Robert Baldyga [Fri, 17 Apr 2015 06:43:09 +0000 (08:43 +0200)]
serial: samsung: fix serial console break
This patch fixes problems with serial console break. When function
s3c64xx_serial_startup() was started while serial console has been working,
it caused lose of characters written to TX FIFO. This effect was particularly
observable with systemd, which closes serial port every time when it's
not currently needed, hence function s3c64xx_serial_startup() is called
quite often there. To fix this problem we avoid resetting TX FIFO if port is
used as serial console.
Example of broken console log:
[ 1086.7 Expecting device dev-ttySAC1.device...
[ 1086.[ OK ] Reached target Paths.
[ 1086.756416] s[ OK ] Reached target Swap.
[ 1086.776413] systemd[1]: Reached target Swap.
[ 1086.776642] systemd[1]: Starting Root Slice.
[ 5.53403[ OK ] Created slice Root Slice.
[ 5.548433] systemd[1]: Create[ OK ] Created slice User and Session Slice.
[ 5.568414] sys[ OK ] Listening on /dev/initctl Compatibility Named Pipe.
[ 5.588388] s[ OK ] Listening on Delayed Shutdown Socket.
[ 5.608376] sy[ OK ] Listening on Journal Socket (/dev/log).
[ 5.628361] [ OK ] Listening on udev Kernel Socket.
[ 5.648357] s[ OK ] Listening on udev Control Socket.
[ 5.668353] s[ OK ] Listening on Journal Socket.
[ 5.688366] systemd[1]: Listeni[ OK ] Created slice System Slice.
[ 5.708393] Mounting Temporary Directory...
[ 7139.067436] Starting prepare device daemon...
[ 7139.091726] sy Starting Generate environment from /etc/profile.d...
[ 5.792867] system Starting Create Static Device Nodes in /dev...
[ 7848.718 Mounting Debug File System...
[ 7848.7384 Mounting Configuration File System...
[ 5.852 Starting Apply Kernel Variables...
[ 5.8720 Starting Setup Virtual Console...
[ 7848.798 Starting udev Coldplug all Devices...
[ 7848.817 Starting Journal Service...
[ OK ] Started Journal Service.
[ 7848.854222] s[ OK ] Reached target Slices.
Starting Remount Root and Kernel File Systems...
[ OK ] Mounted Configuration File System.
tty/serial: at91: maxburst was missing for dma transfers
Maxburst was not set when doing the dma slave configuration. This value
is checked by the recently introduced xdmac. It causes an error when
doing the slave configuration and so prevents from using dma.
Do not probe all serial drivers by of_serial.c which are using
device_type = "serial"; property. Only drivers which have valid
compatible strings listed in the driver should be probed.
When PORT_UNKNOWN is setup probe will fail anyway.
Arnd quotation about driver historical background:
"when I wrote that driver initially, the idea was that it would
get used as a stub to hook up all other serial drivers but after
that, the common code learned to create platform devices from DT"
This patch fix the problem with on the system with xilinx_uartps and
16550a where of_serial failed to register for xilinx_uartps and because
of irq_dispose_mapping() removed irq_desc. Then when xilinx_uartps was asking
for irq with request_irq() EINVAL is returned.
Michal Simek [Mon, 13 Apr 2015 14:34:21 +0000 (16:34 +0200)]
serial: xilinx: Use platform_get_irq to get irq description structure
For systems with CONFIG_SERIAL_OF_PLATFORM=y and device_type =
"serial"; property in DT of_serial.c driver maps and unmaps IRQ (because
driver probe fails). Then a driver is called but irq mapping is not
created that's why driver is failing again in again on request_irq().
Based on this use platform_get_irq() instead of platform_get_resource()
which is doing irq_desc allocation and driver itself can request IRQ.
Warning(drivers/tty/serial/serial_core.c:1778): No description found for parameter 'putchar'
Warning(drivers/tty/serial/serial_core.c:1778): Excess function parameter 'write' description in 'uart_console_write'
It added some sanity checks to ignore potential garbage in CDC headers but
also introduced a potential infinite loop. This can happen at the first
loop iteration (elength = 0 in that case) if the description isn't a
DT_CS_INTERFACE or later if 'buffer[0]' is zero.
It should also be noted that the wrong length was being added to 'buffer'
in case 'buffer[1]' was not a DT_CS_INTERFACE descriptor, since elength was
assigned after that check in the loop.
A specially crafted USB device could be used to trigger this infinite loop.
Ivan T. Ivanov [Mon, 27 Apr 2015 12:27:36 +0000 (15:27 +0300)]
Revert "usb: host: ehci-msm: Use devm_ioremap_resource instead of devm_ioremap"
This reverts commit 70843f623b58 ("usb: host: ehci-msm: Use
devm_ioremap_resource instead of devm_ioremap") and commit e507bf577e5a ("host: ehci-msm: remove duplicate check on resource"),
because msm_otg and this driver are using same address space to
access AHB mode and USB command registers.
Li Jun [Sun, 12 Apr 2015 09:51:02 +0000 (17:51 +0800)]
usb: chipidea: otg: remove mutex unlock and lock while stop and start role
Wrongly release mutex lock during otg_statemachine may result in re-enter
otg_statemachine, which is not allowed, we should do next state transtition
after previous one completed.
Hans de Goede [Tue, 21 Apr 2015 09:20:31 +0000 (11:20 +0200)]
uas: Add US_FL_MAX_SECTORS_240 flag
The usb-storage driver sets max_sectors = 240 in its scsi-host template,
for uas we do not want to do that for all devices, but testing has shown
that some devices need it.
This commit adds a US_FL_MAX_SECTORS_240 flag for such devices, and
implements support for it in uas.c, while at it it also adds support
for US_FL_MAX_SECTORS_64 to uas.c.
Hans de Goede [Tue, 21 Apr 2015 09:20:30 +0000 (11:20 +0200)]
uas: Allow uas_use_uas_driver to return usb-storage flags
uas_use_uas_driver may set some US_FL_foo flags during detection, currently
these are stored in a local variable and then throw away, but these may be
of interest to the caller, so add an extra parameter to (optionally) return
the detected flags, and use this in the uas driver.
ALU64_DIV instruction should be dividing 64-bit by 64-bit,
whereas do_div() does 64-bit by 32-bit divide.
x64 and arm64 JITs correctly implement 64 by 64 unsigned divide.
llvm BPF backend emits code assuming that ALU64_DIV does 64 by 64.
