Philippe Longepe [Tue, 10 Nov 2015 01:40:47 +0000 (17:40 -0800)]
cpufreq: intel_pstate: Add separate support for Airmont cores
There are two flavors of Atom cores to be supported by intel_pstate,
Silvermont and Airmont, so make the driver distinguish between them by
adding separate frequency tables.
Separate the CPU defaults params for each of them and match the CPU IDs
against them as appropriate.
Philippe Longepe [Tue, 10 Nov 2015 01:40:46 +0000 (17:40 -0800)]
cpufreq: intel_pstate: Replace BYT with ATOM
Rename symbol and function names starting with "BYT" or "byt" to
start with "ATOM" or "atom", respectively, so as to make it clear
that they may apply to Atom in general and not just to Baytrail
(the goal is to support several Atoms architectures eventually).
Revert "cpufreq: intel_pstate: Use ACPI perf configuration"
Revert commit 37afb0003242 (cpufreq: intel_pstate: Use ACPI perf
configuration) that is reported to cause a regression to happen
on a system where invalid data are returned by the ACPI _PSS object.
Since that commit makes assumptions regarding the _PSS output
correctness that may turn out to be overly optimistic in general,
there is a concern that it may introduce regression on more
systems, so it's better to revert it now and we'll revisit the
underlying issue in the next cycle with a more robust solution.
Conflicts:
drivers/cpufreq/intel_pstate.c
Fixes: 37afb0003242 (cpufreq: intel_pstate: Use ACPI perf configuration) Reported-by: Borislav Petkov <[email protected]> Signed-off-by: Rafael J. Wysocki <[email protected]>
Revert "cpufreq: intel_pstate: Avoid calculation for max/min"
Revert commit 4ef451487019 (cpufreq: intel_pstate: Avoid calculation for
max/min) as it depends on commit 37afb0003242 (cpufreq: intel_pstate: Use
ACPI perf configuration) that causes problems to happen and needs to be
reverted.
Glen Lee [Thu, 5 Nov 2015 09:51:23 +0000 (18:51 +0900)]
Revert "Staging: wilc1000: coreconfigurator: Drop unneeded wrapper functions"
The source and destination pointers are misplaced.
This will be like,
ether_addr_copy(data, bssid + ADDR2);
-> ether_addr_copy(bssid, data + ADDR2);
and also to use ether_addr_copy, it has to be proved that src/dst address are
properly aligned(2).
Merge tag 'iio-fixes-for-4.4a' of ssh://ra.kernel.org/pub/scm/linux/kernel/git/jic23/iio into staging-linus
Jonathan writes:
First set of IIO fixes for the 4.4 cycle.
This set does not include those for issues introduced during the merge
window. Fixes of those will follow in a future series.
* ad5064
- Make sure the local i2c_write returns 0 on success rather than the
number of bytes transfered. Otherwise we report an error on all writes.
- Fix a shift for ad5629 and ad5669 which gives incorrect DAC output on
these parts.
* ad7793
- The product ID on the datasheet is wrong. Fix it in the driver.
* IIO_DUMMY_EVGEN
- select IRQ_WORK as a dependency.
* lpc32xx
- make sure clock is prepared before enabling.
* si7020
- data byte order was reversed. Fix it.
* vf610
- Internal temperature calculation was wrong if a different
reference voltage was used. Now use a linear interpolation
function to make it work over the full range.
- Fix a division by zero in the case of a device tree property
not being present (same issue two fixes).
* xilinx XADC
- VREFN scale was wrong - fix it.
This cause an allmodconfig build failure, since our underlying
load-acquire implementation does not handle const types correctly:
include/net/sock.h: In function 'sk_state_load':
./arch/arm64/include/asm/barrier.h:71:3: error: read-only variable '___p1' used as 'asm' output
asm volatile ("ldarb %w0, %1" \
This patch fixes the problem by reusing the trick in READ_ONCE that
loads via a non-const member of an anonymous union. This has the
advantage of allowing us to use smp_load_acquire on packed structures
(e.g. arch_spinlock_t) as well as primitive types.
Mathias Nyman [Wed, 18 Nov 2015 08:48:22 +0000 (10:48 +0200)]
xhci: Fix a race in usb2 LPM resume, blocking U3 for usb2 devices
Clear device initiated resume variables once device is fully up and running
in U0 state.
Resume needs to be signaled for 20ms for usb2 devices before they can be
moved to U0 state.
An interrupt is triggered if a device initiates resume. As we handle the
event in interrupt context we can not sleep for 20ms, so we instead set
a resume flag, a timestamp, and start the roothub polling.
