Convert a number to uint64_t doesn't work

[PingISTL]

Dear support

I need to pass 128 bits service uuid to discovery process as variable, I have an array of 16 bytes of uuid number, it is divided into five parts as show here

BT_UUID_128_ENCODE(w32, w1, w2, w3, w48)

where w48 is 48 bits number to pass to, so I have to define a uint16_t variable(w48) to hold it,  and convert from my byte array (value[ ]) as show below:

uint64_t w48;

w48 = value[10]<<40 + value[11]<<32;   //this line ends up 0

w48 += value[12]<<24+value[13]<<16;

w48 +=value[14]<<8 + value[15];

But the first line doesn't work, it ends up 0, while the other 2 lines works individually. let me know what I can do please?

Thank you!

Ping

[PingISTL]

further - to make it more strict, code is like below:

w48 = (value[10]<<40) & 0xff0000000000 + (value[11]<<32)&0xff00000000;   //this line doesn't work

w48 += (value[12]<<24)&0xff000000)+(value[13]<<16)&0xff0000);  //this works 

w48 +=((value[14]<<8)&0xff00) + value[15];  //this works too

It is just first line when number is over 32 bits have problems, help please!

Ping

[Håkon Alseth]

Hi,

 

The compiler (gcc in my test) should warn you when an overflow occurs, like in this case:

warning: left shift count >= width of type [-Wshift-count-overflow]
   15 |  w48 = (value[10]<<40) + (value[11]<<32);   //this line ends up 0
      |                  ^~
warning: left shift count >= width of type [-Wshift-count-overflow]
   15 |  w48 = (value[10]<<40) + (value[11]<<32);   //this line ends up 0

Can you try casting value[10] and [11] to uint64_t to see if this fixes the issue?

 

Kind regards,

Håkon

[PingISTL]

thank you, Hakon

Casting individual value[10] and value[11] works, but not overall result, my compiler doesn't give out any warnings.

Ping

[Håkon Alseth]

Hi Ping,

 

Can you share the content of the value[10..15] as well as the full content of "w48"?

PingISTL said:

Casting individual value[10] and value[11] works, but not overall result, my compiler doesn't give out any warnings.

I'm not sure what you mean, were you able to successfully populate the upper 32 bits ?

 

Kind regards,

Håkon

[PingISTL]

Sorry for that, I didn't make myself clear, your solution works. code as below:

w48 = ((uint64_t)Val[14]<<8)+(uint64_t)Val[15];
w48 += ((uint64_t)Val[12]<<24)+((uint64_t)Val[13]<<16);
w48 +=  ((uint64_t)Val[10]<<40)+((uint64_t)Val[11]<<32);

Thank you!

Ping