Fixes: 89aa075832b0 ("net: sock: allow eBPF programs to be attached to sockets") Reported-by: Michael Holzheu <[email protected]> Acked-by: Daniel Borkmann <[email protected]> Signed-off-by: Alexei Starovoitov <[email protected]> Signed-off-by: David S. Miller <[email protected]>
net: netcp: remove call to netif_carrier_(on/off) for MAC to Phy interface
Currently when interface type is MAC to Phy, netif_carrier_(on/off)
is called which is not needed as Phy lib already updates the carrier
status to net stack. This is needed only for other interface types
1) mlx4 doesn't check fully for supported valid RSS hash function, fix
from Amir Vadai
2) Off by one in ibmveth_change_mtu(), from David Gibson
3) Prevent altera chip from reporting false error interrupts in some
circumstances, from Chee Nouk Phoon
4) Get rid of that stupid endless loop trying to allocate a FIN packet
in TCP, and in the process kill deadlocks. From Eric Dumazet
5) Fix get_rps_cpus() crash due to wrong invalid-cpu value, also from
Eric Dumazet
6) Fix two bugs in async rhashtable resizing, from Thomas Graf
7) Fix topology server listener socket namespace bug in TIPC, from Ying
Xue
8) Add some missing HAS_DMA kconfig dependencies, from Geert
Uytterhoeven
9) bgmac driver intends to force re-polling but does so by returning
the wrong value from it's ->poll() handler. Fix from Rafał Miłecki
10) When the creater of an rhashtable configures a max size for it,
don't bark in the logs and drop insertions when that is exceeded.
Fix from Johannes Berg
11) Recover from out of order packets in ppp mppe properly, from Sylvain
Rochet
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
bnx2x: really disable TPA if 'disable_tpa' option is set
net:treewide: Fix typo in drivers/net
net/mlx4_en: Prevent setting invalid RSS hash function
mdio-mux-gpio: use new gpiod_get_array and gpiod_put_array functions
netfilter; Add some missing default cases to switch statements in nft_reject.
ppp: mppe: discard late packet in stateless mode
ppp: mppe: sanity error path rework
net/bonding: Make DRV macros private
net: rfs: fix crash in get_rps_cpus()
altera tse: add support for fixed-links.
pxa168: fix double deallocation of managed resources
net: fix crash in build_skb()
net: eth: altera: Resolve false errors from MSGDMA to TSE
ehea: Fix memory hook reference counting crashes
net/tg3: Release IRQs on permanent error
net: mdio-gpio: support access that may sleep
inet: fix possible panic in reqsk_queue_unlink()
rhashtable: don't attempt to grow when at max_size
bgmac: fix requests for extra polling calls from NAPI
tcp: avoid looping in tcp_send_fin()
...
Michal Schmidt [Mon, 27 Apr 2015 15:20:38 +0000 (17:20 +0200)]
bnx2x: really disable TPA if 'disable_tpa' option is set
bnx2x's 'disable_tpa=1' module option is not respected properly and TPA
(transparent packet aggregation) remains enabled. Even though the
module option causes LRO to be disabled, TPA is enabled in GRO mode.
Additionally, disabling GRO via ethtool then has no effect. One can
still observe tpa_* statistics increase and large packets being received
in tcpdump.
The bug was an unintended consequence of commit aebf6244cd39 "bnx2x: Be
more forgiving toward SW GRO".
Fix it by following the bp->disable_tpa flag when initializing fp's.
Amir Vadai [Mon, 27 Apr 2015 10:40:56 +0000 (13:40 +0300)]
net/mlx4_en: Prevent setting invalid RSS hash function
mlx4_en_check_rxfh_func() was checking for hardware support before
setting a known RSS hash function, but didn't do any check before
setting unknown RSS hash function. Need to make it fail on such values.
In this occasion, moved the actual setting of the new value from the
check function into mlx4_en_set_rxfh().
Fixes: 947cbb0 ("net/mlx4_en: Support for configurable RSS hash function") Signed-off-by: Amir Vadai <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Rojhalat Ibrahim [Mon, 27 Apr 2015 08:37:31 +0000 (10:37 +0200)]
mdio-mux-gpio: use new gpiod_get_array and gpiod_put_array functions
Use the new gpiod_get_array and gpiod_put_array functions
(added to mainline in the v4.1 merge window) for obtaining and
disposing of GPIO descriptors.
David S. Miller [Mon, 27 Apr 2015 17:20:34 +0000 (13:20 -0400)]
netfilter; Add some missing default cases to switch statements in nft_reject.
This fixes:
====================
net/netfilter/nft_reject.c: In function ‘nft_reject_dump’:
net/netfilter/nft_reject.c:61:2: warning: enumeration value ‘NFT_REJECT_TCP_RST’ not handled in switch [-Wswitch]
switch (priv->type) {
^
net/netfilter/nft_reject.c:61:2: warning: enumeration value ‘NFT_REJECT_ICMPX_UNREACH’ not handled in switch [-Wswi\
tch]
net/netfilter/nft_reject_inet.c: In function ‘nft_reject_inet_dump’:
net/netfilter/nft_reject_inet.c:105:2: warning: enumeration value ‘NFT_REJECT_TCP_RST’ not handled in switch [-Wswi\
tch]
switch (priv->type) {
^
====================
The 3w-9xxx driver needs to tear down the dma mappings before returning
the command to the midlayer, as there is no guarantee the sglist and
count are valid after that point. Also remove the dma mapping helpers
which have another inherent race due to the request_id index.
The 3w-xxxx driver needs to tear down the dma mappings before returning
the command to the midlayer, as there is no guarantee the sglist and
count are valid after that point. Also remove the dma mapping helpers
which have another inherent race due to the request_id index.
The 3w-sas driver needs to tear down the dma mappings before returning
the command to the midlayer, as there is no guarantee the sglist and
count are valid after that point. Also remove the dma mapping helpers
which have another inherent race due to the request_id index.
Mike Christie [Tue, 21 Apr 2015 03:42:24 +0000 (22:42 -0500)]
SCSI: add 1024 max sectors black list flag
This works around a issue with qnap iscsi targets not handling large IOs
very well.
