Comment 7 for bug 591969

This is what I sent to C de-Avillez some minutes ago (somehow it seemed like he sent me a private fwd email). so copypaste:

> Making dd give a better diagnostic for e.g., "bs=3G" on an i686
> system might be more work than it's worth.
Well, this way it's okay that dd does not accept 3G as block size, but then
man dd would document that according to the appropriate behavior.
Like this:
BYTES may be followed by the following multiplicative suffixes: c =1, w =2,
b=512, kB =1000, K =1024, MB =1000*1000, M=1024*1024, xM =M
GB=1000*1000*1000, G =1024*1024*1024, and so on for T, P, E, Z, Y.
The same applies to BLOCKS with the practical exception your
machine(?)/kernel(?)/... needs to be able to allocate memory for (at
least(???)) one(?) block, so, you can expect not to be able to use block sizes
greater or equal to 2G on a 32 bit architecture, and 16E on a 64 bit
architecture.

or something like this. what I read now in man dd is a partial documentation
which would (and had) me expect bs=3G to be OK in all cases, however it isn't
ok in my case.

Thank you for your reply anyways.