Roland Dreier [Tue, 12 Jun 2012 01:23:16 +0000 (18:23 -0700)]
qla2xxx: Don't leak commands we give up on in qlt_do_work()
If we go to the "out_term:" exit path in qlt_do_work(), we call
qlt_send_term_exchange() with a NULL cmd, which means that it can't
possibly free the cmd for us. Add an explicit call to free the
command memory, so we don't leak the allocation.
This will also fix warnings about "BUG qla_tgt_cmd_cachep: Objects
remaining on kmem_cache_close" from slub when unloading the qla2xxx
target module.
Roland Dreier [Tue, 12 Jun 2012 01:23:15 +0000 (18:23 -0700)]
qla2xxx: Don't crash if we can't find cmd for failed CTIO
In qlt_do_ctio_completion(), there's no point in calling
qlt_term_ctio_exchange() with a NULL cmd -- all that it does is crash
in a NULL pointer dereference, since it does
qlt_send_term_exchange(vha, cmd, &cmd->atio, 1);
and dereferencing &cmd->atio is a bad idea if cmd itself is NULL.
If we really need to do this, we could take the values from the
failed CTIO we're processing, but it's not clear if it's worth
the replumbing to do that.
Roland Dreier [Tue, 5 Jun 2012 06:37:34 +0000 (23:37 -0700)]
tcm_qla2xxx: Don't insert nacls without sessions into the btree
When we create an explicit node ACL in tcm_qla2xxx_make_nodeacl(),
there is a call to tcm_qla2xxx_setup_nacl_from_rport(), which puts the
node ACL into the lport_fcport_map even though there is no session yet
for the initiator. Since the only time we remove entries from this
map is when we free a session, this means that if we later delete this
node ACL without the initiator ever creating a session, we'll leave
the nacl pointer in the btree pointing at freed memory.
This is especially bad if that initiator later does send us a command
that would cause us to create a dynamic ACL and session: we'll find
the stale freed nacl pointer in the btree and end up with use-after-free.
We could add more code to clear the btree entry when deleting the
explicit nacl, but the original insertion is pointless: without a
session attached, we'll just have to update the entry when a session
appears anyway. So we can just delete tcm_qla2xxx_setup_nacl_from_rport()
and the code that calls it.
Roland Dreier [Tue, 5 Jun 2012 06:24:51 +0000 (23:24 -0700)]
target: Return error to initiator if SET TARGET PORT GROUPS emulation fails
The error paths in target_emulate_set_target_port_groups() are all
essentially "rc = -EINVAL; goto out;" but the code at "out:" ignores
rc and always returns success. This means that even if eg explicit
ALUA is turned off, the initiator will always see a good SCSI status
for SET TARGET PORT GROUPS.
Fix this by returning rc as is intended. It appears this bug was
added by the following patch:
tcm_qla2xxx: Clear session s_id + loop_id earlier during shutdown
This patch adds a new tcm_qla2xxx_clear_sess_lookup() call to clear session
specific s_id + loop_id entries used for se_node_acl pointer lookup ahead
of releasing se_session within the process context workqueue callback in
tcm_qla2xxx_free_session().
It makes the call in existing tcm_qla2xxx_clear_nacl_from_fcport_map()
code invoked from qlt_unreg_sess() in interrupt context w/ hardware_lock
held, ahead of the process context callback into qlt_free_session_done()
-> tcm_qla2xxx_free_session().
We are doing this to address a race between incoming ATIO or TMR packets
using stale se_node_acl pointer once session shutdown has been invoked via
qlt_unreg_sess() in qla_target.c LLD code, and when the entire tcm_qla2xxx
endpoint has not been forced into shutdown w/ echo 0 > ../$QLA2XXX_PORT/enable
CROSS_COMPILE must be setup before using e.g. cc-option (and a few other
as-*, cc-*, ld-* macros), else they will check against the wrong compiler
when cross-compiling, and may invoke the cross compiler with wrong or
suboptimal compiler options.
Joern Engel [Fri, 18 May 2012 20:58:23 +0000 (13:58 -0700)]
tcm_qla2xxx: Convert to TFO->put_session() usage
This patch converts tcm_qla2xxx code to use an internal kref_put() for
se_session->sess_kref in order to ensure that qla_hw_data->hardware_lock
can be held while calling qlt_unreg_sess() for the final put.
