Hint for autodir (closes RC) and comments about linux-kernel-headers
- To: email@example.com
- Subject: Hint for autodir (closes RC) and comments about linux-kernel-headers
- From: "Francesco P. Lovergine" <firstname.lastname@example.org>
- Date: Mon, 11 Dec 2006 12:17:12 +0100
- Message-id: <20061211111712.GA7385@mithrandir>
After about 20 days of work with both autodir and autofs4 upstream
a final fix for autodir is available (closes #399454).
Incidentally 0.9.8 is almost the same of 0.9.7 (integrating my previous fix
for an header file and an initial trial to fix the above bug) at upstream level
and we are quite confident it does not impact stability in respect with 0.9.7.
Just in case I could provide a 1:0.9.7-1 if required (0.9.7 was the last available
version in testing AFAIK).
About the fix, as communicated in #debian-kernel, just FYI:
<frankie> FYI: #399454 is essentially a bug in the auto_fs4.h header
file, as resulted by talking with autofs and autodir upstreams (API
breakage due to a change in a union used both in v4 and
v5). The issue potentially impact any program built on
2.6.17+ and depending on auto_fs4.h.
<frankie> any program which still uses v4 protocol, indeed
<frankie> autofs upstream is fixing on his side for the kernel, but I
wonder if we need to fix as well etch linux-kernel-headers
<frankie> of course this is not a problem for debian binaries, but i
think developers would not appreciate a broken header in etch for their
buildings and the issue is quite obscure
No answer :-(
This is an issue at kernel headers level, but Ian Kent
has not still provided a final fix for mainstream kernel.
Incidentally not every program which uses auto_fs4.h and the kernel
module is exposed to the breakage, and probably very few programs use autofs4.
Maybe documenting all the issue (but where? linux-kernel-headers is probably
the best choice) and providing the old auto_fs4.h is at this time the most
reasonable thing to do.
Francesco P. Lovergine