linux-kernel - [PATCH 0/2] serial: 8250: Workaround to avoid irq=0 for console

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 5 Jun 2015 09:55:48 +0000
From:	Taichi Kageyama <t-kageyama@...jp.nec.com>
To:	"gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>
CC:	"linux-serial@...r.kernel.org" <linux-serial@...r.kernel.org>,
	"jslaby@...e.cz" <jslaby@...e.cz>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"prarit@...hat.com" <prarit@...hat.com>,
	Naoya Horiguchi <n-horiguchi@...jp.nec.com>
Subject: [PATCH 0/2] serial: 8250: Workaround to avoid irq=0 for console

This patch set provides a workaround to avoid the following problem.
It's based on Linux 4.1-rc3 mainstream kernel.
I've tested this patch set on x86-64 machine and KVM.

Problem
--------------------------
There're cases where autoconfig_irq() fails during boot.
In these cases, the console doesn't work in interrupt mode,
the mode cannot be changed anymore, and "input overrun"
(which can make operation mistakes) happens easily.
This problem happens with high rate every boot once it occurs
because the boot sequence is always almost same.
I saw the original problem on RHEL6.6.

Fix
--------------------------
serial: 8250: Fix autoconfig_irq() to avoid race conditions
  Fix one of autoconfig_irq() failure case.
serial: 8250: Allow to skip autoconfig_irq() for a console
  Provide a workaround to avoid other autoconfig_irq() failure cases.

Conditions of Reproduction
--------------------------
- Build with CONFIG_SERIAL_8250_DETECT_IRQ.
- Need non-PnP console serial or PnP console with no CONFIG_SERIAL_8250_PNP
- Kick printk() repeatedly on the CPU which can handle an interrupt
   from a console serial during autoconfig_irq(). The CPU is basically cpu0.
- Disable the interrupt of the CPU for longer during autoconfig_irq().

Note
--------------------------
Ideally, I think autoconfig_irq() should be fixed completely,
but it's hard from the following points
as long as auto_irq algorithm is used.
  - How long it should wait for an interrrupt
  - How to assure the interrupt of the CPU enabled during auto_irq
  - How to know which CPU can handle an interrupt from a serial.

Do you have any other idea?
In my opinion, providing a workaround is better than
applying big changes to the old function for legacy devices.

Taichi Kageyama (2):
   serial: 8250: Fix autoconfig_irq() to avoid race conditions
   serial: 8250: Allow to skip autoconfig_irq() for a console

  drivers/tty/serial/8250/8250_core.c |   23 +++++++++++++++++++++++
  1 files changed, 23 insertions(+), 0 deletions(-)