Paul Mundt [Fri, 25 May 2012 04:02:48 +0000 (13:02 +0900)]
sh: switch to generic strncpy_from_user().
This kills off the special sh32/64 versions and adopts the generic
version. It should be possible to optimize this for SH-4A unaligned
loads, but this is a corner case that can be supported incrementally.
Paul Bolle [Fri, 8 Jun 2012 20:49:17 +0000 (22:49 +0200)]
sh: Kill off last dead UBC header
Commit 7025bec9125b0a02edcaf22c2dce753bf2c95480 ("sh: Kill off dead UBC
headers.") skipped arch/sh/include/cpu-sh2a/cpu/ubc.h. Since nothing is
using that header either, kill it off too.
Laurent Pinchart [Tue, 12 Jun 2012 22:28:24 +0000 (00:28 +0200)]
serial: sh-sci: Make probe fail for ports that exceed the maximum count
The driver supports a maximum number of ports configurable at compile
time. Make sure the probe() method fails when registering a port that
exceeds the maximum instead of returning success without registering the
port.
This fixes a crash at system suspend time, when the driver tried to
suspend a non-registered port using the UART core.
Tony Zelenoff [Tue, 5 Jun 2012 13:58:04 +0000 (17:58 +0400)]
USB: mos7840: Fix compilation of usb serial driver
The __devinitconst section can't be referenced
from usb_serial_device structure. Thus removed it as
it done in other mos* device drivers.
Error itself:
WARNING: drivers/usb/serial/mos7840.o(.data+0x8): Section mismatch in reference
from the variable moschip7840_4port_device to the variable
.devinit.rodata:id_table
The variable moschip7840_4port_device references
the variable __devinitconst id_table
Johan Hovold [Tue, 29 May 2012 16:22:48 +0000 (18:22 +0200)]
USB: option: fix memory leak
Fix memory leak introduced by commit 383cedc3bb435de7a2 ("USB: serial:
full autosuspend support for the option driver") which allocates
usb-serial data but never frees it.
Johan Hovold [Tue, 29 May 2012 15:57:52 +0000 (17:57 +0200)]
USB: option: fix port-data abuse
Commit 8b4c6a3ab596961b78465 ("USB: option: Use generic USB wwan code")
moved option port-data allocation to usb_wwan_startup but still cast the
port data to the old struct...
Alan Cox [Tue, 22 May 2012 19:45:13 +0000 (20:45 +0100)]
USB: mct_u232: Fix incorrect TIOCMSET return
The low level helper returns 1 on success. The ioctl should however return
0. As this is the only user of the helper return, make the helper return 0 or
an error code.
Andrew Bird [Mon, 28 May 2012 11:43:06 +0000 (12:43 +0100)]
USB: option: Updated Huawei K4605 has better id
Later firmwares for this device now have proper subclass and
protocol info so we can identify it nicely without needing to use
the blacklist. I'm not removing the old 0xff matching as there
may be devices in the field that still need that.
Julia Lawall [Tue, 12 Jun 2012 14:17:25 +0000 (10:17 -0400)]
drivers/ide/ide-cs.c: adjust suspicious bit operation
IO_DATA_PATH_WIDTH_8 is 0, so a bit-and with it is always false. The
value IO_DATA_PATH_WIDTH covers the bits of the IO_DATA_PATH constants, so
first pick those bits and then make the test using !=.
This problem was found using Coccinelle (http://coccinelle.lip6.fr/).
Eric Dumazet [Tue, 12 Jun 2012 06:03:51 +0000 (06:03 +0000)]
bonding: Fix corrupted queue_mapping
In the transmit path of the bonding driver, skb->cb is used to
stash the skb->queue_mapping so that the bonding device can set its
own queue mapping. This value becomes corrupted since the skb->cb is
also used in __dev_xmit_skb.
When transmitting through bonding driver, bond_select_queue is
called from dev_queue_xmit. In bond_select_queue the original
skb->queue_mapping is copied into skb->cb (via bond_queue_mapping)
and skb->queue_mapping is overwritten with the bond driver queue.
