The newly added dmaengine support in the omap2 nand driver
potentially causes an undefined return value from the
omap_nand_probe function when dmaengine_slave_config
reports an error. Let's handle this by returning the
same error back to the caller.
Without this patch, building omap2plus_defconfig results in:
drivers/mtd/nand/omap2.c: In function 'omap_nand_probe':
drivers/mtd/nand/omap2.c:1154:6: warning: 'err' may be used uninitialized in this function [-Wuninitialized]
Arnd Bergmann [Sat, 4 Aug 2012 10:33:24 +0000 (10:33 +0000)]
gpio: em: do not discard em_gio_irq_domain_cleanup
The newly added gpio-em driver marks its em_gio_irq_domain_cleanup
function as __devexit, which would lead to that function being
discarded in case CONFIG_HOTPLUG is disabled. However, the function
is also called by the error handling logic em_gio_probe, which
would cause a jump into a NULL pointer if it was removed from the
kernel or module.
Without this patch, building kzm9d_defconfig results in:
WARNING: drivers/gpio/built-in.o(.devinit.text+0x330): Section mismatch in reference from the function em_gio_probe() to the function .devexit.text:em_gio_irq_domain_cleanup()
The function __devinit em_gio_probe() references
a function __devexit em_gio_irq_domain_cleanup().
This is often seen when error handling in the init function
uses functionality in the exit path.
The fix is often to remove the __devexit annotation of
em_gio_irq_domain_cleanup() so it may be used outside an exit section.
Arnd Bergmann [Sat, 4 Aug 2012 10:28:33 +0000 (10:28 +0000)]
ARM: exynos: exynos_pm_add_dev_to_genpd may be unused
exynos_pm_add_dev_to_genpd is used if one or more out of a large
number of Kconfig symbols are enabled. However the new
exynos_defconfig selects none of those, so the function becomes
unused. Marking it so lets the compiler automatically discard
it.
Without this patch, building exynos_defconfig results in:
arch/arm/mach-exynos/pm_domains.c:118:123: warning: 'exynos_pm_add_dev_to_genpd' defined but not used [-Wunused-function]
Kees Cook [Fri, 10 Aug 2012 02:01:26 +0000 (19:01 -0700)]
Yama: higher restrictions should block PTRACE_TRACEME
The higher ptrace restriction levels should be blocking even
PTRACE_TRACEME requests. The comments in the LSM documentation are
misleading about when the checks happen (the parent does not go through
security_ptrace_access_check() on a PTRACE_TRACEME call).
Eric Dumazet [Thu, 9 Aug 2012 13:56:06 +0000 (13:56 +0000)]
ipv4: tcp: unicast_sock should not land outside of TCP stack
commit be9f4a44e7d41cee (ipv4: tcp: remove per net tcp_sock) added a
selinux regression, reported and bisected by John Stultz
selinux_ip_postroute_compat() expect to find a valid sk->sk_security
pointer, but this field is NULL for unicast_sock
It turns out that unicast_sock are really temporary stuff to be able
to reuse part of IP stack (ip_append_data()/ip_push_pending_frames())
Fact is that frames sent by ip_send_unicast_reply() should be orphaned
to not fool LSM.
Note IPv6 never had this problem, as tcp_v6_send_response() doesnt use a
fake socket at all. I'll probably implement tcp_v4_send_response() to
remove these unicast_sock in linux-3.7
Yuval Mintz [Thu, 9 Aug 2012 04:37:26 +0000 (04:37 +0000)]
bnx2x: Fix recovery flow cleanup during probe
During probe, every function probed clears the recovery registers from
all functions on its path - thus signaling that given a future recovery
event, there will be no need to wait for those functions.
This is a flawed behaviour - each function should only be responsible
for its own bit.
Since this registers are handled during the load/unload routines,
this cleanup is removed altogether.
Yuval Mintz [Thu, 9 Aug 2012 04:37:25 +0000 (04:37 +0000)]
bnx2x: fix unload previous driver flow when flr-capable
The existing previous driver unload flow is flawed, causing the probe of
functions reaching the 'uncommon fork' in flr-capable devices to fail.
This patch resolves this, as well as fixing the flow for hypervisors which
disable flr capabilities from functions as they pass them as PDA to VMs,
as we cannot base the flow on the pci configuration space.
This is a fix for bug, introduced in 3.4 kernel by commit 1ab5ecb90cb6a3df1476e052f76a6e8f6511cb3d ("tun: don't hold network
namespace by tun sockets"), which, among other things, replaced simple
sock_put() by sk_release_kernel(). Below is sequence, which leads to
oops for non-persistent devices:
Alexander Block [Thu, 2 Aug 2012 23:16:20 +0000 (17:16 -0600)]
Btrfs: remove mnt_want_write call in btrfs_mksubvol
We got a recursive lock in mksubvol because the caller already held
a lock. I think we got into this due to a merge error. Commit a874a63
removed the mnt_want_write call from btrfs_mksubvol and added a
replacement call to mnt_want_write_file in btrfs_ioctl_snap_create_transid.
Commit e7848683 however tried to move all calls to mnt_want_write above
i_mutex. So somewhere while merging this, it got mixed up. The
solution is to remove the mnt_want_write call completely from
mksubvol.
