kernel/kernel-generic.git
10 years agodrm/radeon/atom: select the proper number of lanes in transmitter setup
Alex Deucher [Thu, 6 Mar 2014 23:09:52 +0000 (18:09 -0500)]
drm/radeon/atom: select the proper number of lanes in transmitter setup

commit d03874c881a049a50e12f285077ab1f9fc2686e1 upstream.

We need to check for DVI vs. HDMI when setting up duallink since
HDMI is single link only.  Fixes 4k modes on newer asics.

bug:
https://bugs.freedesktop.org/show_bug.cgi?id=75223

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofs/proc/base.c: fix GPF in /proc/$PID/map_files
Artem Fetishev [Mon, 10 Mar 2014 22:49:45 +0000 (15:49 -0700)]
fs/proc/base.c: fix GPF in /proc/$PID/map_files

commit 70335abb2689c8cd5df91bf2d95a65649addf50b upstream.

The expected logic of proc_map_files_get_link() is either to return 0
and initialize 'path' or return an error and leave 'path' uninitialized.

By the time dname_to_vma_addr() returns 0 the corresponding vma may have
already be gone.  In this case the path is not initialized but the
return value is still 0.  This results in 'general protection fault'
inside d_path().

Steps to reproduce:

  CONFIG_CHECKPOINT_RESTORE=y

    fd = open(...);
    while (1) {
        mmap(fd, ...);
        munmap(fd, ...);
    }

  ls -la /proc/$PID/map_files

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=68991

Signed-off-by: Artem Fetishev <artem_fetishev@epam.com>
Signed-off-by: Aleksandr Terekhov <aleksandr_terekhov@epam.com>
Reported-by: <wiebittewas@gmail.com>
Acked-by: Pavel Emelyanov <xemul@parallels.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiscsi-target: Fix iscsit_get_tpg_from_np tpg_state bug
Nicholas Bellinger [Wed, 26 Feb 2014 11:09:41 +0000 (03:09 -0800)]
iscsi-target: Fix iscsit_get_tpg_from_np tpg_state bug

commit a2a99cea5ec7c1e47825559f0e75a4efbcf8aee3 upstream.

This patch fixes a bug in iscsit_get_tpg_from_np() where the
tpg->tpg_state sanity check was looking for TPG_STATE_FREE,
instead of != TPG_STATE_ACTIVE.

The latter is expected during a normal TPG shutdown once the
tpg_state goes into TPG_STATE_INACTIVE in order to reject any
new incoming login attempts.

Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiser-target: Fix post_send_buf_count for RDMA READ/WRITE
Nicholas Bellinger [Thu, 27 Feb 2014 17:05:03 +0000 (09:05 -0800)]
iser-target: Fix post_send_buf_count for RDMA READ/WRITE

commit b6b87a1df604678ed1be40158080db012a99ccca upstream.

This patch fixes the incorrect setting of ->post_send_buf_count
related to RDMA WRITEs + READs where isert_rdma_rw->send_wr_num
was not being taken into account.

This includes incrementing ->post_send_buf_count within
isert_put_datain() + isert_get_dataout(), decrementing within
__isert_send_completion() + isert_response_completion(), and
clearing wr->send_wr_num within isert_completion_rdma_read()

This is necessary because even though IB_SEND_SIGNALED is
not set for RDMA WRITEs + READs, during a QP failure event
the work requests will be returned with exception status
from the TX completion queue.

Acked-by: Sagi Grimberg <sagig@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiser-target: Ignore completions for FRWRs in isert_cq_tx_work
Nicholas Bellinger [Thu, 27 Feb 2014 15:02:48 +0000 (07:02 -0800)]
iser-target: Ignore completions for FRWRs in isert_cq_tx_work

commit 9bb4ca68fc48eea618b1af1c1cde2a251ae32d1b upstream.

This patch changes IB_WR_FAST_REG_MR + IB_WR_LOCAL_INV related
work requests to include a ISER_FRWR_LI_WRID value in order to
signal isert_cq_tx_work() that these requests should be ignored.

This is necessary because even though IB_SEND_SIGNALED is not
set for either work request, during a QP failure event the work
requests will be returned with exception status from the TX
completion queue.

v2 changes:
 - Rename ISER_FRWR_LI_WRID -> ISER_FASTREG_LI_WRID (Sagi)

Acked-by: Sagi Grimberg <sagig@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiscsi/iser-target: Fix isert_conn->state hung shutdown issues
Nicholas Bellinger [Mon, 3 Feb 2014 20:54:39 +0000 (12:54 -0800)]
iscsi/iser-target: Fix isert_conn->state hung shutdown issues

commit defd884845297fd5690594bfe89656b01f16d87e upstream.

This patch addresses a couple of different hug shutdown issues
related to wait_event() + isert_conn->state.  First, it changes
isert_conn->conn_wait + isert_conn->conn_wait_comp_err from
waitqueues to completions, and sets ISER_CONN_TERMINATING from
within isert_disconnect_work().

Second, it splits isert_free_conn() into isert_wait_conn() that
is called earlier in iscsit_close_connection() to ensure that
all outstanding commands have completed before continuing.

Finally, it breaks isert_cq_comp_err() into seperate TX / RX
related code, and adds logic in isert_cq_rx_comp_err() to wait
for outstanding commands to complete before setting ISER_CONN_DOWN
and calling complete(&isert_conn->conn_wait_comp_err).

Acked-by: Sagi Grimberg <sagig@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiscsi/iser-target: Use list_del_init for ->i_conn_node
Nicholas Bellinger [Mon, 3 Feb 2014 20:53:51 +0000 (12:53 -0800)]
iscsi/iser-target: Use list_del_init for ->i_conn_node

commit 5159d763f60af693a3fcec45dce2021f66e528a4 upstream.

There are a handful of uses of list_empty() for cmd->i_conn_node
within iser-target code that expect to return false once a cmd
has been removed from the per connect list.

This patch changes all uses of list_del -> list_del_init in order
to ensure that list_empty() returns false as expected.

Acked-by: Sagi Grimberg <sagig@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoACPI / sleep: Add extra checks for HW Reduced ACPI mode sleep states
Rafael J. Wysocki [Thu, 13 Mar 2014 21:11:39 +0000 (22:11 +0100)]
ACPI / sleep: Add extra checks for HW Reduced ACPI mode sleep states

commit a4e90bed511220ff601d064c9e5d583e91308f65 upstream.

If the HW Reduced ACPI mode bit is set in the FADT, ACPICA uses
the optional sleep control and sleep status registers for making
the system enter sleep states (including S5), so it is not possible
to use system sleep states or power it off using ACPI if the HW
Reduced ACPI mode bit is set and those registers are not available.

For this reason, add a new function, acpi_sleep_state_supported(),
checking if the HW Reduced ACPI mode bit is set and whether or not
system sleep states are usable in that case in addition to checking
the return value of acpi_get_sleep_type_data() and make the ACPI
sleep setup routines use that function to check the availability of
system sleep states.

Among other things, this prevents the kernel from attempting to
use ACPI for powering off HW Reduced ACPI systems without the sleep
control and sleep status registers, because ACPI power off doesn't
have a chance to work on them.  That allows alternative power off
mechanisms that may actually work to be used on those systems.  The
affected machines include Dell Venue 8 Pro, Asus T100TA, Haswell
Desktop SDP and Ivy Bridge EP Demo depot.

References: https://bugzilla.kernel.org/show_bug.cgi?id=70931
Reported-by: Adam Williamson <awilliam@redhat.com>
Tested-by: Aubrey Li <aubrey.li@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoNFSv4: nfs4_stateid_is_current should return 'true' for an invalid stateid
Trond Myklebust [Wed, 5 Mar 2014 13:44:23 +0000 (08:44 -0500)]
NFSv4: nfs4_stateid_is_current should return 'true' for an invalid stateid

commit e1253be0ece1a95a02c7f5843194877471af8179 upstream.

When nfs4_set_rw_stateid() can fails by returning EIO to indicate that
the stateid is completely invalid, then it makes no sense to have it
trigger a retry of the READ or WRITE operation. Instead, we should just
have it fall through and attempt a recovery.

This fixes an infinite loop in which the client keeps replaying the same
bad stateid back to the server.

Reported-by: Andy Adamson <andros@netapp.com>
Link: http://lkml.kernel.org/r/1393954269-3974-1-git-send-email-andros@netapp.com
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoNFS: Fix a delegation callback race
Trond Myklebust [Mon, 3 Mar 2014 03:03:12 +0000 (22:03 -0500)]
NFS: Fix a delegation callback race

commit 755a48a7a4eb05b9c8424e3017d947b2961a60e0 upstream.

The clean-up in commit 36281caa839f ended up removing a NULL pointer check
that is needed in order to prevent an Oops in
nfs_async_inode_return_delegation().

Reported-by: "Yan, Zheng" <zheng.z.yan@intel.com>
Link: http://lkml.kernel.org/r/5313E9F6.2020405@intel.com
Fixes: 36281caa839f (NFSv4: Further clean-ups of delegation stateid validation)
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agospi: spi-ath79: fix initial GPIO CS line setup
Gabor Juhos [Sun, 2 Mar 2014 19:54:42 +0000 (20:54 +0100)]
spi: spi-ath79: fix initial GPIO CS line setup

commit 61d1cf163c8653934cc8cd5d0b2a562d0990c265 upstream.

The 'ath79_spi_setup_cs' function initializes the chip
select line of a given SPI device in order to make sure
that the device is inactive.

If the SPI_CS_HIGH bit is set for a given device, it
means that the CS line of that device is active HIGH
so it must be set to LOW initially. In case of GPIO
CS lines, the 'ath79_spi_setup_cs' function does the
opposite of that due to the wrong GPIO flags.

Fix the code to use the correct GPIO flags.

Reported-by: Ronald Wahl <ronald.wahl@raritan.com>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agospi: spi-imx: spi_imx_remove: do not disable disabled clocks
Philippe De Muyter [Thu, 27 Feb 2014 09:16:15 +0000 (10:16 +0100)]
spi: spi-imx: spi_imx_remove: do not disable disabled clocks

commit fd40dccb1a170ba689664481a3de83617b7194d2 upstream.

Currently, at module removal, one gets the following warnings:
------------[ cut here ]------------
WARNING: at drivers/clk/clk.c:780 clk_disable+0x18/0x24()
Modules linked in: spi_imx(-) [last unloaded: ev76c560]
CPU: 1 PID: 16337 Comm: rmmod Tainted: G        W    3.10.17-80548-g90191eb-dirty #33
[<80013b4c>] (unwind_backtrace+0x0/0xf8) from [<800115dc>] (show_stack+0x10/0x14)
[<800115dc>] (show_stack+0x10/0x14) from [<800257b8>] (warn_slowpath_common+0x4c/0x68)
[<800257b8>] (warn_slowpath_common+0x4c/0x68) from [<800257f0>] (warn_slowpath_null+0x1c/0x24)
[<800257f0>] (warn_slowpath_null+0x1c/0x24) from [<803f60ec>] (clk_disable+0x18/0x24)
[<803f60ec>] (clk_disable+0x18/0x24) from [<7f02c9cc>] (spi_imx_remove+0x54/0x9c [spi_imx])
[<7f02c9cc>] (spi_imx_remove+0x54/0x9c [spi_imx]) from [<8025868c>] (platform_drv_remove+0x18/0x1c)
[<8025868c>] (platform_drv_remove+0x18/0x1c) from [<80256f60>] (__device_release_driver+0x70/0xcc)
[<80256f60>] (__device_release_driver+0x70/0xcc) from [<80257770>] (driver_detach+0xcc/0xd0)
[<80257770>] (driver_detach+0xcc/0xd0) from [<80256d90>] (bus_remove_driver+0x7c/0xc0)
[<80256d90>] (bus_remove_driver+0x7c/0xc0) from [<80068668>] (SyS_delete_module+0x144/0x1f8)
[<80068668>] (SyS_delete_module+0x144/0x1f8) from [<8000e080>] (ret_fast_syscall+0x0/0x30)
---[ end trace 1f5df9ad54996300 ]---
------------[ cut here ]------------
WARNING: at drivers/clk/clk.c:780 clk_disable+0x18/0x24()
Modules linked in: spi_imx(-) [last unloaded: ev76c560]
CPU: 1 PID: 16337 Comm: rmmod Tainted: G        W    3.10.17-80548-g90191eb-dirty #33
[<80013b4c>] (unwind_backtrace+0x0/0xf8) from [<800115dc>] (show_stack+0x10/0x14)
[<800115dc>] (show_stack+0x10/0x14) from [<800257b8>] (warn_slowpath_common+0x4c/0x68)
[<800257b8>] (warn_slowpath_common+0x4c/0x68) from [<800257f0>] (warn_slowpath_null+0x1c/0x24)
[<800257f0>] (warn_slowpath_null+0x1c/0x24) from [<803f60ec>] (clk_disable+0x18/0x24)
[<803f60ec>] (clk_disable+0x18/0x24) from [<7f02c9e8>] (spi_imx_remove+0x70/0x9c [spi_imx])
[<7f02c9e8>] (spi_imx_remove+0x70/0x9c [spi_imx]) from [<8025868c>] (platform_drv_remove+0x18/0x1c)
[<8025868c>] (platform_drv_remove+0x18/0x1c) from [<80256f60>] (__device_release_driver+0x70/0xcc)
[<80256f60>] (__device_release_driver+0x70/0xcc) from [<80257770>] (driver_detach+0xcc/0xd0)
[<80257770>] (driver_detach+0xcc/0xd0) from [<80256d90>] (bus_remove_driver+0x7c/0xc0)
[<80256d90>] (bus_remove_driver+0x7c/0xc0) from [<80068668>] (SyS_delete_module+0x144/0x1f8)
[<80068668>] (SyS_delete_module+0x144/0x1f8) from [<8000e080>] (ret_fast_syscall+0x0/0x30)
---[ end trace 1f5df9ad54996301 ]---

Since commit 9e556dcc55774c9a1032f32baa0e5cfafede8b70, "spi: spi-imx: only
enable the clocks when we start to transfer a message", clocks are always
disabled except when transmitting messages.  There is thus no need to
disable them at module removal.

