RE: [Xen-users] [SEMI-SOLVED] XenSerialConsole help + GRUB2 (Machine locks up on boot)

[Mike Viau <viaum@xxxxxxxxxxxxxxx>]

> Fri, 16 Apr 2010 18:19:59 +0300 <pasik@xxxxxx> wrote:
>
> On Fri, Apr 16, 2010 at 10:36:10AM -0400, Mike Viau wrote:
> > > Thu, 15 Apr 2010 22:37:14 +0300 <pasik@xxxxxx> wrote:
> > >
> > > On Thu, Apr 15, 2010 at 03:24:03PM -0400, Mike Viau wrote:
> > > > > > >
> > > > > >
> > > > > > Attached is grub.cfg.
> > > > > >
> > > > > > Please notice:
> > > > > >
> > > > > > menuentry "Xen 4.0.0 / PVOPS 2.6.32-4-xen-amd64 with a serial console"
> > > > {
> > > > > > insmod ext2
> > > > > > set root=(hd2,1)
> > > > > > search --fs-uuid --set 13f83e98-6e2e-48a5-992e-b266d17bbfd3
> > > > > > multiboot /xen.gz dom0_mem=2048M loglvl=all guest_loglvl=all
> > > > sync_console
> > > > > > console_to_ring com1=115200,8n1,0x1000,0 console=com1
> > > > > > module /vmlinuz-2.6.32-4-xen-amd64
> > > > > > root=UUID=a8129b5a-f8cc-460e-bda4-93cc547b9f7e ro console=hvc0
> > > > > > earlyprintk=xen
> > > > > > module /initrd.img-2.6.32-4-xen-amd64
> > > > > > }
> > > > > >
> > > > > >
> > > > > > I have tried using both 0x1000 and 0x1008 with IRQ 21 and 22 and 0
> > > > > >
> > > > >
> > > > > So do you now get Xen hypervisor boot messages?
> > > >
> > > > Yes! And I have access to the hyper-visor by hitting ctrl+a three
> > times.
> > > >
> > >
> > >
> > > Good. So what was the problem with Xen boot messages earlier? What did
> > you change
> > > in grub settings to get it working?
> > >
> > >
> >
> > I was just not able to get any messages through the serial console at
> > first. I was hoping to just setup my grub config as seen about with the
> > I/O address and IRQ for my PCI serial port card and start getting the Xen
> > boot messages across the serial link.
> >
> > The grub configuration above worked perfectly after I patched my kernel to
> > support my netmos 9865 serial card [*].
>
> Dom0 Linux kernel (or the drivers in it) don't affect Xen _hypervisor_ serial console
> logging at all..

I'll admit it possible, but I noticed it working right off away after booting my patched kernel. Can someone please confirm without a doubt the individual serial port drivers are compiled into the Xen Hypervisor itself?

Again after the Xen hypervisor as well as the Dom0 boot messages appeared after booting the same grub configuration with the new kernel I thought the Xen hypervisor might look into the dom0 kernel image for the serial drivers.

Additionally I was not getting any /dev/ttyS# entries, after using the newly patched kernel I was able to get /dev/ttyS0 from the second physical serial port as the first phyiscal serial port was being used for logging. The card has two physical serial ports in total.

>
> You must have had wrong options in grub.conf earlier..
>
> > So really no changes were really
> > needed by I later decided to remove the "sync_console" and
> > "console_to_ring" options. I also removed "earlyprintk=xen" from the dom0
> > kernel line and replaced "console=hvc0" with "console=hvc0 console=tty0"so
> > that I can get the messages on my screen and serial console.
> >
> > [*] I downloaded the current 2.6.33 stable pvops kernel sources and wrote
> > in the patches. The top of my git log output:
> >
> > commit 0cc94d21a154f2360daf886e7ec0a25fa6f67452
> > Author: root <root@xxxxxxxxxxxxx>
> > Date: Wed Apr 14 19:37:45 2010 -0400
> >
> > 8250_pci: add support for MCS9865 / SYBA 6x Serial Port Card
> >
> > commit d4454747e32cb45d27fe1fae024436660fd6774f
> > Merge: 90d4e45... 19f00f0...
> > Author: Jeremy Fitzhardinge <jeremy.fitzhardinge@xxxxxxxxxx>
> > Date: Fri Apr 2 11:29:33 2010 -0700
> >
> > Merge remote branch 'stable-2.6.33/master' into xen/stable-2.6.33.x
> >
> > * stable-2.6.33/master: (156 commits)
> > Linux 2.6.33.2
> > pata_via: fix VT6410/6415/6330 detection issue
> > GFS2: Skip check for mandatory locks when unlocking
> > classmate-laptop: use a single MODULE_DEVICE_TABLE to get correct
> > aliases
> > x86: Fix sched_clock_cpu for systems with unsynchronized TSC
> > b43: Workaround circular locking in hw-tkip key update callback
> > Bluetooth: Fix kernel crash on L2CAP stress tests
> > Bluetooth: Fix potential bad memory access with sysfs files
> > ahci: use BIOS date in broken_suspend list
> > quota: Fix warning when a delayed write happens before quota is
> > enabled
> > quota: manage reserved space when quota is not active [v2]
> > tcp: Fix tcp_make_synack()
> > tcp: Fix OOB POLLIN avoidance.
> > tcp: Fix tcp_mark_head_lost() with packets == 0
> > skbuff: remove unused dma_head & dma_maps fields
> > net: Potential null skb->dev dereference
> > r8169: offical fix for CVE-2009-4537 (overlength frame DMAs)
> > netlink: fix unaligned access in nla_get_be64()
> > netfilter: ctnetlink: fix reliable event delivery if message
> > building fail
> > netlink: fix NETLINK_RECV_NO_ENOBUFS in netlink_set_err()
> > ...
> >
> > My final question, is it worth submitting that latest patch to xen-devel?
> > Is there a better channel to use? What steps would I take?
> >
>
> You should send the patches to lkml I think. If they're already in the git tree of Linus,
> then they could be easily backported to 2.6.32.x (stable) and then send them for inclusion
> in the next upstream (kernel.org) 2.6.32.x stable release.

I'll do what I can to get the patches into Linus git trees. I have contacted the manufacture (moschip) about the issue but they seem to be taking their time with creating a revised working version of their driver (based on 8250 anyways) in which they plan on posting to their website as apposed to submitting to the Linux kernel team anyways.

>
> Then they'll get to Jeremy's tree automatically when Jeremy pulls from the upstream tree.
>
> -- Pasi
>

---

Stay in touch. Get Hotmail & Messenger on your phone.

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users