Subsequently in dev_queue_xmit, __dev_xmit_skb is called which writes
the packet length into skb->cb, thereby overwriting the stashed
queue mappping. In bond_dev_queue_xmit (called from hard_start_xmit),
the queue mapping for the skb is set to the stashed value which is now
the skb length and hence is an invalid queue for the slave device.
If we want to save skb->queue_mapping into skb->cb[], best place is to
add a field in struct qdisc_skb_cb, to make sure it wont conflict with
other layers (eg : Qdiscc, Infiniband...)
This patchs also makes sure (struct qdisc_skb_cb)->data is aligned on 8
bytes :
netem qdisc for example assumes it can store an u64 in it, without
misalignment penalty.
Note : we only have 20 bytes left in (struct qdisc_skb_cb)->data[].
The largest user is CHOKe and it fills it.
Weiping Pan [Sun, 10 Jun 2012 23:00:20 +0000 (23:00 +0000)]
bonding:record primary when modify it via sysfs
If we modify primary via sysfs and it is not a valid slave,
we should record it for future use, and this behavior is the same with
bond_check_params().
Pin 0x1b was connected to the front panel connector, which according to
the HDA standard should contain a mic and a headphone. In this case,
the headphone was listed as "line out" by BIOS.
Linus Torvalds [Tue, 12 Jun 2012 15:48:17 +0000 (18:48 +0300)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Pull drm fixes from Dave Airlie:
"As per your -rc2 announce, this is small and urgent only,
The radeon one is for a regression in 3.4 so we need this one in your
tree so we can send the stable one out, code in 3.4 broke some old
userspaces. The max props increase fixes spew being seen on a few
machines. And a ttm regression to fix some accounting issues that
affect vmwgfx."
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/ttm: Fix buffer object metadata accounting regression v2
drm: increase DRM_OBJECT_MAX_PROPERTY to 24
drm/radeon: fix tiling and command stream checking on evergreen v3
A regression was introduced in the 3.3 rc series, commit
"drm/ttm: simplify memory accounting for ttm user v2",
causing the metadata of buffer objects created using the ttm_bo_create()
function to be accounted twice.
That causes massive leaks with the vmwgfx driver running for example
SpecViewperf Catia-03 test 2, eventually killing the app.
Furthermore, the same commit introduces a regression where
metadata accounting is leaked if a buffer object is
initialized with an illegal size. This is also fixed with this commit.
v2: Fixed an error path and removed an unused variable.
Paulo Zanoni [Tue, 12 Jun 2012 14:27:01 +0000 (11:27 -0300)]
drm: increase DRM_OBJECT_MAX_PROPERTY to 24
Before Kernel 3.5, no one was checking for the return value of
drm_connector_attach_property, so we never noticed that we were unable
to create some properties. Commit "drm: WARN() when
drm_connector_attach_property fails" added a WARN when we fail to
create a property, and the transition from "connector properties" to
"object properties" changed the warning message a little bit.
On i915 machines with many TV connectors we hit the maximum number of
properties (since each TV connector uses a lot of properties), so we
get a few backtraces in our logs. This commit increases the maximum
number of properties to 24 hoping we'll have enough room for
everybody.
Chris suggested that we convert this code to "lists", but I believe
this conversion can come after we make sure people's dmesgs are not
spammed by our driver.
Dan Carpenter [Tue, 12 Jun 2012 07:37:39 +0000 (10:37 +0300)]
NFS: add an endian notation for sparse
This is supposed to be a __be32 value. Sparse complains a lot:
fs/nfs/callback_xdr.c:699:30: warning: incorrect type in initializer (different base types)
fs/nfs/callback_xdr.c:699:30: expected unsigned int [unsigned] status
fs/nfs/callback_xdr.c:699:30: got restricted __be32 const [usertype] csr_status
fs/nfs/callback_xdr.c:715:9: warning: cast to restricted __be32
fs/nfs/callback_xdr.c:716:16: warning: incorrect type in return expression (different base types)
fs/nfs/callback_xdr.c:716:16: expected restricted __be32
fs/nfs/callback_xdr.c:716:16: got unsigned int [unsigned] status
Dong Aisheng [Fri, 8 Jun 2012 13:33:12 +0000 (21:33 +0800)]
pinctrl: pinctrl-imx: fix incorrect debug message of maps
After create config map, the new_map pointer becomes point to
PIN_MAP_TYPE_CONFIGS_PIN map rather than PIN_MAP_TYPE_MUX_GROUP map any more.
