 |
|
|
MESC TOOL NEWS:
MESCT-NC308_1-000716D
NC308WA, NC30WA, NC79WA, NC77WA
Precautions
|
|---|
Please take note of the following problems in using C compilers (with assembler and integrated development environment) NC308WA for the M16C/80 series, NC30WA for the M16C/60 and M16C/20 series, NC79WA for the 7900 series, and NC77WA for the 77xx series of MCUs:
- Assigning the address of a function to a variable of type pointer to another function results
in a System Error being generated if the types of both functions are different from each other.
- Folding floating-point constants may cause incorrect subtract operations to be performed.
- Folding constants of type unsigned long may cause incorrect divide and modulus operations to be performed.
-
Problem on Pointers Pointing to Functions
Assigning the address of a function to a variable of type pointer
to another function results in a System Error being generated
if the types of both functions are different from each other.
- 1.1 Versions Concerned
- NC308WA V.1.00 Release 1--V.3.00 Release 1
- NC30WA V.2.00 Release 1--V.4.00 Release 1
- NC79WA V.2.00 Release 1--V.4.00 Release 1
- NC77WA V.5.00 Release 1--V.5.20 Release 2
- 1.2 Conditions
- This problem occurs if the following two conditions are satisfied:
- (1)A variable is declared to be of type pointer to a function.
- (2)The address of a function is assigned to the variable in (1) of type
pointer to another function where the types of both functions are different from each other.
- 1.3 Example
--------------------------------------------------------------------
extern int *func(int);
void ff(void)
{
int *(*fp1)(int); /* Condition (1) */
int (*fp2)(int); /* Condition (1) */
fp1 = func; /* No problem because the types are the same */
fp2 = func; /* Condition (2); System Error occurs because
the types are different from each other */
}
--------------------------------------------------------------------
- 1.4 Workaround
- Define the type of the function pointed to by a variable of type pointer
as the same as that of the function whose address is assigned to the above variable.
- 1.5 Schedule of Fixing Problem
- We plan to fix this problem in our next release.
NC308WA, NC30WA, NC79WA, NC77WA Precautions
MESCT-NC308_1-000716D
-
Problem on Folding of Floating-Point Constants
Folding of floating-point constants may cause incorrect subtract operations
to be performed.
- 2.1 Versions Concerned
- NC308WA V.1.00 Release 1--V.3.00 Release 1
- NC30WA V.2.00 Release 1--V.4.00 Release 1
- NC79WA V.2.00 Release 1--V.4.00 Release 1
- NC77WA V.4.00 Release 1--V.5.20 Release 2
- 2.2 Conditions
- This problem occurs if the following two conditions are satisfied:
- (1)An operation is performed between a variable and a constant of type float
or double, regardless of whether their types are the same or not.
- (2)The operation has any of the following four patterns:
- A variable + constant 1 - constant 2
- A variable - constant 1 - constant 2
- Constant 1 + a variable - constant 2
- Constant 1 - a variable - constant 2
- Note: This problem occurs when any of the above patterns comprises
a part of a long expression, too.
- 2.3 Example
--------------------------------------------------------------------
void func(void)
{
double d1, d2;
d1 = d2 + 1.0 - 1.0; /* Interpreted as d1 = d2 + 2.0 */
d1 = (d2 + 1.0 - 1.0) / d2; /* Interpreted as d1 = (d2 + 2.0)/
d2 */
}
--------------------------------------------------------------------
- 2.4 Workaround
- Instead of describing operations between constants, use the result of them.
- 2.5 Schedule of Fixing Problem
- We plan to fix this problem in our next release.
NC308WA, NC30WA, NC79WA, NC77WA Precautions
MESCT-NC308_1-000716D
- Problem on Folding Constant of Type Unsigned Long
Folding constants of type unsigned long may cause incorrect divide and modulus operations to be performed.
- 3.1 Versions Concerned
- NC308WA V.1.00 Release 1--V.2.00 Release 2
- NC30WA V.2.00 Release 1--V.3.20 Release 2
- NC79WA V.2.00 Release 1--V.3.20 Release 2
- NC77WA V.3.00 Release 1--V.5.20 Release 2
- 3.2 Conditions
- This problem occurs if the following three conditions are satisfied:
- (1)A divide or modulus operation is performed between constants.
(2)One or both of the constants are of type unsigned long(Casted or suffixed with UL).
(3) One or both of the constants are within a range of 0x80000000L--0x0ffffffff
- 3.3 Example
--------------------------------------------------------------------
void func(void)
{
/* The result is 0x00000000H though 0x5555555H expected. */
unsigned long ul = (0x0ffffffffUL / 48UL);
...
}
--------------------------------------------------------------------
- 3.4 Workaround
- Instead of describing operations between constants, use the result of them.
- 3.5 Schedule of Fixing Problem
- We plan to fix this problem in our next release.
|
|
