The core i386 NLKD adjustments to pre-existing code.
Signed-Off-By: Jan Beulich <[email protected]>
(actual patch attached)
On Wed, Nov 09, 2005 at 03:11:51PM +0100, Jan Beulich wrote:
> The core i386 NLKD adjustments to pre-existing code.
>
> Signed-Off-By: Jan Beulich <[email protected]>
>
> (actual patch attached)
If your code doesn't work with 4k stacks you have a problem because
8k stacks will soon be removed (my goal is 2.6.16, perhaps one or two
releases later).
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
>>> Adrian Bunk <[email protected]> 09.11.05 20:00:17 >>>
>On Wed, Nov 09, 2005 at 03:11:51PM +0100, Jan Beulich wrote:
>> The core i386 NLKD adjustments to pre-existing code.
>>
>> Signed-Off-By: Jan Beulich <[email protected]>
>>
>> (actual patch attached)
>
>If your code doesn't work with 4k stacks you have a problem because
>8k stacks will soon be removed (my goal is 2.6.16, perhaps one or two
>releases later).
It's not that it doesn't work with them, but chances of stack overflow
are too high for my taste.
Jan
On Thu, Nov 10, 2005 at 09:04:58AM +0100, Jan Beulich wrote:
> >>> Adrian Bunk <[email protected]> 09.11.05 20:00:17 >>>
> >On Wed, Nov 09, 2005 at 03:11:51PM +0100, Jan Beulich wrote:
> >> The core i386 NLKD adjustments to pre-existing code.
> >>
> >> Signed-Off-By: Jan Beulich <[email protected]>
> >>
> >> (actual patch attached)
> >
> >If your code doesn't work with 4k stacks you have a problem because
> >8k stacks will soon be removed (my goal is 2.6.16, perhaps one or two
>
> >releases later).
>
> It's not that it doesn't work with them, but chances of stack overflow
> are too high for my taste.
If there's a chance of a stack overflow the stack usage has to be
reduced until the chance goes down to 0.
> Jan
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
>If there's a chance of a stack overflow the stack usage has to be
>reduced until the chance goes down to 0.
How does one reduce stack usage in the presence of recursion driven by
user input (referring to expression evaluation)?
Also, NLKD has an extension to the (simplistic) pt_regs frame
(including e.g. floating point state) and may be used to debug itself
(i.e. there may be more than one frame on the stack at a time).
Jan
On 2005-11-10T12:52:07, Jan Beulich <[email protected]> wrote:
> >If there's a chance of a stack overflow the stack usage has to be
> >reduced until the chance goes down to 0.
> How does one reduce stack usage in the presence of recursion driven by
> user input (referring to expression evaluation)?
Recursion removal is a pretty standard technique and featured in almost
all introductionary computer science texts. A quick google query finds
http://portal.acm.org/citation.cfm?id=359344
All recursive algorithms can be expressed non-recursively, although it
might not always be as nice. Or you can put an upper limit on the
allowed complexity of queries.
Sincerely,
Lars Marowsky-Br?e <[email protected]>
--
High Availability & Clustering
SUSE Labs, Research and Development
SUSE LINUX Products GmbH - A Novell Business -- Charles Darwin
"Ignorance more frequently begets confidence than does knowledge"