[PATCH][BZ #14719] Return EAI_SYSTEM from getaddrinfo if we run out of fds

[Dmitry V. Levin]

On Tue, Apr 02, 2013 at 12:48:44PM +0530, Siddhesh Poyarekar wrote:
> On 1 April 2013 22:41, Dmitry V. Levin <ldv@altlinux.org> wrote:
> > $ ip a
> > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
> >     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> >     inet 127.0.0.1/8 scope host lo
> > $ ip r
> 
> OK, so bring down network *and* unload networking modules.

Or, alternatively, unshare network using CLONE_NEWNET and
create loopback interface there.

> I got an EAI_SYSTEM when I linked against a stale build of last
> month's code (March 26th to be specific).  I did a clean build there
> and it has been giving me EAI_NONAME ever since.  So I'm still not
> able to reproduce the problem the way you are.  Also, are you sure
> that the behaviour before the patch was EAI_AGAIN and not EAI_NONAME?

Sorry, it was actually EAI_NONAME and not EAI_AGAIN, I've been misled
by comment in test-getaddrinfo.c

> I've been getting EAI_NONAME for everything since 2.15 and it seems
> like the right thing to do to me.

I agree, EAI_NONAME would be the right thing.
Unfortunately, the patch makes it return EAI_SYSTEM.


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20130402/4787bd1c/attachment.sig>