Fixes: 9e556dcc55774 (spi: spi-imx: only enable the clocks when we start to transfer a message)
Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Acked-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agospi: fsl-dspi: Fix getting correct address for master
Axel Lin [Fri, 14 Feb 2014 04:49:12 +0000 (12:49 +0800)]
spi: fsl-dspi: Fix getting correct address for master

commit 017145fef567430789e40f6a22a90ce2a766370b upstream.

Current code set platform drvdata to dspi. However, the code in dspi_suspend()
and dspi_resume() assumes the drvdata is the address of master.
Fix it by setting platform drvdata to master.

Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agospi: coldfire-qspi: Fix getting correct address for *mcfqspi
Axel Lin [Fri, 14 Feb 2014 01:53:00 +0000 (09:53 +0800)]
spi: coldfire-qspi: Fix getting correct address for *mcfqspi

commit ee73b4c6e3fc0755a91752ab8eebc8e070038b53 upstream.

dev_get_drvdata() returns the address of master rather than mcfqspi.

Fixes: af361079 (spi/coldfire-qspi: Drop extra calls to spi_master_get in suspend/resume functions)
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agolibata: use wider match for blacklisting Crucial M500
Tejun Heo [Mon, 10 Mar 2014 15:13:43 +0000 (11:13 -0400)]
libata: use wider match for blacklisting Crucial M500

commit 83493d7e782d2630f1a55def14a79f0e7c4faac3 upstream.

We're now blacklisting "Crucial_CT???M500SSD1" and
"Crucial_CT???M500SSD3".  Also, "Micron_M500*" is blacklisted which is
about the same devices as the crucial branded ones.  Let's merge the
two Crucial M500 entries and widen the match to
"Crucial_CT???M500SSD*" so that we don't have to fiddle with new
entries for similar devices.

Signed-off-by: Tejun Heo <tj@kernel.org>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agolibata: add ATA_HORKAGE_BROKEN_FPDMA_AA quirk for Seagate Momentus SpinPoint M8 ...
Michele Baldessari [Fri, 7 Mar 2014 16:34:29 +0000 (16:34 +0000)]
libata: add ATA_HORKAGE_BROKEN_FPDMA_AA quirk for Seagate Momentus SpinPoint M8 (2BA30001)

commit b28a613e9138e4b3a64649bd60b13436f4b4b49b upstream.

Via commit 87809942d3fa "libata: add ATA_HORKAGE_BROKEN_FPDMA_AA quirk
for Seagate Momentus SpinPoint M8" we added a quirk for disks named
"ST1000LM024 HN-M101MBB" with firmware revision "2AR10001".

As reported on https://bugzilla.redhat.com/show_bug.cgi?id=1073901,
we need to also add firmware revision 2BA30001 as it is broken as well.

Reported-by: Nicholas <arealityfarbetween@googlemail.com>
Signed-off-by: Michele Baldessari <michele@acksyn.org>
Tested-by: Guilherme Amadio <guilherme.amadio@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agolibata: disable queued TRIM for Crucial M500 mSATA SSDs
Marios Andreopoulos [Mon, 3 Mar 2014 16:19:59 +0000 (18:19 +0200)]
libata: disable queued TRIM for Crucial M500 mSATA SSDs

commit 2564338b13e6e132ee224edb63e1e872adf431f4 upstream.

Queued TRIM commands cause problems and silent file system corruption
on Crucial M500 SSDs. This patch disables them for the mSATA model of
the drive.

Signed-off-by: Marios Andreopoulos <opensource@andmarios.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=71371
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofirewire: don't use PREPARE_DELAYED_WORK
Tejun Heo [Fri, 7 Mar 2014 15:19:57 +0000 (10:19 -0500)]
firewire: don't use PREPARE_DELAYED_WORK

commit 70044d71d31d6973665ced5be04ef39ac1c09a48 upstream.

PREPARE_[DELAYED_]WORK() are being phased out.  They have few users
and a nasty surprise in terms of reentrancy guarantee as workqueue
considers work items to be different if they don't have the same work
function.

firewire core-device and sbp2 have been been multiplexing work items
with multiple work functions.  Introduce fw_device_workfn() and
sbp2_lu_workfn() which invoke fw_device->workfn and
sbp2_logical_unit->workfn respectively and always use the two
functions as the work functions and update the users to set the
->workfn fields instead of overriding work functions using
PREPARE_DELAYED_WORK().

This fixes a variety of possible regressions since a2c1c57be8d9
"workqueue: consider work function when searching for busy work items"
due to which fw_workqueue lost its required non-reentrancy property.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Cc: linux1394-devel@lists.sourceforge.net
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofirewire: ohci: fix probe failure with Agere/LSI controllers
Stefan Richter [Thu, 6 Mar 2014 19:39:04 +0000 (20:39 +0100)]
firewire: ohci: fix probe failure with Agere/LSI controllers

commit 0ca49345b6f489e95f8d6edeb0b092e257475b2a upstream.

Since commit bd972688eb24
"firewire: ohci: Fix 'failed to read phy reg' on FW643 rev8",
there is a high chance that firewire-ohci fails to initialize LSI née
Agere controllers.
https://bugzilla.kernel.org/show_bug.cgi?id=65151

Peter Hurley points out the reason:  IEEE 1394a:2000 clause 5A.1 (or
IEEE 1394:2008 clause 17.2.1) say:  "The PHY shall insure that no more
than 10 ms elapse from the reassertion of LPS until the interface is
reset.  The link shall not assert LReq until the reset is complete."
In other words, the link needs to give the PHY at least 10 ms to get
the interface operational.

With just the msleep(1) in bd972688eb24, the first read_phy_reg()
during ohci_enable() may happen before the phy-link interface reset was
finished, and fail.  Due to the high variability of msleep(n) with small
n, this failure was not fully reproducible, and not apparent at all with
low CONFIG_HZ setting.

On the other hand, Peter can no longer reproduce the issue with FW643
rev8.  The read phy reg failures that happened back then may have had an
unrelated cause.  So, just revert bd972688eb24, except for the valid
comment on TSB82AA2 cards.

Reported-by: Mikhail Gavrilov
Reported-by: Jay Fenlason <fenlason@redhat.com>
Reported-by: Clemens Ladisch <clemens@ladisch.de>
Reported-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agofirewire: net: fix use after free
Stefan Richter [Tue, 18 Feb 2014 21:25:15 +0000 (22:25 +0100)]
firewire: net: fix use after free

commit 8987583366ae9e03c306c2b7d73bdb952df1d08d upstream.

Commit 8408dc1c14c1 "firewire: net: use dev_printk API" introduced a
use-after-free in a failure path.  fwnet_transmit_packet_failed(ptask)
may free ptask, then the dev_err() call dereferenced it.  The fix is
straightforward; simply reorder the two calls.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotracing: Do not add event files for modules that fail tracepoints
Steven Rostedt (Red Hat) [Wed, 26 Feb 2014 18:37:38 +0000 (13:37 -0500)]
tracing: Do not add event files for modules that fail tracepoints

commit 45ab2813d40d88fc575e753c38478de242d03f88 upstream.

If a module fails to add its tracepoints due to module tainting, do not
create the module event infrastructure in the debugfs directory. As the events
will not work and worse yet, they will silently fail, making the user wonder
why the events they enable do not display anything.

Having a warning on module load and the events not visible to the users
will make the cause of the problem much clearer.

Link: http://lkml.kernel.org/r/20140227154923.265882695@goodmis.org
Fixes: 6d723736e472 "tracing/events: add support for modules to TRACE_EVENT"
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoACPI / EC: Clear stale EC events on Samsung systems
Kieran Clancy [Fri, 28 Feb 2014 14:12:28 +0000 (00:42 +1030)]
ACPI / EC: Clear stale EC events on Samsung systems

commit ad332c8a45330d170bb38b95209de449b31cd1b4 upstream.

A number of Samsung notebooks (530Uxx/535Uxx/540Uxx/550Pxx/900Xxx/etc)
continue to log events during sleep (lid open/close, AC plug/unplug,
battery level change), which accumulate in the EC until a buffer fills.
After the buffer is full (tests suggest it holds 8 events), GPEs stop
being triggered for new events. This state persists on wake or even on
power cycle, and prevents new events from being registered until the EC
is manually polled.

This is the root cause of a number of bugs, including AC not being
detected properly, lid close not triggering suspend, and low ambient
light not triggering the keyboard backlight. The bug also seemed to be
responsible for performance issues on at least one user's machine.

Juan Manuel Cabo found the cause of bug and the workaround of polling
the EC manually on wake.

The loop which clears the stale events is based on an earlier patch by
Lan Tianyu (see referenced attachment).

This patch:
 - Adds a function acpi_ec_clear() which polls the EC for stale _Q
   events at most ACPI_EC_CLEAR_MAX (currently 100) times. A warning is
   logged if this limit is reached.
 - Adds a flag EC_FLAGS_CLEAR_ON_RESUME which is set to 1 if the DMI
   system vendor is Samsung. This check could be replaced by several
   more specific DMI vendor/product pairs, but it's likely that the bug
   affects more Samsung products than just the five series mentioned
   above. Further, it should not be harmful to run acpi_ec_clear() on
   systems without the bug; it will return immediately after finding no
   data waiting.
 - Runs acpi_ec_clear() on initialisation (boot), from acpi_ec_add()
 - Runs acpi_ec_clear() on wake, from acpi_ec_unblock_transactions()

References: https://bugzilla.kernel.org/show_bug.cgi?id=44161
References: https://bugzilla.kernel.org/show_bug.cgi?id=45461
References: https://bugzilla.kernel.org/show_bug.cgi?id=57271
References: https://bugzilla.kernel.org/attachment.cgi?id=126801
Suggested-by: Juan Manuel Cabo <juanmanuel.cabo@gmail.com>
Signed-off-by: Kieran Clancy <clancy.kieran@gmail.com>
Reviewed-by: Lan Tianyu <tianyu.lan@intel.com>
Reviewed-by: Dennis Jansen <dennis.jansen@web.de>
Tested-by: Kieran Clancy <clancy.kieran@gmail.com>
Tested-by: Juan Manuel Cabo <juanmanuel.cabo@gmail.com>
Tested-by: Dennis Jansen <dennis.jansen@web.de>
Tested-by: Maurizio D'Addona <mauritiusdadd@gmail.com>
Tested-by: San Zamoyski <san@plusnet.pl>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoACPI / resources: ignore invalid ACPI device resources
Zhang Rui [Thu, 27 Feb 2014 03:37:15 +0000 (11:37 +0800)]
ACPI / resources: ignore invalid ACPI device resources

commit b355cee88e3b1a193f0e9a81db810f6f83ad728b upstream.

ACPI table may export resource entry with 0 length.
But the current code interprets this kind of resource in a wrong way.
It will create a resource structure with
res->end = acpi_resource->start + acpi_resource->len - 1;

This patch fixes a problem on my machine that a platform device fails
to be created because one of its ACPI IO resource entry (start = 0,
end = 0, length = 0) is translated into a generic resource with
start = 0, end = 0xffffffff.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agocpuset: fix a race condition in __cpuset_node_allowed_softwall()
Li Zefan [Thu, 27 Feb 2014 10:19:36 +0000 (18:19 +0800)]
cpuset: fix a race condition in __cpuset_node_allowed_softwall()

commit 99afb0fd5f05aac467ffa85c36778fec4396209b upstream.

It's not safe to access task's cpuset after releasing task_lock().
Holding callback_mutex won't help.

Signed-off-by: Li Zefan <lizefan@huawei.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agocpuset: fix a locking issue in cpuset_migrate_mm()
Li Zefan [Thu, 27 Feb 2014 10:19:03 +0000 (18:19 +0800)]
cpuset: fix a locking issue in cpuset_migrate_mm()

commit 4729583006772b9530404bc1bb7c3aa4a10ffd4d upstream.

I can trigger a lockdep warning:

  # mount -t cgroup -o cpuset xxx /cgroup
  # mkdir /cgroup/cpuset
  # mkdir /cgroup/tmp
  # echo 0 > /cgroup/tmp/cpuset.cpus
  # echo 0 > /cgroup/tmp/cpuset.mems
  # echo 1 > /cgroup/tmp/cpuset.memory_migrate
  # echo $$ > /cgroup/tmp/tasks
  # echo 1 > /cgruop/tmp/cpuset.mems

  ===============================
  [ INFO: suspicious RCU usage. ]
  3.14.0-rc1-0.1-default+ #32 Not tainted
  -------------------------------
  include/linux/cgroup.h:682 suspicious rcu_dereference_check() usage!
  ...
    [<ffffffff81582174>] dump_stack+0x72/0x86
    [<ffffffff810b8f01>] lockdep_rcu_suspicious+0x101/0x140
    [<ffffffff81105ba1>] cpuset_migrate_mm+0xb1/0xe0
  ...

We used to hold cgroup_mutex when calling cpuset_migrate_mm(), but now
we hold cpuset_mutex, which causes task_css() to complain.

This is not a false-positive but a real issue.

Holding cpuset_mutex won't prevent a task from migrating to another
cpuset, and it won't prevent the original task->cgroup from destroying
during this change.

Fixes: 5d21cc2db040 (cpuset: replace cgroup_mutex locking with cpuset internal locking)
Signed-off-by: Li Zefan <lizefan@huawei.com>
Sigend-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agogenirq: Remove racy waitqueue_active check
Chuansheng Liu [Mon, 24 Feb 2014 03:29:50 +0000 (11:29 +0800)]
genirq: Remove racy waitqueue_active check

commit c685689fd24d310343ac33942e9a54a974ae9c43 upstream.

We hit one rare case below:

T1 calling disable_irq(), but hanging at synchronize_irq()
always;
The corresponding irq thread is in sleeping state;
And all CPUs are in idle state;

After analysis, we found there is one possible scenerio which
causes T1 is waiting there forever:
CPU0                                       CPU1
 synchronize_irq()
  wait_event()
    spin_lock()
                                           atomic_dec_and_test(&threads_active)
      insert the __wait into queue
    spin_unlock()
                                           if(waitqueue_active)
    atomic_read(&threads_active)
                                             wake_up()

Here after inserted the __wait into queue on CPU0, and before
test if queue is empty on CPU1, there is no barrier, it maybe
cause it is not visible for CPU1 immediately, although CPU0 has
updated the queue list.
It is similar for CPU0 atomic_read() threads_active also.

