openjdk team mailing list archive

Thread
Date

Bug#574355: openjdk-6-jdk: eclipse networking failure

To: Luca Tettamanti <kronos.it@xxxxxxxxx>, 574355@xxxxxxxxxxxxxxx
From: Torsten Werner <mail.twerner@xxxxxxxxxxxxxx>
Date: Thu, 18 Mar 2010 07:16:53 +0100
In-reply-to: <20100317150410.GA5230@nb-core2.darkstar.lan>
Reply-to: Torsten Werner <mail.twerner@xxxxxxxxxxxxxx>, 574355@xxxxxxxxxxxxxxx
Resent-cc: OpenJDK Team <openjdk@xxxxxxxxxxxxxxxxxxx>, Torsten Werner <twerner@xxxxxxxxxx>
Resent-date: Thu, 18 Mar 2010 06:18:03 +0000
Resent-date: Thu, 18 Mar 2010 06:18:05 +0000
Resent-from: Torsten Werner <mail.twerner@xxxxxxxxxxxxxx>
Resent-message-id: <handler.574355.B574355.126889302425150@xxxxxxxxxxxxxxx>
Resent-sender: Debian BTS <debbugs@xxxxxxxxxxxxxxxx>
Resent-to: debian-bugs-dist@xxxxxxxxxxxxxxxx

Hi,

On Wed, Mar 17, 2010 at 4:04 PM, Luca Tettamanti <kronos.it@xxxxxxxxx> wrote:
> Instead I suspect that it's caused by the workaround for that bug
> (ipv4-mapped-ipv6-addresses.diff)

yes and that will be fixed with the next upload.

But there is an inherent IPv6 problem with eclipse itself. It executes

socket(PF_INET6, SOCK_STREAM, IPPROTO_IP) = 78
bind(78, {sa_family=AF_INET6, sin6_port=htons(0), inet_pton(AF_INET6,
"::", &sin6_addr), sin6_flowinfo=0, sin6_scope_id=0}, 28) = 0

before connecting to an IPv4 address. This is bound to fail with
net.ipv6.bindv6only switched on. (But it will work again with
net.ipv6.bindv6only switched off.)

Cheers,
Torsten