Re: 5.3-RELEASE TODO

From: Kris Kennaway <kris_at_obsecurity.org>
Date: Fri, 24 Sep 2004 00:56:39 -0700
On Fri, Sep 24, 2004 at 01:41:22AM -0600, Scott Long wrote:

>  |--------------------+-----------+-------------+-------------------------|
>  |                    |           |             |There are reports of     |
>  |                    |           |             |applications wedging in  |
>  |                    |           |             |poll() and select() while|
>  |                    |           |             |running the network stack|
>  |                    |           |             |without the Giant lock. A|
>  |                    |           |             |recent sleepq change     |
>  |                    |           |             |appears to have caused   |
>  |                    |           |             |some of the observed     |
>  |                    |           |             |problems to go away      |
>  |poll()/select()     |           |             |(others are difficult to |
>  |application wedge   |In progress|Robert Watson|test for due to recent   |
>  |reports with        |           |             |SMP instability). A fix  |
>  |debug.mpsafenet=1   |           |             |has been committed to CVS|
>  |                    |           |             |HEAD and merged to       |
>  |                    |           |             |RELENG_5 and appears to  |
>  |                    |           |             |resolve problems with    |
>  |                    |           |             |poll(); we are waiting   |
>  |                    |           |             |for feedback that it has |
>  |                    |           |             |corrected the reported   |
>  |                    |           |             |problems with select()   |
>  |                    |           |             |also before moving this  |
>  |                    |           |             |to "testing" status.     |
>  |--------------------+-----------+-------------+-------------------------|

This seems to have gone away for me. 

>  |--------------------+-----------+-------------+-------------------------|
>  |                    |           |             |Kris Kennaway has        |
>  |                    |           |             |reported problems with   |
>  |                    |           |             |boot time panic's, mutex |
>  |boot time panic     |Not done   |-            |Giant not owned at       |
>  |                    |           |             |kern/vfs_subr.c:1365. See|
>  |                    |           |             |"Re: 5.3-RELEASE TODO"   |
>  |                    |           |             |thread in -current_at_.     |
>  |--------------------+-----------+-------------+-------------------------|

So does this.

>  |------------------+-------------+----------------+----------------------|
>  |                  |             |                | The LOR reported in  |
>  |                  |             |                | PR kern/55175 needs  |
>  | filedesc LOR     | Not done    | --             | to be fixed.         |
>  |                  |             |                | Filedesc locking     |
>  |                  |             |                | needs to be heavily  |
>  |                  |             |                | reviewed in general. |

I think the LOR was fixed a long time ago by reverting the code change
that caused it.

>  |-------------------+---------------+--------------+---------------------|
>  |                   |               |              | SCHED_ULE provides  |
>  |                   |               |              | better              |
>  |                   |               |              | interactivity,      |
>  |                   |               |              | higher performance, |
>  | SCHED_ULE as the  | Needs testing | Jeff         | and the ability to  |
>  | default scheduler |               | Roberson     | support pinning and |
>  |                   |               |              | affinity. Basic HTT |
>  |                   |               |              | scheduling policies |
>  |                   |               |              | should be in place  |
>  |                   |               |              | for 5.3 also.       |
>  |-------------------+---------------+--------------+---------------------|

This is OBE (although ULE is still known to be broken).

Kris


Received on Fri Sep 24 2004 - 05:56:23 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:13 UTC