Arnd Bergmann [Sat, 4 Aug 2012 10:48:12 +0000 (10:48 +0000)]
ARM: imx: gpmi-nand depends on mxs-dma
It is not currently possible to build the gpmi-nand driver without
also building the mxs-dma driver. Clarify this Kconfig and enable
both in the defconfig file so we can build it again with both enabled.
drivers/built-in.o: In function `gpmi_dma_filter':
clk-fixed-factor.c:(.text+0xafc18): undefined reference to `mxs_dma_is_apbh'
make[1]: *** [vmlinux] Error 1
make: *** [sub-make] Error 2
Arnd Bergmann [Sat, 4 Aug 2012 10:31:24 +0000 (10:31 +0000)]
ARM: integrator: include <linux/export.h>
Without this patch, building integrator_defconfig results in:
arch/arm/mach-integrator/core.c:150:1: warning: data definition has no type or storage class [enabled by default]
arch/arm/mach-integrator/core.c:150:1: warning: type defaults to 'int' in declaration of 'EXPORT_SYMBOL' [-Wimplicit-int]
arch/arm/mach-integrator/core.c:150:1: warning: parameter names (without types) in function declaration [enabled by default]
Arnd Bergmann [Sat, 4 Aug 2012 07:52:19 +0000 (07:52 +0000)]
ARM: s3c24xx: use new PWM driver
The samsung PWM driver has moved to the new PWM subsystem, which
changed the Kconfig symbol for that driver, but the rx1950 and
gta02 boards still uses the old one.
Without this patch, building s3c2410_defconfig results in:
arch/arm/mach-s3c24xx/built-in.o: In function `rx1950_lcd_power':
arch/arm/mach-s3c24xx/mach-rx1950.c:430: undefined reference to `pwm_config'
arch/arm/mach-s3c24xx/mach-rx1950.c:431: undefined reference to `pwm_disable'
arch/arm/mach-s3c24xx/mach-rx1950.c:437: undefined reference to `pwm_config'
arch/arm/mach-s3c24xx/mach-rx1950.c:438: undefined reference to `pwm_enable'
arch/arm/mach-s3c24xx/built-in.o: In function `rx1950_backlight_exit':
arch/arm/mach-s3c24xx/mach-rx1950.c:504: undefined reference to `pwm_free'
arch/arm/mach-s3c24xx/built-in.o: In function `rx1950_backlight_init':
arch/arm/mach-s3c24xx/mach-rx1950.c:487: undefined reference to `pwm_request'
Arnd Bergmann [Sat, 4 Aug 2012 04:59:34 +0000 (04:59 +0000)]
ARM: sa1100: include linux/io.h in hackkit leds code
The sa1100 definition of the io_p2v macro has changed in v3.6, and this one
file stopped working because of that.
Without this patch, building hackkit_defconfig results in:
arch/arm/mach-sa1100/leds-hackkit.c: In function 'hackkit_leds_event':
arch/arm/mach-sa1100/leds-hackkit.c:39:4: error: implicit declaration of function 'IOMEM' [-Werror=implicit-function-declaration]
The EETI touchscreen asserts its IRQ line as soon as it has data in its
internal buffers. The line is automatically deasserted once all data has
been read via I2C. Hence, the driver has to monitor the GPIO line and
cannot simply rely on the interrupt handler reception.
In the current implementation of the driver, irq_to_gpio() is used to
determine the GPIO number from the i2c_client's IRQ value.
As irq_to_gpio() is not available on all platforms, this patch changes
this and makes the driver ignore the passed in IRQ. Instead, a GPIO is
added to the platform_data struct and gpio_to_irq is used to derive the
IRQ from that GPIO. If this fails, bail out. The driver is only able to
work in environments where the touchscreen GPIO can be mapped to an
IRQ.
Without this patch, building raumfeld_defconfig results in:
drivers/input/touchscreen/eeti_ts.c: In function 'eeti_ts_irq_active':
drivers/input/touchscreen/eeti_ts.c:65:2: error: implicit declaration of function 'irq_to_gpio' [-Werror=implicit-function-declaration]
Arnd Bergmann [Sun, 5 Aug 2012 14:58:37 +0000 (14:58 +0000)]
ARM: pxa: remove irq_to_gpio from ezx-pcap driver
The irq_to_gpio function was removed from the pxa platform
in linux-3.2, and this driver has been broken since.
There is actually no in-tree user of this driver that adds
this platform device, but the driver can and does get enabled
on some platforms.
Without this patch, building ezx_defconfig results in:
drivers/mfd/ezx-pcap.c: In function 'pcap_isr_work':
drivers/mfd/ezx-pcap.c:205:2: error: implicit declaration of function 'irq_to_gpio' [-Werror=implicit-function-declaration]
Stephen Warren [Fri, 3 Aug 2012 23:45:21 +0000 (17:45 -0600)]
ARM: tegra: more regulator fixes for Harmony
Commit 3d55c29 "ARM: tegra: harmony: add regulator supply name and its
input supply" was supposed to fix all the problems with regulators on
Harmony. However, it appears that I only tested it when booting using
board files, not when booting using device tree. This change fixes two
problems with regulators when booting using device tree:
1) That patch only created the vdd_sys regulator when booting using a
board file. Since this is the root of the whole regulator tree, this
caused no regulators to successfully initialize when booting using
device tree. The registration of vdd_sys is moved to fix this.
