will
be accessed via pointers, i.e. the actual address of an element will be
computed first and then used to access the desired element.
Example 3 - open array
======================
SV:
typedef struct {int i; ... } MyType;
extern void foo(input MyType i [][]);
// 2-dimensional unsized unpacked array of MyType
MyType a_10x5 [11:20][6:2];
MyType a_64x8 [64:1][-1:-8];
foo(a_10x5);
foo(a_64x8);
C:
#include "svc_bin.h"
typedef struct {int i; ... } MyType;
void foo(svHandle h)
{
MyType my_value;
int i, j;
int lo1 = svUnpackedLow(h, 1);
int hi1 = svUnpackedHigh(h, 1);
int lo2 = svUnpackedLow(h, 2);
int hi2 = svUnpackedHigh(h, 2);
for (i = lo1; i <= hi1; i++) {
for (j = lo2; j <= hi2; j++) {
my_value = *(MyType *)svGetArrElemPtr2(h, i, j);
...
*(MyType *)svGetArrElemPtr2(h, i, j) = my_value;
...
}
...
}
}
Example 4 - open array
======================
SV:
typedef struct { ... } MyType;
extern void foo(input MyType i [], output MyType o []);
MyType source [11:20];
MyType target [11:20];
foo(source, target);
C:
#include "svc_bin.h"
typedef struct ... } MyType;
void foo(svHandle hin, svHandle hout)
{
int count = svUnpackedLength(hin, 1);
MyType *s = (MyType *)svGetArrayPtr(hin);
MyType *d = (MyType *)svGetArrayPtr(hout);
if (s && d) { /* both arrays have C layout */
/* an efficient solution using poiter arithmetics */
while (count--)
*d++ = *s++;
/* even more efficient:
memcpy(d, s, svSizeOfArray(hin));
*/
} else { /* less efficient yet implementation independent */
int i = svUnpackedLow(hin, 1);
int j = svUnpackedLow(hout, 1);
while (i <= svUnpackedHigh(hin, 1)) {
*(MyType *)svGetArrElemPtr1(hout, j++) =
*(MyType *)svGetArrElemPtr1(hin, i++);
}
}
}
------------------------ End Of C Layer --------------------------------------
==============================================================================
Andrzej I. Litwiniuk, PhD Principal Engineer VCS R&D
Synopsys, Inc TEL: (508) 263-8056
154 Crane Meadow Road, Suite 300, FAX: (508) 263-8069
Marlboro, MA 01752, USA
==============================================================================
This archive was generated by hypermail 2b28 : Fri Jan 10 2003 - 07:51:48 PST