The roothub code will later move the port to U0 when it finds a port in
resume state with the resume flag set, and timestamp passed by 20ms.
A host initiated resume is however not done in interrupt context, and
host initiated resume code will directly signal resume, wait 20ms and then
move the port to U0.
These two codepaths can race, if we are in the middle of a host initated
resume, while sleeping for 20ms, we may handle a port event and find the
port in resume state. The port event handling code will assume the resume
was device initiated and set the resume flag and timestamp.
Root hub code will however not catch the port in resume state again as the
host initated resume code has already moved the port to U0.
The resume flag and timestamp will remain set for this port preventing port
from suspending again (LPM setting port to U3)
Fix this for now by always clearing the device initated resume parameters
once port is in U0
Lu Baolu [Wed, 18 Nov 2015 08:48:21 +0000 (10:48 +0200)]
usb: xhci: fix checking ep busy for CFC
Function ep_ring_is_processing() checks the dequeue pointer
in endpoint context to know whether an endpoint is busy with
processing TRBs. This is not correct since dequeue pointer
field in an endpoint context is only valid when the endpoint
is in Halted or Stopped states. This buggy code causes audio
noise when playing sound with USB headset connected to host
controllers which support CFC (one of xhci 1.1 features).
This patch should exist in stable kernel since v4.3.
Rajmohan Mani [Wed, 18 Nov 2015 08:48:20 +0000 (10:48 +0200)]
xhci: Workaround to get Intel xHCI reset working more reliably
Existing Intel xHCI controllers require a delay of 1 mS,
after setting the CMD_RESET bit in command register, before
accessing any HC registers. This allows the HC to complete
the reset operation and be ready for HC register access.
Without this delay, the subsequent HC register access,
may result in a system hang, very rarely.
Verified CherryView / Braswell platforms go through over
5000 warm reboot cycles (which was not possible without
this patch), without any xHCI reset hang.
Linus Torvalds [Wed, 18 Nov 2015 16:59:29 +0000 (08:59 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Pull s390 fixes from Martin Schwidefsky:
"Assorted bug fixes, the mlock2 system call gets added, and one
improvement. The boot from dasd devices is now possible from a wider
range of devices"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390: remove SALIPL loader
s390: wire up mlock2 system call
s390: remove g5 elf platform support
s390: avoid cache aliasing under z/VM and KVM
s390/sclp: _sclp_wait_int(): retain full PSW mask
s390/zcrypt: Fix initialisation when zcrypt is built-in
s390/zcrypt: Fix kernel crash on systems without AP bus support
s390: add support for ipl devices in subchannel sets > 0
s390/ipl: fix out of bounds access in scpdata_write
s390/pci_dma: improve debugging of errors during dma map
s390/pci_dma: handle dma table failures
s390/pci_dma: unify label of invalid translation table entries
s390/syscalls: remove system call number calculation
s390/cio: simplify css_generate_pgid
s390/diag: add a s390 prefix to the diagnose trace point
s390/head: fix error message on unsupported hardware
Linus Torvalds [Wed, 18 Nov 2015 16:43:29 +0000 (08:43 -0800)]
Merge tag 'hwmon-for-linus-v4.4-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging
Pull hwmon fixes from Guenter Roeck:
"Fix build issues in scpi and ina2xx drivers, update scpi driver to
support recent firmware, and fix an uninitialized variable warning in
applesmc driver"
drm/i915: Consider SPLL as another shared pll, v2.
When diagnosing a unrelated bug for someone on irc, it would seem the hardware can
be brought up by the BIOS with the embedded displayport using the SPLL for spread spectrum.
Right now this is not handled well in i915, and it calculates the crtc needs to
be reprogrammed on the first modeset without SSC, but the SPLL itself was kept
active. Fix this by exposing SPLL as a shared pll that will not be returned
by intel_get_shared_dpll; you have to know it exists to use it.
Changes since v1:
- Create a separate dpll_hw_state.spll for spll, and use
separate pll functions for spll.
Laura Abbott [Thu, 12 Nov 2015 20:21:10 +0000 (12:21 -0800)]
arm64: Fix R/O permissions in mark_rodata_ro
The permissions in mark_rodata_ro trigger a build error
with STRICT_MM_TYPECHECKS. Fix this by introducing
PAGE_KERNEL_ROX for the same reasons as PAGE_KERNEL_RO.
