[uClibc]long long functions in stdlib

[Beaulieu, James A.]

Actually, yes, I am trying to get the complete system into two megs of 
flash, and I have the system there, but the more I save, the better I am. 
In it, I have to include pretty much a whole dsl/atm router 
configuration. There are a bunch of pieces that I have been working on, 
and just "hacking to work" rather then fixing. When I get some time 
several weeks from now, I will revisit these problem areas, and try to 
send in proper patches. For now with the long long support, I just 
removed those two functions functionality to get things working.

Perhaps the short term solution is to replace the long long __a with a 
long __a1, __a2 inside of the structure, and rewrite the functions to 
just use the modified __a1 and __a2. It doesn't seem to be accessed 
outside of the *rand* functions anyways.

The long long was just the latest portion. The other posts are just other 
areas that I have found.

BTW, half of the ATM library and utilities support is working with uClibc.

Thanks.

Jim

On Tuesday 09 July 2002 06:02 pm, Manuel Novoa III wrote:
> Hello,
>
> On Tue, Jul 09, 2002 at 11:10:55AM -0400, Beaulieu, James A. wrote:
> > Greetings.
> >
> > When you build without long long support, the two stdlib functions:
> > __drand48_iterate()
> > srand48_r()
> > fail with they operate on the drand48_data structure. This structure
> > properly has the long long __a variable removed, but the above
> > functions try to operate on __a anyways. There has to be a rework of
> > the functions to support the lack of a long long construct.
>
> Yes.  I started dealing with "no long long" issues last week and
> haven't gotten back to it.  I probably won't get to it for a couple
> of weeks, as it is stuff I'm doing in my spare time (connected with
> my bcc/elks port).  There are a number of places that need fixing.
>
> Do you have a pressing need to build without long long support?
>
> Manuel

-- 
James A. Beaulieu
Telogy Networks
A Texas Instruments Company