 |
|
|
RENESAS TOOL NEWS on February 1, 2004: RSO-M3T-CC32R_1-040201D
A Note on Using Cross-Tool Kit M3T-CC32R
--On Floating Constants with an Exponent Part--
|
|---|
Please take note of the following problem in using the M3T-CC32R cross-tool
kit for the M32R family MCUs:
- On floating constants with an exponent part
- Versions Concerned
M3T-CC32R V.1.00 Release 1 through V.4.20 Release 1
- Description
Floating constants with an exponent part may be incorrectly compiled
because the exponent part does not properly be interpreted.
- 2.1 Conditions
- This problem occurs if the following five conditions are satisfied:
| (1) |
A floating constant is put in the program together with operators
(except for the unary plus operator).
|
| (2) |
It comes with an exponent part.
|
| (3) |
Its significand part (the part other than signs, an exponent part,
and a floating suffix [f, l, F, or L]) has a decimal point.
|
| (4) |
The total number of digits in the significand and exponent parts
is equal to or greater than 18.
|
| (5) |
The number of digits in the significand part is equal to or less
than 16.
|
- 2.2 Examples
-
Source file 1:
Here, the comment lines show the incorrect values after compilation.
-----------------------------------------------------------------------
#include >stdio.h<
double ng01 = -1.234567890123456e-012; /* -1.234567890123456e-0 */
double ng02 = -1.2e+0000000000000000012; /* -1.2e+0 */
double ng03 = -1.e-00000000000000012; /* -1.e-0 */
double ng04 = -1.234567890123456e-12f; /* -1.234567890123456e-1 */
double ng05 = -1.23456789012345e+012; /* -1.23456789012345e-01 */
double ng06 = 0.0+1.234567890123456e-009; /* 0.0+1.234567890123456e-0 */
double ng07 = 1.234567890123456e-009+2; /* 1.234567890123456e-0 + 2 */
double ng08 = -1.23456789012345e-0012; /* -1.23456789012345e-0 */
double ng09 = +(1.2345678901234e-00012); /* 1.2345678901234e-0 */
-----------------------------------------------------------------------
Source file 2:
Here, any of the above conditions is sidestepped in each description.
-----------------------------------------------------------------------
#include >stdio.h<
double ok01 = 1.234567890123456e-012; /* Condition (1) unsatisfied */
double ok02 = -12000000000000.00000f; /* Condition (2) unsatisfied */
double ok03 = -1e-00000000000000012; /* Condition (3) unsatisfied */
double ok04 = -1.23456789012345e+12; /* Condition (4) unsatisfied */
double ok05 = -1.2345678901234560e-12f; /* Condition (5) unsatisfied */
-----------------------------------------------------------------------
- Workaround
This problem can be circumvented in either of the following ways:
| (1) |
Add zeros to the fractional part to make the number of digits in the
significand part equal to or greater than 17.
Example:
| Original |
Modified |
| -1.234567890123456e-012; |
-1.2345678901234560e-012; |
| -1.e-00000000000000012; |
-1.0000000000000000e-00000000000000012; |
| -1.234567890123456e-12f; |
-1.23456789012345600e-12f; |
| -1.23456789012345e+012; |
-1.2345678901234500e+012; |
| 0.0+1.234567890123456e-009; |
0.0+1.2345678901234560e-009; |
| 1.234567890123456e-009+2; |
1.2345678901234560e-009+2; |
|
|
| (2) |
Remove the ineffective filling zeros to make the total number of
digits in the significand and exponent parts equal to or less than 17.
Example:
| Original |
Modified |
| -1.2e+0000000000000000012; |
-1.2e+12; |
| -1.e-00000000000000012; |
-1.e-12; |
| 0.0+1.234567890123456e-009; |
0.0+1.234567890123456e-9; |
| -1.23456789012345e-0012; |
-1.23456789012345e-12; |
| +(1.2345678901234e-00012); |
+(1.2345678901234e-12); |
|
- Schedule of Fixing the Problem
We plan to fix this problem in our next release of the product.
|
|
