Re: [hatari-devel] MODE SENSE patch

[Uwe Seimet <Uwe.Seimet@xxxxxxxxx>]

By the way, maybe the difference between Linux-68K and Intel-Linux is
related to the byte order? Maybe somewhere in the Hatari code SCSI CDB or
result data are swapped on Intel-Linux where they should not be swapped
because they are already in the correct (i.e. big endian = SCSI) byte order.

> Hi,
> 
> > >> Uh, that's a bit involved, as it was from m68k Linux.  Quick instructions...
> > > 
> > > I'm afraid I am not prepared to spend time on testing on Linux-68k. In this
> > > case I suggest that you simply revert all of my changes.
> > 
> > Sorry, I should have been more clear that this was *not* a regression.
> > 
> > My reply was about letting people know that your SENSE patch did not 
> > resolve that (already known)  issue, and the attached output from your 
> > "full CDB" patch hopefully providing more info, in case somebody will 
> > look at implementing what Linux expects.
> 
> I see :). I indeed thought you were talking about a regression. Without
> knowing the details when looking at the code that displays this message I
> would have expected the message to be independent of the OS (i.e. Intel or
> 68K Linux). You should consider extending the error message so that it
> display the number of bytes returned, e.g.
> 
> Log_Printf(LOG_WARN, "HDC: *** Strange REQUEST SENSE: %d bytes ***!\n"i,
>   nRetLen);
> 
> This may help with analyzing the problem.
> 
> Best regards
> 
> Uwe
> 
>