[Samba] cli_pipe: return critical error. Error was Call returnedzerobyte(EOF), hilfe!

[Kurt Pfeifle]

Dragan Krnic wrote:

> Hi Kurt,
> 
>      I've read the new document. It's a very fine 
>      piece of technical documentation. However, it 
>      doesn't help me in the present predicament.
>      Going now by this book I'm only getting more 
>      frustrated. What are your versions of Samba,
>      CUPS(?), OS, and Printer again?
> 
> 2.2.8a,

I had to go back to 2.2.7a because of some printing problems
occuring here (acknowledged by Gerry in a mail to me, saying
he'll fix them for 2.2.9), but can't remember exactly which
one (need to look it up again).

> 1.1.18, SuSE 8.2 w/2.4.20-62, Lexmark C720
> 
>      I'm stuck!
> 
>   I would try to reproduce it here, if you tell me 
>   where I can get the driver files. Do you see a 
>   possibility to place them somewhere for me to 
>   download.
> 
> I can upload the bunch on an FTP server and send you 
> access details per separate mail but I don't think you 
> can reproduce the problem. It has nothing to do with 
> the data. I'm actually looking for pointers as to 
> where to look next.

Please do so.

>      For example adding driver with rcpclient 
>      invariably ends in "Segmentation fault"
>      (Signal 11)
> 
>   I am currently seeing Segmentation faults (on 
>   2.2.7a) also with various rpcclient commands (for 
>   ex. "level" 2 or 3 subcommands for enumprinters 
>   and enumdrivers...). However, adddriver|setdriver 
>   work....
> 
>      immediately
>      after pop_sec_ctx, whose push counterpart was 
>      invoked before a long series of opening and
>      closing all normal files in W32X86/3.
> 
>   Have you tried to install the Windows NT driver 
>   which goes into "W32X86/2/"
> 
> No. What difference would it make? 

Weeeellll, don't know. I personally have only 4 or 5 of these
"3" drivers ever installed via rpcclient (was going OK), but
~ 100-200 of the old "2" sort. The "2" are old-style and accepted
by 2K/XP only in a sort of compatibility mode -- they are running
in Kernel Mode, whereas the "3" run in non-Kernel Mode (or so I
believe). So it *might* be a difference, in that rpcclient may
have a bug related to the one type of drivers, but not to the
other. It was just wild speculation and a desperate attempt to
give some advice....

But now that we have the guilty one....

> My clients are
> all W2Ks.
> 
>     This is the command I used:
>        rpcclient -d 4 -Uroot -c 'adddriver \
>        "Windows NT x86" \
>        "lp:\
> 
>   Just a crazy idea: try to name the thing to something
>   other than "lp"....
> 
> Doesn't make any difference.
> 
>        lexpsnt3.dll:\
>        OPC720.PPD:\
>        LEXPSNTU.DLL:\
>        LEXPSNT.HLP:\
>        NULL:\

here is missing:

          RAW:
>        LEXPSLNG.DLL,LEXDRVX.DLL,LEXFNTRC.DLL,\
>        LEXPSHOW.HLP,LEXPSNT.CNT,PTGUIW32.DLL,\
>        PTRESW32.DLL,PTAPIW32.DLL,LEXMV95.HLP,\
>        DRVNPANT.DLL,LEXDRVIN.EXE,LEXPSNT.ALL,\
>        OPTRASBM.DLL,SC1275BM.DLL,K1220BM.DLL,\
>        OC1200BM.DLL,OC40BM.DLL,OC45BM.DLL,\
>        E855BM.DLL,W810BM.DLL,W820BM.DLL,M410BM.DLL,\
>        T61XBM.DLL,T62XBM.DLL,C710BM.DLL,C720BM.DLL,\
>        C750BM.DLL,C910BM.DLL,E32XBM.DLL,LMPCLTHK.DLL,\
>        LEXCFI.DLL,LEXEDF.DLL,LEXPPDS.ZIP,PTZIPW32.DLL,\
>        LEXUNST.ZIP"' p90
>   Gosh! That is a long list (34) of "Dependentfiles"! 
>   Did they integrate a sound driver to tell you that 
>   the job is ready?!
> 
> It eMails a snapshot of the piled paper begging to be 
> picked up from the tray when the job is done. 

Are you serious? Or just kidding? And this takes 34
additional dependent driver files?? How much in kB gives that?

> Seriously, Kurt, isn't there a backdoor to tdb files?

Not that I know. Some of our great developers need to
step forward on that. This question is too difficult
for a little user like me to answer....

> Can't I just poke the necessary entries with an hex
> editor? It's just a replacement for registries. The
> rpcclient utility has the problem, not samba.

I believe it is neither - nor. It is rather one of the
undersigned...  ;-)

> Cheers
> Dragan
> 
> 

Cheers,
Kurt