Re: linux-headers upgrade to 3.0.39

----- Mail original -----
> De: "Eric Oberlander" <[email protected]>
> À: "Olaf Westrik" <[email protected]>
> Cc: "IPCop devel" <[email protected]>
> Envoyé: Vendredi 12 Octobre 2012 15:14:46
> Objet: Re: [IPCop-devel] linux-headers upgrade to 3.0.39
> 
> > looking into any remaining issues for 2.1.0.
> > Let us set a deadline of August 18th for 2.1.0-rc1 and (if all goes
> > well) a release of 2.1.0 on September 1st.
> 
> How are we doing on the release candidate?
> 
> It would be good to get the update out before the clocks go back, so
> the latest tzdata is available.
> 
Sorry, I make the planning shift a lot.

I make everyday some progress toward an upgrade to linux-3.4 kernel.
I am building with last 3.4.13 actually.
I will comment wanpipe build until the issues to build with 3.4 are fixed.

syslinux-4.05 doesn't compile with 3.4
syslinux-4.06 should be released at the weekend or early next week.
pre12 have been released a few days away and 4.06-pre13 yesterday.
The alternative is to copy the needed patch from fedora syslinux package tree with 4.05.

I encounter some issues with util-linux-2.22 that have been fixed with util-linux-2.22.1 just released a few days ago.
There was some corner cases badly handled by libmount, now the default on 2.22.
When the kernel running on the building machine is older than 2.6.37, a few issues appear with mount/umount.
I discovered that while building 2.22 and running coreutils tests on a machine that happen to run a 2.6.32 kernel.

While doing that, I learned how to play with loop module.
This case is special as it is possible to trigger loop module load from inside our building chroot (using losetup).

That let me fix one parted-3.1 test that was failing after I fixed /etc/mtab to look better.

I then experimented how to run mdadm and util-linux tests during the build.
The trick is to copy the required modules inside our build tree before the build, so they are available for modprobe during the build. That would work for mdadm actually but not for util-linux without other changes (as modprobe is build after util-linux).

I am unsure of the value of those tests using the machine kernel when the running kernel version is very different from the kernel we build against.
But that should be the right way if running the same kernel version as the one we build.
What I know is that one mdadm test fail : 01raid6integ (and test suite stop).

I will try to commit all my tree work this week-end. From memory, I have in my tree:
- les-451
- linux-3.4.13
- util-linux-2.22.1
- parted-3.1
- psmisc-22.20
- a few glibc-2.11 patches
- the /run change (not yet fully ready actually)
- touch /etc/fstab or fsck complain verbosely during mdadm tests

What I have not planned to upgrade is :
- replace module-init-tool by kmod (no time to work on this)
- replace libusb by libusbx (no time to work on this, no urgency too and I didn't know libusbx exist a few days ago)
- udev upgrade (all distro stay with 175 if I trust the linux-kernel 'udev breakage' topic)
  our udev-166 may not be worst than any recent for our need.

I don't know for :
- patch-2.7.1
- iproute2-3.6.0

Gilles




------------------------------------------------------------------------------
Don't let slow site performance ruin your business. Deploy New Relic APM
Deploy New Relic app performance management and know exactly
what is happening inside your Ruby, Python, PHP, Java, and .NET app
Try New Relic at no cost today and get our sweet Data Nerd shirt too!
http://p.sf.net/sfu/newrelic-dev2dev
_______________________________________________
IPCop-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ipcop-devel