linux-kernel - [PATCH 0/2] x86/intel_rdt: Fix cache pseudo-locking error path cleanup

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]

Message-Id: <cover.1530421026.git.reinette.chatre@intel.com>
Date:   Sat, 30 Jun 2018 22:03:01 -0700
From:   Reinette Chatre <reinette.chatre@...el.com>
To:     tglx@...utronix.de, fenghua.yu@...el.com, tony.luck@...el.com,
        vikas.shivappa@...ux.intel.com
Cc:     gavin.hindman@...el.com, jithu.joseph@...el.com,
        dave.hansen@...el.com, mingo@...hat.com, hpa@...or.com,
        x86@...nel.org, linux-kernel@...r.kernel.org,
        Reinette Chatre <reinette.chatre@...el.com>
Subject: [PATCH 0/2] x86/intel_rdt: Fix cache pseudo-locking error path cleanup

Dear Maintainers,

A bug exists in the error handling code during pseudo-lock region creation.
When an error occurs early during pseudo-lock region creation the
pseudo_lock_region struct is not cleaned up properly but remains associated
with the resource group (since it remains in pseudo-locksetup mode).
This partially initialized struct causes problems when other areas
need to obtain resource group data - when partially initialized the
resource group is treated as a pseudo-locked region.

Following is an example of the error being encountered. First a
pseudo-locked region of larger than 4MB is attempted. This fails early
because of lack for support. Since this is not cleaned up properly,
a subsequent attempt fails because it is (incorrectly) believed that
a pseudo-locked region already exists, also the bit_usage file
reports incorrect data.

# mount -t resctrl resctrl /sys/fs/resctrl
# cd /sys/fs/resctrl/
# mkdir p1
# echo 'L3:1=0xffff0' > schemata
# echo pseudo-locksetup > p1/mode
# echo 'L3:1=0xf' > p1/schemata
-bash: echo: write error: Argument list too long
# cat info/last_cmd_status
requested region exceeds maximum size
# echo 'L3:1=0x1' > p1/schemata
-bash: echo: write error: Invalid argument
# cat info/last_cmd_status
pseudo-locked region in hierarchy
# cat info/L3/bit_usage
0=XXSSSSSSSSSSSSSSSSSS;1=XXSSSSSSSSSSSSSSPPPP

After the fixes in this series have been applied:
# mount -t resctrl resctrl /sys/fs/resctrl/
# cd /sys/fs/resctrl/
# mkdir p1
# echo pseudo-locksetup > p1/mode
# echo 'L3:1=0xffff0' > schemata
# echo 'L3:1=0xf' > p1/schemata
-bash: echo: write error: Argument list too long
# cat info/last_cmd_status
requested region exceeds maximum size
# cat info/L3/bit_usage
0=XXSSSSSSSSSSSSSSSSSS;1=XXSSSSSSSSSSSSSS0000
# echo 'L3:1=0x1' > p1/schemata
# cat info/L3/bit_usage
0=XXSSSSSSSSSSSSSSSSSS;1=XXSSSSSSSSSSSSSS000P

Reinette Chatre (2):
  x86/intel_rdt: Move pseudo_lock_region_clear
  x86/intel_rdt: Fix cleanup of plr structure on error

 arch/x86/kernel/cpu/intel_rdt_pseudo_lock.c | 68 ++++++++++++---------
 1 file changed, 40 insertions(+), 28 deletions(-)

-- 
2.17.0