[PATCH] hdparm utilizes shared memory for no good reason ?

[Rob Landley]

On Tuesday 16 May 2006 5:08 pm, Paul Fox wrote:
>  > On Tue May 16 2006 15:23, Tito wrote:
>  > > Hi, attached are 2 patches to resolve this hdparm shared memory
>  > > problem.
>
>  ...
>
>  > But this is BB not a hard drive diagnostic/benchmark tool - you only
>  > need go/no-go measurements of an option change.  Userspace buffers
>  > should be fine.
>
> if the options are left in, and the results are arguably suspect, then
> the tool should print a big fat warning that it's not the same as
> the "real" hdparm.

Calm down.  This is just a timing loop to tell you how fast your hard drive 
can read data.  People generally run this kind of thing 2 or three times to 
get any sort of number anyway.  Something else could be using the hard drive 
while you run the test.

I'd delete a tool before I added a "big fat warning" to it.  It either works 
or it doesn't work.  Having a tool and then warning people not to use it is 
just _stupid_.  Benchmarking disk throughput is something you could do this 
with dd.

Also, there is no "real" hdparm.  That's totally the wrong _conceptual_ 
approach.  If the busybox version can't do the job then we don't have the 
tool, but if there's just an implementation that you can't even DERIVE a spec 
from then I'd argue nobody has any business using the tool at all.

Rob
-- 
Never bet against the cheap plastic solution.