The target returns:
VPD INQUIRY: Block limits page (SBC)
Maximum compare and write length: 1 blocks
Optimal transfer length granularity: 1 blocks
Maximum transfer length: 4294967295 blocks
Optimal transfer length: 4294967295 blocks
Maximum prefetch, xdread, xdwrite transfer length: 0 blocks
Maximum unmap LBA count: 8388607
Maximum unmap block descriptor count: 1
Optimal unmap granularity: 16383
Unmap granularity alignment valid: 0
Unmap granularity alignment: 0
Maximum write same length: 0xffffffff blocks
Maximum atomic transfer length: 0
Atomic alignment: 0
Atomic transfer length granularity: 0
and it is *sometimes* able to handle at least one IO of size up to 8 MB. We
have seen in traces where it will sometimes work, but other times it
looks like it fails and it looks like it returns failures if we send
multiple large IOs sometimes. Also it looks like it can return 2 different
errors. It will sometimes send iscsi reject errors indicating out of
resources or it will send invalid cdb illegal requests check conditions.
And then when it sends iscsi rejects it does not seem to handle retries
when there are command sequence holes, so I could not just add code to
try and gracefully handle that error code.
The problem is that we do not have a good contact for the company,
so we are not able to determine under what conditions it returns
which error and why it sometimes works.
So, this patch just adds a new black list flag to set targets like this to
the old max safe sectors of 1024. The max_hw_sectors changes added in 3.19
caused this regression, so I also ccing stable.
The task migration notifier was originally introduced in order to support
the pvclock vsyscall with non-synchronized TSC, but KVM only supports it
with synchronized TSC. Hence, on KVM the race condition is only needed
due to a bad implementation on the host side, and even then it's so rare
that it's mostly theoretical.
As far as KVM is concerned it's possible to fix the host, avoiding the
additional complexity in the vDSO and the (re)introduction of the task
migration notifier.
Xen, on the other hand, hasn't yet implemented vsyscall support at
all, so we do not care about its plans for non-synchronized TSC.
Radim Krčmář [Wed, 25 Mar 2015 11:08:14 +0000 (12:08 +0100)]
kvm: x86: fix kvmclock update protocol
The kvmclock spec says that the host will increment a version field to
an odd number, then update stuff, then increment it to an even number.
The host is buggy and doesn't do this, and the result is observable
when one vcpu reads another vcpu's kvmclock data.
There's no good way for a guest kernel to keep its vdso from reading
a different vcpu's kvmclock data, but we don't need to care about
changing VCPUs as long as we read a consistent data from kvmclock.
(VCPU can change outside of this loop too, so it doesn't matter if we
return a value not fit for this VCPU.)
Basically snd_emux_detach_seq() doesn't need a protection of
emu->register_mutex as it's already being unregistered. So, we can
get rid of this for avoiding the deadlock.
Some models provide too long string for the shortname that has 32bytes
including the terminator, and it results in a non-terminated string
exposed to the user-space. This isn't too critical, though, as the
string is stopped at the succeeding longname string.
This patch fixes such entries by dropping "SB" prefix (it's enough to
fit within 32 bytes, so far). Meanwhile, it also changes strcpy()
with strlcpy() to make sure that this kind of problem won't happen in
future, too.
Buffers allocated by dma_alloc_coherent() are always zeroed on Alpha,
ARM (32bit), MIPS, PowerPC, x86/x86_64 and probably other architectures.
It turned out that some drivers rely on this 'feature'. Allocated buffer
might be also exposed to userspace with dma_mmap() call, so clearing it
is desired from security point of view to avoid exposing random memory
to userspace. This patch unifies dma_alloc_coherent() behavior on ARM64
architecture with other implementations by unconditionally zeroing
allocated buffer.
Since several interrupt controllers including GIC support both edge and
level triggered interrupts, it's useful to provide that information in
/proc/interrupts even on ARM64 similar to ARM and PPC.
This is based on Geert Uytterhoeven's commit 7c07005eea96 ("ARM: 8339/1:
Enable CONFIG_GENERIC_IRQ_SHOW_LEVEL")
Andre Przywara [Mon, 20 Apr 2015 10:14:19 +0000 (11:14 +0100)]
arm64: add missing data types in smp_load_acquire/smp_store_release
Commit 8053871d0f7f ("smp: Fix smp_call_function_single_async()
locking") introduced a call to smp_load_acquire() with a u16 argument,
but we only cared about u32 and u64 types in that function so far.
This resulted in a compiler warning fortunately, pointing at an
uninitialized use. Due to the implementation structure the compiler
misses that bug in the smp_store_release(), though.
Add the u16 and u8 variants using ldarh/stlrh and ldarb/stlrb,
respectively. Together with the compiletime_assert_atomic_type() check
this should cover all cases now.
The mute-LED mode control has the fixed on/off states that are
supposed to remain on/off regardless of the master switch. However,
this doesn't work actually because the vmaster hook is called in the
vmaster code itself.
This patch fixes it by calling the hook indirectly after checking the
mute LED mode.
ALSA: hda - Fix click noise at start on Dell XPS13
Dell XPS13 produces a click noise at boot up, and Gabriele spotted out
that it's triggered by the initial pin control of the mic (NID 0x19).
This has to be set to Hi-Z Vref while the driver initializes to Vref
80% as a normal mic.
This patch fixes the generic parser code not to override the target
vref if it has been already set by the driver, and adds a proper
initialization of the target vref for this pin in the Realtek driver
side.
drivers: sh: Disable PM runtime for multi-platform r8a73a4 with genpd
If the default PM domain using PM_CLK is used for PM runtime, the real PM
domain(s) cannot be registered from DT later.
Hence do not enable it when running a multi-platform kernel with genpd
support on an r8a73a4. The R-Mobile PM domain driver will take care of
PM runtime management of the module clocks.
The default PM domain is still needed for:
- platforms without genpd support,
- the legacy (non-DT) case, where genpd may take over later, except
for the C5 "always on" PM domain.
drivers: sh: Disable PM runtime for multi-platform sh73a0 with genpd
If the default PM domain using PM_CLK is used for PM runtime, the real PM
domain(s) cannot be registered from DT later.
Hence do not enable it when running a multi-platform kernel with genpd
support on an sh73a0. The R-Mobile PM domain driver will take care of
PM runtime management of the module clocks.
The default PM domain is still needed for:
- platforms without genpd support,
- the legacy (non-DT) case, where genpd may take over later, except
for the C5 "always on" PM domain.