2) When booting use DT, the regulator core sets has_full_constraints,
which in turn causes the core to turn off any regulators not marked
as always on. Some of the affected regulators are required for basic
system operation. To solve this, add always on constraints to all
relevant regulators. This doesn't affect booting using a board file
since nothing sets has_full_constraints in that case.
Emil Tantilov [Wed, 1 Aug 2012 08:12:21 +0000 (08:12 +0000)]
e1000e: fix panic while dumping packets on Tx hang with IOMMU
This patch resolves a "BUG: unable to handle kernel paging request at ..."
oops while dumping packet data. The issue occurs with IOMMU enabled due to
the address provided by phys_to_virt().
This patch avoids phys_to_virt() by using skb->data and the address of the
pages allocated for Rx.
Emil Tantilov [Sat, 28 Jul 2012 05:07:48 +0000 (05:07 +0000)]
igb: fix panic while dumping packets on Tx hang with IOMMU
This patch resolves a "BUG: unable to handle kernel paging request at ..."
oops while dumping packet data. The issue occurs with IOMMU enabled due to
the address provided by phys_to_virt().
This patch avoids phys_to_virt() by making using skb->data and the address
of the pages allocated for Rx.
Aaro Koskinen [Tue, 7 Aug 2012 12:44:05 +0000 (14:44 +0200)]
ARM: dma-mapping: fix incorrect freeing of atomic allocations
Commit e9da6e9905e639b0f842a244bc770b48ad0523e9 (ARM: dma-mapping:
remove custom consistent dma region) changed the way atomic allocations
are handled. However, arm_dma_free() was not modified accordingly, and
as a result freeing of atomic allocations does not work correctly when
CMA is disabled. Memory is leaked and following WARNINGs are seen:
[ 57.698911] ------------[ cut here ]------------
[ 57.753518] WARNING: at arch/arm/mm/dma-mapping.c:263 arm_dma_free+0x88/0xe4()
[ 57.811473] trying to free invalid coherent area: e0848000
[ 57.867398] Modules linked in: sata_mv(-)
[ 57.921373] [<c000d270>] (unwind_backtrace+0x0/0xf0) from [<c0015430>] (warn_slowpath_common+0x50/0x68)
[ 58.033924] [<c0015430>] (warn_slowpath_common+0x50/0x68) from [<c00154dc>] (warn_slowpath_fmt+0x30/0x40)
[ 58.152024] [<c00154dc>] (warn_slowpath_fmt+0x30/0x40) from [<c000dc18>] (arm_dma_free+0x88/0xe4)
[ 58.219592] [<c000dc18>] (arm_dma_free+0x88/0xe4) from [<c008fa30>] (dma_pool_destroy+0x100/0x148)
[ 58.345526] [<c008fa30>] (dma_pool_destroy+0x100/0x148) from [<c019a64c>] (release_nodes+0x144/0x218)
[ 58.475782] [<c019a64c>] (release_nodes+0x144/0x218) from [<c0197e10>] (__device_release_driver+0x60/0xb8)
[ 58.614260] [<c0197e10>] (__device_release_driver+0x60/0xb8) from [<c0198608>] (driver_detach+0xd8/0xec)
[ 58.756527] [<c0198608>] (driver_detach+0xd8/0xec) from [<c0197c54>] (bus_remove_driver+0x7c/0xc4)
[ 58.901648] [<c0197c54>] (bus_remove_driver+0x7c/0xc4) from [<c004bfac>] (sys_delete_module+0x19c/0x220)
[ 59.051447] [<c004bfac>] (sys_delete_module+0x19c/0x220) from [<c0009140>] (ret_fast_syscall+0x0/0x2c)
[ 59.207996] ---[ end trace 0745420412c0325a ]---
[ 59.287110] ------------[ cut here ]------------
[ 59.366324] WARNING: at arch/arm/mm/dma-mapping.c:263 arm_dma_free+0x88/0xe4()
[ 59.450511] trying to free invalid coherent area: e0847000
[ 59.534357] Modules linked in: sata_mv(-)
[ 59.616785] [<c000d270>] (unwind_backtrace+0x0/0xf0) from [<c0015430>] (warn_slowpath_common+0x50/0x68)
[ 59.790030] [<c0015430>] (warn_slowpath_common+0x50/0x68) from [<c00154dc>] (warn_slowpath_fmt+0x30/0x40)
[ 59.972322] [<c00154dc>] (warn_slowpath_fmt+0x30/0x40) from [<c000dc18>] (arm_dma_free+0x88/0xe4)
[ 60.070701] [<c000dc18>] (arm_dma_free+0x88/0xe4) from [<c008fa30>] (dma_pool_destroy+0x100/0x148)
[ 60.256817] [<c008fa30>] (dma_pool_destroy+0x100/0x148) from [<c019a64c>] (release_nodes+0x144/0x218)
[ 60.445201] [<c019a64c>] (release_nodes+0x144/0x218) from [<c0197e10>] (__device_release_driver+0x60/0xb8)
[ 60.634148] [<c0197e10>] (__device_release_driver+0x60/0xb8) from [<c0198608>] (driver_detach+0xd8/0xec)
[ 60.823623] [<c0198608>] (driver_detach+0xd8/0xec) from [<c0197c54>] (bus_remove_driver+0x7c/0xc4)
[ 61.013268] [<c0197c54>] (bus_remove_driver+0x7c/0xc4) from [<c004bfac>] (sys_delete_module+0x19c/0x220)
[ 61.203472] [<c004bfac>] (sys_delete_module+0x19c/0x220) from [<c0009140>] (ret_fast_syscall+0x0/0x2c)
[ 61.393390] ---[ end trace 0745420412c0325b ]---
Aaro Koskinen [Tue, 7 Aug 2012 12:39:25 +0000 (14:39 +0200)]
ARM: dma-mapping: fix atomic allocation alignment
The alignment mask is calculated incorrectly. Fixing the calculation
makes strange hangs/lockups disappear during the boot with Amstrad E3
and 3.6-rc1 kernel.