From Ard:
"PAGE_KERNEL_EXEC has PTE_WRITE set as well, making the range
writeable under the ARMv8.1 DBM feature, that manages the
dirty bit in hardware (writing to a page with the PTE_RDONLY
and PTE_WRITE bits both set will clear the PTE_RDONLY bit in that case)"
Ard Biesheuvel [Mon, 16 Nov 2015 12:12:48 +0000 (13:12 +0100)]
arm64: crypto: reduce priority of core AES cipher
The asynchronous, merged implementations of AES in CBC, CTR and XTS
modes are preferred when available (i.e., when instantiating ablkciphers
explicitly). However, the synchronous core AES cipher combined with the
generic CBC mode implementation will produce a 'cbc(aes)' blkcipher that
is callable asynchronously as well. To prevent this implementation from
being used when the accelerated asynchronous implemenation is also
available, lower its priority to 250 (i.e., below the asynchronous
module's priority of 300).
Ard Biesheuvel [Tue, 17 Nov 2015 08:53:31 +0000 (09:53 +0100)]
arm64: use non-global mappings for UEFI runtime regions
As pointed out by Russell King in response to the proposed ARM version
of this code, the sequence to switch between the UEFI runtime mapping
and current's actual userland mapping (and vice versa) is potentially
unsafe, since it leaves a time window between the switch to the new
page tables and the TLB flush where speculative accesses may hit on
stale global TLB entries.
So instead, use non-global mappings, and perform the switch via the
ordinary ASID-aware context switch routines.
Mika Kuoppala [Tue, 17 Nov 2015 16:14:26 +0000 (18:14 +0200)]
drm/i915: Fix gpu frequency change tracing
With gen < 9 we have had always 50Mhz units as our hw
ratio. With gen >= 9 the hw ratio changed to 16.667Mhz (50/3).
The result was that our gpu frequency tracing started to output
values 3 times larger than expected due to hardcoded scaling
value. Fix this by using Use intel_gpu_freq() when generating Mhz
value from ratio for 'intel_gpu_freq_change' trace event.
Li Jun [Tue, 13 Oct 2015 10:23:31 +0000 (18:23 +0800)]
usb: chipidea: debug: disable usb irq while role switch
Since the ci->role will be set after the host role start is complete, there
will be nobody cared irq during start host if usb irq enabled. This error
can be reproduced on i.mx6 sololite EVK board by:
1. disable otg id irq(IDIE) and disable all real otg properties of usbotg1
in dts.
2. boot up the board with ID cable and usb device connected.
3. echo gadget > /sys/kernel/debug/ci_hdrc.0/role
4. echo host > /sys/kernel/debug/ci_hdrc.0/role
5. irq 212: nobody cared.
Peter Chen [Wed, 16 Sep 2015 01:35:06 +0000 (09:35 +0800)]
ARM: dts: imx27.dtsi: change the clock information for usb
For imx27, it needs three clocks to let the controller work,
the old code is wrong, and usbmisc has not included clock handling
code any more. Without this patch, it will cause below data
abort when accessing usbmisc registers.
usbcore: registered new interface driver usb-storage
Unhandled fault: external abort on non-linefetch (0x008) at 0xf4424600
pgd = c0004000
[f4424600] *pgd=10000452(bad)
Internal error: : 8 [#1] PREEMPT ARM
Modules linked in:
CPU: 0 PID: 1 Comm: swapper Not tainted 4.1.0-next-20150701-dirty #3089
Hardware name: Freescale i.MX27 (Device Tree Support)
task: c7832b60 ti: c783e000 task.ti: c783e000
PC is at usbmisc_imx27_init+0x4c/0xbc
LR is at usbmisc_imx27_init+0x40/0xbc
pc : [<c03cb5c0>] lr : [<c03cb5b4>] psr: 60000093
sp : c783fe08 ip : 00000000 fp : 00000000
r10: c0576434 r9 : 0000009c r8 : c7a773a0
r7 : 01000000 r6 : 60000013 r5 : c7a776f0 r4 : c7a773f0
r3 : f4424600 r2 : 00000000 r1 : 00000001 r0 : 00000001
Flags: nZCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel
Control: 0005317f Table: a0004000 DAC: 00000017
Process swapper (pid: 1, stack limit = 0xc783e190)
Stack: (0xc783fe08 to 0xc7840000)
Peter Chen [Wed, 16 Sep 2015 01:40:51 +0000 (09:40 +0800)]
usb: chipidea: imx: refine clock operations to adapt for all platforms
Some i.mx platforms need three clocks to let controller work, but
others only need one, refine clock operation to adapt for all
platforms, it fixes a regression found at i.mx27.
