public inbox for linux-mtd@lists.infradead.org
 help / color / mirror / Atom feed
* Re: Fw: Top 100 inline functions (make allyesconfig) was Re: [ANNOUNCE] pahole and other DWARF2 utilities
       [not found] <20061104132050.4950866b.akpm@osdl.org>
@ 2006-11-05  6:32 ` David Woodhouse
  2006-11-05 16:43   ` Arnaldo Carvalho de Melo
  0 siblings, 1 reply; 2+ messages in thread
From: David Woodhouse @ 2006-11-05  6:32 UTC (permalink / raw)
  To: Andrew Morton; +Cc: lwn, linux-mtd, linux-kernel, Arnaldo Carvalho de Melo

On Sat, 2006-11-04 at 13:20 -0800, Andrew Morton wrote:
> cfi_build_cmd() is nutty. 

Damn right it's nutty. Imagine the number of ways you can wire up 1-8
flash chips in either 8-bit or 16-bit mode to a bus which is between 8
and 64 bits wide. Deal with it in software, with a "chip driver"
abstraction which knows what data to put at which address on each
_chip_, and which needs to calculate the corresponding bus data/address.

In the sensible case where you build in support for what you have -- one
interleave, one mode, one bus size -- it's simple. And that's why it's
inline. This is one of the cases where 'allyesconfig' just doesn't make
much sense.

I'm not entirely averse to taking it out-of-line, but show me data on
the interesting case rather than the allyesconfig case.

And tell me about it in about two weeks' time when the sky stops falling
on my head and I get to go home. :)

-- 
dwmw2

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Fw: Top 100 inline functions (make allyesconfig) was Re: [ANNOUNCE] pahole and other DWARF2 utilities
  2006-11-05  6:32 ` Fw: Top 100 inline functions (make allyesconfig) was Re: [ANNOUNCE] pahole and other DWARF2 utilities David Woodhouse
@ 2006-11-05 16:43   ` Arnaldo Carvalho de Melo
  0 siblings, 0 replies; 2+ messages in thread
From: Arnaldo Carvalho de Melo @ 2006-11-05 16:43 UTC (permalink / raw)
  To: David Woodhouse; +Cc: Andrew Morton, lwn, linux-mtd, linux-kernel

On Sun, Nov 05, 2006 at 02:32:17PM +0800, David Woodhouse wrote:
> On Sat, 2006-11-04 at 13:20 -0800, Andrew Morton wrote:
> > cfi_build_cmd() is nutty. 
> 
> Damn right it's nutty. Imagine the number of ways you can wire up 1-8
> flash chips in either 8-bit or 16-bit mode to a bus which is between 8
> and 64 bits wide. Deal with it in software, with a "chip driver"
> abstraction which knows what data to put at which address on each
> _chip_, and which needs to calculate the corresponding bus data/address.
> 
> In the sensible case where you build in support for what you have -- one
> interleave, one mode, one bus size -- it's simple. And that's why it's
> inline. This is one of the cases where 'allyesconfig' just doesn't make
> much sense.
> 
> I'm not entirely averse to taking it out-of-line, but show me data on
> the interesting case rather than the allyesconfig case.

See my Answer to Adrian, it was just extreme testing the tools, I'll use
config files shipped by distros to get even more interesting and down to
earth results.

- Arnaldo

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2006-11-05 16:44 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <20061104132050.4950866b.akpm@osdl.org>
2006-11-05  6:32 ` Fw: Top 100 inline functions (make allyesconfig) was Re: [ANNOUNCE] pahole and other DWARF2 utilities David Woodhouse
2006-11-05 16:43   ` Arnaldo Carvalho de Melo

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox