Johnny Billquist wrote:
> Michael L. Hitch wrote:
>> On Sat, 14 Apr 2007, Johnny Billquist wrote:
>>
>>> Hmm. Just checked the log, and Matt committed something yesterday... 
>>> Maybe I was using an old version...? I'll do some more testing.
>>
>>   I think I can say that wasn't it.  If I mount an nfs filesystem 
>> with -r 2048 -w 2048 -I 2048, it gets NULLs in the file.  If I mount 
>> with the default sizes (8192), it works fine.
>
> Yep. Same problem still here as well.
> Interesting enough, looking at the resulting file, every other 4K gets 
> clobbered. Now, the VAX uses a 4K page size these days.
> So it's every other page that gets bad. I wonder if there is something 
> that is assuming 8K of something in NFS underneath?
>
> It would be very interesting to try this on some other architecture to 
> see if the problem exists there as well. Preferrably some architecture 
> with a page size of less than 8K.
>
> This might not be VAX specific...
>
>     Johnny
>

(Apologies if this has already been discussed, still catching up on mail 
after easter ...)

It would be interesting to know how it fairs with 
-o,-r=4096,-w=4096,-I=4096.

-Tobias