David S. Miller [Mon, 27 Apr 2015 03:25:13 +0000 (23:25 -0400)]
Merge branch 'ppp_mppe_desync'
Sylvain Rochet says:
====================
ppp: mppe: fixes MPPE desync on links which don't guarantee packet ordering
I am currently having an issue with PPP over L2TP (UDP) and MPPE in
stateless mode (default mode), UDP does not guarantee packet ordering so
we might get out of order packet. MPPE needs to be continuously synched
so we should drop late UDP packet.
I added a printk on the number of time we rekeyed in MPPE decompressor,
this is what we currently have if we receive a slightly out of order UDP
packet:
[1731001.049206] mppe_decompress[1]: ccount 1559
[1731001.049216] mppe_decompress[1]: rekeyed 1 times
[1731001.049228] mppe_decompress[1]: ccount 1560
[1731001.049232] mppe_decompress[1]: rekeyed 1 times
[1731001.050170] mppe_decompress[1]: ccount 1562
[1731001.050182] mppe_decompress[1]: rekeyed 2 times
[1731001.050191] mppe_decompress[1]: ccount 1561
[1731001.062576] mppe_decompress[1]: rekeyed 4095 times
^^^^
This is obviously wrong, we missed packet 1561 and we already rekeyed 2
times for 1562 we previously received, we can't recover the decryption
key we need for 1561, we should drop it instead of rekeying 4095 times.
This patch series drop any packet with are not within the 4096/2 forward
window.
====================
When PPP is used over a link which does not guarantee packet ordering,
we might get late MPPE packets. This is a problem because MPPE must be
kept synchronized and the current implementation does not drop them and
rekey 4095 times instead of 0, which is wrong.
In order to prevent rekeying about a whole count space times (~ 4095
times), drop packets which are not within the forward 4096/2 window and
increase sanity error counter.
dmaengine: usb-dmac: Protect PM-only functions to kill warning
If CONFIG_PM=n:
drivers/dma/sh/usb-dmac.c:677: warning: ‘usb_dmac_runtime_suspend’ defined but not used
drivers/dma/sh/usb-dmac.c:688: warning: ‘usb_dmac_runtime_resume’ defined but not used
Protect the unused functions by #ifdef CONFIG_PM to fix this.
The bonding modules currently defines four macros with
general names that pollute the global namespace:
DRV_VERSION
DRV_RELDATE
DRV_NAME
DRV_DESCRIPTION
Fixing that by defining a private bonding_priv.h
header files which includes those defines.
Andy Lutomirski [Sun, 26 Apr 2015 23:47:59 +0000 (16:47 -0700)]
x86_64, asm: Work around AMD SYSRET SS descriptor attribute issue
AMD CPUs don't reinitialize the SS descriptor on SYSRET, so SYSRET with
SS == 0 results in an invalid usermode state in which SS is apparently
equal to __USER_DS but causes #SS if used.
Work around the issue by setting SS to __KERNEL_DS __switch_to, thus
ensuring that SYSRET never happens with SS set to NULL.
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Pull intel drm fixes from Dave Airlie.
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/i915: vlv: fix save/restore of GFX_MAX_REQ_COUNT reg
drm/i915: Workaround to avoid lite restore with HEAD==TAIL
drm/i915: cope with large i2c transfers
Pull intel iommu updates from David Woodhouse:
"This lays a little of the groundwork for upcoming Shared Virtual
Memory support — fixing some bogus #defines for capability bits and
adding the new ones, and starting to use the new wider page tables
where we can, in anticipation of actually filling in the new fields
therein.
It also allows graphics devices to be assigned to VM guests again.
This got broken in 3.17 by disallowing assignment of RMRR-afflicted
devices. Like USB, we do understand why there's an RMRR for graphics
devices — and unlike USB, it's actually sane. So we can make an
exception for graphics devices, just as we do USB controllers.
Finally, tone down the warning about the X2APIC_OPT_OUT bit, due to
persistent requests. X2APIC_OPT_OUT was added to the spec as a nasty
hack to allow broken BIOSes to forbid us from using X2APIC when they
do stupid and invasive things and would break if we did.
Someone noticed that since Windows doesn't have full IOMMU support for
DMA protection, setting the X2APIC_OPT_OUT bit made Windows avoid
initialising the IOMMU on the graphics unit altogether.
This means that it would be available for use in "driver mode", where
the IOMMU registers are made available through a BAR of the graphics
device and the graphics driver can do SVM all for itself.
So they started setting the X2APIC_OPT_OUT bit on *all* platforms with
SVM capabilities. And even the platforms which *might*, if the
planets had been aligned correctly, possibly have had SVM capability
but which in practice actually don't"
* git://git.infradead.org/intel-iommu:
iommu/vt-d: support extended root and context entries
iommu/vt-d: Add new extended capabilities from v2.3 VT-d specification
iommu/vt-d: Allow RMRR on graphics devices too
iommu/vt-d: Print x2apic opt out info instead of printing a warning
iommu/vt-d: kill bogus ecap_niotlb_iunits()
Merge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux
Pull i2c fixes from Wolfram Sang:
"This has a mixture of merge window cleanups and bugfixes"
* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
i2c: st: add include for pinctrl
i2c: mux: use proper dev when removing "channel-X" symlinks
i2c: digicolor: remove duplicate include
i2c: Mark adapter devices with pm_runtime_no_callbacks
i2c: pca-platform: fix broken email address
i2c: mxs: fix broken email address
i2c: rk3x: report number of messages transmitted
Merge branch 'for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs
Pull btrfs fixes from Chris Mason:
"Filipe hit two problems in my block group cache patches. We finalized
the fixes last week and ran through more tests"
* 'for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
Btrfs: prevent list corruption during free space cache processing
Btrfs: fix inode cache writeout
Dave Airlie [Mon, 27 Apr 2015 00:35:15 +0000 (10:35 +1000)]
Merge tag 'drm-intel-next-fixes-2015-04-25' of git://anongit.freedesktop.org/drm-intel into drm-fixes
three fixes for i915.
* tag 'drm-intel-next-fixes-2015-04-25' of git://anongit.freedesktop.org/drm-intel:
drm/i915: vlv: fix save/restore of GFX_MAX_REQ_COUNT reg
drm/i915: Workaround to avoid lite restore with HEAD==TAIL
drm/i915: cope with large i2c transfers
Merge tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client updates from Trond Myklebust:
"Another set of mainly bugfixes and a couple of cleanups. No new
functionality in this round.
