Suspecting a bug in plugin_ops.h

Suspecting a bug in plugin_ops.h

[Soulard, MathieuX]

Hi all,


I'm using alsalib and I checked its code because of one issue I have.

I did not yet solved my issue, but I think I saw a bug in alsa-lib that I would like to share with you.

It is located in [alsa-lib]/src/pcm/plugin_ops.h in the inline function sx24s().

static inline u_int32_t sx24s(u_int32_t x)
{
        if(x&0x00008000)
                return x|0x000000FF;
        return x&0xFFFFFF00;
}

I think that 0x00008000 should be replaced by 0x00000800.



Thanks
Regards

Mathieu

Re: Suspecting a bug in plugin_ops.h

[Clemens Ladisch]

Soulard, MathieuX wrote:
> static inline u_int32_t sx24s(u_int32_t x)
> {
>         if(x&0x00008000)
>                 return x|0x000000FF;
>         return x&0xFFFFFF00;
> }
> 
> I think that 0x00008000 should be replaced by 0x00000800.

This function is indented to sign-extended a 24-bit value to 32 bits,
operating in 'wrong' endianness.

Why do you think that the sign bit is in the highest bit of the second-
highest nibble?


Regards,
Clemens

Re: Suspecting a bug in plugin_ops.h

[Soulard, MathieuX]

> > static inline u_int32_t sx24s(u_int32_t x)
> > {
> >         if(x&0x00008000)
> >                 return x|0x000000FF;
> >         return x&0xFFFFFF00;
> > }
> >
> > I think that 0x00008000 should be replaced by 0x00000800.
> 
> This function is indented to sign-extended a 24-bit value to 32 bits,
> operating in 'wrong' endianness.
> 
> Why do you think that the sign bit is in the highest bit of the second-
> highest nibble?

Well... because I think I made a mistake !
Actually I missed the endianness here, I thought about another format.

Thanks for your prompt response.


Thanks
Regards

Mathieu