Thus using new_map pointer to display the MUX_GROUP info is not correct.
Using map pointer instead to show the correct MUX_GROUP map info.
Original the debug message is:
imx6q-pinctrl 20e0000.iomuxc: maps: function Yp group MX6Q_PAD_SD3_CMD num 12
After fix it is:
imx6q-pinctrl 20e0000.iomuxc: maps: function usdhc3 group usdhc3grp-1 num 11
Devendra Naga [Fri, 8 Jun 2012 19:22:11 +0000 (00:52 +0530)]
pinctrl: pinctrl-imx: free allocated pinctrl_map structure only once and use kernel facilities for IMX_PMX_DUMP
a) as we allocate the pinctrl_map structure at imx_dt_node_to_map at line 167, anyway
if its an element, or a num_elements * (sizeof(type)) elements allocated to one single
pointer must be freed only once.
CASE. A)
as new_map is not moved and allocated like,
for (i = 0; i < MAX_ELEMS; i++) {
new_map[i] = kmalloc(numelems * size, GFP_KERNEL);
}
its freed as
for (i = 0; i < MAX_ELEMS; i++) {
kfree(new_map[i]);
}
CASE. B)
and its allocated like
new_map = kmalloc(numelems * size, GFP_KERNEL);
it just needs kfree not as case A's.
b) use KERN_DEBUG facility for the IMX_PMX_DUMP macro.
Linus Walleij [Tue, 12 Jun 2012 10:43:06 +0000 (12:43 +0200)]
pinctrl: nomadik: fix up typo
Commit a60b57eddaa8af6c02cf7bbeb58ebf82881f08ac
"drivers/gpio: gpio-nomadik: Add support for irqdomains"
changed GPIO offset calculations to have this form:
(gpio % NMK_GPIO_PER_CHIP) except in this one place for
setting sleep mode, where the conversion was all wrong, and
instead mod:ing the GPIO with the IRQ base which does not
make any sense.
Linus Torvalds [Tue, 12 Jun 2012 04:39:58 +0000 (07:39 +0300)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Pull m68knommu from Greg Ungerer:
"This contains five fixes. Four fix build problems introduced by
recent clean up and merging of the m68k timer and ptrace code. The
other fixes the 528x ColdFire CPU QSPI base address definition, missed
in the ColdFire QSPI cleanup."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
m68k: make syscall_trace_enter/leave exist for non-MMU classic m68k types
m68knommu: fix 68360 local setting of timer interrupt handler
m68knommu: fix 68328 local setting of timer interrupt handler
m68k: fix inclusion of arch_gettimeoffset for non-MMU 68k classic CPU types
m68knommu: m528x qspi definition fix
Axel Lin [Fri, 8 Jun 2012 23:41:42 +0000 (07:41 +0800)]
leds: Make LEDS_ASIC3 and LEDS_RENESAS_TPU depend on LEDS_CLASS=y
Otherwise, I got below build error when CONFIG_LEDS_CLASS=m.
LD init/built-in.o
drivers/built-in.o: In function `asic3_led_probe':
clkdev.c:(.devinit.text+0x4680): undefined reference to `led_classdev_register'
drivers/built-in.o: In function `r_tpu_probe':
clkdev.c:(.devinit.text+0x4838): undefined reference to `led_classdev_register'
drivers/built-in.o: In function `asic3_led_remove':
clkdev.c:(.devexit.text+0x564): undefined reference to `led_classdev_unregister'
drivers/built-in.o: In function `r_tpu_remove':
clkdev.c:(.devexit.text+0x5a0): undefined reference to `led_classdev_unregister'
make: *** [vmlinux] Error 1
Rafal Prylowski [Mon, 28 May 2012 15:35:54 +0000 (23:35 +0800)]
leds: don't disable blinking when writing the same value to delay_on or delay_off
Function led_set_software_blink() assumes that blink timer is still running,
but commit 488bc35bf40df89d37486c1826b178a2fba36ce7 introduced disabling
of blink timer before each call to led_set_software_blink().
Correct led_software_blink():
1) remove protection against reprogramming blink timer to the same values,
because it only disables blinking now,
2) remove unnecessary call to led_stop_software_blink().
