Novice XC8 question

[DTJ]

Playing with a p18lf25k22, XC8 on MPLABx


When I press a button it triggers an interrupt and sets a flag bit in a register (two bits actually, B7 and B0):

Like this: Button Press --> Interrupt ---> 0b10000010 ==> flag_register

In the main code loop I then test for the flag and if it's set I run some other code:


                if ((status_flags & 0x00000010)) //if button pressed do this code...
                {
                   
                   __delay_ms(30);
                   __delay_ms(30);
                   //do some stuff.......

                 }

My problem is the execution never passes the "if" test, it is as if the "if" always tests false.

If I step through the program and mouse over the status_flags variable it shows the contents as 0x82 = 0b10000010 which is what
I expect. I'm trying to test if bit1 is set. Why doesn't it work?

Thanks.   

[AndyC_772]

Your code says 0x00000010, not 0b00000010.

You're comparing against the value 16, not 2.

[DTJ]

 

Oh FFS! Thank you very much!

So obvious, I've been looking at it for hours......

[Howardlong]

Been there, done that...

[AndyC_772]

Me too! 

[dannyf]

Binary notation isn't terribly helpful, as this example shows.

[DTJ]

I just find using binary for flag registers more intuitive.

[mikerj]

I just find using binary for flag registers more intuitive.

Don't use "Magic" number like this.  Define your bitmask using the BV() macro (which may already be defined in the compiler library with a leading underscore) and it's all much less error prone

Code: [Select]

#define BV(x) (1U<<x)

/* Bit mask for bit 3 */
#define MY_BITMASK BV(3)


[dannyf]

Code: [Select]

#define BV(x) (1U<<x)

Seems to be a very dangerous way of writing a macro.

[mikerj]

Code: [Select]

#define BV(x) (1U<<x)

Seems to be a very dangerous way of writing a macro.

Seems to be a very typical way of writing a macro.  How would you write it, given the use to which it will be put?

[dannyf]

Seems to be a very typical way of writing a macro.

Just because it is popular or typical doesn't mean it is right.

The mistake is very basic, probably the first thing they tell you NOT to do when learning to write a macro.

Someone like your caliber will figure it out quickly, I am sure.

[dannyf]

"#define BV(x) (1U<<x)"

It shows how important it is to be able to go back to basics, no matter how good you think you are.

The most dangerous code is one that's often overlooked and one that works most of the time - because of that, it is widely used yet least suspected.

When it fails, it tends to fail spectacularly.

Every once in a while, great work like that happens to highly accomplished people who may have done X projects, sold Y dollars, and written Z lines of flawless code.

I guess being always humble has its places.

[Kalvin]

Code: [Select]

#define BV(x) (1U<<x)

Seems to be a very dangerous way of writing a macro.

Yes, quite understandable for a beginner doing this as a not knowing any better, but I just hate when I see this in so called professional code. The lack of p********* is just a sign of ignorance or just plain lazyness.

[dannyf]

Agreed. Another example is to load up the 24bit systick counter with a value of 25M.

If the person who wrote that piece of garbage had run his code, he would realize that mistake instantly.

[Kalvin]

Code: [Select]

#define BV(x) (1U<<x)

Seems to be a very dangerous way of writing a macro.

Seems to be a very typical way of writing a macro.  How would you write it, given the use to which it will be put?

You may want to see this page, for example: http://www.cs.yale.edu/homes/aspnes/pinewiki/C%282f%29Macros.html "1. Macros with arguments". You should find the answer and the rationale there. While at it, see also the part "3.3. Multiple statements in one macro" how to combine multiple statements in a macro.

[mikerj]

You may want to see this page, for example: http://www.cs.yale.edu/homes/aspnes/pinewiki/C%282f%29Macros.html "1. Macros with arguments". You should find the answer and the rationale there. While at it, see also the part "3.3. Multiple statements in one macro" how to combine multiple statements in a macro.

I'm very well aware of the dangers of function-like macros, I'm simply waiting for the great DannyF to show us all how this should be correctly implemented given it's intended use.  I suspect it will be an extended wait.

[macboy]

You may want to see this page, for example: http://www.cs.yale.edu/homes/aspnes/pinewiki/C%282f%29Macros.html "1. Macros with arguments". You should find the answer and the rationale there. While at it, see also the part "3.3. Multiple statements in one macro" how to combine multiple statements in a macro.

I'm very well aware of the dangers of function-like macros, I'm simply waiting for the great DannyF to show us all how this should be correctly implemented given it's intended use.  I suspect it will be an extended wait.

I too would expect no less. A six-line response without even a hint at the suggested problem or a solution.

Here's a better way:

Code: [Select]

#define BV(x) (1U<<(x))The parenthesis around x ensure that the entire expression passed as x will be first evaluated, then bitshifted left. If the parenthesis are not there, imagine what happens when you call "BV(i-1)".