hdparm utilizes shared memory for no good reason ?

[Rob Landley]

On Tuesday 16 May 2006 7:49 am, Tito wrote:
> On Tuesday 16 May 2006 08:19, Mike Frysinger wrote:
> > was trying to use `hdparm -t` on a uClinux host when it failed with:
> > could not allocate sharedmem buf
> >
> > this is because hdparm uses shm* functions to get a buffer of memory ...
> > but afaict, there's no reason a buffer created on the stack/heap wouldnt
> > work just as well
> >
> > could someone enlighten me as to why the shm* stuff is needed ?
> > -mike
>
> It is what  the real hdparm does,  after Rob applies my pending shrinkage
> and bugfix patch I can try to write a fix for that ...

Sorry, got distracted by md5sum yesterday.

Applied, svn 15089.

> if you haven't one 
> already. Please notice that the patch contains an update for the timing
> functions to the code of hdparm 6.6+, but i'm sure it uses the shm* stuff
> the same.

I only reviewed about 1/3 of it (it's big but seemed sane) and then made sure 
it compiled and ran make bloatcheck against it.  At this point, you know more 
about hdparm than I do.

P.S.  My ideal size info for a patch is "make bloatcheck" on a defconfig 
before and after the patch is applied.  It's not a requirement, but if you're 
going to provide size info anyway that's what I find most useful.  FYI.

> Ciao,
> Tito

Rob
-- 
Never bet against the cheap plastic solution.