Chris Brand [Tue, 7 Aug 2012 12:01:14 +0000 (14:01 +0200)]
ARM: mm: fix MMU mapping of CMA regions
Fix dma_contiguous_remap() so that it continues through all the
regions, even after encountering one that is outside lowmem.
Without this change, if you have two CMA regions, the first outside
lowmem and the seocnd inside lowmem, only the second one will get
set up in the MMU. Data written to that region then doesn't get
automatically flushed from the cache into memory.
Signed-off-by: Chris Brand <[email protected]>
[extended patch subject with 'fix' word] Signed-off-by: Marek Szyprowski <[email protected]>
Dave Airlie [Wed, 8 Aug 2012 23:54:49 +0000 (09:54 +1000)]
Merge branch 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel into drm-next
Daniel writes:
"- Regression fixer for an OOPS at boot when i915.ko is built-in and
CONFIG_PM=n, introduce in 3.5 (patch from Hunt Xu)
- Regression fixer for occlusion query failures, the required w/a wasn't
applied in all cases (thanks to Eric for tracking this on down).
- dmar vs. dma_buf imprt fix (Dave Airlie)
- 2 patches to fight down forcewake issues on snb. This is the stuff I've
talked about 2 weeks ago already, it's a minefield. Investigation still
going on, but afaict this is the best we have for now.
- a few minor things to keep coverty&compiler happy (Alan, Davendra,
Stéphane)
- tons of hsw pci ids - this one is a bit late because internal approval
sometimes takes a while, but ppl in charge finally agreed that world+dog
already knows about ult and crw haswell variants ;-)
Wrt regressions I'm aware of:
- the power regression due to semaphores=1. Ben is running around with a
killawatt, unfortunately we have a hard time reproducing this one. And
this /shouldn't/ increase power usage. Ben has turned up a few odds bits
though already.
- the lvds fix in 3.6-rc1 broke a backlight after lid close/open (but can
be resurrected with a modeset cycle). I guess we anger the bios - I'm
still looking into this one.
- gmbus broke edid reading on an odd-ball monitor, we need to fall-back.
Due to vacation (both mine&the reporter's) this is stalling for a final
patch and a tested-by on it. But issue is fully diagnosed."
* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel:
drm/i915: correctly order the ring init sequence
drm/i915: add more Haswell PCI IDs
drm/i915: make rc6 in sysfs functions conditional
drm/i915: Workaround hang with BSD and forcewake on SandyBridge
drm/i915: Make intel_panel_get_backlight static.
i915: don't map imported dma-bufs for dmar.
drm/i915: remove unused variable
drm/i915: Don't forget to apply SNB PIPE_CONTROL GTT workaround.
drm/i915: fix forcewake related hangs on snb
i915: Remove silly test
i915: fix error path leak in intel_sdvo_write_cmd
vlv: it might be wise if we initialised the flag value...
Arnd Bergmann [Wed, 8 Aug 2012 04:47:24 +0000 (04:47 +0000)]
net/stmmac: mark probe function as __devinit
Driver probe functions are generally __devinit so they will be
discarded after initialization for non-hotplug kernels.
This was found by a new warning after patch 6a228452d "stmmac: Add
device-tree support" adds a new __devinit function that is called
from stmmac_pltfr_probe.
Without this patch, building socfpga_defconfig results in:
WARNING: drivers/net/ethernet/stmicro/stmmac/stmmac.o(.text+0x5d4c): Section mismatch in reference from the function stmmac_pltfr_probe() to the function .devinit.text:stmmac_probe_config_dt()
The function stmmac_pltfr_probe() references
the function __devinit stmmac_probe_config_dt().
This is often because stmmac_pltfr_probe lacks a __devinit
annotation or the annotation of stmmac_probe_config_dt is wrong.
The #ifdefs regarding CONFIG_ARCH_LPC32XX_MII_SUPPORT and
CONFIG_ARCH_LPC32XX_IRAM_FOR_NET are obsolete since the symbols have been
removed from Kconfig and replaced by devicetree based configuration.
Jesper Juhl [Tue, 7 Aug 2012 11:56:26 +0000 (11:56 +0000)]
cdc-phonet: Don't leak in usbpn_open
We allocate memory for 'req' with usb_alloc_urb() and then test
'if (!req || rx_submit(pnd, req, GFP_KERNEL | __GFP_COLD))'.
If we enter that branch due to '!req' then there is no problem. But if
we enter the branch due to 'req' being != 0 and the 'rx_submit()' call
being false, then we'll leak the memory we allocated.
