Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.kernel > #75473

Bug#998035: linux-image-5.10.0-9-amd64: Debian 11 in Xen PV DomU crashes intel_pmc_core on boot, DomU zombiefies.

Cross-posted to 2 groups.

Show all headers | View raw

[Multipart message — attachments visible in raw view] - view raw

Hi ,

I found a new workaround ...

* With PV and set "maxmem=" with a 4+ GB.

Example:

kernel      = '/usr/lib/grub-xen/grub-x86_64-xen.bin'
extra       = '(hd1)/boot/grub/grub.cfg'
root        = ''

vcpus       = '1'
memory      = '1024'
maxmem      = '4096'


This has the same effect of changing "memory" over 4GB .

Regards,


On Thu, 28 Oct 2021 23:02:07 +0200 
=?utf-8?b?RHIuIE5hZ3kgRWxlbcOpciBLw6Fyb2x5?= <eknagy@omikk.bme.hu> wrote:
 > Package: linux-image-5.10.0-9-amd64
 > Version: linux-image-5.10.0-9-amd64 and linux-image-5.14.0-0.bpo.2-amd64
 > Severity: important
 >
 > Dear All,
 >
 > I am reporting this bug mostly to help others with the same problem, 
proposing
 > adding a warning to the Debian 11 release notes and hoping for an 
upstream
 > kernel bugfix.
 >
 > Description: Debian 11 Xen PV DomU (RAM<4GB) does not correctly shuts 
down
 > because of a intel_pmc_core module problems on Intel Xeon E3-1230 
(and possibly
 > other Intel CPUs).
 >
 > https://github.com/QubesOS/qubes-issues/issues/6052 seems to be the 
same issue.
 >
 > Workarounds:
 > * Use a Debian 10 kernel in the DomU, which works
 > * Allocate 4+ GB RAM to the DomU
 > * Use PVH instead of PV (needs Xen 4.9+, and is the preferred way 
since Xen
 > 4.10)
 >
 > Please note:
 > * Backports kernel (linux-image-5.14.0-0.bpo.2-amd64) suffers from 
the same
 > problem.
 > * Debian 10 Dom0 Xen 4.11.4+107-gef32c7afa2-1 beheaves the same way
 > * Debian 9 Dom0 Xen 4.8.5.final+shim4.10.4-1+deb9u12 used PVHv1, 
which differs
 > from PVHv2 used by Xen 4.09+
 >
 > Test case:
 > * Install Debian 10 or Debian 11, install Xen, create a PV config as 
below and
 > upon startup "BUG: unable to handle page fault for address" is 
displayed and it
 > fails to stop with "poweroff" later.
 > kernel = "/usr/lib/grub-xen/grub-x86_64-xen.bin"
 > extra = '(hd1)/boot/grub/grub.cfg'
 > * Change PV to PVH and it works correctly:
 > kernel = "/root/xen/images/debian11/vmlinuz-5.10.0-9-amd64"
 > ramdisk = "/root/xen/images/debian11/initrd.img-5.10.0-9-amd64"
 > type = 'pvh'
 >
 > The full bug in my case:
 > [ 3.088164] BUG: unable to handle page fault for address: 
ffffc9004049b818
 > [ 3.088175] #PF: supervisor read access in kernel mode
 > [ 3.088179] #PF: error_code(0x0000) - not-present page
 > [ 3.088183] PGD 7fbd9067 P4D 7fbd9067 PUD 5186067 PMD 5303067 PTE 0
 > [ 3.088191] Oops: 0000 [#1] SMP NOPTI
 > [ 3.088195] CPU: 0 PID: 201 Comm: systemd-udevd Not tainted 
5.10.0-9-amd64
 > #1 Debian 5.10.70-1
 > [ 3.088204] RIP: e030:pmc_core_probe+0x136/0x410 [intel_pmc_core]
 > [ 3.088209] Code: c0 48 c7 c7 48 a6 3c c0 e8 c7 25 d2 c0 48 8b 05 b0 
7a 00
 > 00 48 c7 83 88 00 00 00 20 a6 3c c0 48 63 40 50 48 03 05 92 7a 00 00 
<8b> 00 48
 > 8b 15 91 7a 00 00 48 c7 c7 e0 54 3c c0 8b 4a 54 ba 01 00
 > [ 3.088222] RSP: e02b:ffffc9004026fc30 EFLAGS: 00010286
 > [ 3.088226] RAX: ffffc9004049b818 RBX: ffff88800b028400 RCX:
 > 00000000fe002000
 > [ 3.088232] RDX: ffffffffc03ca600 RSI: ffffffffc03c41f6 RDI:
 > ffffffffc03ca648
 > [ 3.088238] RBP: ffff88800b028410 R08: 0000000000000000 R09:
 > 00000000fe001fff
 > [ 3.088244] R10: 0000000000007ff0 R11: ffff888008e01740 R12:

Back to linux.debian.kernel | Previous | Next — Previous in thread | Find similar | Unroll thread

Thread

Bug#998035: linux-image-5.10.0-9-amd64: Debian 11 in Xen PV DomU crashes intel_pmc_core on boot, DomU zombiefies. Dr. Nagy Elemér Károly <eknagy@omikk.bme.hu> - 2021-10-28 23:10 +0200
  Bug#998035: linux-image-5.10.0-9-amd64: Debian 11 in Xen PV DomU crashes intel_pmc_core on boot, DomU zombiefies. "Thierry Jaouen (via Free.fr)" <tjaouen@free.fr> - 2022-06-04 12:40 +0200

csiph-web