[PATCH 0/1][XENIAL] fix lockup in hvc driver on ppc64el

Colin King colin.king at canonical.com
Fri Aug 18 12:22:00 UTC 2017


From: Colin Ian King <colin.king at canonical.com>

BugLink: http://bugs.launchpad.net/bugs/1711401

[SRU REQUEST][XENIAL]

When running stress-ng --dev stressor accessing /dev/hvc1 causes the
stressor to lock up and never terminal.

[FIX]
Upstream commmit:

>From bbc3dfe8805de86874b1a1b1429a002e8670043e
Mon Sep 17 00:00:00 2001
From: Sam Mendoza-Jonas <sam at mendozajonas.com>
Date: Mon, 11 Jul 2016 13:38:57 +1000
Subject: [PATCH] tty/hvc: Use IRQF_SHARED for OPAL hvc consoles

[TESTING]
Without the fix, stress-ng --dev run as root hangs, and one gets an irq
failed error and an oops:

[ 2096.476610] hvc_open: request_irq failed with rc -16.
[ 2096.476634] Unable to handle kernel paging request for data at address 0x000000a8
[ 2096.476641] Faulting instruction address: 0xc000000000b0d8a4
[ 2096.476647] Oops: Kernel access of bad area, sig: 11 [#1]
[ 2096.476650] SMP NR_CPUS=2048 NUMA PowerNV

With the fix, the stressor runs w/o any issues

[REGRESSION POTENTIAL]
This is specific to one driver for ppc64el, so the risk potential is
limited. The fix has limited regression potential as it essentially
marks a flag to allow interrupt to be shared and fundamentally this
is a very minor change to the interrupt handling functionality in the driver.

Sam Mendoza-Jonas (1):
  tty/hvc: Use IRQF_SHARED for OPAL hvc consoles

 drivers/tty/hvc/hvc_console.h | 1 +
 drivers/tty/hvc/hvc_irq.c     | 9 +++++++--
 drivers/tty/hvc/hvc_opal.c    | 3 +++
 3 files changed, 11 insertions(+), 2 deletions(-)

-- 
2.11.0





More information about the kernel-team mailing list