Deal with the leak by always calling 'usb_free_urb(req)' when entering
the branch. If 'req' happens to be 0 then the call is harmless, if it
is not 0 then we free the memory we allocated but don't need.
Jesper Juhl [Tue, 7 Aug 2012 08:32:34 +0000 (08:32 +0000)]
batman-adv: Fix mem leak in the batadv_tt_local_event() function
Memory is allocated for 'tt_change_node' with kmalloc().
'tt_change_node' may go out of scope really being used for anything
(except have a few members initialized) if we hit the 'del:' label.
This patch makes sure we free the memory in that case.
Paolo Valente [Tue, 7 Aug 2012 07:27:25 +0000 (07:27 +0000)]
sched: add missing group change to qfq_change_class
[Resending again, as the text was corrupted by the email client]
To speed up operations, QFQ internally divides classes into
groups. Which group a class belongs to depends on the ratio between
the maximum packet length and the weight of the class. Unfortunately
the function qfq_change_class lacks the steps for changing the group
of a class when the ratio max_pkt_len/weight of the class changes.
For example, when the last of the following three commands is
executed, the group of class 1:1 is not correctly changed:
tc disc add dev XXX root handle 1: qfq
tc class add dev XXX parent 1: qfq classid 1:1 weight 1
tc class change dev XXX parent 1: classid 1:1 qfq weight 4
Not changing the group of a class does not affect the long-term
bandwidth guaranteed to the class, as the latter is independent of the
maximum packet length, and correctly changes (only) if the weight of
the class changes. In contrast, if the group of the class is not
updated, the class is still guaranteed the short-term bandwidth and
packet delay related to its old group, instead of the guarantees that
it should receive according to its new weight and/or maximum packet
length. This may also break service guarantees for other classes.
This patch adds the missing operations.
David S. Miller [Wed, 8 Aug 2012 22:55:02 +0000 (15:55 -0700)]
Merge branch 'fixes-for-3.6' of git://gitorious.org/linux-can/linux-can
Marc Kleine-Budde says:
====================
here's a fix intended for the v3.6 release cycle. Oliver noticed and
fixed that the flags definition for the new canfd_frame contains
redundant and confusing information.
====================
Sonic Zhang [Mon, 6 Aug 2012 16:08:36 +0000 (16:08 +0000)]
drivers: net: irda: bfin_sir: fix compile error
Bit IREN is replaced by UMOD_IRDA and UMOD_MASK since blackfin 60x added, but
this driver didn't update which will cause bfin_sir build error:
drivers/net/irda/bfin_sir.c:161:9: error: 'IREN' undeclared (first use in this
function)
drivers/net/irda/bfin_sir.c:435:18: error: 'IREN' undeclared (first use in
this function)
drivers/net/irda/bfin_sir.c:521:11: error: 'IREN' undeclared (first use in
this function)
Arnd Bergmann [Sat, 4 Aug 2012 07:12:00 +0000 (07:12 +0000)]
usb/ohci-omap: remove unused variable
Commit c2e935a7d "USB: move transceiver from ehci_hcd and ohci_hcd to
hcd and rename it as phy" removed the last use of the "ohci" variable
in the usb_hcd_omap_remove function, but left the variable in place
unused.
Without this patch, building omap1_defconfig results in:
In file included from drivers/usb/host/ohci-hcd.c:1013:0:
drivers/usb/host/ohci-omap.c: In function 'usb_hcd_omap_remove':
drivers/usb/host/ohci-omap.c:406:19: warning: unused variable 'ohci' [-Wunused-variable]
Arnd Bergmann [Sat, 4 Aug 2012 06:20:49 +0000 (06:20 +0000)]
mfd/asic3: fix asic3_mfd_probe return value
In commit 4f304245b "mfd: Set asic3 DS1WM clock_rate", a possible
path through asic3_mfd_probe was introduced that would lead to
an unpredictable return value, if everything succeeds but there
are pdata->leds is NULL. This was reported correctly by gcc.
Without this patch, building magician_defconfig results in:
drivers/mfd/asic3.c: In function 'asic3_mfd_probe':
drivers/mfd/asic3.c:940:2: warning: 'ret' may be used uninitialized in this function [-Wuninitialized]
Revert "NMI watchdog: fix for lockup detector breakage on resume"
Revert commit 45226e9 (NMI watchdog: fix for lockup detector breakage
on resume) which breaks resume from system suspend on my SH7372
Mackerel board (by causing a NULL pointer dereference to happen) and
is generally wrong, because it abuses the CPU hotplug functionality
in a shamelessly blatant way.
The original issue should be addressed through appropriate syscore
resume callback instead.
PM: Make dev_pm_get_subsys_data() always return 0 on success
Commits 1d5fcfec22 (PM / Domains: Add device domain data reference
counter) and 62d4490294 (PM / Domains: Allow device callbacks to be
added at any time) added checks for the return value of
dev_pm_get_subsys_data(), but those checks were incorrect, because
that function returned 1 on success in some cases.
Since all of the existing users of dev_pm_get_subsys_data() don't use
the positive value returned by it on success, change its definition
so that it always returns 0 when successful.
NeilBrown [Sun, 5 Aug 2012 20:56:20 +0000 (22:56 +0200)]
RTC: Avoid races between RTC alarm wakeup and suspend.
