On Thu, Apr 1, 2010 at 9:13 AM, Garrett D'Amore
<garrett.dam...@oracle.com> wrote:
> On 03/31/10 10:47 PM, Alan Coopersmith wrote:
>>
>> Chris Pickett wrote:
>>
>>>
>>> But I think ulimit -p should not be restricted to certain buffer
>>> sizes, it should be the script author to pick a safe value.
>>> If we use a global kernel tunable for setting the default PIPE_BUF it
>>> could spew a warning to /var/adm/messages if the size is not a
>>> multiple of 5120.
>>>
>>
>> I thought the issue was that POSIX allowed you to make this change
>> globally or per-filesystem, but not per-process:
>> http://mail.opensolaris.org/pipermail/shell-discuss/2009-July/000868.html
>>
>> And you couldn't allow shrinking PIPE_BUF without breaking binary
>> compatibility with apps compiled with the existing value, but could
>> allow increasing it.
>>
>>
>
> Ah, I was only thinking about pipes created with pipe(). FIFOs and other
> such beasties would need to observe the system wide limit. I think
> adjusting *those* values may well be out of scope.
Why should this be out of scope for a kernel tunable? IMO I already
proved that increasing the buffer gives a nice performance benefit.
Chris
--
^---^
(@)v(@) Chris Pickett
| / IT consultant
===m==m=== pkch...@users.sourceforge.net
_______________________________________________
perf-discuss mailing list
perf-discuss@opensolaris.org