So we'd need one smp_mb() before waitqueue_active.that, but removing
the waitqueue_active() check solves it as wel l and it makes
things simple and clear.

Signed-off-by: Chuansheng Liu <chuansheng.liu@intel.com>
Cc: Xiaoming Wang <xiaoming.wang@intel.com>
Link: http://lkml.kernel.org/r/1393212590-32543-1-git-send-email-chuansheng.liu@intel.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoRevert "xhci 1.0: Limit arbitrarily-aligned scatter gather."
Mathias Nyman [Fri, 7 Mar 2014 15:06:57 +0000 (17:06 +0200)]
Revert "xhci 1.0: Limit arbitrarily-aligned scatter gather."

commit e2ed511400d41e0d136089d5a55ceab57c6a2426 upstream.

This reverts commit 247bf557273dd775505fb9240d2d152f4f20d304.

This commit, together with commit 3804fad45411b48233b48003e33a78f290d227c8
"USBNET: ax88179_178a: enable tso if usb host supports sg dma" were
origially added to get xHCI 1.0 hosts and usb ethernet ax88179_178a devices
working together with scatter gather. xHCI 1.0 hosts pose some requirement on how transfer
buffers are aligned, setting this requirement for 1.0 hosts caused USB 3.0 mass
storage devices to fail more frequently.

USB 3.0 mass storage devices used to work before 3.14-rc1.  Theoretically,
the TD fragment rules could have caused an occasional disk glitch.
Now the devices *will* fail, instead of theoretically failing.
>From a user perspective, this looks like a regression; the USB device obviously
fails on 3.14-rc1, and may sometimes silently fail on prior kernels.

The proper soluition is to implement the TD fragment rules required, but for now
this patch needs to be reverted to get USB 3.0 mass storage devices working at the
level they used to.

Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agousb: Make DELAY_INIT quirk wait 100ms between Get Configuration requests
Julius Werner [Tue, 4 Mar 2014 19:27:38 +0000 (11:27 -0800)]
usb: Make DELAY_INIT quirk wait 100ms between Get Configuration requests

commit d86db25e53fa69e3e97f3b55dd82a70689787c5d upstream.

The DELAY_INIT quirk only reduces the frequency of enumeration failures
with the Logitech HD Pro C920 and C930e webcams, but does not quite
eliminate them. We have found that adding a delay of 100ms between the
first and second Get Configuration request makes the device enumerate
perfectly reliable even after several weeks of extensive testing. The
reasons for that are anyone's guess, but since the DELAY_INIT quirk
already delays enumeration by a whole second, wating for another 10th of
that isn't really a big deal for the one other device that uses it, and
it will resolve the problems with these webcams.

Signed-off-by: Julius Werner <jwerner@chromium.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agousb: Add device quirk for Logitech HD Pro Webcams C920 and C930e
Julius Werner [Tue, 4 Mar 2014 18:52:39 +0000 (10:52 -0800)]
usb: Add device quirk for Logitech HD Pro Webcams C920 and C930e

commit e0429362ab15c46ea4d64c3f8c9e0933e48a143a upstream.

We've encountered a rare issue when enumerating two Logitech webcams
after a reboot that doesn't power cycle the USB ports. They are spewing
random data (possibly some leftover UVC buffers) on the second
(full-sized) Get Configuration request of the enumeration phase. Since
the data is random this can potentially cause all kinds of odd behavior,
and since it occasionally happens multiple times (after the kernel
issues another reset due to the garbled configuration descriptor), it is
not always recoverable. Set the USB_DELAY_INIT quirk that seems to work
around the issue.

Signed-off-by: Julius Werner <jwerner@chromium.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoRevert "USBNET: ax88179_178a: enable tso if usb host supports sg dma"
Mathias Nyman [Fri, 7 Mar 2014 15:06:58 +0000 (17:06 +0200)]
Revert "USBNET: ax88179_178a: enable tso if usb host supports sg dma"

commit 469d417b68958a064c09e7875646c97c6e783dfc upstream.

This reverts commit 3804fad45411b48233b48003e33a78f290d227c8.

This commit, together with commit 247bf557273dd775505fb9240d2d152f4f20d304
"xhci 1.0: Limit arbitrarily-aligned scatter gather." were
origially added to get xHCI 1.0 hosts and usb ethernet ax88179_178a devices
working together with scatter gather. xHCI 1.0 hosts pose some requirement on how transfer
buffers are aligned, setting this requirement for 1.0 hosts caused USB 3.0 mass
storage devices to fail more frequently.

USB 3.0 mass storage devices used to work before 3.14-rc1.  Theoretically,
the TD fragment rules could have caused an occasional disk glitch.
Now the devices *will* fail, instead of theoretically failing.
>From a user perspective, this looks like a regression; the USB device obviously
fails on 3.14-rc1, and may sometimes silently fail on prior kernels.

The proper soluition is to implement the TD fragment rules for xHCI 1.0 hosts,
but for now, revert this patch until scatter gather can be properly supported.

Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86_pkg_temp_thermal: Do not expose as a hwmon device
Jean Delvare [Sun, 2 Mar 2014 14:33:35 +0000 (15:33 +0100)]
x86_pkg_temp_thermal: Do not expose as a hwmon device

commit 79786880a47a8c5b4c8146c03432b3387a07a169 upstream.

The temperature value reported by x86_pkg_temp_thermal is already
reported by the coretemp driver. So, do not expose this thermal zone
as a hwmon device, because it would be redundant.

Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Eduardo Valentin <eduardo.valentin@ti.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86/amd/numa: Fix northbridge quirk to assign correct NUMA node
Daniel J Blueman [Thu, 13 Mar 2014 11:43:01 +0000 (19:43 +0800)]
x86/amd/numa: Fix northbridge quirk to assign correct NUMA node

commit 847d7970defb45540735b3fb4e88471c27cacd85 upstream.

For systems with multiple servers and routed fabric, all
northbridges get assigned to the first server. Fix this by also
using the node reported from the PCI bus. For single-fabric
systems, the northbriges are on PCI bus 0 by definition, which
are on NUMA node 0 by definition, so this is invarient on most
systems.

Tested on fam10h and fam15h single and multi-fabric systems and
candidate for stable.

Signed-off-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Steffen Persvold <sp@numascale.com>
Acked-by: Borislav Petkov <bp@suse.de>
Link: http://lkml.kernel.org/r/1394710981-3596-1-git-send-email-daniel@numascale.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86: fix compile error due to X86_TRAP_NMI use in asm files
Linus Torvalds [Sat, 8 Mar 2014 02:58:40 +0000 (18:58 -0800)]
x86: fix compile error due to X86_TRAP_NMI use in asm files

commit b01d4e68933ec23e43b1046fa35d593cefcf37d1 upstream.

It's an enum, not a #define, you can't use it in asm files.

Introduced in commit 5fa10196bdb5 ("x86: Ignore NMIs that come in during
early boot"), and sadly I didn't compile-test things like I should have
before pushing out.

My weak excuse is that the x86 tree generally doesn't introduce stupid
things like this (and the ARM pull afterwards doesn't cause me to do a
compile-test either, since I don't cross-compile).

Cc: Don Zickus <dzickus@redhat.com>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86: Ignore NMIs that come in during early boot
H. Peter Anvin [Fri, 7 Mar 2014 23:05:20 +0000 (15:05 -0800)]
x86: Ignore NMIs that come in during early boot

commit 5fa10196bdb5f190f595ebd048490ee52dddea0f upstream.

Don Zickus reports:

A customer generated an external NMI using their iLO to test kdump
worked.  Unfortunately, the machine hung.  Disabling the nmi_watchdog
made things work.

I speculated the external NMI fired, caused the machine to panic (as
expected) and the perf NMI from the watchdog came in and was latched.
My guess was this somehow caused the hang.

   ----

It appears that the latched NMI stays latched until the early page
table generation on 64 bits, which causes exceptions to happen which
end in IRET, which re-enable NMI.  Therefore, ignore NMIs that come in
during early execution, until we have proper exception handling.

Reported-and-tested-by: Don Zickus <dzickus@redhat.com>
Link: http://lkml.kernel.org/r/1394221143-29713-1-git-send-email-dzickus@redhat.com
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoregulator: core: Replace direct ops->enable usage
Markus Pargmann [Thu, 20 Feb 2014 16:36:03 +0000 (17:36 +0100)]
regulator: core: Replace direct ops->enable usage

commit 30c219710358c5cca2f8bd2e9e547c6aadf7cf8b upstream.

There are some direct ops->enable in the regulator core driver. This is
a potential issue as the function _regulator_do_enable() handles gpio
regulators and the normal ops->enable calls. These gpio regulators are
simply ignored when ops->enable is called directly.

One possible bug is that boot-on and always-on gpio regulators are not
enabled on registration.

This patch replaces all ops->enable calls by _regulator_do_enable.

[Handle missing enable operations -- broonie]

Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoARM: 7991/1: sa1100: fix compile problem on Collie
Linus Walleij [Tue, 25 Feb 2014 21:41:41 +0000 (22:41 +0100)]
ARM: 7991/1: sa1100: fix compile problem on Collie

commit 052450fdc55894a39fbae93d9bbe43947956f663 upstream.

Due to a problem in the MFD Kconfig it was not possible to
compile the UCB battery driver for the Collie SA1100 system,
in turn making it impossible to compile in the battery driver.
(See patch "mfd: include all drivers in subsystem menu".)

After fixing the MFD Kconfig (separate patch) a compile error
appears in the Collie battery driver due to the <mach/collie.h>
implicitly requiring <mach/hardware.h> through <linux/gpio.h>
via <mach/gpio.h> prior to commit
40ca061b "ARM: 7841/1: sa1100: remove complex GPIO interface".

Fix this up by including the required header into
<mach/collie.h>.

Cc: Andrea Adami <andrea.adami@gmail.com>
Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agopowerpc: Align p_dyn, p_rela and p_st symbols
Anton Blanchard [Mon, 3 Mar 2014 21:31:24 +0000 (08:31 +1100)]
powerpc: Align p_dyn, p_rela and p_st symbols

commit a5b2cf5b1af424ee3dd9e3ce6d5cea18cb927e67 upstream.

The 64bit relocation code places a few symbols in the text segment.
These symbols are only 4 byte aligned where they need to be 8 byte
aligned. Add an explicit alignment.

Signed-off-by: Anton Blanchard <anton@samba.org>
Tested-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agopowerpc/tm: Fix crash when forking inside a transaction
Michael Neuling [Mon, 3 Mar 2014 03:21:40 +0000 (14:21 +1100)]
powerpc/tm: Fix crash when forking inside a transaction

commit 621b5060e823301d0cba4cb52a7ee3491922d291 upstream.

When we fork/clone we currently don't copy any of the TM state to the new
thread.  This results in a TM bad thing (program check) when the new process is
switched in as the kernel does a tmrechkpt with TEXASR FS not set.  Also, since
R1 is from userspace, we trigger the bad kernel stack pointer detection.  So we
end up with something like this:

   Bad kernel stack pointer 0 at c0000000000404fc
   cpu 0x2: Vector: 700 (Program Check) at [c00000003ffefd40]
       pc: c0000000000404fc: restore_gprs+0xc0/0x148
       lr: 0000000000000000
       sp: 0
      msr: 9000000100201030
     current = 0xc000001dd1417c30
     paca    = 0xc00000000fe00800   softe: 0        irq_happened: 0x01
       pid   = 0, comm = swapper/2
   WARNING: exception is not recoverable, can't continue

The below fixes this by flushing the TM state before we copy the task_struct to
the clone.  To do this we go through the tmreclaim patch, which removes the
checkpointed registers from the CPU and transitions the CPU out of TM suspend
mode.  Hence we need to call tmrechkpt after to restore the checkpointed state
and the TM mode for the current task.

To make this fail from userspace is simply:
tbegin
li r0, 2
sc
<boom>

Kudos to Adhemerval Zanella Neto for finding this.

Signed-off-by: Michael Neuling <mikey@neuling.org>
cc: Adhemerval Zanella Neto <azanella@br.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agopinctrl: sunxi: use chained_irq_{enter, exit} for GIC compatibility
Chen-Yu Tsai [Mon, 10 Feb 2014 16:22:37 +0000 (00:22 +0800)]
pinctrl: sunxi: use chained_irq_{enter, exit} for GIC compatibility

commit 905a5117e79367b7e58ae046d12ca9961f048c89 upstream.

On tha Allwinner A20 SoC, the external interrupts on the pin controller
device are connected to the GIC. Without chained_irq_{enter, exit},
external GPIO interrupts, such as used by mmc core card detect, cause
the system to hang.

This issue was first encountered during my attempt to get out-of-band
interrupts for WiFi on the Cubietruck working. With David's new series
of sunci-mci using mmc slot-gpio for (GPIO interrupt based) card
detection, removing the SD card also causes my Cubietruck to hang. This
problem should extend to all Allwinner A20 based boards.

With this fix, the system no longer hangs when I remove or insert the
SD card. /proc/interrupts show that the interrupt has correctly fired.
However the system still does not detect card removal/insertion. I
believe this is another unrelated issue.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoALSA: hda - Fix loud click noise with IdeaPad 410Y
Takashi Iwai [Fri, 7 Mar 2014 07:37:19 +0000 (08:37 +0100)]
ALSA: hda - Fix loud click noise with IdeaPad 410Y

commit 9b745ab897199c2af9f21ca9681ef86d5b971002 upstream.

