Consulting on my next project

[NorthGuy]

Deleted. Duplicate.

 

[NorthGuy]

Head&0x1FFF - Tail&0x1FFF

No. Just this:

Head - Tail

BTW: It's not a good idea to add to the existing posts because people won't read the same post twice.

 

[electroRF]

Thank you NorthGuy

Are you sure about having Head always larger than Tail?

At some point, Head will "overflow" and will start counting from 0 again, while Tail is yet to overflow and therefore be larger than Head, right?

 

[NorthGuy]

Thank you NorthGuy

Are you sure about having Head always larger than Tail?

At some point, Head will "overflow" and will start counting from 0 again, while Tail is yet to overflow and therefore be larger than Head, right?

If they're 32-bit numbers, they only overflow at 4294967296, but if they do overflow, it doesn't really matter. You simply define Size as unsigned int.

unsigned int Size;
 
Size = Head - Tail;

If Tail is 0xffffff80 and Head overflowed to 0x10, the difference will be calculated correctly as 0x90.

 

[electroRF]

Hi again NorthGuy.

Thank you friend

If Size is unsigned int.
Lets take the case where there were (2^32 + 100) bytes written, so
Head = 100
And say that
Tail = 4294967196 (which is 2^32 - 100)

e.g. Size should be 200.

Size = Head - Tail = 100 - 4294967196 = -4294967096

Does it really turn out to be 200?

 

[NorthGuy]

Does it really turn out to be 200?

Yes. If you assign -4294967096 to a 32-bit int, you'll get 200.

The binary presentation of -4294967096 is FFFFFFFF000000C8. Chopping of past 32-bit gives you 0xc8 = 200.

 

[electroRF]

Got you

Thank you NorthGuy!

You helped me so much, your support is very valuable!