Highlights include:
Stable patches:
- Fix a regression in /proc/self/mountstats
- Fix the pNFS flexfiles O_DIRECT support
- Fix high load average due to callback thread sleeping
Bugfixes:
- Various patches to fix the pNFS layoutcommit support
- Do not cache pNFS deviceids unless server notifications are enabled
- Fix a SUNRPC transport reconnection regression
- make debugfs file creation failure non-fatal in SUNRPC
- Another fix for circular directory warnings on NFSv4 "junctioned"
mountpoints
- Fix locking around NFSv4.2 fallocate() support
- Truncating NFSv4 file opens should also sync O_DIRECT writes
- Prevent infinite loop in rpcrdma_ep_create()
Features:
- Various improvements to the RDMA transport code's handling of
memory registration
- Various code cleanups"
* tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (55 commits)
fs/nfs: fix new compiler warning about boolean in switch
nfs: Remove unneeded casts in nfs
NFS: Don't attempt to decode missing directory entries
Revert "nfs: replace nfs_add_stats with nfs_inc_stats when add one"
NFS: Rename idmap.c to nfs4idmap.c
NFS: Move nfs_idmap.h into fs/nfs/
NFS: Remove CONFIG_NFS_V4 checks from nfs_idmap.h
NFS: Add a stub for GETDEVICELIST
nfs: remove WARN_ON_ONCE from nfs_direct_good_bytes
nfs: fix DIO good bytes calculation
nfs: Fetch MOUNTED_ON_FILEID when updating an inode
sunrpc: make debugfs file creation failure non-fatal
nfs: fix high load average due to callback thread sleeping
NFS: Reduce time spent holding the i_mutex during fallocate()
NFS: Don't zap caches on fallocate()
xprtrdma: Make rpcrdma_{un}map_one() into inline functions
xprtrdma: Handle non-SEND completions via a callout
xprtrdma: Add "open" memreg op
xprtrdma: Add "destroy MRs" memreg op
xprtrdma: Add "reset MRs" memreg op
...
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull fourth vfs update from Al Viro:
"d_inode() annotations from David Howells (sat in for-next since before
the beginning of merge window) + four assorted fixes"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
RCU pathwalk breakage when running into a symlink overmounting something
fix I_DIO_WAKEUP definition
direct-io: only inc/dec inode->i_dio_count for file systems
fs/9p: fix readdir()
VFS: assorted d_backing_inode() annotations
VFS: fs/inode.c helpers: d_inode() annotations
VFS: fs/cachefiles: d_backing_inode() annotations
VFS: fs library helpers: d_inode() annotations
VFS: assorted weird filesystems: d_inode() annotations
VFS: normal filesystems (and lustre): d_inode() annotations
VFS: security/: d_inode() annotations
VFS: security/: d_backing_inode() annotations
VFS: net/: d_inode() annotations
VFS: net/unix: d_backing_inode() annotations
VFS: kernel/: d_inode() annotations
VFS: audit: d_backing_inode() annotations
VFS: Fix up some ->d_inode accesses in the chelsio driver
VFS: Cachefiles should perform fs modifications on the top layer only
VFS: AF_UNIX sockets should call mknod on the top layer only
Merge tag 'pm+acpi-4.1-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull more power management and ACPI updates from Rafael Wysocki:
"These are fixes mostly (intel_pstate, ACPI core, ACPI EC driver,
cpupower tool), a new CPU ID for the Intel RAPL driver and one
intel_pstate driver improvement that didn't make it to my previous
pull requests due to timing.
Specifics:
- Fix a build warning in the intel_pstate driver showing up in
non-SMP builds (Borislav Petkov)
- Change one of the intel_pstate's P-state selection parameters for
Baytrail and Cherrytrail CPUs to significantly improve performance
at the cost of a small increase in energy consumption (Kristen
Carlson Accardi)
- Fix a NULL pointer dereference in the ACPI EC driver due to an
unsafe list walk in the query handler removal routine (Chris
Bainbridge)
- Get rid of a false-positive lockdep warning in the ACPI container
hot-remove code (Rafael J Wysocki)
- Prevent the ACPI device enumeration code from creating device
objects of a wrong type in some cases (Rafael J Wysocki)
- Add Skylake processors support to the Intel RAPL power capping
driver (Brian Bian)
- Drop the stale MAINTAINERS entry for the ACPI dock driver that is
regarded as part of the ACPI core and maintained along with it now
(Chao Yu)
- Fix cpupower tool breakage caused by a library API change in libpci
3.3.0 (Lucas Stach)"
* tag 'pm+acpi-4.1-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
ACPI / scan: Add a scan handler for PRP0001
ACPI / scan: Annotate physical_node_lock in acpi_scan_is_offline()
ACPI / EC: fix NULL pointer dereference in acpi_ec_remove_query_handler()
MAINTAINERS: remove maintainship entry of docking station driver
powercap / RAPL: Add support for Intel Skylake processors
cpufreq: intel_pstate: Fix an annoying !CONFIG_SMP warning
intel_pstate: Change the setpoint for Atom params
cpupower: fix breakage from libpci API change
Pull crypto fixes from Herbert Xu:
"This push fixes a build problem with img-hash under non-standard
configurations and a serious regression with sha512_ssse3 which can
lead to boot failures"
* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: img-hash - CRYPTO_DEV_IMGTEC_HASH should depend on HAS_DMA
crypto: x86/sha512_ssse3 - fixup for asm function prototype change
Merge tag 'platform-drivers-x86-v4.1-1' of git://git.infradead.org/users/dvhart/linux-platform-drivers-x86
Pull x86 platform driver updates from Darren Hart:
"This series includes significant updates to the toshiba_acpi driver
and the reintroduction of the dell-laptop keyboard backlight additions
I had to revert previously. Also included are various fixes for
typos, warnings, correctness, and minor bugs.