Lenovo IdeaPad 410Y with ALC282 codec makes loud click noises at boot
and shutdown.  Also, it wrongly misdetects the acpi_thinkpad hook.
This patch adds a device-specific fixup for disabling the shutup
callback that is the cause of the click noise and also avoiding the
thinpad_helper calls.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=71511
Reported-and-tested-by: Guilherme Amadio <guilherme.amadio@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoALSA: hda - Add missing loopback merge path for AD1884/1984 codecs
Takashi Iwai [Wed, 5 Mar 2014 10:52:24 +0000 (11:52 +0100)]
ALSA: hda - Add missing loopback merge path for AD1884/1984 codecs

commit c5eda4c1bf6214332c46fb2f4e7c42a85e5e5643 upstream.

The mixer widget (NID 0x20) of AD1884 and AD1984 codecs isn't
connected directly to the actual I/O paths but only via another mixer
widget (NID 0x21).  We need a similar fix as we did for AD1882.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoALSA: hda - Added inverted digital-mic handling for Acer TravelMate 8371
Marius Knaust [Mon, 3 Mar 2014 00:48:58 +0000 (01:48 +0100)]
ALSA: hda - Added inverted digital-mic handling for Acer TravelMate 8371

commit a6b92b6650d010d58b6e6fe42c6271266e0b1134 upstream.

Signed-off-by: Marius Knaust <marius.knaust@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoALSA: usb-audio: Add quirk for Logitech Webcam C500
Takashi Iwai [Wed, 5 Mar 2014 11:34:39 +0000 (12:34 +0100)]
ALSA: usb-audio: Add quirk for Logitech Webcam C500

commit e805ca8b0a9b6c91099c0eaa4b160a1196a4ae25 upstream.

Logitech C500 (046d:0807) needs the same workaround like other
Logitech Webcams.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoALSA: oxygen: Xonar DG(X): capture from I2S channel 1, not 2
Roman Volkov [Fri, 24 Jan 2014 12:18:14 +0000 (16:18 +0400)]
ALSA: oxygen: Xonar DG(X): capture from I2S channel 1, not 2

commit 3dd77654fb1d7f68b9739f3039bad8dbbc0739f8 upstream.

Actually CS4245 connected to the I2S channel 1 for
capture, not channel 2. Otherwise capturing and
playback does not work for CS4245.

Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/radeon: TTM must be init with cpu-visible VRAM, v2
Lauri Kasanen [Fri, 28 Feb 2014 18:50:23 +0000 (20:50 +0200)]
drm/radeon: TTM must be init with cpu-visible VRAM, v2

commit 14eedc32a3c0ec9dd70448a73763ee21feae3111 upstream.

Without this, a bo may get created in the cpu-inaccessible vram.
Before the CP engines get setup, all copies are done via cpu memcpy.

This means that the cpu tries to read from inaccessible memory, fails,
and the radeon module proceeds to disable acceleration.

Doing this has no downsides, as the real VRAM size gets set as soon as the
CP engines get init.

This is a candidate for 3.14 fixes.

v2: Add comment on why the function is used

Signed-off-by: Lauri Kasanen <cand@gmx.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: save and copy AP's VHT capability info correctly
Amitkumar Karwar [Wed, 5 Mar 2014 02:43:14 +0000 (18:43 -0800)]
mwifiex: save and copy AP's VHT capability info correctly

commit d51246481c7f28bbfa1f814ded2da65e531cd4b2 upstream.

While preparing association request, intersection of device's
VHT capability information and corresponding field advertised
by AP is used.

This patch fixes a couple errors while saving and copying vht_cap
and vht_oper fields from AP's beacon.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: copy AP's HT capability info correctly
Amitkumar Karwar [Wed, 5 Mar 2014 02:43:13 +0000 (18:43 -0800)]
mwifiex: copy AP's HT capability info correctly

commit c99b1861c232e1f641f13b8645e0febb3712cc71 upstream.

While preparing association request, intersection of device's HT
capability information and corresponding fields advertised by AP
is used.

This patch fixes an error while copying this field from AP's
beacon.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: do not advertise usb autosuspend support
Bing Zhao [Thu, 27 Feb 2014 04:11:22 +0000 (20:11 -0800)]
mwifiex: do not advertise usb autosuspend support

commit adb07df1e039e9fe43e66aeea8b4771f83659dbb upstream.

As many Surface Pro I & II users have found out, the mwifiex_usb
doesn't support usb autosuspend, and it has caused some system
stability issues.

Bug 69661 - mwifiex_usb on MS Surface Pro 1 is unstable
Bug 60815 - Interface hangs in mwifiex_usb
Bug 64111 - mwifiex_usb USB8797 crash failed to get signal
      information

USB autosuspend get triggered when Surface Pro's AC power is
removed or powertop enables power saving on USB8797 device.
Driver's suspend handler is called here, but resume handler
won't be called until the AC power is put back on or powertop
disables power saving for USB8797.

We need to refactor the suspend/resume handlers to support
usb autosuspend properly. For now let's just remove it.

Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: fix cmd and Tx data timeout issue for PCIe cards
Amitkumar Karwar [Tue, 18 Feb 2014 23:41:56 +0000 (15:41 -0800)]
mwifiex: fix cmd and Tx data timeout issue for PCIe cards

commit 1c97560f6d751a620978504a4a888c631192b71a upstream.

We are sending sleep confirm done interrupt in the middle of
sleep handshake. There is a corner case when Tx done interrupt
is received from firmware during sleep handshake due to which
host and firmware power states go out of sync causing cmd and
Tx data timeout problem.

Hence sleep confirm done interrupt is sent at the end of sleep
handshake to fix the problem.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: add NULL check for PCIe Rx skb
Amitkumar Karwar [Tue, 18 Feb 2014 23:41:55 +0000 (15:41 -0800)]
mwifiex: add NULL check for PCIe Rx skb

commit bb8e6a1ee881d131e404f0f1f5e8dc9281002771 upstream.

We may get a NULL pointer here if skb allocation for Rx packet
was failed earlier.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomwifiex: clean pcie ring only when device is present
Avinash Patil [Tue, 18 Feb 2014 23:41:54 +0000 (15:41 -0800)]
mwifiex: clean pcie ring only when device is present

commit 4f7ba432202c8330cc03ab959c6228d0de5dc4a3 upstream.

Write io memory to clean PCIe buffer only when PCIe device is
present else this results into crash because of invalid memory
access.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiwlwifi: disable TX AMPDU by default for iwldvm
Emmanuel Grumbach [Wed, 12 Feb 2014 13:15:05 +0000 (15:15 +0200)]
iwlwifi: disable TX AMPDU by default for iwldvm

commit 205e2210daa975d92ace485a65a31ccc4077fe1a upstream.

NICs supported by iwldvm don't handle well TX AMPDU.
Disable it by default, still leave the possibility to
the user to force enable it with a debug parameter.

NICs supported by iwlmvm don't suffer from the same issue,
leave TX AMPDU enabled by default for these.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiwlwifi: fix TX status for aggregated packets
Johannes Berg [Tue, 25 Feb 2014 09:37:15 +0000 (10:37 +0100)]
iwlwifi: fix TX status for aggregated packets

commit 143582c6847cb285b361804c613127c25de60ca4 upstream.

Only the first packet is currently handled correctly, but then
all others are assumed to have failed which is problematic. Fix
this, marking them all successful instead (since if they're not
then the firmware will have transmitted them as single frames.)

This fixes the lost packet reporting.

Also do a tiny variable scoping cleanup.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
[Add the dvm part]
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiwlwifi: dvm: clear IWL_STA_UCODE_INPROGRESS when assoc fails
Emmanuel Grumbach [Tue, 18 Feb 2014 08:30:18 +0000 (10:30 +0200)]
iwlwifi: dvm: clear IWL_STA_UCODE_INPROGRESS when assoc fails

commit ec6f678c74dbdb06a6a775bbb00f1d26c17c404b upstream.

We set IWL_STA_UCODE_INPROGRESS flag when we add a station
and clear it when we send the LQ command for it. But the LQ
command is sent only when the association succeeds.
If the association doesn't succeed, we would leave this flag
set and that wouldn't indicate the station entry as vacant.

This probably fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1065663

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoiwlwifi: mvm: change of listen interval from 70 to 10
Max Stepanov [Sun, 16 Feb 2014 14:36:57 +0000 (16:36 +0200)]
iwlwifi: mvm: change of listen interval from 70 to 10

commit e7eb65cac0720df8b3946af7f7a9dc363cf0a814 upstream.

Some APs reject STA association request if a listen interval value exceeds
a threshold of 10. Thus, for example, Cisco APs may deny STA associations
returning status code 12 (Association denied due to reason outside the scope
of 802.11 standard) in the association response frame.

Fixing the issue by setting the default IWL_CONN_MAX_LISTEN_INTERVAL value
from 70 to 10.

Signed-off-by: Max Stepanov <Max.Stepanov@intel.com>
Reviewed-by: Alexander Bondar <alexander.bondar@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoath9k: fix invalid descriptor discarding
Felix Fietkau [Mon, 24 Feb 2014 21:26:06 +0000 (22:26 +0100)]
ath9k: fix invalid descriptor discarding

commit b7b146c9c9a0248cc57da71244f672ebc54bbef1 upstream.

Only set sc->rx.discard_next to rx_stats->rs_more when actually
discarding the current descriptor.

Also, fix a detection of broken descriptors:
First the code checks if the current descriptor is not done.
Then it checks if the next descriptor is done.
Add a check that afterwards checks the first descriptor again, because
it might have been completed in the mean time.

This fixes a regression introduced in
commit 723e711356b5a8a95728a890e254e8b0d47b55cf
"ath9k: fix handling of broken descriptors"

Reported-by: Marco André Dinis <marcoandredinis@gmail.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoath9k: fix ps-poll responses under a-mpdu sessions
Felix Fietkau [Sat, 22 Feb 2014 12:48:19 +0000 (13:48 +0100)]
ath9k: fix ps-poll responses under a-mpdu sessions

commit 558ff225de80ac95b132d3a115ddadcd64498b4f upstream.

When passing tx frames to the U-APSD queue for powersave poll responses,
the ath_atx_tid pointer needs to be passed to ath_tx_setup_buffer for
proper sequence number accounting.

This fixes high latency and connection stability issues with ath9k
running as AP and a few kinds of mobile phones as client, when PS-Poll
is heavily used

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoath9k: protect tid->sched check
Stanislaw Gruszka [Wed, 19 Feb 2014 12:15:17 +0000 (13:15 +0100)]
ath9k: protect tid->sched check

commit 21f8aaee0c62708654988ce092838aa7df4d25d8 upstream.

We check tid->sched without a lock taken on ath_tx_aggr_sleep(). That
is race condition which can result of doing list_del(&tid->list) twice
(second time with poisoned list node) and cause crash like shown below:

[424271.637220] BUG: unable to handle kernel paging request at 00100104
[424271.637328] IP: [<f90fc072>] ath_tx_aggr_sleep+0x62/0xe0 [ath9k]
...
[424271.639953] Call Trace:
[424271.639998]  [<f90f6900>] ? ath9k_get_survey+0x110/0x110 [ath9k]
[424271.640083]  [<f90f6942>] ath9k_sta_notify+0x42/0x50 [ath9k]
[424271.640177]  [<f809cfef>] sta_ps_start+0x8f/0x1c0 [mac80211]
[424271.640258]  [<c10f730e>] ? free_compound_page+0x2e/0x40
[424271.640346]  [<f809e915>] ieee80211_rx_handlers+0x9d5/0x2340 [mac80211]
[424271.640437]  [<c112f048>] ? kmem_cache_free+0x1d8/0x1f0
[424271.640510]  [<c1345a84>] ? kfree_skbmem+0x34/0x90
[424271.640578]  [<c10fc23c>] ? put_page+0x2c/0x40
[424271.640640]  [<c1345a84>] ? kfree_skbmem+0x34/0x90
[424271.640706]  [<c1345a84>] ? kfree_skbmem+0x34/0x90
[424271.640787]  [<f809dde3>] ? ieee80211_rx_handlers_result+0x73/0x1d0 [mac80211]
[424271.640897]  [<f80a07a0>] ieee80211_prepare_and_rx_handle+0x520/0xad0 [mac80211]
[424271.641009]  [<f809e22d>] ? ieee80211_rx_handlers+0x2ed/0x2340 [mac80211]
[424271.641104]  [<c13846ce>] ? ip_output+0x7e/0xd0
[424271.641182]  [<f80a1057>] ieee80211_rx+0x307/0x7c0 [mac80211]
[424271.641266]  [<f90fa6ee>] ath_rx_tasklet+0x88e/0xf70 [ath9k]
[424271.641358]  [<f80a0f2c>] ? ieee80211_rx+0x1dc/0x7c0 [mac80211]
[424271.641445]  [<f90f82db>] ath9k_tasklet+0xcb/0x130 [ath9k]

Bug report:
https://bugzilla.kernel.org/show_bug.cgi?id=70551

Reported-and-tested-by: Max Sydorenko <maxim.stargazer@gmail.com>
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoath9k: Fix ETSI compliance for AR9462 2.0
Sujith Manoharan [Fri, 14 Feb 2014 02:45:20 +0000 (08:15 +0530)]
ath9k: Fix ETSI compliance for AR9462 2.0

commit b3050248c167871ca52cfdb2ce78aa2460249346 upstream.

The minimum CCA power threshold values have to be adjusted
for existing cards to be in compliance with new regulations.
Newer cards will make use of the values obtained from EEPROM,
support for this was added earlier. To make sure that cards
that are already in use and don't have proper values in EEPROM,
do not violate regulations, use the initvals instead.

Reported-by: Jeang Daniel <dyjeong@qca.qualcomm.com>
Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: clear sequence/fragment number in QoS-null frames
Johannes Berg [Tue, 4 Mar 2014 12:46:53 +0000 (13:46 +0100)]
mac80211: clear sequence/fragment number in QoS-null frames

commit 864a6040f395464003af8dd0d8ca86fed19866d4 upstream.

Avoid leaking data by sending uninitialized memory and setting an
invalid (non-zero) fragment number (the sequence number is ignored
anyway) by setting the seq_ctrl field to zero.

Fixes: 3f52b7e328c5 ("mac80211: mesh power save basics")
Fixes: ce662b44ce22 ("mac80211: send (QoS) Null if no buffered frames")
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: fix association to 20/40 MHz VHT networks
Johannes Berg [Thu, 27 Feb 2014 19:47:53 +0000 (20:47 +0100)]
mac80211: fix association to 20/40 MHz VHT networks

commit cb664981607a6b5b3d670ad57bbda893b2528d96 upstream.