If an RTC alarm fires just as suspend is happening, it is possible for
suspend to complete and the alarm to be missed.
To avoid the race, we must register the event with the PM core.
As the event is made visible to userspace through a thread which is
only scheduled by the interrupt, we need a pm_stay_awake/pm_relax
pair preventing suspend from the interrupt until the thread completes
its work.
This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as
it provides suspend protection for all RTCs that use rtc_update_irq.
1) Missed rcu_assign_pointer() in mac80211 scanning, from Johannes
Berg.
2) Allow devices to limit the number of segments that an individual
TCP TSO packet can use at a time, to deal with device and/or driver
specific limitations. From Ben Hutchings.
3) Fix unexpected hard IPSEC expiration after setting the date. From
Fan Du.
4) Memory leak fix in bxn2x driver, from Jesper Juhl.
5) Fix two memory leaks in libertas driver, from Daniel Drake.
6) Fix deref of out-of-range array index in packet scheduler generic
actions layer. From Hiroaki SHIMODA.
7) Fix TX flow control errors in mlx4 driver, from Yevgeny Petrilin.
8) Fix CRIS eth_v10.c driver build, from Randy Dunlap.
9) Fix wrong SKB freeing in LLC protocol layer, from Sorin Dumitru.
10) The IP output path checks neigh lookup errors incorrectly, it needs
to use IS_ERR(). From Vasiliy Kulikov.
11) An estimator leak leads to deref of freed memory in timer handler,
fix from Hiroaki SHIMODA.
12) TCP early demux in ipv6 needs to use DST cookies in order to
validate the RX route properly. Fix from Eric Dumazet.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (43 commits)
net: ipv6: fix TCP early demux
net: Use PTR_RET rather than if(IS_ERR(.. [1]
net_sched: act: Delete estimator in error path.
ip: fix error handling in ip_finish_output2()
llc: free the right skb
ixp4xx_eth: fix ptp_ixp46x build failure
drivers/atm/iphase.c: fix error return code
tcp_output: fix sparse warning for tcp_wfree
drivers/net/phy/mdio-mux-gpio.c: drop devm_kfree of devm_kzalloc'd data
batman-adv: select an internet gateway if none was chosen
mISDN: Bugfix for layer2 fixed TEI mode
igb: don't break user visible strings over multiple lines in igb_ethtool.c
igb: correct hardware type (i210/i211) check in igb_loopback_test()
igb: Fix for failure to init on some 82576 devices.
cris: fix eth_v10.c build error
cdc-ncm: tag Ericsson WWAN devices (eg F5521gw) with FLAG_WWAN
isdnloop: fix and simplify isdnloop_init()
hyperv: Move wait completion msg code into rndis_filter_halt_device()
net/mlx4_core: Remove port type restrictions
net/mlx4_en: Fixing TX queue stop/wake flow
...
Linus Torvalds [Wed, 8 Aug 2012 17:04:40 +0000 (20:04 +0300)]
Merge tag 'pinctrl-fixes-for-v3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl
Pull pinctrl fixes from Linus Walleij:
- Move MXS pinctrl registration to poscore_initcall
- Fix up various devm_* managed resources code paths
- Fix one function group in the Nomadik driver
- Update MAINTAINERS
* tag 'pinctrl-fixes-for-v3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
drivers/pinctrl/pinctrl-nomadik.c: drop devm_kfree of devm_kzalloc'd data
MAINTAINERS: fix a few pinctrl related entries
pinctrl-sirf: remove devm_kfree at error path
pinctrl/nomadik: fix hsi function group list
pinctrl/pinctrl-u300: remove unneeded devm_kfree call
pinctrl: mxs: register driver at postcore_initcall time
Linus Torvalds [Wed, 8 Aug 2012 17:01:45 +0000 (20:01 +0300)]
Merge tag 'iommu-fixes-v3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu
Pull IOMMU fixes from Joerg Roedel:
"These patches fix a couple of issues. First of all a few problems
with ACS on x86 introduced in the last merge window, where ACS did not
work on AMD and a NULL pointer dereference when there ran against
SR-IOV devices.
The patches fallen out of coccinelle checks fix a possible invalid
memory reference and a possible memory leak. The other patches mostly
fix build errors and warnings and a wrong return value."
* tag 'iommu-fixes-v3.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
iommu/amd: Fix ACS path checking
iommu/intel: Fix ACS path checking
iommu/amd: Fix pci_request_acs() call-place
iommu/exynos: Fix build error
iommu/tegra: smmu: Fix error initial value at domain_init
iommu/tegra: smmu: Cleanup with lesser nest
iommu: Add missing forward declaration in include file
iommu: Include linux/types.h
iommu/intel: add missing free_domain_mem
iommu/tegra: remove invalid reference to list iterator variable
Linus Torvalds [Wed, 8 Aug 2012 16:59:52 +0000 (19:59 +0300)]
Merge tag 'sound-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound fixes from Takashi Iwai:
"Containing only a few really small/trivial fixes. The only urgent fix
is a regression fix of HDMI codec probing, introduced in 3.6-rc1. The
rest are HD-audio specific fixes and a copule of minor bug fixes in
PCM core and the old emu10k1."
