FreeTDS Development Group

[Chris Eleveld <ihermit2 AT yahoo.com>]

I just ran a test.  Maybe we need a check against the
protocol level.  I'm using TDS version 4.6 protocol
from AIX with FreeTDS to Solaris with Sybase.

Does anyone have an eclectic set of servers and a big
endian client?

Maybe we need the change to depend on wether you
enable the mssql stuff or another config flag?  for
TDSVER set to 4.2, 4.6, and 5.0 however with the
servers I have to chat to I don't get any messages
unless I use the if I have.  But, both machines in any
of my configurations are big endian. 

--- Michael Horowitz
<michael.horowitz AT storagenetworks.com> wrote:
> Hey Chris,
> 
> There was a thread about the big endian issue 2-3
> weeks ago(I am
> connecting from a Solaris box).  I came across the
> same problem.  I
> initially just erased the second swap, but I was
> informed that would cause
> problems when "emulate little endian" wasn't on. 
> So, I conditioned the
> second swap on "if (!tds->emul_little_endian)".  I
> noticed you did "if
> (tds->emul_little_endian)" in your patch, are you
> sure you didn't mean the
> other way?  tds_get_int does the swap if
> emul_little_endian is true, so
> you only want to swap it if tds_get_int didn't swap
> it already, which is
> if emul_little_endian is false.
> 
> In my experience with MS SQL Server 7.0, I MUST use
> emul_little_endian,
> otherwise the datetime fields have their bytes
> backwards and the "swap
> broken dates" seems to do nothing for me.
> 
> I tested this fix on MS SQL Server 7.0 and 2000 with
> mixed results.
> This is what I wrote in a previous email about it:
> "Actually, I tried SQL Server 2000 without
> "emulate-little-endian" and
> with
> that swap removed and the datetime fields are fine
> and so are the error
> messages.
> With SQL Server 7.0 with the same setup, the
> datetime fields are wrong,
> and the error messages are messed up.
> 
> I tried using the following instead of completely
> removing the swap:
> if (!tds->emul_little_endian) {
>        tds_swap_bytes(&rc, 4);
> }
> That fixed the error messages on SQL Server 7.0, but
> then they were
> screwed up on SQL Server 2000."
> 
> Anyway, it seems that you have had a similar problem
> with the big endian
> issue as me, so I wanted to chyme in with my
> experience with it.  While
> either of the fixes seems to work under specific
> configurations, I am not
> sure what the final fix should be to make it work
> under all platforms and
> with all types of SQL Servers both MS and Sybase.
> 
> Do you have any other info that might help with
> this?
> 
> Thanks,
> 
> Mike
> 
> 
> > below is a patch that includes what I did for the
> > strange serve on my network.  More importantly
> however
> > are the changes for big endian platforms towards
> the
> > end with messages.  The symptom I was seeing was
> that
> > errors and messages were not being handled. 
> trackign
> > through the code this appeared to be happenning
> > because of a double endian swap.  I have also had
> to
> > make some changes to the dblib call to the message
> > handler as it did not pass the final argument of
> the
> > line number.  I will send that patch as it is
> rather
> > simple if anyone needs it.
> > 
> > 
> > *** freetds-0.53-orig/src/tds/token.c   Sun Nov 25
> > 19:04:17 2001
> > --- freetds-0.53/src/tds/token.c        Fri Jan 18
> > 14:54:07 2002
> > ***************
> > *** 185,192 ****
> >                                 ack =
> > tds_get_byte(tds);
> >                                 major_ver =
> > tds_get_byte(tds);
> >                                 minor_ver =
> > tds_get_byte(tds);
> > !                               tds_get_n(tds,
> NULL,
> > len-4);
> > !                               tds_get_byte(tds);
> >   #ifdef WORDS_BIGENDIAN
> >   /*
> >                                 if (major_ver==7)
> {
> > --- 185,219 ----
> >                                 ack =
> > tds_get_byte(tds);
> >                                 major_ver =
> > tds_get_byte(tds);
> >                                 minor_ver =
> > tds_get_byte(tds);
> > !                               if ((major_ver ==
> 4)
> > && (minor_ver == 6)) {
> > !                                   int tlen;
> > !                                   char
> > servname[1024];
> > !
> > !                                  
> tds_get_byte(tds);
> > /* teeny version */
> > !                                  
> tds_get_byte(tds);
> > /* tiny version */
> > !                                   tlen =
> > tds_get_byte(tds); /* servname len */
> > !                                   if (tlen >=
> > sizeof(servname))
> > !                                       tlen =
> > sizeof(servname) - 1;
> > !                                   tds_get_n(tds,
> > servname, tlen);
> > !                                   servname[tlen]
> =
> > '\000';
> > !                                   if
> > (!strcmp(servname, "OpenServer")) {
> > !                                       if (ack ==
> 0)
> > succeed=TDS_SUCCEED;
> > !                                   }
> > !                                   /* eat the
> rest of
> > the data as below */
> > !                                   tds_get_n(tds,
> > NULL, len -
> > !                                              
> > 1/*ack*/ -
> > !                                              
> > 4/*major.minor.teeny.tiny*/ -
> > !                                              
> > 1/*servnamelen*/ -
> > !                                              
> > tlen/*servname*/);
> > !                               } else {
> > !                                   /*
> > !                                    * shouldn't
> these
> > 2 just be len - 3?
> > !                                    * after all
> we
> > haev read 3 bytes of the
> > !                                    * total
> length.
> > !                                    */
> > !                                   tds_get_n(tds,
> > NULL, len-4);
> > !                                  
> tds_get_byte(tds);
> > !                               }
> >   #ifdef WORDS_BIGENDIAN
> >   /*
> >                                 if (major_ver==7)
> {
> > ***************
> > *** 1134,1143 ****
> > 
> >         /* message number */
> >                 /* swap bytes on bigendian systems
> > (mlilback, 11/7/01) */
> >         rc = tds_get_int(tds);
> >   #ifdef WORDS_BIGENDIAN
> > !       tds_swap_bytes(&rc, 4);
> > ! #endif
> >         tds->msg_info->msg_number = rc;
> > 
> >         /* msg state */
> > --- 1161,1176 ----
> > 
> >         /* message number */
> >                 /* swap bytes on bigendian systems
> > (mlilback, 11/7/01) */
> > +               /*
> > +                * we have to be careful not to
> double
> > swap here.
> > +                * tds_get_int will swap out bytes
> for
> > us if emul_little_endian
> 
=== message truncated ===


__________________________________________________
Do You Yahoo!?
Send FREE video emails in Yahoo! Mail!
http://promo.yahoo.com/videomail/