Specifics:
dell-laptop:
- add support for keyboard backlight.
toshiba_acpi:
- adaptive keyboard, hotkey, USB sleep and charge, and backlight
updates. Update sysfs documentation.
toshiba_bluetooth:
- fix enabling/disabling loop on recent devices
apple-gmux:
- lock iGP IO to protect from vgaarb changes
* tag 'platform-drivers-x86-v4.1-1' of git://git.infradead.org/users/dvhart/linux-platform-drivers-x86: (25 commits)
toshiba_acpi: Do not register vendor backlight when acpi_video bl is available
MAINTAINERS: Add me on list of Dell laptop drivers
platform: x86: dell-laptop: Add support for keyboard backlight
Documentation/ABI: Update sysfs-driver-toshiba_acpi entry
toshiba_acpi: Fix pr_* messages from USB Sleep Functions
toshiba_acpi: Update and fix USB Sleep and Charge modes
wmi: Use bool function return values of true/false not 1/0
toshiba_bluetooth: Fix enabling/disabling loop on recent devices
toshiba_bluetooth: Clean up *_add function and disable BT device at removal
toshiba_bluetooth: Add three new functions to the driver
toshiba_acpi: Fix the enabling of the Special Functions
toshiba_acpi: Use the Hotkey Event Type function for keymap choosing
toshiba_acpi: Add Hotkey Event Type function and definitions
x86/wmi: delete unused wmi_data_lock mutex causing gcc warning
apple-gmux: lock iGP IO to protect from vgaarb changes
MAINTAINERS: Add missing Toshiba devices and add myself as maintainer
toshiba_acpi: Update events in toshiba_acpi_notify
intel-oaktrail: Fix trivial typo in comment
thinkpad_acpi: off by one in adaptive_keyboard_hotkey_notify_hotkey()
thinkpad_acpi: signedness bugs getting current_mode
...
Merge tag 'chrome-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/olof/chrome-platform
Pull chrome platform updates from Olof Johansson:
"Here's a set of updates to the Chrome OS platform drivers for this
merge window.
Main new things this cycle is:
- Driver changes to expose the lightbar to users. With this, you can
make your own blinkenlights on Chromebook Pixels.
- Changes in the way that the atmel_mxt trackpads are probed. The
laptop driver is trying to be smart and not instantiate the devices
that don't answer to probe. For the trackpad that can come up in
two modes (bootloader or regular), this gets complicated since the
driver already knows how to handle the two modes including the
actual addresses used. So now the laptop driver needs to know more
too, instantiating the regular address even if the bootloader one
is the probe that passed.
- mfd driver improvements by Javier Martines Canillas, and a few
bugfixes from him, kbuild and myself"
* tag 'chrome-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/olof/chrome-platform:
platform/chrome: chromeos_laptop - instantiate Atmel at primary address
platform/chrome: cros_ec_lpc - Depend on X86 || COMPILE_TEST
platform/chrome: cros_ec_lpc - Include linux/io.h header file
platform/chrome: fix platform_no_drv_owner.cocci warnings
platform/chrome: cros_ec_lightbar - fix duplicate const warning
platform/chrome: cros_ec_dev - fix Unknown escape '%' warning
platform/chrome: Expose Chrome OS Lightbar to users
platform/chrome: Create sysfs attributes for the ChromeOS EC
mfd: cros_ec: Instantiate ChromeOS EC character device
platform/chrome: Add Chrome OS EC userspace device interface
platform/chrome: Add cros_ec_lpc driver for x86 devices
mfd: cros_ec: Add char dev and virtual dev pointers
mfd: cros_ec: Use fixed size arrays to transfer data with the EC
Merge tag 'cris-for-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/jesper/cris
Pull arch/cris updates from Jesper Nilsson:
"Some much needed love for the CRIS-port.
There's a bunch of changes this time, giving the CRISv32 port a bit of
modern makeover with device-tree, irq domain and gpiolib support, and
more switchover to generic frameworks.
Some small fixes and removal of the theoretical SMP support brings up
the rear"
* tag 'cris-for-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/jesper/cris:
cris: fix integer overflow in ELF_ET_DYN_BASE
CRISv32: use GENERIC_SCHED_CLOCK
CRISv32: use MMIO clocksource
CRISv32: use generic clockevents
CRIS: use generic headers via Kbuild
CRIS: use generic cmpxchg.h
CRIS: use generic atomic.h
CRIS: use generic atomic bitops
CRISv10: remove redundant macros from system.h
CRIS: remove SMP code
CRISv32: don't enable irqs in INIT_THREAD
CRISv32: handle multiple signals
CRISv32: prevent bogus restarts on sigreturn
CRISv32: don't attempt syscall restart on irq exit
Add binding documentation for CRIS
CRIS: add Axis 88 board device tree
CRISv32: add device tree support
CRISv32: add irq domains support
CRIS: enable GPIOLIB
Merge tag 'powerpc-4.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux
Pull powerpc fixes from Michael Ellerman:
- fix for mm_dec_nr_pmds() from Scott.
- fixes for oopses seen with KVM + THP from Aneesh.
- build fixes from Aneesh & Shreyas.
* tag 'powerpc-4.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux:
powerpc/mm: Fix build error with CONFIG_PPC_TRANSACTIONAL_MEM disabled
powerpc/kvm: Fix ppc64_defconfig + PPC_POWERNV=n build error
powerpc/mm/thp: Return pte address if we find trans_splitting.
powerpc/mm/thp: Make page table walk safe against thp split/collapse
KVM: PPC: Remove page table walk helpers
KVM: PPC: Use READ_ONCE when dereferencing pte_t pointer
powerpc/hugetlb: Call mm_dec_nr_pmds() in hugetlb_free_pmd_range()
Eric Dumazet [Sat, 25 Apr 2015 16:35:24 +0000 (09:35 -0700)]
net: rfs: fix crash in get_rps_cpus()
Commit 567e4b79731c ("net: rfs: add hash collision detection") had one
mistake :
RPS_NO_CPU is no longer the marker for invalid cpu in set_rps_cpu()
and get_rps_cpu(), as @next_cpu was the result of an AND with
rps_cpu_mask
This bug showed up on a host with 72 cpus :
next_cpu was 0x7f, and the code was trying to access percpu data of an
non existent cpu.
In a follow up patch, we might get rid of compares against nr_cpu_ids,
if we init the tables with 0. This is silly to test for a very unlikely
condition that exists only shortly after table initialization, as
we got rid of rps_reset_sock_flow() and similar functions that were
writing this RPS_NO_CPU magic value at flow dismantle : When table is
old enough, it never contains this value anymore.
Andreas Oetken [Sat, 25 Apr 2015 16:07:52 +0000 (18:07 +0200)]
altera tse: add support for fixed-links.
Add support for fixed-links in configurations without PHY.
(e.g. connection to a switch, SGMII point to point, SFPs)
Check: Documentation/devicetree/bindings/net/fixed-link.txt. Signed-off-by: Andreas Oetken <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull second batch of KVM changes from Paolo Bonzini:
"This mostly includes the PPC changes for 4.1, which this time cover
Book3S HV only (debugging aids, minor performance improvements and
some cleanups). But there are also bug fixes and small cleanups for
ARM, x86 and s390.