When a VHT network uses 20 or 40 MHz as per the HT operation
information, the channel center frequency segment 0 field in
the VHT operation information is reserved, so ignore it.

This fixes association with such networks when the AP puts 0
into the field, previously we'd disconnect due to an invalid
channel with the message
wlan0: AP VHT information is invalid, disable VHT

Fixes: f2d9d270c15ae ("mac80211: support VHT association")
Reported-by: Tim Nelson <tim.l.nelson@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: don't validate unchanged AP bandwidth while tracking
Johannes Berg [Fri, 21 Feb 2014 19:34:34 +0000 (20:34 +0100)]
mac80211: don't validate unchanged AP bandwidth while tracking

commit 963a1852fbac4f75a2d938fa2e734ef1e6d4c044 upstream.

The MLME code in mac80211 must track whether or not the AP changed
bandwidth, but if there's no change while tracking it shouldn't do
anything, otherwise regulatory updates can make it impossible to
connect to certain APs if the regulatory database doesn't match the
information from the AP. See the precise scenario described in the
code.

This still leaves some possible problems with CSA or if the AP
actually changed bandwidth, but those cases are less common and
won't completely prevent using it.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=70881

Reported-and-tested-by: Nate Carlson <kernel@natecarlson.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: fix AP powersave TX vs. wakeup race
Emmanuel Grumbach [Thu, 20 Feb 2014 07:22:11 +0000 (09:22 +0200)]
mac80211: fix AP powersave TX vs. wakeup race

commit 1d147bfa64293b2723c4fec50922168658e613ba upstream.

There is a race between the TX path and the STA wakeup: while
a station is sleeping, mac80211 buffers frames until it wakes
up, then the frames are transmitted. However, the RX and TX
path are concurrent, so the packet indicating wakeup can be
processed while a packet is being transmitted.

This can lead to a situation where the buffered frames list
is emptied on the one side, while a frame is being added on
the other side, as the station is still seen as sleeping in
the TX path.

As a result, the newly added frame will not be send anytime
soon. It might be sent much later (and out of order) when the
station goes to sleep and wakes up the next time.

Additionally, it can lead to the crash below.

Fix all this by synchronising both paths with a new lock.
Both path are not fastpath since they handle PS situations.

In a later patch we'll remove the extra skb queue locks to
reduce locking overhead.

BUG: unable to handle kernel
NULL pointer dereference at 000000b0
IP: [<ff6f1791>] ieee80211_report_used_skb+0x11/0x3e0 [mac80211]
*pde = 00000000
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
EIP: 0060:[<ff6f1791>] EFLAGS: 00210282 CPU: 1
EIP is at ieee80211_report_used_skb+0x11/0x3e0 [mac80211]
EAX: e5900da0 EBX: 00000000 ECX: 00000001 EDX: 00000000
ESI: e41d00c0 EDI: e5900da0 EBP: ebe458e4 ESP: ebe458b0
 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
CR0: 8005003b CR2: 000000b0 CR3: 25a78000 CR4: 000407d0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
Process iperf (pid: 3934, ti=ebe44000 task=e757c0b0 task.ti=ebe44000)
iwlwifi 0000:02:00.0: I iwl_pcie_enqueue_hcmd Sending command LQ_CMD (#4e), seq: 0x0903, 92 bytes at 3[3]:9
Stack:
 e403b32c ebe458c4 00200002 00200286 e403b338 ebe458cc c10960bb e5900da0
 ff76a6ec ebe458d8 00000000 e41d00c0 e5900da0 ebe458f0 ff6f1b75 e403b210
 ebe4598c ff723dc1 00000000 ff76a6ec e597c978 e403b758 00000002 00000002
Call Trace:
 [<ff6f1b75>] ieee80211_free_txskb+0x15/0x20 [mac80211]
 [<ff723dc1>] invoke_tx_handlers+0x1661/0x1780 [mac80211]
 [<ff7248a5>] ieee80211_tx+0x75/0x100 [mac80211]
 [<ff7249bf>] ieee80211_xmit+0x8f/0xc0 [mac80211]
 [<ff72550e>] ieee80211_subif_start_xmit+0x4fe/0xe20 [mac80211]
 [<c149ef70>] dev_hard_start_xmit+0x450/0x950
 [<c14b9aa9>] sch_direct_xmit+0xa9/0x250
 [<c14b9c9b>] __qdisc_run+0x4b/0x150
 [<c149f732>] dev_queue_xmit+0x2c2/0xca0

Reported-by: Yaara Rozenblum <yaara.rozenblum@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Reviewed-by: Stanislaw Gruszka <sgruszka@redhat.com>
[reword commit log, use a separate lock]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomac80211: send control port protocol frames to the VO queue
Felix Fietkau [Tue, 11 Feb 2014 15:02:47 +0000 (16:02 +0100)]
mac80211: send control port protocol frames to the VO queue

commit 1bf4bbb4024dcdab5e57634dd8ae1072d42a53ac upstream.

Improves reliability of wifi connections with WPA, since authentication
frames are prioritized over normal traffic and also typically exempt
from aggregation.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agorapidio/tsi721: fix tasklet termination in dma channel release
Alexandre Bounine [Mon, 3 Mar 2014 23:38:36 +0000 (15:38 -0800)]
rapidio/tsi721: fix tasklet termination in dma channel release

commit 04379dffdd4da820d51a1566ad2e86f3b1ad97ed upstream.

This patch is a modification of the patch originally proposed by
Xiaotian Feng <xtfeng@gmail.com>: https://lkml.org/lkml/2012/11/5/413
This new version disables DMA channel interrupts and ensures that the
tasklet wil not be scheduled again before calling tasklet_kill().

Unfortunately the updated patch was not released at that time due to
planned rework of Tsi721 mport driver to use threaded interrupts (which
has yet to happen).  Recently the issue was reported again:
https://lkml.org/lkml/2014/2/19/762.

Description from the original Xiaotian's patch:

 "Some drivers use tasklet_disable in device remove/release process,
  tasklet_disable will inc tasklet->count and return.  If the tasklet is
  not handled yet under some softirq pressure, the tasklet will be
  placed on the tasklet_vec, never have a chance to be excuted.  This
  might lead to a heavy loaded ksoftirqd, wakeup with pending_softirq,
  but tasklet is disabled.  tasklet_kill should be used in this case."

This patch is applicable to kernel versions starting from v3.5.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Xiaotian Feng <xtfeng@gmail.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Mike Galbraith <bitbucket@online.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosched: Fix double normalization of vruntime
George McCollister [Tue, 18 Feb 2014 23:56:51 +0000 (17:56 -0600)]
sched: Fix double normalization of vruntime

commit 791c9e0292671a3bfa95286bb5c08129d8605618 upstream.

dequeue_entity() is called when p->on_rq and sets se->on_rq = 0
which appears to guarentee that the !se->on_rq condition is met.
If the task has done set_current_state(TASK_INTERRUPTIBLE) without
schedule() the second condition will be met and vruntime will be
incorrectly adjusted twice.

In certain cases this can result in the task's vruntime never increasing
past the vruntime of other tasks on the CFS' run queue, starving them of
CPU time.

This patch changes switched_from_fair() to use !p->on_rq instead of
!se->on_rq.

I'm able to cause a task with a priority of 120 to starve all other
tasks with the same priority on an ARM platform running 3.2.51-rt72
PREEMPT RT by writing one character at time to a serial tty (16550 UART)
in a tight loop. I'm also able to verify making this change corrects the
problem on that platform and kernel version.

Signed-off-by: George McCollister <george.mccollister@gmail.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1392767811-28916-1-git-send-email-george.mccollister@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomemcg: fix endless loop in __mem_cgroup_iter_next()
Hugh Dickins [Mon, 3 Mar 2014 23:38:24 +0000 (15:38 -0800)]
memcg: fix endless loop in __mem_cgroup_iter_next()

commit ce48225fe3b1b0d1fc9fceb96ac3d8a879e45114 upstream.

Commit 0eef615665ed ("memcg: fix css reference leak and endless loop in
mem_cgroup_iter") got the interaction with the commit a few before it
d8ad30559715 ("mm/memcg: iteration skip memcgs not yet fully
initialized") slightly wrong, and we didn't notice at the time.

It's elusive, and harder to get than the original, but for a couple of
days before rc1, I several times saw a endless loop similar to that
supposedly being fixed.

This time it was a tighter loop in __mem_cgroup_iter_next(): because we
can get here when our root has already been offlined, and the ordering
of conditions was such that we then just cycled around forever.

Fixes: 0eef615665ed ("memcg: fix css reference leak and endless loop in mem_cgroup_iter").
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Greg Thelen <gthelen@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoocfs2 syncs the wrong range...
Al Viro [Mon, 10 Feb 2014 20:18:55 +0000 (15:18 -0500)]
ocfs2 syncs the wrong range...

commit 1b56e98990bcdbb20b9fab163654b9315bf158e8 upstream.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoocfs2: fix quota file corruption
Jan Kara [Mon, 3 Mar 2014 23:38:32 +0000 (15:38 -0800)]
ocfs2: fix quota file corruption

commit 15c34a760630ca2c803848fba90ca0646a9907dd upstream.

Global quota files are accessed from different nodes.  Thus we cannot
cache offset of quota structure in the quota file after we drop our node
reference count to it because after that moment quota structure may be
freed and reallocated elsewhere by a different node resulting in
corruption of quota file.

Fix the problem by clearing dq_off when we are releasing dquot structure.
We also remove the DB_READ_B handling because it is useless -
DQ_ACTIVE_B is set iff DQ_READ_B is set.

Signed-off-by: Jan Kara <jack@suse.cz>
Cc: Goldwyn Rodrigues <rgoldwyn@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm: include VM_MIXEDMAP flag in the VM_SPECIAL list to avoid m(un)locking
Vlastimil Babka [Mon, 3 Mar 2014 23:38:27 +0000 (15:38 -0800)]
mm: include VM_MIXEDMAP flag in the VM_SPECIAL list to avoid m(un)locking

commit 9050d7eba40b3d79551668f54e68fd6f51945ef3 upstream.

Daniel Borkmann reported a VM_BUG_ON assertion failing:

  ------------[ cut here ]------------
  kernel BUG at mm/mlock.c:528!
  invalid opcode: 0000 [#1] SMP
  Modules linked in: ccm arc4 iwldvm [...]
   video
  CPU: 3 PID: 2266 Comm: netsniff-ng Not tainted 3.14.0-rc2+ #8
  Hardware name: LENOVO 2429BP3/2429BP3, BIOS G4ET37WW (1.12 ) 05/29/2012
  task: ffff8801f87f9820 ti: ffff88002cb44000 task.ti: ffff88002cb44000
  RIP: 0010:[<ffffffff81171ad0>]  [<ffffffff81171ad0>] munlock_vma_pages_range+0x2e0/0x2f0
  Call Trace:
    do_munmap+0x18f/0x3b0
    vm_munmap+0x41/0x60
    SyS_munmap+0x22/0x30
    system_call_fastpath+0x1a/0x1f
  RIP   munlock_vma_pages_range+0x2e0/0x2f0
  ---[ end trace a0088dcf07ae10f2 ]---

because munlock_vma_pages_range() thinks it's unexpectedly in the middle
of a THP page.  This can be reproduced with default config since 3.11
kernels.  A reproducer can be found in the kernel's selftest directory
for networking by running ./psock_tpacket.

The problem is that an order=2 compound page (allocated by
alloc_one_pg_vec_page() is part of the munlocked VM_MIXEDMAP vma (mapped
by packet_mmap()) and mistaken for a THP page and assumed to be order=9.

The checks for THP in munlock came with commit ff6a6da60b89 ("mm:
accelerate munlock() treatment of THP pages"), i.e.  since 3.9, but did
not trigger a bug.  It just makes munlock_vma_pages_range() skip such
compound pages until the next 512-pages-aligned page, when it encounters
a head page.  This is however not a problem for vma's where mlocking has
no effect anyway, but it can distort the accounting.

Since commit 7225522bb429 ("mm: munlock: batch non-THP page isolation
and munlock+putback using pagevec") this can trigger a VM_BUG_ON in
PageTransHuge() check.

This patch fixes the issue by adding VM_MIXEDMAP flag to VM_SPECIAL, a
list of flags that make vma's non-mlockable and non-mergeable.  The
reasoning is that VM_MIXEDMAP vma's are similar to VM_PFNMAP, which is
already on the VM_SPECIAL list, and both are intended for non-LRU pages
where mlocking makes no sense anyway.  Related Lkml discussion can be
found in [2].

 [1] tools/testing/selftests/net/psock_tpacket
 [2] https://lkml.org/lkml/2014/1/10/427

Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Reported-by: Daniel Borkmann <dborkman@redhat.com>
Tested-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Cc: John David Anglin <dave.anglin@bell.net>
Cc: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
Cc: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Carsten Otte <cotte@de.ibm.com>
Cc: Jared Hulbert <jaredeh@gmail.com>
Tested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm: page_alloc: exempt GFP_THISNODE allocations from zone fairness
Johannes Weiner [Mon, 3 Mar 2014 23:38:41 +0000 (15:38 -0800)]
mm: page_alloc: exempt GFP_THISNODE allocations from zone fairness

commit 27329369c9ecf37771b2a65202cbf5578cff3331 upstream.

Jan Stancek reports manual page migration encountering allocation
failures after some pages when there is still plenty of memory free, and
bisected the problem down to commit 81c0a2bb515f ("mm: page_alloc: fair
zone allocator policy").

The problem is that GFP_THISNODE obeys the zone fairness allocation
batches on one hand, but doesn't reset them and wake kswapd on the other
hand.  After a few of those allocations, the batches are exhausted and
the allocations fail.

Fixing this means either having GFP_THISNODE wake up kswapd, or
GFP_THISNODE not participating in zone fairness at all.  The latter
seems safer as an acute bugfix, we can clean up later.