* tag 'sound-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ALSA: hda - Fix double quirk for Quanta FL1 / Lenovo Ideapad
ALSA: hda - Fix ugly debug prints with CONFIG_SND_VERBOSE_PRINTK=y
ALSA: hda - remove redundant auto quirks for conexant 506x
ALSA: hda - remove quirk for Dell Vostro 1015
ALSA: hda - add dock support for Thinkpad X230
ALSA: hda - Fix regression of HDMI codec probing
ALSA: hda - add dock support for Thinkpad T430s
ALSA: emu10k1: Avoid access to invalid pages when period=1
ALSA: PCM: Fix possible memory leaks in the error path
Heiko Carstens [Wed, 8 Aug 2012 07:32:20 +0000 (09:32 +0200)]
s390/compat: fix mmap compat system calls
The native 31 bit and the compat behaviour for the mmap system calls differ:
In native 31 bit mode the passed in address for the mmap system call will be
unmodified passed to sys_mmap_pgoff().
In compat mode however the passed in address will be modified with
compat_ptr() which masks out the most significant bit.
The result is that in native 31 bit mode each mmap request (with MAP_FIXED)
will fail where the most significat bit is set, while in compat mode it
may succeed.
This odd behaviour was introduced with d3815898 "[S390] mmap: add missing
compat_ptr conversion to both mmap compat syscalls".
To restore a consistent behaviour accross native and compat mode this
patch functionally reverts the above mentioned commit.
Jiri Kosina [Thu, 2 Aug 2012 07:11:23 +0000 (09:11 +0200)]
s390: do not clobber personality flags in sys_32_personality()
There are multiple errors in how sys_32_personality() handles personality
flags stored in top three bytes.
- directly comparing current->personality against PER_LINUX32 doesn't work
in cases when any of the personality flags stored in the top three bytes
are used.
- directly forcefully setting personality to PER_LINUX32 or PER_LINUX
discards any flags stored in the top three bytes
Fix the first one by properly using personality() macro to compare only
PER_MASK bytes.
Fix the second one by setting only the bits that should be set, instead of
overwriting the whole value.
Michael Holzheu [Tue, 31 Jul 2012 09:01:30 +0000 (11:01 +0200)]
s390/sclp_sdias: Add missing break and "fall through"
If the last block of the HSA is read, EVSTATE_ALL_STORED is returned
by SCLP. Because of a missing break in the switch statement two trace
entries are written in this case: "all stored" and "part stored".
This patch adds the missing break and also adds a "fall through"
comment to improve the readability.
Daniel Vetter [Tue, 7 Aug 2012 07:54:14 +0000 (09:54 +0200)]
drm/i915: correctly order the ring init sequence
We may only start to set up the new register values after having
confirmed that the ring is truely off. Otherwise the hw might lose the
newly written register values. This is caught later on in the init
sequence, when we check whether the register writes have stuck.
Arnd Bergmann [Tue, 7 Aug 2012 09:22:14 +0000 (11:22 +0200)]
Merge tag 'imx-fixes' of git://git.pengutronix.de/git/imx/linux-2.6 into fixes
ARM i.MX fixes for 3.6-rc
* tag 'imx-fixes' of git://git.pengutronix.de/git/imx/linux-2.6:
i.MX27: Fix emma-prp and csi clocks.
ARM: clk-imx31: Fix the keypad clock name
ARM: dts: imx27-3ds.dts: Fix serial console node
Oliver Hartkopp [Mon, 6 Aug 2012 16:02:29 +0000 (18:02 +0200)]
canfd: remove redundant CAN FD flag
The first idea of the CAN FD implementation started with a new struct
canfd_frame to be used for both CAN FD frames and legacy CAN frames.
The now mainlined implementation supports both CAN frame types simultaneously
and distinguishes them only by their required sizes: CAN_MTU and CANFD_MTU.
Only the struct canfd_frame contains a flags element which is needed for the
additional CAN FD information. As CAN FD implicitly means that the 'Extened
Data Length' mode is enabled the formerly defined CANFD_EDL bit became
redundant and also confusing as an unset bit would be an error and would
always need to be tested.
This patch removes the obsolete CANFD_EDL bit and clarifies the documentation
for the use of struct canfd_frame and the CAN FD relevant flags.
Stefan Assmann [Tue, 7 Aug 2012 07:45:57 +0000 (00:45 -0700)]
igb: add delay to allow igb loopback test to succeed on 8086:10c9
Some 8086:10c9 NICs have a problem completing the ethtool loopback test.
The result looks like this:
ethtool -t eth1
The test result is FAIL
The test extra info:
Register test (offline) 0
Eeprom test (offline) 0
Interrupt test (offline) 0
Loopback test (offline) 13
Link test (on/offline) 0
A bisect clearly points to commit a95a07445ee97a2fef65befafbadcc30ca1bd145.
However that seems to only trigger the bug. While adding some printk the
problem disappeared, so this might be a timing issue. After some trial and
error I discovered that adding a small delay just before igb_write_phy_reg()
in igb_integrated_phy_loopback() allows the loopback test to succeed.
I was unable to figure out the root cause so far but I expect it to be
somewhere in the following executing path
igb_integrated_phy_loopback
->igb_write_phy_reg_igp
->igb_write_phy_reg_mdic
->igb_acquire_phy_82575
->igb_acquire_swfw_sync_82575
The problem could only be observed on 8086:10c9 NICs so far and not all
of them show the behaviour. I did not restrict the workaround to this
type of NIC as it should do no harm to other igb NICs.