The task_migration_notifier revert and real fix is still pending
review, but I'll send it as soon as possible after -rc1"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (29 commits)
KVM: arm/arm64: check IRQ number on userland injection
KVM: arm: irqfd: fix value returned by kvm_irq_map_gsi
KVM: VMX: Preserve host CR4.MCE value while in guest mode.
KVM: PPC: Book3S HV: Use msgsnd for signalling threads on POWER8
KVM: PPC: Book3S HV: Translate kvmhv_commence_exit to C
KVM: PPC: Book3S HV: Streamline guest entry and exit
KVM: PPC: Book3S HV: Use bitmap of active threads rather than count
KVM: PPC: Book3S HV: Use decrementer to wake napping threads
KVM: PPC: Book3S HV: Don't wake thread with no vcpu on guest IPI
KVM: PPC: Book3S HV: Get rid of vcore nap_count and n_woken
KVM: PPC: Book3S HV: Move vcore preemption point up into kvmppc_run_vcpu
KVM: PPC: Book3S HV: Minor cleanups
KVM: PPC: Book3S HV: Simplify handling of VCPUs that need a VPA update
KVM: PPC: Book3S HV: Accumulate timing information for real-mode code
KVM: PPC: Book3S HV: Create debugfs file for each guest's HPT
KVM: PPC: Book3S HV: Add ICP real mode counters
KVM: PPC: Book3S HV: Move virtual mode ICP functions to real-mode
KVM: PPC: Book3S HV: Convert ICS mutex lock to spin lock
KVM: PPC: Book3S HV: Add guest->host real mode completion counters
KVM: PPC: Book3S HV: Add helpers for lock/unlock hpte
...
pxa168: fix double deallocation of managed resources
Commit 43d3ddf87a57 ("net: pxa168_eth: add device tree support") starts
to use managed resources by adding devm_clk_get() and
devm_ioremap_resource(), but it leaves explicit iounmap() and clock_put()
in pxa168_eth_remove() and in failure handling code of pxa168_eth_probe().
As a result double free can happen.
The patch removes explicit resource deallocation. Also it converts
clk_disable() to clk_disable_unprepare() to make it symmetrical with
clk_prepare_enable().
Found by Linux Driver Verification project (linuxtesting.org).
Yang Dongsheng [Thu, 9 Apr 2015 04:08:43 +0000 (12:08 +0800)]
Btrfs: fill ->last_trans for delayed inode in btrfs_fill_inode.
We need to fill inode when we found a node for it in delayed_nodes_tree.
But we did not fill the ->last_trans currently, it will cause the test
of xfstest/generic/311 fail. Scenario of the 311 is shown as below:
Problem:
(1). test_fd = open(fname, O_RDWR|O_DIRECT)
(2). pwrite(test_fd, buf, 4096, 0)
(3). close(test_fd)
(4). drop_all_caches() <-------- "echo 3 > /proc/sys/vm/drop_caches"
(5). test_fd = open(fname, O_RDWR|O_DIRECT)
(6). fsync(test_fd);
<-------- we did not get the correct log entry for the file
Reason:
When we re-open this file in (5), we would find a node
in delayed_nodes_tree and fill the inode we are lookup with the
information. But the ->last_trans is not filled, then the fsync()
will check the ->last_trans and found it's 0 then say this inode
is already in our tree which is commited, not recording the extents
for it.
Fix:
This patch fill the ->last_trans properly and set the
runtime_flags if needed in this situation. Then we can get the
log entries we expected after (6) and generic/311 passed.
btrfs: unlock i_mutex after attempting to delete subvolume during send
Whenever the check for a send in progress introduced in commit 521e0546c970 (btrfs: protect snapshots from deleting during send) is
hit, we return without unlocking inode->i_mutex. This is easy to see
with lockdep enabled:
[ +0.000059] ================================================
[ +0.000028] [ BUG: lock held when returning to user space! ]
[ +0.000029] 4.0.0-rc5-00096-g3c435c1 #93 Not tainted
[ +0.000026] ------------------------------------------------
[ +0.000029] btrfs/211 is leaving the kernel with locks still held!
[ +0.000029] 1 lock held by btrfs/211:
[ +0.000023] #0: (&type->i_mutex_dir_key){+.+.+.}, at: [<ffffffff8135b8df>] btrfs_ioctl_snap_destroy+0x2df/0x7a0
Omar Sandoval [Tue, 24 Feb 2015 10:47:05 +0000 (02:47 -0800)]
btrfs: fix race on ENOMEM in alloc_extent_buffer
Consider the following interleaving of overlapping calls to
alloc_extent_buffer:
Call 1:
- Successfully allocates a few pages with find_or_create_page
- find_or_create_page fails, goto free_eb
- Unlocks the allocated pages
Call 2:
- Calls find_or_create_page and gets a page in call 1's extent_buffer
- Finds that the page is already associated with an extent_buffer
- Grabs a reference to the half-written extent_buffer and calls
mark_extent_buffer_accessed on it
mark_extent_buffer_accessed will then try to call mark_page_accessed on
a null page and panic.
The fix is to decrement the reference count on the half-written
extent_buffer before unlocking the pages so call 2 won't use it. We
should also set exists = NULL in the case that we don't use exists to
avoid accidentally returning a freed extent_buffer in an error case.
# make device tree with one big hole
for i in `seq 1 1 100`; do
fallocate -l 1g $mntpath/$i
done
sync
for i in `seq 1 1 95`; do
rm $mntpath/$i
done
sync
# wait cleaner thread remove unused block group
sleep 300
fallocate -l 1g $mntpath/aaa
# failed to allocate new chunk
fallocate -l 1g $mntpath/bbb
Above script will make device tree with one big hole, and can only allocate
just one chunk in a transaction, so failed to allocate new chunk for $mntpath/bbb
Chris Mason [Sat, 18 Apr 2015 12:22:48 +0000 (05:22 -0700)]
Btrfs: don't check for delalloc_bytes in cache_save_setup
Now that we're doing free space cache writeback outside the critical
section in the commit, there is a bigger window for delalloc_bytes to
be added after a cache has been written. find_free_extent may do this
without putting the block group back into the dirty list, and also
without a transaction running.
