Re: [Sqlrelay-discussion] sqlrsh from linux to solaris

[Firstworks/4access <dav...@fi...>]

Yeah, that's definitely the problem.  It's a known bug and it'll be
fixed in the next release.  I'm going to make a prerelease soon that has
the fix in it too.

The problem was that way back when I first started writing SQL Relay, I
decided to use short, long and long long data types, assuming that they
were 16,32 and 64 bits long on all platforms.  This turned out to be
untrue but I didn't notice until amd64 came out.  A long is 64 bits on
amd64 and so is a long long.  So, if this code gets compiled on amd64:

unsigned long	a=5;
write(fd,&a,sizeof(unsigned long));

Then it will try to write a 64 bit number.  Where the same code compiled
on a 32 bit machine (or more precisely, a machine where a long is 32
bits) will try to write a 32 bit number.  If "fd" in the above example
is a network socket open between a 32 and 64 bit machine, all heck
breaks loose when one machine writes a 32 bit number and the other
machine expects to read a 64 bit number or vice-versa.

Recently I switched all the datatypes to int16_t, int32_t, int64_t,
uint16_t, etc.  It should fix the problem.

Dave
dav...@fi...


On Tue, 2005-06-14 at 20:22 +0200, bogusz wrote:
> Hello,
> 
> Im getting following error when trying connect form linux sqlrsh to solaris
> sqlrelay with oracle:
> 
> root@solaris # cat sqlr-listener.26895
> 06/14/2005 18:43:06 CEST listener [26895] : getting authentication...
> 
> 06/14/2005 18:43:06 CEST listener [26895] : authentication failed: password
> size is wrong
> 
> Local connection from Solaris works fine.
> 
> My Linux box is SuSe 9 Enterprise x86_64 and Oracle box is Solaris 9.
> 
> I found function responsible for this error:
> 
> bool sqlrconnection::getPasswordFromClient() {
>         unsigned long size=0;
>         if (clientsock->read(&size)==sizeof(unsigned long) &&
>                 size<=(unsigned long)USERSIZE &&
>                 (unsigned
> long)(clientsock->read(passwordbuffer,size))==size) {
>                 passwordbuffer[size]=(char)NULL;
>                 return true;
>         }
>         #ifdef SERVER_DEBUG
>         debugPrint("connection",1,
>                 "authentication failed: password size is wrong");
>         #endif
>         return false;
> }
> 
> Could it be that long formats in Sparc and AMD Opteron architectures cause
> problem?
> 
> Regards
> 
> bogusz
> 
> 
> 
> -------------------------------------------------------
> SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
> from IBM. Find simple to follow Roadmaps, straightforward articles,
> informative Webcasts and more! Get everything you need to get up to
> speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
> _______________________________________________
> Sqlrelay-discussion mailing list
> Sql...@li...
> https://lists.sourceforge.net/lists/listinfo/sqlrelay-discussion
>