Reported-by: Jan Stancek <jstancek@redhat.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agozram: avoid null access when fail to alloc meta
Minchan Kim [Mon, 3 Mar 2014 23:38:34 +0000 (15:38 -0800)]
zram: avoid null access when fail to alloc meta

commit db5d711e2db776f18219b033e5dc4fb7e4264dd7 upstream.

zram_meta_alloc could fail so caller should check it.  Otherwise, your
system will hang.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agodrm/ttm: don't oops if no invalidate_caches()
Rob Clark [Wed, 12 Mar 2014 14:59:37 +0000 (10:59 -0400)]
drm/ttm: don't oops if no invalidate_caches()

commit 9ef7506f7eff3fc42724269f62e30164c141661f upstream.

A few of the simpler TTM drivers (cirrus, ast, mgag200) do not implement
this function.  Yet can end up somehow with an evicted bo:

  BUG: unable to handle kernel NULL pointer dereference at           (null)
  IP: [<          (null)>]           (null)
  PGD 16e761067 PUD 16e6cf067 PMD 0
  Oops: 0010 [#1] SMP
  Modules linked in: bnep bluetooth rfkill fuse ip6t_rpfilter ip6t_REJECT ipt_REJECT xt_conntrack ebtable_nat ebtable_broute bridge stp llc ebtable_filter ebtables ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw ip6table_filter ip6_tables iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack iptable_mangle iptable_security iptable_raw iptable_filter ip_tables sg btrfs zlib_deflate raid6_pq xor dm_queue_length iTCO_wdt iTCO_vendor_support coretemp kvm dcdbas dm_service_time microcode serio_raw pcspkr lpc_ich mfd_core i7core_edac edac_core ses enclosure ipmi_si ipmi_msghandler shpchp acpi_power_meter mperf nfsd auth_rpcgss nfs_acl lockd uinput sunrpc dm_multipath xfs libcrc32c ata_generic pata_acpi sr_mod cdrom
   sd_mod usb_storage mgag200 syscopyarea sysfillrect sysimgblt i2c_algo_bit lpfc drm_kms_helper ttm crc32c_intel ata_piix bfa drm ixgbe libata i2c_core mdio crc_t10dif ptp crct10dif_common pps_core scsi_transport_fc dca scsi_tgt megaraid_sas bnx2 dm_mirror dm_region_hash dm_log dm_mod
  CPU: 16 PID: 2572 Comm: X Not tainted 3.10.0-86.el7.x86_64 #1
  Hardware name: Dell Inc. PowerEdge R810/0H235N, BIOS 0.3.0 11/14/2009
  task: ffff8801799dabc0 ti: ffff88016c884000 task.ti: ffff88016c884000
  RIP: 0010:[<0000000000000000>]  [<          (null)>]           (null)
  RSP: 0018:ffff88016c885ad8  EFLAGS: 00010202
  RAX: ffffffffa04e94c0 RBX: ffff880178937a20 RCX: 0000000000000000
  RDX: 0000000000000000 RSI: 0000000000240004 RDI: ffff880178937a00
  RBP: ffff88016c885b60 R08: 00000000000171a0 R09: ffff88007cf171a0
  R10: ffffea0005842540 R11: ffffffff810487b9 R12: ffff880178937b30
  R13: ffff880178937a00 R14: ffff88016c885b78 R15: ffff880179929400
  FS:  00007f81ba2ef980(0000) GS:ffff88007cf00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000000 CR3: 000000016e763000 CR4: 00000000000007e0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
  Stack:
   ffffffffa0306fae ffff8801799295c0 0000000000260004 0000000000000001
   ffff88016c885b60 ffffffffa0307669 00ff88007cf17738 ffff88017cf17700
   ffff880178937a00 ffff880100000000 ffff880100000000 0000000079929400
  Call Trace:
   [<ffffffffa0306fae>] ? ttm_bo_handle_move_mem+0x54e/0x5b0 [ttm]
   [<ffffffffa0307669>] ? ttm_bo_mem_space+0x169/0x340 [ttm]
   [<ffffffffa0307bd7>] ttm_bo_move_buffer+0x117/0x130 [ttm]
   [<ffffffff81130001>] ? perf_event_init_context+0x141/0x220
   [<ffffffffa0307cb1>] ttm_bo_validate+0xc1/0x130 [ttm]
   [<ffffffffa04e7377>] mgag200_bo_pin+0x87/0xc0 [mgag200]
   [<ffffffffa04e56c4>] mga_crtc_cursor_set+0x474/0xbb0 [mgag200]
   [<ffffffff811971d2>] ? __mem_cgroup_commit_charge+0x152/0x3b0
   [<ffffffff815c4182>] ? mutex_lock+0x12/0x2f
   [<ffffffffa0201433>] drm_mode_cursor_common+0x123/0x170 [drm]
   [<ffffffffa0205231>] drm_mode_cursor_ioctl+0x41/0x50 [drm]
   [<ffffffffa01f5ca2>] drm_ioctl+0x502/0x630 [drm]
   [<ffffffff815cbab4>] ? __do_page_fault+0x1f4/0x510
   [<ffffffff8101cb68>] ? __restore_xstate_sig+0x218/0x4f0
   [<ffffffff811b4445>] do_vfs_ioctl+0x2e5/0x4d0
   [<ffffffff8124488e>] ? file_has_perm+0x8e/0xa0
   [<ffffffff811b46b1>] SyS_ioctl+0x81/0xa0
   [<ffffffff815d05d9>] system_call_fastpath+0x16/0x1b
  Code:  Bad RIP value.
  RIP  [<          (null)>]           (null)
   RSP <ffff88016c885ad8>
  CR2: 0000000000000000

Signed-off-by: Rob Clark <rclark@redhat.com>
Reviewed-by: Jérôme Glisse <jglisse@redhat.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Cc: stable@vger.kernel.org
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomemcg: reparent charges of children before processing parent
Filipe Brandenburger [Mon, 3 Mar 2014 23:38:25 +0000 (15:38 -0800)]
memcg: reparent charges of children before processing parent

commit 4fb1a86fb5e4209a7d4426d4e586c58e9edc74ac upstream.

Sometimes the cleanup after memcg hierarchy testing gets stuck in
mem_cgroup_reparent_charges(), unable to bring non-kmem usage down to 0.

There may turn out to be several causes, but a major cause is this: the
workitem to offline parent can get run before workitem to offline child;
parent's mem_cgroup_reparent_charges() circles around waiting for the
child's pages to be reparented to its lrus, but it's holding
cgroup_mutex which prevents the child from reaching its
mem_cgroup_reparent_charges().

Further testing showed that an ordered workqueue for cgroup_destroy_wq
is not always good enough: percpu_ref_kill_and_confirm's call_rcu_sched
stage on the way can mess up the order before reaching the workqueue.

Instead, when offlining a memcg, call mem_cgroup_reparent_charges() on
all its children (and grandchildren, in the correct order) to have their
charges reparented first.

Fixes: e5fca243abae ("cgroup: use a dedicated workqueue for cgroup destruction")
Signed-off-by: Filipe Brandenburger <filbranden@google.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Tejun Heo <tj@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: <stable@vger.kernel.org> [v3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotcp: syncookies: reduce mss table to four values
Florian Westphal [Fri, 20 Sep 2013 20:32:56 +0000 (22:32 +0200)]
tcp: syncookies: reduce mss table to four values

commit 086293542b991fb88a2e41ae7b4f82ac65a20e1a upstream.

Halve mss table size to make blind cookie guessing more difficult.
This is sad since the tables were already small, but there
is little alternative except perhaps adding more precise mss information
in the tcp timestamp.  Timestamps are unfortunately not ubiquitous.

Guessing all possible cookie values still has 8-in 2**32 chance.

Reported-by: Jakob Lell <jakob@jakoblell.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotcp: syncookies: reduce cookie lifetime to 128 seconds
Florian Westphal [Fri, 20 Sep 2013 20:32:55 +0000 (22:32 +0200)]
tcp: syncookies: reduce cookie lifetime to 128 seconds

commit 8c27bd75f04fb9cb70c69c3cfe24f4e6d8e15906 upstream.

We currently accept cookies that were created less than 4 minutes ago
(ie, cookies with counter delta 0-3).  Combined with the 8 mss table
values, this yields 32 possible values (out of 2**32) that will be valid.

Reducing the lifetime to < 2 minutes halves the guessing chance while
still providing a large enough period.

While at it, get rid of jiffies value -- they overflow too quickly on
32 bit platforms.

getnstimeofday is used to create a counter that increments every 64s.
perf shows getnstimeofday cost is negible compared to sha_transform;
normal tcp initial sequence number generation uses getnstimeofday, too.

Reported-by: Jakob Lell <jakob@jakoblell.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoxfrm: prevent ipcomp scratch buffer race condition
Michal Kubecek [Thu, 17 Oct 2013 13:07:40 +0000 (15:07 +0200)]
xfrm: prevent ipcomp scratch buffer race condition

commit 12e3594698f6c3ab6ebacc79f2fb2ad2bb5952b5 upstream.

In ipcomp_compress(), sortirq is enabled too early, allowing the
per-cpu scratch buffer to be rewritten by ipcomp_decompress()
(called on the same CPU in softirq context) between populating
the buffer and copying the compressed data to the skb.

v2: as pointed out by Steffen Klassert, if we also move the
local_bh_disable() before reading the per-cpu pointers, we can
get rid of get_cpu()/put_cpu().

v3: removed ipcomp_decompress part (as explained by Herbert Xu,
it cannot be called from process context), get rid of cpu
variable (thanks to Eric Dumazet)

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoFix mountpoint reference leakage in linkat
Oleg Drokin [Fri, 31 Jan 2014 20:41:58 +0000 (15:41 -0500)]
Fix mountpoint reference leakage in linkat

commit d22e6338db7f613dd4f6095c190682fcc519e4b7 upstream.

Recent changes to retry on ESTALE in linkat
(commit 442e31ca5a49e398351b2954b51f578353fdf210)
introduced a mountpoint reference leak and a small memory
leak in case a filesystem link operation returns ESTALE
which is pretty normal for distributed filesystems like
lustre, nfs and so on.
Free old_path in such a case.

[AV: there was another missing path_put() nearby - on the previous
goto retry]

[js: the second path_put is not in 3.12 yet, hunk removed]

Signed-off-by: Oleg Drokin: <green@linuxhacker.ru>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet: sctp: fix sctp_sf_do_5_1D_ce to verify if we/peer is AUTH capable
Daniel Borkmann [Mon, 3 Mar 2014 16:23:04 +0000 (17:23 +0100)]
net: sctp: fix sctp_sf_do_5_1D_ce to verify if we/peer is AUTH capable

[ Upstream commit ec0223ec48a90cb605244b45f7c62de856403729 ]

RFC4895 introduced AUTH chunks for SCTP; during the SCTP
handshake RANDOM; CHUNKS; HMAC-ALGO are negotiated (CHUNKS
being optional though):

  ---------- INIT[RANDOM; CHUNKS; HMAC-ALGO] ---------->
  <------- INIT-ACK[RANDOM; CHUNKS; HMAC-ALGO] ---------
  -------------------- COOKIE-ECHO -------------------->
  <-------------------- COOKIE-ACK ---------------------

A special case is when an endpoint requires COOKIE-ECHO
chunks to be authenticated:

  ---------- INIT[RANDOM; CHUNKS; HMAC-ALGO] ---------->
  <------- INIT-ACK[RANDOM; CHUNKS; HMAC-ALGO] ---------
  ------------------ AUTH; COOKIE-ECHO ---------------->
  <-------------------- COOKIE-ACK ---------------------

RFC4895, section 6.3. Receiving Authenticated Chunks says:

  The receiver MUST use the HMAC algorithm indicated in
  the HMAC Identifier field. If this algorithm was not
  specified by the receiver in the HMAC-ALGO parameter in
  the INIT or INIT-ACK chunk during association setup, the
  AUTH chunk and all the chunks after it MUST be discarded
  and an ERROR chunk SHOULD be sent with the error cause
  defined in Section 4.1. [...] If no endpoint pair shared
  key has been configured for that Shared Key Identifier,
  all authenticated chunks MUST be silently discarded. [...]

  When an endpoint requires COOKIE-ECHO chunks to be
  authenticated, some special procedures have to be followed
  because the reception of a COOKIE-ECHO chunk might result
  in the creation of an SCTP association. If a packet arrives
  containing an AUTH chunk as a first chunk, a COOKIE-ECHO
  chunk as the second chunk, and possibly more chunks after
  them, and the receiver does not have an STCB for that
  packet, then authentication is based on the contents of
  the COOKIE-ECHO chunk. In this situation, the receiver MUST
  authenticate the chunks in the packet by using the RANDOM
  parameters, CHUNKS parameters and HMAC_ALGO parameters
  obtained from the COOKIE-ECHO chunk, and possibly a local
  shared secret as inputs to the authentication procedure
  specified in Section 6.3. If authentication fails, then
  the packet is discarded. If the authentication is successful,
  the COOKIE-ECHO and all the chunks after the COOKIE-ECHO
  MUST be processed. If the receiver has an STCB, it MUST
  process the AUTH chunk as described above using the STCB
  from the existing association to authenticate the
  COOKIE-ECHO chunk and all the chunks after it. [...]

Commit bbd0d59809f9 introduced the possibility to receive
and verification of AUTH chunk, including the edge case for
authenticated COOKIE-ECHO. On reception of COOKIE-ECHO,
the function sctp_sf_do_5_1D_ce() handles processing,
unpacks and creates a new association if it passed sanity
checks and also tests for authentication chunks being
present. After a new association has been processed, it
invokes sctp_process_init() on the new association and
walks through the parameter list it received from the INIT
chunk. It checks SCTP_PARAM_RANDOM, SCTP_PARAM_HMAC_ALGO
and SCTP_PARAM_CHUNKS, and copies them into asoc->peer
meta data (peer_random, peer_hmacs, peer_chunks) in case
sysctl -w net.sctp.auth_enable=1 is set. If in INIT's
SCTP_PARAM_SUPPORTED_EXT parameter SCTP_CID_AUTH is set,
peer_random != NULL and peer_hmacs != NULL the peer is to be
assumed asoc->peer.auth_capable=1, in any other case
asoc->peer.auth_capable=0.