With the patch below the loopback test succeeded 500 times in a row
using a NIC that would otherwise fail.
Tushar Dave [Wed, 1 Aug 2012 02:11:15 +0000 (02:11 +0000)]
e1000e: 82571 Tx Data Corruption during Tx hang recovery
A bus trace shows that while executing e1000e_down, TCTL is cleared except
for the PSP bit. This occurs while in the middle of fetching a TSO packet
since the Tx packet buffer is full at that point. Before the device is
reset, the e1000_watchdog_task starts to run from the middle (it was
apparently pre-empted earlier, although that is not in the trace) and sets
TCTL.EN. At that point, 82571 transmits the corrupted packet, apparently
because TCTL.MULR was cleared in the middle of fetching a packet, which is
forbidden.
Driver should just clear TCTL.EN in e1000_reset_hw_82571 instead of
clearing the entire register, so as not to change any settings in the
middle of fetching a packet.
Tushar Dave [Tue, 31 Jul 2012 02:02:43 +0000 (02:02 +0000)]
e1000e: NIC goes up and immediately goes down
Found that commit d478eb44 was a bad commit.
If the link partner is transmitting codeword (even if NULL codeword),
then the RXCW.C bit will be set so check for RXCW.CW is unnecessary.
Ref: RH BZ 840642
Shawn Guo [Sun, 5 Aug 2012 06:01:28 +0000 (14:01 +0800)]
ARM: dts: imx: add alias for gpio
Add alias for gpio nodes, so that gpio driver can identify the port
number and then specify a sensible gpio base rather than using the
one dynamically allocated by gpio core.
Shawn Guo [Sun, 5 Aug 2012 06:01:26 +0000 (14:01 +0800)]
gpio/mxc: specify gpio base for device tree probe
Currently, unlike the non-DT probe where the gpio base is specified
with pdev->id, the DT probe uses the base dynamically allocated by
gpio core, which uses a completely different numbering scheme. This
causes two issues to user space applications which access sysfs entry
/sys/class/gpio/gpioN.
* It breaks the compatibility with user space applications between
non-DT and DT kernels.
* It's not intuitive and sometimes hard for users to map the Linux
gpio number to the actual hardware pin.
Use alias to identify the gpio port/bank, and then the gpio base
can be specified with port id to solve above issues. If alias is not
defined in device tree, the base number dynamically allocated by gpio
core will be used.
With commit 3e11f7b (gpio/generic: initialize basic_mmio_gpio shadow
variables properly) in place, the shadow variables initialization is
being done in generic driver bgpio_init call.
Remove the redundant shadow variables initialization from gpio-mxc
driver.
Hiroaki SHIMODA [Mon, 6 Aug 2012 05:45:48 +0000 (05:45 +0000)]
net_sched: act: Delete estimator in error path.
Some action modules free struct tcf_common in their error path
while estimator is still active. This results in est_timer()
dereference freed memory.
Add gen_kill_estimator() in ipt, pedit and simple action.
Vasiliy Kulikov [Mon, 6 Aug 2012 03:55:29 +0000 (03:55 +0000)]
ip: fix error handling in ip_finish_output2()
__neigh_create() returns either a pointer to struct neighbour or PTR_ERR().
But the caller expects it to return either a pointer or NULL. Replace
the NULL check with IS_ERR() check.
Karsten Keil [Sat, 4 Aug 2012 00:14:25 +0000 (00:14 +0000)]
mISDN: Bugfix for layer2 fixed TEI mode
If a fixed TEI is used, the initial state of the layer 2 statmachine need to be
4 (TEI assigned). This was true only for Point to Point connections, but not
for the other fixed TEIs. It was not found before, because usually only the
TEI 0 is used as fixed TEI for PtP mode, but if you try X31 packet mode
connections with SAPI 16, TEI 1, it did fail.
Johannes Berg [Sun, 5 Aug 2012 16:31:46 +0000 (18:31 +0200)]
iwlwifi: disable greenfield transmissions as a workaround
There's a bug that causes the rate scaling to get stuck
when it has to use single-stream rates with a peer that
can do GF and SGI; the two are incompatible so we can't
use them together, but that causes the algorithm to not
work at all, it always rejects updates.
Disable greenfield for now to prevent that problem.
Daniel Drake [Thu, 2 Aug 2012 17:41:48 +0000 (18:41 +0100)]
cfg80211: process pending events when unregistering net device
libertas currently calls cfg80211_disconnected() when it is being
brought down. This causes an event to be allocated, but since the
wdev is already removed from the rdev by the time that the event
processing work executes, the event is never processed or freed.
http://article.gmane.org/gmane.linux.kernel.wireless.general/95666
Fix this leak, and other possible situations, by processing the event
queue when a device is being unregistered. Thanks to Johannes Berg for
the suggestion.
If the device was not found in a list of found devices names of which
are pending.This may happen in a case when HCI Remote Name Request
was sent as a part of incoming connection establishment procedure.
Hence there is no need to continue resolving a next name as it will
be done upon receiving another Remote Name Request Complete Event.
This will fix a kernel crash when trying to use this entry to resolve
the next name.
projects / linux.git / log