Greg Ungerer [Wed, 23 May 2012 03:43:52 +0000 (13:43 +1000)]
m68k: make syscall_trace_enter/leave exist for non-MMU classic m68k types
The assembler entry code calls directly to the syscall_trace_enter() and
syscall_trace_leave() functions. But currently they are conditionaly
compiled out for the non-MMU classic m68k CPU types (so 68328 for example),
resulting in a link error:
LD vmlinux
arch/m68k/platform/68328/built-in.o: In function `do_trace':
(.text+0x1c): undefined reference to `syscall_trace_enter'
arch/m68k/platform/68328/built-in.o: In function `do_trace':
(.text+0x4c): undefined reference to `syscall_trace_leave'
Change the conditional check that includes these functions to be true for
the !defined(CONFIG_MMU) case as well.
Greg Ungerer [Wed, 23 May 2012 03:32:55 +0000 (13:32 +1000)]
m68knommu: fix 68360 local setting of timer interrupt handler
Compiling for 68360 based targets fails with:
arch/m68k/platform/68360/config.c: In function ‘hw_tick’:
arch/m68k/platform/68360/config.c:55:2: error: implicit declaration of function ‘arch_timer_interrupt’
arch/m68k/platform/68360/config.c: At top level:
arch/m68k/platform/68360/config.c:64:6: error: conflicting types for ‘hw_timer_init’
arch/m68k/include/asm/machdep.h:36:13: note: previous declaration of ‘hw_timer_init’ was here
Changes made to hw_timer_init() didn't get updated in the 68328 timer code.
So process and call the "handler" arg that is now passed into that
hw_timer_init() function.
Greg Ungerer [Wed, 23 May 2012 03:27:18 +0000 (13:27 +1000)]
m68knommu: fix 68328 local setting of timer interrupt handler
Compiling for 68328 based targets fails with:
arch/m68k/platform/68328/timers.c: In function ‘hw_tick’:
arch/m68k/platform/68328/timers.c:65:2: error: implicit declaration of function ‘arch_timer_interrupt’
arch/m68k/platform/68328/timers.c: At top level:
arch/m68k/platform/68328/timers.c:102:6: error: conflicting types for ‘hw_timer_init’
arch/m68k/include/asm/machdep.h:36:13: note: previous declaration of ‘hw_timer_init’ was here
Changes made to hw_timer_init() didn't get updated in the 68328 timer code.
So process and call the "handler" arg that is now passed into that
hw_timer_init() function.
Greg Ungerer [Wed, 23 May 2012 01:32:45 +0000 (11:32 +1000)]
m68k: fix inclusion of arch_gettimeoffset for non-MMU 68k classic CPU types
When building for non-MMU based classic 68k CPU types (like the 68328 for
example) you get a compilation error:
CC arch/m68k/kernel/time.o
arch/m68k/kernel/time.c:91:5: error: redefinition of ‘arch_gettimeoffset’
include/linux/time.h:145:19: note: previous definition of ‘arch_gettimeoffset’ was here
The arch_gettimeoffset() code is included when building for these CPU types,
but it shouldn't be. Those machine types do not have
CONFIG_ARCH_USES_GETTIMEOFFSET set.
The fix is simply to conditionally include the arch_gettimeoffset() code on
that same config setting that specifies its use or not.
Paul Bolle [Sat, 9 Jun 2012 22:51:47 +0000 (22:51 +0000)]
sparc: remove two unused headers
Nothing includes these two headers. None of the macros they define are
used anywhere in the tree. This was also the case in v2.6.12-rc2 and,
presumably, every release in between. These two headers can safely be
removed.
Chris Metcalf [Thu, 7 Jun 2012 10:45:02 +0000 (10:45 +0000)]
tilegx network driver: initial support
This change adds support for the tilegx network driver based on the
GXIO IORPC support in the tilegx software stack, using the on-chip
mPIPE packet processing engine.
Stefan Roese [Thu, 7 Jun 2012 23:21:05 +0000 (23:21 +0000)]
net: stmmac: Fix clock en-/disable calls
clk_{un}prepare is mandatory for platforms using common clock framework.
Since these drivers are used by SPEAr platform, which supports common
clock framework, add clk_{un}prepare() support for them. Otherwise
the clocks are not correctly en-/disabled and ethernet support doesn't
work.
