clone 478717 -1
reassign -1 linux-2.6
retitle -1 [hppa] Serious instability issues with NPTL
block 478717 by -1
thanks

While trying to reproduce the issue at hand, I only managed to make paer
(running kernel 2.6.26-1-parisc) deadlock, like what was described with
2.6.18. So, contrary to previous claims, the deadlock issue is not
fixed, which makes the issue basically impossible to debug until the
kernel is fixed.

I don’t know of a PA-RISC emulator, so this only leaves people with
physical access to such a machine able to fix the kernel wrt. threading
support.

I fear that the ruby1.9 packages that were built with a working 2.6.22
kernel will cause similar crashes on other machines when running
threaded applications, and this could also be the case of other
programs. Looking at the kernel logs on paer confirms this, as several
other programs (especially python) are causing oopses. Leaving these
packages as is, without the ability to make security updates during the
lenny cycle, is just hiding the dust under the carpet.

Quoting #debian-release after the bug sprint affectations:
Np237: Congrats on getting the ruby1.9 bug.
he just wanted a excuse to bake cookies since the very beginning
Funnily he's probably the only one getting a new conclusion: drop anarch. :-P

I thought it was ironic at the moment, but actually this is so true. The
current hppa kernel is not suitable for a stable release, and the kernel
maintainers are not in a position where they can fix it. The sane
conclusion may be therefore to remove hppa from the release
architectures so that we can downgrade this bug.

Cheers,
--
.''`.
: :' : We are debian.org. Lower your prices, surrender your code.
`. `' We will add your hardware and software distinctiveness to
`- our own. Resistance is futile.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iD8DBQBJB4KzrSla4ddfhTMRAky0AKDaNKj+YvbHdCzRemH++b euguQYFgCgwJQ9
Na5872/60/3J3xyNobIWR0Y=
=UzsV
-----END PGP SIGNATURE-----