public inbox for linux-xfs@vger.kernel.org
 help / color / mirror / Atom feed
* on stack space....
@ 2006-12-01 20:28 Eric Sandeen
  0 siblings, 0 replies; only message in thread
From: Eric Sandeen @ 2006-12-01 20:28 UTC (permalink / raw)
  To: xfs

On the issue of stack space & where it goes in xfs...

<arjan> checkstack does not count function call arguments
<arjan> if you pass big-ish structures to functions things get skewed
quickly
<arjan> the caller pushes the arguments
<arjan> just before the 'call'

and also:

<arjan> btw one thing that helps modern gcc is to move variable
declarations into the inner most {} scope possible
<sandeen> does it really help?
<arjan> it'll allow it to share stack slots
<sandeen> older gccs that was actually -worse-
<arjan> yes
<sandeen> xfs was originally largely written that way
<sandeen> how recent?
<arjan> 4.1 and later have this fixed

lots of irix xfs was written this way IIRC, but I think things got moved
out.  Maybe it'd help to move them back in, for things like xfs_bmapi...

although it's a -penalty- for older gcc's still I think.

Anyway, just found that interesting, might help us get some of this
under control without a huge amount of refactoring...

-Eric

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2006-12-01 20:29 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-12-01 20:28 on stack space Eric Sandeen

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