Bug 6589 - getifaddrs returns entries with invalid ifa_name
Summary: getifaddrs returns entries with invalid ifa_name
Status: RESOLVED FIXED
Alias: None
Product: glibc
Classification: Unclassified
Component: libc (show other bugs)
Version: unspecified
: P2 normal
Target Milestone: ---
Assignee: Ulrich Drepper
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-06-04 08:53 UTC by Vinay Sridhar
Modified: 2014-07-03 11:28 UTC (History)
2 users (show)

See Also:
Host:
Target:
Build:
Last reconfirmed:
fweimer: security-

Attachments
Patch to add support for infiniband device addresses (277 bytes, patch)
2008-06-04 08:55 UTC, Vinay Sridhar 		Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Vinay Sridhar 2008-06-04 08:53:50 UTC 
On a system with ipv6 interfaces of Infiniband devices, getifaddrs returns
interface names for the ib entries as junk. The reason for this is that the
sockaddr_ll structure fails to accommodate 20 byte IB device addresses. By
extending the sll_addr field (currently 8 bytes) to 24 bytes, the issue is solved.
Comment 1 Vinay Sridhar 2008-06-04 08:55:56 UTC 
Created attachment 2773 [details]
Patch to add support for infiniband device addresses
Comment 2 Ulrich Drepper 2008-06-13 06:18:03 UTC 
This does not work.

First, the kernel defines the structure exactly as we do.

Second, changing this structure changes the ABI.

How does the kernel deal with this?
Comment 3 Vinay Sridhar 2008-06-19 07:54:23 UTC 
Ulrich, 

I came across this thread that describes how the kernel handles 
device addresses > 8 bytes ::
http://lists.openfabrics.org/pipermail/general/2005-September/010986.html

Thanks,
Vinay
Comment 4 Vinay Sridhar 2008-07-22 11:18:04 UTC 
Ulrich,

Any update with regard to the proposed fix? 

Thanks,
Vinay
Comment 5 Ulrich Drepper 2008-08-13 06:58:48 UTC 
I have no way to test it but I hope it's fixed in cvs.  Reopen if necessary with
detailed description of the problem.
Comment 6 Vinay Sridhar 2008-08-14 08:52:53 UTC 
(In reply to comment #5)
> I have no way to test it but I hope it's fixed in cvs.  Reopen if necessary with
> detailed description of the problem.

Tested with cvs. Yes, its fixed.

Thanks,
Vinay