Jeff Layton [Mon, 11 Jun 2012 14:03:42 +0000 (10:03 -0400)]
rpc_pipefs: allow rpc_purge_list to take a NULL waitq pointer
In the event that we don't have a dentry for a rpc_pipefs pipe, we still
need to allow the queue_timeout job to clean out the queue. There's just
no waitq to wake up in that event.
Andy Adamson [Wed, 30 May 2012 20:12:24 +0000 (16:12 -0400)]
NFSv4 do not send an empty SETATTR compound
Commit 536e43d12b9517bbbf6114cd1a12be27857a4d7a ATTR_OPEN check can result in
an ia_valid with only ATTR_FILE set, and no NFS_VALID_ATTRS attributes to
request from the server.
Sachin Prabhu [Thu, 31 May 2012 16:45:00 +0000 (17:45 +0100)]
NFSv2: EOF incorrectly set on short read
In cases where the server returns fewer bytes then those requested, we
can incorrectly set the eof flag for the file. Fixing this allows the
request to be retried with updated offset and count arguments.
ARM: mm: fix type of the arm_dma_limit global variable
arm_dma_limit stores physical address of maximal address accessible by DMA,
so the phys_addr_t type makes much more sense for it instead of u32. This
patch fixes the following build warning:
arch/arm/mm/init.c:380: warning: comparison of distinct pointer types lacks a cast
Sachin Kamat [Wed, 6 Jun 2012 06:03:35 +0000 (08:03 +0200)]
ARM: dma-mapping: Add missing static storage class specifier
Fixes the following sparse warnings:
arch/arm/mm/dma-mapping.c:231:15: warning: symbol 'consistent_base' was not
declared. Should it be static?
arch/arm/mm/dma-mapping.c:326:8: warning: symbol 'coherent_pool_size' was not
declared. Should it be static?
Wanpeng Li [Sun, 10 Jun 2012 02:50:52 +0000 (10:50 +0800)]
x86/mm: Fix some kernel-doc warnings
Fix kernel-doc warnings in arch/x86/mm/ioremap.c and
arch/x86/mm/pageattr.c, just like this one:
Warning(arch/x86/mm/ioremap.c:204):
No description found for parameter 'phys_addr'
Warning(arch/x86/mm/ioremap.c:204):
Excess function parameter 'offset' description in 'ioremap_nocache'
Ingo Molnar [Mon, 11 Jun 2012 08:30:23 +0000 (10:30 +0200)]
Merge branch 'rcu/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu into core/urgent
Merge RCU fixes from Paul E. McKenney:
" This series has four patches, the major point of which is to eliminate
some slowdowns (including boot-time slowdowns) resulting from some
RCU_FAST_NO_HZ changes. The issue with the changes is that posting timers
from the idle loop has no effect if the CPU has entered dyntick-idle
mode because the CPU has already computed its wakeup time, and posting
a timer does not cause it to be recomputed. The short-term fix is for
RCU to precompute the timeout value so that the CPU's calculation is
correct. "
Paul Mundt [Mon, 11 Jun 2012 08:10:16 +0000 (17:10 +0900)]
clocksource: sh_tmu: Use clockevents_config_and_register().
This switches over to the now exported clockevents_config() and
clockevents_config_and_register() helpers. This knocks off a
long-standing TMU TODO item.
Roland Dreier [Sun, 10 Jun 2012 20:05:24 +0000 (20:05 +0000)]
net: Reorder initialization in ip_route_output to fix gcc warning
If I build with W=1, for every file that includes <net/route.h>, I get the warning
include/net/route.h: In function 'ip_route_output':
include/net/route.h:135:3: warning: initialized field overwritten [-Woverride-init]
include/net/route.h:135:3: warning: (near initialization for 'fl4') [-Woverride-init]
(This is with "gcc (Debian 4.6.3-1) 4.6.3")
A fix seems pretty trivial: move the initialization of .flowi4_tos
earlier. As far as I can tell, this has no effect on code generation.
Paul Mundt [Mon, 11 Jun 2012 05:29:58 +0000 (14:29 +0900)]
bug.h: need linux/kernel.h for TAINT_WARN.
asm-generic/bug.h uses taint flags that are only defined in
linux/kernel.h, resulting in build failures on platforms that
don't include linux/kernel.h some other way:
arch/sh/include/asm/thread_info.h:172:2: error: 'TAINT_WARN' undeclared (first use in this function)
Caused by commit edd63a2763bd ("set_restore_sigmask() is never called
without SIGPENDING (and never should be)").
