Re: [PATCH][BZ #16009] fix memory handling in strxfrm_l

[Leonhard Holz <leonhard dot holz at web dot de>]

Here are the updated diffs. Tests look ok but do not complete because of 
an (hopefully unrelated) compile error in string/tester.c.

Am 31.12.2014 um 10:47 schrieb Siddhesh Poyarekar:
> On Wed, Dec 31, 2014 at 10:36:11AM +0100, Leonhard Holz wrote:
> > I want to object here. The needed cache size can be in a region where the
> > malloc will not fail but hurt the system performance considerable
> > (swapping), especially as the malloced size is five times the string size. I
> > think that low level functions like strxfrm should not carry such risks.
>
> Fair enough.  In that case just mention in a comment that the size
> you've chosen is arbitrary and only ensures that it is small enough to
> be contained on the stack.  If someone wants to prove that the size is
> optimal (or another size is optimal) then they can do that on top of
> your change.
>
> Siddhesh

Attachment: sort-test.sh.diff
Description: Text document

Attachment: strxfrm_l.c.diff
Description: Text document

Attachment: xfrm-test.c.diff
Description: Text document