FreeTDS Development Group

[Brian Macy <bmacy AT siterock.com>]

>From what I can tell...
- except for the length of the packet, multi-byte data is in *client*
byte order. I've yet to figure out how the server knows this though.
- the code in freetds/src/read.c tds_get_smallint function is wrong...
it should *always* have the LITTLE_ENDIAN behavior. My suggested
implementation follows.

TDS_SMALLINT tds_get_smallint(TDSSOCKET *tds)
{
unsigned char bytes[2];

        tds_get_n(tds, bytes, 2);
        return *(TDS_SMALLINT*)bytes;
}
/*
** Get a 32 int from the server
*/
TDS_INT tds_get_int(TDSSOCKET *tds)
{
unsigned char bytes[4];

        tds_get_n(tds, bytes, 4);
        return *(TDS_INT*)bytes;
}

- if sizeof(long) != 4 or sizeof(short) != 2, you are just screwed
outright... you can hack freetds/include/tds.h to fix some of it but I
know there are other places in the code that would need to change

- There is a bug in token.c tds_process_cancel... the status is a 2 byte
*not* 1 byte value...as such Perl DBI::finish() and ctlib ct_cancel will
be broken on big endian machines. My suggested fix follows.

int tds_process_cancel(TDSSOCKET *tds)
{
int marker, cancelled=0;

        do {
                marker=tds_get_byte(tds);
                if (marker==TDS_DONE_TOKEN) {
                        TDS_SMALLINT status = tds_get_smallint(tds);
                        cancelled = status & 0x20;
                        tds_get_n(tds,NULL,6);
                } else {
                        tds_process_default_tokens(tds,marker);
                }
        } while (!cancelled);
        tds->state = TDS_COMPLETED;
}

Brian Macy

Justin Cragin wrote:
> 
> I am acctualy having a problem related to this, I have freetds working with
> my app on linux, but when trying to run it on an ultra sparc with sol7 my
> app segfaults, I traced it to the second call of _ct_bind_data() when I am
> getting a debugging print of
>  1inside _ct_bind_data() setting colum size for 32 = -14961268
> And then I segfault on the line
> 
> *((CS_SMALLINT *)curcol->colum_nullbind)=0;
> 
> Is this an endian issue?  If so do you know how I might fix it?
> 
>   thanks, justin cragin
> 
> ---
> You are currently subscribed to freetds as: bmacy AT siterock.com
> To unsubscribe, forward this message to $subst('Email.Unsub')