Checking for delalloc_bytes in cache_save_setup means we might leave the
cache marked as written without invalidating it. Consistency checks
during mount will toss the cache, but it's better to get rid of the
check in cache_save_setup and let it get invalidated by the checks
already done during cache write out.
Filipe Manana [Sat, 25 Apr 2015 17:31:05 +0000 (18:31 +0100)]
Btrfs: fix deadlock when starting writeback of bg caches
While starting the writes of the dirty block group caches, if we don't
find a block group item in the extent tree we were leaving without
releasing our path, running delayed references and then looping again to
process any new dirty block groups. However this second iteration of the
loop could cause a deadlock because it tries to lock some other extent
tree node/leaf which another task already locked and it's blocked because
it's waiting for a lock on some node/leaf that is in our path that was not
released before.
We could also deadlock when running the delayed references - as we could
end up trying to lock the same nodes/leafs that we have in our local path
(with a different lock type).
This happens because in btrfs_start_dirty_block_groups() we splice the
transaction's list of dirty block groups into a local list and then we
keep extracting the first element of the list without holding the
cache_write_mutex mutex. This means that before we acquire that mutex
the first block group on the list might be removed by a conurrent task
running btrfs_remove_block_group(). So make sure we extract the first
element (and test the list emptyness) while holding that mutex.
Fixes: 1bbc621ef284 ("Btrfs: allow block group cache writeout
outside critical section in commit")
Fixes: 79930f5892e ("net: do not deplete pfmemalloc reserve") Signed-off-by: Eric Dumazet <[email protected]> Reported-by: Sasha Levin <[email protected]> Signed-off-by: David S. Miller <[email protected]>
net: eth: altera: Resolve false errors from MSGDMA to TSE
This patch resolves false errors from MSGDMA in TX mSGDMA MM to ST
mode, and is a continuation of the patch recently submitted by Andrea
Oetken. The MSGDMA had a logic bug that masked detection of this issue
prior to Quartus 14.1/Build 164. When the MSGDMA logic bug was addressed
in Quartus 14.1/Build 164, the driver problem was exposed.
The problem is corrected by making sure MSGDMA_DESC_CTL_TR_ERR_IRQ is not
set for any of the transmit DMA descriptors, and only used for receive
descriptors.
Michael Ellerman [Fri, 24 Apr 2015 05:52:32 +0000 (15:52 +1000)]
ehea: Fix memory hook reference counting crashes
The recent commit to only register the EHEA memory hotplug hooks on
adapter probe has a few problems.
Firstly the reference counting is wrong for multiple adapters, in that
the hooks are registered multiple times. Secondly the check in the tear
down path is backward. Finally the error path doesn't decrement the
count.
The multiple registration of the hooks is the biggest problem, as it
leads to oopses when the system is rebooted, and/or errors during memory
hotplug, eg:
Fixes: aa183323312d ("ehea: Register memory hotplug, reboot and crash hooks on adapter probe") Signed-off-by: Michael Ellerman <[email protected]> Signed-off-by: David S. Miller <[email protected]>
When having permanent EEH error, the PCI device will be removed
from the system. For this case, we shouldn't set pcierr_recovery
to true wrongly, which blocks the driver to release the allocated
interrupts and their handlers. Eventually, we can't disable MSI
or MSIx successfully because of the MSI or MSIx interrupts still
have associated interrupt actions, which is turned into following
stack dump.
platform/chrome: chromeos_laptop - instantiate Atmel at primary address
The new Atmel MXT driver expects i2c client's address contain the
primary (main address) of the chip, and calculates the expected
bootloader address form the primary address. Unfortunately chrome_laptop
does probe the devices and if touchpad (or touchscreen, or both) comes
up in bootloader mode the i2c device gets instantiated with the
bootloader address which confuses the driver.
To work around this issue let's probe the primary address first. If the
device is not detected at the primary address we'll probe alternative
addresses as "dummy" devices. If any of them are found, destroy the
dummy client and instantiate client with proper name at primary address
still.
Al Viro [Fri, 24 Apr 2015 19:47:07 +0000 (15:47 -0400)]
RCU pathwalk breakage when running into a symlink overmounting something
Calling unlazy_walk() in walk_component() and do_last() when we find
a symlink that needs to be followed doesn't acquire a reference to vfsmount.
That's fine when the symlink is on the same vfsmount as the parent directory
(which is almost always the case), but it's not always true - one _can_
manage to bind a symlink on top of something. And in such cases we end up
with excessive mntput().
direct-io: only inc/dec inode->i_dio_count for file systems
do_blockdev_direct_IO() increments and decrements the inode
->i_dio_count for each IO operation. It does this to protect against
truncate of a file. Block devices don't need this sort of protection.
For a capable multiqueue setup, this atomic int is the only shared
state between applications accessing the device for O_DIRECT, and it
presents a scaling wall for that. In my testing, as much as 30% of
system time is spent incrementing and decrementing this value. A mixed
read/write workload improved from ~2.5M IOPS to ~9.6M IOPS, with
better latencies too. Before:
In other setups, Robert Elliott reported seeing good performance
improvements:
https://lkml.org/lkml/2015/4/3/557
The more applications accessing the device, the worse it gets.
Add a new direct-io flags, DIO_SKIP_DIO_COUNT, which tells
do_blockdev_direct_IO() that it need not worry about incrementing
or decrementing the inode i_dio_count for this caller.
Johannes Berg [Wed, 22 Apr 2015 09:55:14 +0000 (11:55 +0200)]
fs/9p: fix readdir()
Al Viro's IOV changes broke 9p readdir() because the new code
didn't abort the read when it returned nothing. The original
code checked if the combined error/length was <= 0 but in the
new code that accidentally got changed to just an error check.
Add back the return from the function when nothing is read.
Cc: Al Viro <[email protected]> Fixes: e1200fe68f20 ("9p: switch p9_client_read() to passing struct iov_iter *") Signed-off-by: Johannes Berg <[email protected]> Signed-off-by: Al Viro <[email protected]>
Chris Mason [Fri, 24 Apr 2015 18:00:00 +0000 (11:00 -0700)]
Btrfs: prevent list corruption during free space cache processing
__btrfs_write_out_cache is holding the ctl->tree_lock while it prepares
a list of bitmaps to record in the free space cache. It was dropping
the lock while it worked on other components, which made a window for
free_bitmap() to free the bitmap struct without removing it from the
list.
This changes things to hold the lock the whole time, and also makes sure
we hold the lock during enospc cleanup.
projects / linux.git / log