Linus Torvalds [Mon, 11 Jun 2012 03:57:43 +0000 (06:57 +0300)]
Merge tag 'regmap-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap
Pull regmap fixes from Mark Brown:
"Nothing too exciting - a cleanup for debugfs in error handling and a
fix for the padding (which has only just acquired real use) and
exporting a function that's supposed to be usable by drivers."
* tag 'regmap-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: Export regmap_reinit_cache()
regmap: Fix the size calculation for map->format.buf_size
regmap: clean up debugfs if regmap_init fails
Linus Torvalds [Mon, 11 Jun 2012 03:53:48 +0000 (06:53 +0300)]
Merge tag 'regulator-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator
Pull regulator fixes from Mark Brown:
"A couple of small fixes, plus larger fixes for the gpio-regulator
driver the most recent changes for which had apparently not been
tested at all in -next (or elsewhere from the looks of it)."
* tag 'regulator-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
regulator: core: Properly handle the case min_uV < rdev->desc->min_uV in map_voltage_linear
regulator: max8649: fix missing regmap in rdev
regulator: gpio-regulator: populate selector from set_voltage
regulator: gpio-regulator: Fix finding of smallest value
regulator: gpio-regulator: do not pass drvdata pointer as reference
regulator: anatop: Use correct __devexit_p annotation
regulator: palmas: Fix wrong kfree calls
Eric Dumazet [Tue, 5 Jun 2012 22:35:24 +0000 (22:35 +0000)]
virtio-net: fix a race on 32bit arches
commit 3fa2a1df909 (virtio-net: per cpu 64 bit stats (v2)) added a race
on 32bit arches.
We must use separate syncp for rx and tx path as they can be run at the
same time on different cpus. Thus one sequence increment can be lost and
readers spin forever.
françois romieu [Sat, 9 Jun 2012 10:53:16 +0000 (10:53 +0000)]
r8169: avoid NAPI scheduling delay.
While reworking the r8169 driver a few months ago to perform the
smallest amount of work in the irq handler, I took care of avoiding
any irq mask register operation in the slow work dedicated user
context thread. The slow work thread scheduled an extra round of NAPI
work which would ultimately set the irq mask register as required,
thus keeping such irq mask operations in the NAPI handler.
It would eventually race with the irq handler and delay NAPI execution
for - assuming no further irq - a whole ksoftirqd period. Mildly a
problem for rare link changes or corner case PCI events.
The race was always lost after the last bh disabling lock had been
removed from the work thread and people started wondering where those
pesky "NOHZ: local_softirq_pending 08" messages came from.
Actually the irq mask register _can_ be set up directly in the slow
work thread.
Jerome Glisse [Sat, 9 Jun 2012 14:57:41 +0000 (10:57 -0400)]
drm/radeon: fix tiling and command stream checking on evergreen v3
Fix regresson since the introduction of command stream checking on
evergreen (thread referenced below). Issue is cause by ddx allocating
bo with formula width*height*bpp while programming the GPU command
stream with ALIGN(height, 8). In some case (where page alignment does
not hide the extra size bo should be according to height alignment)
the kernel will reject the command stream.
This patch reprogram the command stream to slice - 1 (slice is
a derivative value from height) which avoid rejecting the command
stream while keeping the value of command stream checking from a
security point of view.
This patch also fix wrong computation of layer size for 2D tiled
surface. Which should fix issue when 2D color tiling is enabled.
This dump the radeon KMS_DRIVER_MINOR so userspace can know if
they are on a fixed kernel or not.
Paul Pluzhnikov [Sat, 9 Jun 2012 14:53:03 +0000 (07:53 -0700)]
net: Make linux/tcp.h C++ friendly (trivial)
I originally sent this patch to <[email protected]>, but Jiri Kosina did
not feel that this is fully appropriate for the trivial tree.
Using linux/tcp.h from C++ results in:
cat t.cc
#include <linux/tcp.h>
int main() { }
g++ -c t.cc
In file included from t.cc:1:
/usr/include/linux/tcp.h:72: error: '__u32 __fswab32(__u32)' cannot appear in a constant-expression
/usr/include/linux/tcp.h:72: error: a function call cannot appear in a constant-expression
...
