[OpenAFS-devel] MTU probing in src/rx/SOLARIS/rx_knet.c
Derrick Brashear
shadow@gmail.com
Fri, 7 Sep 2007 11:32:02 -0400
------=_Part_350_4384692.1189179122468
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
On 9/7/07, Dale Ghent <daleg@umbc.edu> wrote:
>
>
> This is related to getting the Solaris client working on s10u4...
>
> One of the places where we're using the Solaris Private macro
> ILL_START_WALK_ALL [1] is in the rx_knet.c for Solaris. I see that
> it's used to find the MTU for the interface a packet was received on.
> It looks like it's called for every packet received (!)
>
> I looked at rx_knet.c in the other OS-specific dirs and there's no
> twiddling of the UDP packet size based on the MTU being done in
> those. So I'm wondering... there seems to be a Solaris-specific
> purpose around diddling with the MTU. Is this a work-around due to
> some unique way Solaris handles UDP frags or something? Could, er, it
> be done away with?
>
> [1] ILL_START_WALK_ALL changed with s10u4, where an additional arg
> was added to it. It's called in src/afs/afs_server.c and src/rx/
> SOLARIS/rx_knet.c
did i never commit the changes for this? i never figured out how to
configure for it, but i knew this was coming.
(from looking at mtu discovery, of all things)
------=_Part_350_4384692.1189179122468
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
<br><br><div><span class="gmail_quote">On 9/7/07, <b class="gmail_sendername">Dale Ghent</b> <<a href="mailto:daleg@umbc.edu">daleg@umbc.edu</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>This is related to getting the Solaris client working on s10u4...<br><br>One of the places where we're using the Solaris Private macro<br>ILL_START_WALK_ALL [1] is in the rx_knet.c for Solaris. I see that<br>it's used to find the MTU for the interface a packet was received on.
<br>It looks like it's called for every packet received (!)<br><br>I looked at rx_knet.c in the other OS-specific dirs and there's no<br>twiddling of the UDP packet size based on the MTU being done in<br>those. So I'm wondering... there seems to be a Solaris-specific
<br>purpose around diddling with the MTU. Is this a work-around due to<br>some unique way Solaris handles UDP frags or something? Could, er, it<br>be done away with?<br><br>[1] ILL_START_WALK_ALL changed with s10u4, where an additional arg
<br>was added to it. It's called in src/afs/afs_server.c and src/rx/<br>SOLARIS/rx_knet.c</blockquote><div><br>did i never commit the changes for this? i never figured out how to configure for it, but i knew this was coming.
<br><br>(from looking at mtu discovery, of all things)<br><br></div><br></div><br>
------=_Part_350_4384692.1189179122468--