Now, if in sctp_sf_do_5_1D_ce() chunk->auth_chunk is
available, we set up a fake auth chunk and pass that on to
sctp_sf_authenticate(), which at latest in
sctp_auth_calculate_hmac() reliably dereferences a NULL pointer
at position 0..0008 when setting up the crypto key in
crypto_hash_setkey() by using asoc->asoc_shared_key that is
NULL as condition key_id == asoc->active_key_id is true if
the AUTH chunk was injected correctly from remote. This
happens no matter what net.sctp.auth_enable sysctl says.

The fix is to check for net->sctp.auth_enable and for
asoc->peer.auth_capable before doing any operations like
sctp_sf_authenticate() as no key is activated in
sctp_auth_asoc_init_active_key() for each case.

Now as RFC4895 section 6.3 states that if the used HMAC-ALGO
passed from the INIT chunk was not used in the AUTH chunk, we
SHOULD send an error; however in this case it would be better
to just silently discard such a maliciously prepared handshake
as we didn't even receive a parameter at all. Also, as our
endpoint has no shared key configured, section 6.3 says that
MUST silently discard, which we are doing from now onwards.

Before calling sctp_sf_pdiscard(), we need not only to free
the association, but also the chunk->auth_chunk skb, as
commit bbd0d59809f9 created a skb clone in that case.

I have tested this locally by using netfilter's nfqueue and
re-injecting packets into the local stack after maliciously
modifying the INIT chunk (removing RANDOM; HMAC-ALGO param)
and the SCTP packet containing the COOKIE_ECHO (injecting
AUTH chunk before COOKIE_ECHO). Fixed with this patch applied.

Fixes: bbd0d59809f9 ("[SCTP]: Implement the receive and verification of AUTH chunk")
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Vlad Yasevich <yasevich@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoip_tunnel:multicast process cause panic due to skb->_skb_refdst NULL pointer
Xin Long [Mon, 3 Mar 2014 12:18:36 +0000 (20:18 +0800)]
ip_tunnel:multicast process cause panic due to skb->_skb_refdst NULL pointer

[ Upstream commit 10ddceb22bab11dab10ba645c7df2e4a8e7a5db5 ]

when ip_tunnel process multicast packets, it may check if the packet is looped
back packet though 'rt_is_output_route(skb_rtable(skb))' in ip_tunnel_rcv(),
but before that , skb->_skb_refdst has been dropped in iptunnel_pull_header(),
so which leads to a panic.

fix the bug: https://bugzilla.kernel.org/show_bug.cgi?id=70681

Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotg3: Don't check undefined error bits in RXBD
Michael Chan [Fri, 28 Feb 2014 23:05:10 +0000 (15:05 -0800)]
tg3: Don't check undefined error bits in RXBD

[ Upstream commit d7b95315cc7f441418845a165ee56df723941487 ]

Redefine the RXD_ERR_MASK to include only relevant error bits. This fixes
a customer reported issue of randomly dropping packets on the 5719.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv6: ipv6_find_hdr restore prev functionality
Hans Schillstrom [Thu, 27 Feb 2014 11:57:58 +0000 (12:57 +0100)]
ipv6: ipv6_find_hdr restore prev functionality

[ Upstream commit accfe0e356327da5bd53da8852b93fc22de9b5fc ]

The commit 9195bb8e381d81d5a315f911904cdf0cfcc919b8 ("ipv6: improve
ipv6_find_hdr() to skip empty routing headers") broke ipv6_find_hdr().

When a target is specified like IPPROTO_ICMPV6 ipv6_find_hdr()
returns -ENOENT when it's found, not the header as expected.

A part of IPVS is broken and possible also nft_exthdr_eval().
When target is -1 which it is most cases, it works.

This patch exits the do while loop if the specific header is found
so the nexthdr could be returned as expected.

Reported-by: Art -kwaak- van Breemen <ard@telegraafnet.nl>
Signed-off-by: Hans Schillstrom <hans@schillstrom.com>
CC:Ansis Atteka <aatteka@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosfc: check for NULL efx->ptp_data in efx_ptp_event
Edward Cree [Tue, 25 Feb 2014 13:17:59 +0000 (13:17 +0000)]
sfc: check for NULL efx->ptp_data in efx_ptp_event

[ Upstream commit 8f355e5cee63c2c0c145d8206c4245d0189f47ff ]

If we receive a PTP event from the NIC when we haven't set up PTP state
in the driver, we attempt to read through a NULL pointer efx->ptp_data,
triggering a panic.

Signed-off-by: Edward Cree <ecree@solarflare.com>
Acked-by: Shradha Shah <sshah@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoipv6: reuse ip6_frag_id from ip6_ufo_append_data
Hannes Frederic Sowa [Fri, 21 Feb 2014 01:55:35 +0000 (02:55 +0100)]
ipv6: reuse ip6_frag_id from ip6_ufo_append_data

[ Upstream commit 916e4cf46d0204806c062c8c6c4d1f633852c5b6 ]

Currently we generate a new fragmentation id on UFO segmentation. It
is pretty hairy to identify the correct net namespace and dst there.
Especially tunnels use IFF_XMIT_DST_RELEASE and thus have no skb_dst
available at all.

This causes unreliable or very predictable ipv6 fragmentation id
generation while segmentation.

Luckily we already have pregenerated the ip6_frag_id in
ip6_ufo_append_data and can use it here.

Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agovirtio-net: alloc big buffers also when guest can receive UFO
Jason Wang [Fri, 21 Feb 2014 05:08:04 +0000 (13:08 +0800)]
virtio-net: alloc big buffers also when guest can receive UFO

[ Upstream commit 0e7ede80d929ff0f830c44a543daa1acd590c749 ]

We should alloc big buffers also when guest can receive UFO
packets to let the big packets fit into guest rx buffer.

Fixes 5c5167515d80f78f6bb538492c423adcae31ad65
(virtio-net: Allow UFO feature to be set and advertised.)

Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoneigh: recompute reachabletime before returning from neigh_periodic_work()
Duan Jiong [Thu, 27 Feb 2014 09:14:41 +0000 (17:14 +0800)]
neigh: recompute reachabletime before returning from neigh_periodic_work()

[ Upstream commit feff9ab2e7fa773b6a3965f77375fe89f7fd85cf ]

If the neigh table's entries is less than gc_thresh1, the function
will return directly, and the reachabletime will not be recompute,
so the reachabletime can be guessed.

Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agonet-tcp: fastopen: fix high order allocations
Eric Dumazet [Thu, 20 Feb 2014 18:09:18 +0000 (10:09 -0800)]
net-tcp: fastopen: fix high order allocations

[ Upstream commit f5ddcbbb40aa0ba7fbfe22355d287603dbeeaaac ]

This patch fixes two bugs in fastopen :

1) The tcp_sendmsg(...,  @size) argument was ignored.

   Code was relying on user not fooling the kernel with iovec mismatches

2) When MTU is about 64KB, tcp_send_syn_data() attempts order-5
allocations, which are likely to fail when memory gets fragmented.

Fixes: 783237e8daf13 ("net-tcp: Fast Open client - sending SYN-data")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Tested-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agotun: remove bogus hardware vlan acceleration flags from vlan_features
Fernando Luis Vazquez Cao [Tue, 18 Feb 2014 12:20:09 +0000 (21:20 +0900)]
tun: remove bogus hardware vlan acceleration flags from vlan_features

[ Upstream commit 6671b2240c54585d4afb5286a29f1569fe5e40a8 ]

Even though only the outer vlan tag can be HW accelerated in the transmission
path, in the TUN/TAP driver vlan_features mirrors hw_features, which happens
to have the NETIF_F_HW_VLAN_?TAG_TX flags set. Because of this, during packet
tranmisssion through a stacked vlan device dev_hard_start_xmit, (incorrectly)
assuming that the vlan device supports hardware vlan acceleration, does not
add the vlan header to the skb payload and the inner vlan tags are lost
(vlan_tci contains the outer vlan tag when userspace reads the packet from
the tap device).

Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoveth: Fix vlan_features so as to be able to use stacked vlan interfaces
Toshiaki Makita [Tue, 18 Feb 2014 12:20:08 +0000 (21:20 +0900)]
veth: Fix vlan_features so as to be able to use stacked vlan interfaces

[ Upstream commit 8d0d21f4053c07714802cbe8b1fe26913ec296cc ]

Even if we create a stacked vlan interface such as veth0.10.20, it sends
single tagged frames (tagged with only vid 10).
Because vlan_features of a veth interface has the
NETIF_F_HW_VLAN_[CTAG/STAG]_TX bits, veth0.10 also has that feature, so
dev_hard_start_xmit(veth0.10) doesn't call __vlan_put_tag() and
vlan_dev_hard_start_xmit(veth0.10) overwrites vlan_tci.
This prevents us from using a combination of 802.1ad and 802.1Q
in containers, etc.

Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoSELinux: Increase ebitmap_node size for 64-bit configuration
Waiman Long [Tue, 23 Jul 2013 21:38:41 +0000 (17:38 -0400)]
SELinux: Increase ebitmap_node size for 64-bit configuration

commit a767f680e34bf14a36fefbbb6d85783eef99fd57 upstream.

Currently, the ebitmap_node structure has a fixed size of 32 bytes. On
a 32-bit system, the overhead is 8 bytes, leaving 24 bytes for being
used as bitmaps. The overhead ratio is 1/4.

On a 64-bit system, the overhead is 16 bytes. Therefore, only 16 bytes
are left for bitmap purpose and the overhead ratio is 1/2. With a
3.8.2 kernel, a boot-up operation will cause the ebitmap_get_bit()
function to be called about 9 million times. The average number of
ebitmap_node traversal is about 3.7.

This patch increases the size of the ebitmap_node structure to 64
bytes for 64-bit system to keep the overhead ratio at 1/4. This may
also improve performance a little bit by making node to node traversal
less frequent (< 2) as more bits are available in each node.

Signed-off-by: Waiman Long <Waiman.Long@hp.com>
Acked-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoSELinux: Reduce overhead of mls_level_isvalid() function call
Waiman Long [Tue, 23 Jul 2013 21:38:41 +0000 (17:38 -0400)]
SELinux: Reduce overhead of mls_level_isvalid() function call

commit fee7114298cf54bbd221cdb2ab49738be8b94f4c upstream.

While running the high_systime workload of the AIM7 benchmark on
a 2-socket 12-core Westmere x86-64 machine running 3.10-rc4 kernel
(with HT on), it was found that a pretty sizable amount of time was
spent in the SELinux code. Below was the perf trace of the "perf
record -a -s" of a test run at 1500 users:

  5.04%            ls  [kernel.kallsyms]     [k] ebitmap_get_bit
  1.96%            ls  [kernel.kallsyms]     [k] mls_level_isvalid
  1.95%            ls  [kernel.kallsyms]     [k] find_next_bit

The ebitmap_get_bit() was the hottest function in the perf-report
output.  Both the ebitmap_get_bit() and find_next_bit() functions
were, in fact, called by mls_level_isvalid(). As a result, the
mls_level_isvalid() call consumed 8.95% of the total CPU time of
all the 24 virtual CPUs which is quite a lot. The majority of the
mls_level_isvalid() function invocations come from the socket creation
system call.

Looking at the mls_level_isvalid() function, it is checking to see
if all the bits set in one of the ebitmap structure are also set in
another one as well as the highest set bit is no bigger than the one
specified by the given policydb data structure. It is doing it in
a bit-by-bit manner. So if the ebitmap structure has many bits set,
the iteration loop will be done many times.

The current code can be rewritten to use a similar algorithm as the
ebitmap_contains() function with an additional check for the
highest set bit. The ebitmap_contains() function was extended to
cover an optional additional check for the highest set bit, and the
mls_level_isvalid() function was modified to call ebitmap_contains().

With that change, the perf trace showed that the used CPU time drop
down to just 0.08% (ebitmap_contains + mls_level_isvalid) of the
total which is about 100X less than before.

  0.07%            ls  [kernel.kallsyms]     [k] ebitmap_contains
  0.05%            ls  [kernel.kallsyms]     [k] ebitmap_get_bit
  0.01%            ls  [kernel.kallsyms]     [k] mls_level_isvalid
  0.01%            ls  [kernel.kallsyms]     [k] find_next_bit

The remaining ebitmap_get_bit() and find_next_bit() functions calls
are made by other kernel routines as the new mls_level_isvalid()
function will not call them anymore.

This patch also improves the high_systime AIM7 benchmark result,
though the improvement is not as impressive as is suggested by the
reduction in CPU time spent in the ebitmap functions. The table below
shows the performance change on the 2-socket x86-64 system (with HT
on) mentioned above.

+--------------+---------------+----------------+-----------------+
|   Workload   | mean % change | mean % change  | mean % change   |
|              | 10-100 users  | 200-1000 users | 1100-2000 users |
+--------------+---------------+----------------+-----------------+
| high_systime |     +0.1%     |     +0.9%      |     +2.6%       |
+--------------+---------------+----------------+-----------------+

Signed-off-by: Waiman Long <Waiman.Long@hp.com>
Acked-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agomm: do not walk all of system memory during show_mem
Mel Gorman [Tue, 12 Nov 2013 23:08:15 +0000 (15:08 -0800)]
mm: do not walk all of system memory during show_mem

commit c78e93630d15b5f5774213aad9bdc9f52473a89b upstream.

It has been reported on very large machines that show_mem is taking almost
5 minutes to display information.  This is a serious problem if there is
an OOM storm.  The bulk of the cost is in show_mem doing a very expensive
PFN walk to give us the following information

  Total RAM:       Also available as totalram_pages
  Highmem pages:   Also available as totalhigh_pages
  Reserved pages:  Can be inferred from the zone structure
  Shared pages:    PFN walk required
  Unshared pages:  PFN walk required
  Quick pages:     Per-cpu walk required

Only the shared/unshared pages requires a full PFN walk but that
information is useless.  It is also inaccurate as page pins of unshared
pages would be accounted for as shared.  Even if the information was
accurate, I'm struggling to think how the shared/unshared information
could be useful for debugging OOM conditions.  Maybe it was useful before
rmap existed when reclaiming shared pages was costly but it is less
relevant today.