Attached trivial patch fixes this problem.
Tested:
- the t.cc above compiles with g++ and
- the following program generates the same output before/after
the patch:
#include <linux/tcp.h>
#include <stdio.h>
int main ()
{
#define P(a) printf("%s: %08x\n", #a, (int)a)
P(TCP_FLAG_CWR);
P(TCP_FLAG_ECE);
P(TCP_FLAG_URG);
P(TCP_FLAG_ACK);
P(TCP_FLAG_PSH);
P(TCP_FLAG_RST);
P(TCP_FLAG_SYN);
P(TCP_FLAG_FIN);
P(TCP_RESERVED_BITS);
P(TCP_DATA_OFFSET);
#undef P
return 0;
}
Martin Pelikan [Sat, 9 Jun 2012 19:22:11 +0000 (21:22 +0200)]
x86, um: Correct syscall table type attributes breaking gcc 4.8
The latest GCC 4.8 does some more checking on type attributes that
break the build for ARCH=um -> fill them in. Specifically, the
"asmlinkage" attributes is now tested for consistency.
Bryan Schumaker [Fri, 8 Jun 2012 15:32:56 +0000 (11:32 -0400)]
NFS: Use the NFS_DEFAULT_VERSION for v2 and v3 mounts
Older versions of nfs utils don't always pass a "vers=" mount option for
NFS. This chould lead to attempts at using NFS v0 due to a zeroed out
nfs_parsed_mount_data struct. I solve this by setting the default NFS
version to NFS_DEFAULT_VERSION in the v2 and v3 cases (v4 has already been
taken care of by a similar patch).
Fred Isaman [Fri, 8 Jun 2012 20:48:33 +0000 (16:48 -0400)]
NFS: fix directio refcount bug on commit
This reverts a hunk from commit 04277086577
"NFS: Clean up - Simplify reference counting in fs/nfs/direct.c"
The cleanups in that patch affect the write path, but by the time
processing hits commit the removed reference has been added back by
nfs_scan_commit_list(). Without this reversion, any page that is
sent to commit holds on to an unbalanced reference that is never
freed. The immediate effect is an imbalance over the wire between
OPENs and CLOSEs.
Arnd Bergmann [Tue, 15 May 2012 21:02:59 +0000 (15:02 -0600)]
video/console: automatically select a font
The frame buffer console needs at least one font to be built into
the kernel, so add the necessary Kconfig magic to guarantee that
one of the available font is always on. If a user accidentally
disables all fonts manually, the 8x16 font will be selected
anyway.
Arnd Bergmann [Tue, 15 May 2012 21:02:58 +0000 (15:02 -0600)]
video/ili9320: do not mark exported functions __devexit
No symbol can be exported when the section is discarded - the only
solution I could think of is not to mark symbols as __devexit
when they are exported.
Takashi Iwai [Sat, 9 Jun 2012 06:46:42 +0000 (08:46 +0200)]
vga_switcheroo: Enable/disable audio clients at the right time
The audio clients have to be disabled before disabling the VGA and
switching. Similarly, enabling the audio client should be done at
last. Otherwise the audio-side operation stalls, eventually leading
to Oops or lockups.
Randy Dunlap [Fri, 8 Jun 2012 14:07:19 +0000 (14:07 +0000)]
netdev: fix drivers/net/phy/ kernel-doc warnings
Fix kernel-doc warnings in drivers/net/phy:
Warning(drivers/net/phy/mdio_bus.c:109): No description found for parameter 'mdio_bus_np'
Warning(drivers/net/phy/mdio_bus.c:109): Excess function parameter 'mdio_np' description in 'of_mdio_find_bus'
Randy Dunlap [Fri, 8 Jun 2012 14:01:44 +0000 (14:01 +0000)]
net/core: fix kernel-doc warnings
Fix kernel-doc warnings in net/core:
Warning(net/core/skbuff.c:3368): No description found for parameter 'delta_truesize'
Warning(net/core/filter.c:628): No description found for parameter 'pfp'
Warning(net/core/filter.c:628): Excess function parameter 'sk' description in 'sk_unattached_filter_create'
projects / linux.git / log