[Asterisk-Dev] Solaris Port (was: asterisk-users: Re: Fedora Core 2 and Kernel 2.6)

Olle E. Johansson oej at edvina.net
Thu May 20 14:03:32 MST 2004


John Todd wrote:

> 
> At 10:29 AM -0700 on 5/20/04, Bob Knight wrote:
> [snip]
> 
>> If you hit a wall trying to get intel based boxes to do the job, let 
>> me know.
>> I am working on a SunOS port.  It would be fun to see this running on 
>> a Sun Fire server.
>> Should be able to scale it to 1000+ rooms.  Only problem, servers run 
>> from about 50k to a million.
>> That's like real money.  But it would still be fun.
>>
>> btw: this is not a very pretty port.  The current state of the * 
>> source tree does not lend itself
>> very well to other OS's.  Quite a bit of hacking involved. Something 
>> that I would never
>> want to see checked into cvs.
>>
>> -- 
>> Bob Knight
>> [-w] the work option
>> bk at minusw.com
>> 925-449-9163
> 
> 
> 
> Well, if it can be done with a series of reasonably-documented #define's 
> and whatnot, then I don't see why it wouldn't be committed to CVS.  
> Having Solaris interoperability would be actually a very nice feather in 
> Asterisk's cap, since (for whatever reasons) Solaris is a widely-used 
> and supported telephony platform.
> 
> While I am extremely vague on the details, I recall that there are some 
> "special" instructions in the UltraSparc chipset that handle FFT's and 
> other nifty things, which might also lend themselves to large-scale, 
> fast transcoding.  Now, that might have been video-only - I don't 
> recall...  but I know a guy who used those routines to do real-time 
> encoding of 16 (!!!!) ~150kbps IP video streams for motion-detection 
> frame comparison, video storage, and other nifty features, all in a 
> single 440mhz Netra X1.  Perhaps someone with an actual clue about 
> Solaris and Sparc programming could comment on the viability of this for 
> Asterisk's purposes if a port were to be made...
> 
During my investigations on recursive mutexes (yeah, FreeBSD) I've found
out that Solaris is rumoured not to support recursive mutexes at all.

Another one bites the dust. OS X seems to be in the same position.

FreeBSD supports recursive mutexes, we just need some code.
For Solaris and OS X, we need a emulation layer.

Looking forward to patches for Solaris on the bug tracker!
Need SUN sponsoring Astricon :-)

/O



More information about the asterisk-dev mailing list