The PFN walk could be optimised a bit but why bother as the information is
useless.  This patch deletes the PFN walker and infers the total RAM,
highmem and reserved pages count from struct zone.  It omits the
shared/unshared page usage on the grounds that it is useless.  It also
corrects the reporting of HighMem as HighMem/MovableOnly as ZONE_MOVABLE
has similar problems to HighMem with respect to lowmem/highmem exhaustion.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: David Rientjes <rientjes@google.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoepoll: do not take the nested ep->mtx on EPOLL_CTL_DEL
Jason Baron [Thu, 2 Jan 2014 20:58:54 +0000 (12:58 -0800)]
epoll: do not take the nested ep->mtx on EPOLL_CTL_DEL

commit 4ff36ee94d93ddb4b7846177f1118d9aa33408e2 upstream.

The EPOLL_CTL_DEL path of epoll contains a classic, ab-ba deadlock.
That is, epoll_ctl(a, EPOLL_CTL_DEL, b, x), will deadlock with
epoll_ctl(b, EPOLL_CTL_DEL, a, x).  The deadlock was introduced with
commmit 67347fe4e632 ("epoll: do not take global 'epmutex' for simple
topologies").

The acquistion of the ep->mtx for the destination 'ep' was added such
that a concurrent EPOLL_CTL_ADD operation would see the correct state of
the ep (Specifically, the check for '!list_empty(&f.file->f_ep_links')

However, by simply not acquiring the lock, we do not serialize behind
the ep->mtx from the add path, and thus may perform a full path check
when if we had waited a little longer it may not have been necessary.
However, this is a transient state, and performing the full loop
checking in this case is not harmful.

The important point is that we wouldn't miss doing the full loop
checking when required, since EPOLL_CTL_ADD always locks any 'ep's that
its operating upon.  The reason we don't need to do lock ordering in the
add path, is that we are already are holding the global 'epmutex'
whenever we do the double lock.  Further, the original posting of this
patch, which was tested for the intended performance gains, did not
perform this additional locking.

Signed-off-by: Jason Baron <jbaron@akamai.com>
Cc: Nathan Zimmer <nzimmer@sgi.com>
Cc: Eric Wong <normalperson@yhbt.net>
Cc: Nelson Elhage <nelhage@nelhage.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoepoll: do not take global 'epmutex' for simple topologies
Jason Baron [Tue, 12 Nov 2013 23:10:18 +0000 (15:10 -0800)]
epoll: do not take global 'epmutex' for simple topologies

commit 67347fe4e6326338ee217d7eb826bedf30b2e155 upstream.

When calling EPOLL_CTL_ADD for an epoll file descriptor that is attached
directly to a wakeup source, we do not need to take the global 'epmutex',
unless the epoll file descriptor is nested.  The purpose of taking the
'epmutex' on add is to prevent complex topologies such as loops and deep
wakeup paths from forming in parallel through multiple EPOLL_CTL_ADD
operations.  However, for the simple case of an epoll file descriptor
attached directly to a wakeup source (with no nesting), we do not need to
hold the 'epmutex'.

This patch along with 'epoll: optimize EPOLL_CTL_DEL using rcu' improves
scalability on larger systems.  Quoting Nathan Zimmer's mail on SPECjbb
performance:

"On the 16 socket run the performance went from 35k jOPS to 125k jOPS.  In
addition the benchmark when from scaling well on 10 sockets to scaling
well on just over 40 sockets.

...

Currently the benchmark stops scaling at around 40-44 sockets but it seems like
I found a second unrelated bottleneck."

[akpm@linux-foundation.org: use `bool' for boolean variables, remove unneeded/undesirable cast of void*, add missed ep_scan_ready_list() kerneldoc]
Signed-off-by: Jason Baron <jbaron@akamai.com>
Tested-by: Nathan Zimmer <nzimmer@sgi.com>
Cc: Eric Wong <normalperson@yhbt.net>
Cc: Nelson Elhage <nelhage@nelhage.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoepoll: optimize EPOLL_CTL_DEL using rcu
Jason Baron [Tue, 12 Nov 2013 23:10:16 +0000 (15:10 -0800)]
epoll: optimize EPOLL_CTL_DEL using rcu

commit ae10b2b4eb01bedc91d29d5c5bb9e416fd806c40 upstream.

Nathan Zimmer found that once we get over 10+ cpus, the scalability of
SPECjbb falls over due to the contention on the global 'epmutex', which is
taken in on EPOLL_CTL_ADD and EPOLL_CTL_DEL operations.

Patch #1 removes the 'epmutex' lock completely from the EPOLL_CTL_DEL path
by using rcu to guard against any concurrent traversals.

Patch #2 remove the 'epmutex' lock from EPOLL_CTL_ADD operations for
simple topologies.  IE when adding a link from an epoll file descriptor to
a wakeup source, where the epoll file descriptor is not nested.

This patch (of 2):

Optimize EPOLL_CTL_DEL such that it does not require the 'epmutex' by
converting the file->f_ep_links list into an rcu one.  In this way, we can
traverse the epoll network on the add path in parallel with deletes.
Since deletes can't create loops or worse wakeup paths, this is safe.

This patch in combination with the patch "epoll: Do not take global 'epmutex'
for simple topologies", shows a dramatic performance improvement in
scalability for SPECjbb.

Signed-off-by: Jason Baron <jbaron@akamai.com>
Tested-by: Nathan Zimmer <nzimmer@sgi.com>
Cc: Eric Wong <normalperson@yhbt.net>
Cc: Nelson Elhage <nelhage@nelhage.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
CC: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agox86/dumpstack: Fix printk_address for direct addresses
Jiri Slaby [Fri, 25 Oct 2013 13:06:58 +0000 (15:06 +0200)]
x86/dumpstack: Fix printk_address for direct addresses

commit 5f01c98859073cb512b01d4fad74b5f4e047be0b upstream.

Consider a kernel crash in a module, simulated the following way:

 static int my_init(void)
 {
         char *map = (void *)0x5;
         *map = 3;
         return 0;
 }
 module_init(my_init);

When we turn off FRAME_POINTERs, the very first instruction in
that function causes a BUG. The problem is that we print IP in
the BUG report using %pB (from printk_address). And %pB
decrements the pointer by one to fix printing addresses of
functions with tail calls.

This was added in commit 71f9e59800e5ad4 ("x86, dumpstack: Use
%pB format specifier for stack trace") to fix the call stack
printouts.

So instead of correct output:

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000005
  IP: [<ffffffffa01ac000>] my_init+0x0/0x10 [pb173]

We get:

  BUG: unable to handle kernel NULL pointer dereference at 0000000000000005
  IP: [<ffffffffa0152000>] 0xffffffffa0151fff

To fix that, we use %pS only for stack addresses printouts (via
newly added printk_stack_address) and %pB for regs->ip (via
printk_address). I.e. we revert to the old behaviour for all
except call stacks. And since from all those reliable is 1, we
remove that parameter from printk_address.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: joe@perches.com
Cc: jirislaby@gmail.com
Link: http://lkml.kernel.org/r/1382706418-8435-1-git-send-email-jslaby@suse.cz
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agoSUNRPC: close a rare race in xs_tcp_setup_socket.
NeilBrown [Thu, 31 Oct 2013 05:14:36 +0000 (16:14 +1100)]
SUNRPC: close a rare race in xs_tcp_setup_socket.

commit 93dc41bdc5c853916610576c6b48a1704959c70d upstream.

We have one report of a crash in xs_tcp_setup_socket.
The call path to the crash is:

  xs_tcp_setup_socket -> inet_stream_connect -> lock_sock_nested.

The 'sock' passed to that last function is NULL.

The only way I can see this happening is a concurrent call to
xs_close:

  xs_close -> xs_reset_transport -> sock_release -> inet_release

inet_release sets:
   sock->sk = NULL;
inet_stream_connect calls
   lock_sock(sock->sk);
which gets NULL.

All calls to xs_close are protected by XPRT_LOCKED as are most
activations of the workqueue which runs xs_tcp_setup_socket.
The exception is xs_tcp_schedule_linger_timeout.

So presumably the timeout queued by the later fires exactly when some
other code runs xs_close().

To protect against this we can move the cancel_delayed_work_sync()
call from xs_destory() to xs_close().

As xs_close is never called from the worker scheduled on
->connect_worker, this can never deadlock.

Signed-off-by: NeilBrown <neilb@suse.de>
[Trond: Make it safe to call cancel_delayed_work_sync() on AF_LOCAL sockets]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosched/rt: Remove redundant nr_cpus_allowed test
Shawn Bohrer [Fri, 4 Oct 2013 19:24:53 +0000 (14:24 -0500)]
sched/rt: Remove redundant nr_cpus_allowed test

commit 6bfa687c19b7ab8adee03f0d43c197c2945dd869 upstream.

In 76854c7e8f3f4172fef091e78d88b3b751463ac6 ("sched: Use
rt.nr_cpus_allowed to recover select_task_rq() cycles") an
optimization was added to select_task_rq_rt() that immediately
returns when p->nr_cpus_allowed == 1 at the beginning of the
function.

This makes the latter p->nr_cpus_allowed > 1 check redundant,
which can now be removed.

Signed-off-by: Shawn Bohrer <sbohrer@rgmadvisors.com>
Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Mike Galbraith <mgalbraith@suse.de>
Cc: tomk@rgmadvisors.com
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1380914693-24634-1-git-send-email-shawn.bohrer@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosched/rt: Add missing rmb()
Peter Zijlstra [Tue, 15 Oct 2013 10:35:07 +0000 (12:35 +0200)]
sched/rt: Add missing rmb()

commit 7c3f2ab7b844f1a859afbc3d41925e8a0faba5fa upstream.

While discussing the proposed SCHED_DEADLINE patches which in parts
mimic the existing FIFO code it was noticed that the wmb in
rt_set_overloaded() didn't have a matching barrier.

The only site using rt_overloaded() to test the rto_count is
pull_rt_task() and we should issue a matching rmb before then assuming
there's an rto_mask bit set.

Without that smp_rmb() in there we could actually miss seeing the
rto_mask bit.

Also, change to using smp_[wr]mb(), even though this is SMP only code;
memory barriers without smp_ always make me think they're against
hardware of some sort.

Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: vincent.guittot@linaro.org
Cc: luca.abeni@unitn.it
Cc: bruce.ashfield@windriver.com
Cc: dhaval.giani@gmail.com
Cc: rostedt@goodmis.org
Cc: hgu1972@gmail.com
Cc: oleg@redhat.com
Cc: fweisbec@gmail.com
Cc: darren@dvhart.com
Cc: johan.eker@ericsson.com
Cc: p.faure@akatech.ch
Cc: paulmck@linux.vnet.ibm.com
Cc: raistlin@linux.it
Cc: claudio@evidence.eu.com
Cc: insop.song@gmail.com
Cc: michael@amarulasolutions.com
Cc: liming.wang@windriver.com
Cc: fchecconi@gmail.com
Cc: jkacur@redhat.com
Cc: tommaso.cucinotta@sssup.it
Cc: Juri Lelli <juri.lelli@gmail.com>
Cc: harald.gustafsson@ericsson.com
Cc: nicola.manica@disi.unitn.it
Cc: tglx@linutronix.de
Link: http://lkml.kernel.org/r/20131015103507.GF10651@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
10 years agosched: Assign correct scheduling domain to 'sd_llc'
Mel Gorman [Tue, 17 Dec 2013 09:21:25 +0000 (09:21 +0000)]
sched: Assign correct scheduling domain to 'sd_llc'

commit 5d4cf996cf134e8ddb4f906b8197feb9267c2b77 upstream.

Commit 42eb088e (sched: Avoid NULL dereference on sd_busy) corrected a NULL
dereference on sd_busy but the fix also altered what scheduling domain it
used for the 'sd_llc' percpu variable.

One impact of this is that a task selecting a runqueue may consider
idle CPUs that are not cache siblings as candidates for running.
Tasks are then running on CPUs that are not cache hot.

This was found through bisection where ebizzy threads were not seeing equal
performance and it looked like a scheduling fairness issue. This patch
mitigates but does not completely fix the problem on all machines tested
implying there may be an additional bug or a common root cause. Here are
the average range of performance seen by individual ebizzy threads. It
was tested on top of candidate patches related to x86 TLB range flushing.

4-core machine
    3.13.0-rc3            3.13.0-rc3
       vanilla            fixsd-v3r3
Mean   1        0.00 (  0.00%)        0.00 (  0.00%)
Mean   2        0.34 (  0.00%)        0.10 ( 70.59%)
Mean   3        1.29 (  0.00%)        0.93 ( 27.91%)
Mean   4        7.08 (  0.00%)        0.77 ( 89.12%)
Mean   5      193.54 (  0.00%)        2.14 ( 98.89%)
Mean   6      151.12 (  0.00%)        2.06 ( 98.64%)
Mean   7      115.38 (  0.00%)        2.04 ( 98.23%)
Mean   8      108.65 (  0.00%)        1.92 ( 98.23%)

8-core machine
Mean   1         0.00 (  0.00%)        0.00 (  0.00%)
Mean   2         0.40 (  0.00%)        0.21 ( 47.50%)
Mean   3        23.73 (  0.00%)        0.89 ( 96.25%)
Mean   4        12.79 (  0.00%)        1.04 ( 91.87%)
Mean   5        13.08 (  0.00%)        2.42 ( 81.50%)
Mean   6        23.21 (  0.00%)       69.46 (-199.27%)
Mean   7        15.85 (  0.00%)      101.72 (-541.77%)
Mean   8       109.37 (  0.00%)       19.13 ( 82.51%)
Mean   12      124.84 (  0.00%)       28.62 ( 77.07%)
Mean   16      113.50 (  0.00%)       24.16 ( 78.71%)

It's eliminated for one machine and reduced for another.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: Alex Shi <alex.shi@linaro.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: H Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20131217092124.GV11295@suse.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>