On 07/25/2012 12:09 PM, Jukka Rissanen wrote:
Hi Danny,
On 07/25/2012 09:36 AM, Danny Jeongseok Seo wrote:
This patch set is to skip DHCP discovery if server address already known.
Bug# 2559 - Enhance Wi-Fi connection
<https://bugs.meego.com/show_bug.cgi?id=25559>
Most of the patches do not apply because they are malformed (long lines
are split). You should use git send-email when sending patches, that way
you also get nicely threaded mails where individual patches are replies
to the cover letter.
1. ipconfig: add last dhcp server address in connman_ipconfig, and with
ipconfig_load/save
2. gdhcp/client: add last server address in GDHCPClient, and Extend
g_dhcp_client_start to have additional argument of last server address
A. Update last server address when DHCP process successfully finished
(src/dhcp.c)
B. In g_dhcp_client_start, DHCP discovery is skipped and do
start_request
when dhcp server address provided
If I am reading the RFC 2131 http://tools.ietf.org/html/rfc2131
correctly in chapter 3.2, the client should just broadcast the
DHCPREQUEST, in this case we must also not set the ciaddr field. So this
patch does not comply with the RFC.
I formulated my reply too vaguely. It seems that the server id is indeed
needed in DHCPREQUEST (chapter 9.7 http://tools.ietf.org/html/rfc2132).
And I also looked in the wrong place in client.c as the code does not
use ciaddr in the dhcp request, so no problem here.
So it seems that the patch complies with the RFC after all. But as
Marcel mentioned we should probably check the lease time before trying
to use the same address.
Cheers,
Jukka
_______________________________________________
connman mailing list
connman@connman.net
http://lists.connman.net/listinfo/connman