Ingo Molnar [Wed, 18 Nov 2015 05:56:48 +0000 (06:56 +0100)]
Merge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent
Pull perf/urgent fixes from Arnaldo Carvalho de Melo:
- Do not change the key of an object in a rbtree, this time it was
the one for DSOs lookup by its long_name, and the noticed symptom was
with 'perf buildid-list --with-hits' (Adrian Hunter)
- 'perf inject' is a pipe, events it doesn't touch should be passed
on, PERF_RECORD_LOST wasn't, fix it (Adrian Hunter)
- Make 'perf buildid-list' request event ordering, as it needs to
first get the mmap events to be able to mark wich DSOs had hits
(Adrian Hunter)
- Fix memory leaks on failure in 'perf probe' (Masami Hiramatsu, Wang Nan)
kernel-doc: Make it compatible with Perl versions below 5.12 again
Changeset 4d73270192ec('scripts/kernel-doc: Replacing highlights
hash by an array') broke compatibility of the kernel-doc script with
older versions of perl by using "keys ARRAY" syntax with is available
only on Perl 5.12 or newer, according with:
http://perldoc.perl.org/functions/keys.html
Restore backward compatibility by replacing "foreach my $k (keys ARRAY)"
by a C-like variant: "for (my $k = 0; $k < !ARRAY; $k++)"
Merge tag 'fixes-for-v4.4-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-linus
Felipe writes:
usb: fixes for v4.4-rc2
First round of fixes for this -rc cycle. We have the
usual set of miscellaneous fixes. The important
thing here is support for Intel Broxton SoC on dwc3,
some fixes for Rockchip SoCs on dwc2 and a fix on
dwc3 to let it report lower speeds than
USB_SPEED_SUPER.
1) Fix list tests in netfilter ingress support, from Florian Westphal.
2) Fix reversal of input and output interfaces in ingress hook
invocation, from Pablo Neira Ayuso.
3) We have a use after free in r8169, caught by Dave Jones, fixed by
Francois Romieu.
4) Splice use-after-free fix in AF_UNIX frmo Hannes Frederic Sowa.
5) Three ipv6 route handling bug fixes from Martin KaFai Lau:
a) Don't create clone routes not managed by the fib6 tree
b) Don't forget to check expiration of DST_NOCACHE routes.
c) Handle rt->dst.from == NULL properly.
6) Several AF_PACKET fixes wrt transport header setting and SKB
protocol setting, from Daniel Borkmann.
7) Fix thunder driver crash on shutdown, from Pavel Fedin.
8) Several Mellanox driver fixes (max MTU calculations, use of correct
DMA unmap in TX path, etc.) from Saeed Mahameed, Tariq Toukan, Doron
Tsur, Achiad Shochat, Eran Ben Elisha, and Noa Osherovich.
9) Several mv88e6060 DSA driver fixes (wrong bit definitions for
certain registers, etc.) from Neil Armstrong.
10) Make sure to disable preemption while updating per-cpu stats of ip
tunnels, from Jason A. Donenfeld.
11) Various ARM64 bpf JIT fixes, from Yang Shi.
12) Flush icache properly in ARM JITs, from Daniel Borkmann.
13) Fix masking of RX and TX interrupts in ravb driver, from Masaru
Nagai.
14) Fix netdev feature propagation for devices not implementing
->ndo_set_features(). From Nikolay Aleksandrov.
15) Big endian fix in vmxnet3 driver, from Shrikrishna Khare.
16) RAW socket code increments incorrect SNMP counters, fix from Ben
Cartwright-Cox.
17) IPv6 multicast SNMP counters are bumped twice, fix from Neil Horman.
18) Fix handling of VLAN headers on stacked devices when REORDER is
disabled. From Vlad Yasevich.
19) Fix SKB leaks and use-after-free in ipvlan and macvlan drivers, from
Sabrina Dubroca.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (83 commits)
MAINTAINERS: Update Mellanox's Eth NIC driver entries
net/core: revert "net: fix __netdev_update_features return.." and add comment
af_unix: take receive queue lock while appending new skb
rtnetlink: fix frame size warning in rtnl_fill_ifinfo
net: use skb_clone to avoid alloc_pages failure.
packet: Use PAGE_ALIGNED macro
packet: Don't check frames_per_block against negative values
net: phy: Use interrupts when available in NOLINK state
phy: marvell: Add support for 88E1540 PHY
arm64: bpf: make BPF prologue and epilogue align with ARM64 AAPCS
macvlan: fix leak in macvlan_handle_frame
ipvlan: fix use after free of skb
ipvlan: fix leak in ipvlan_rcv_frame
vlan: Do not put vlan headers back on bridge and macvlan ports
vlan: Fix untag operations of stacked vlans with REORDER_HEADER off
via-velocity: unconditionally drop frames with bad l2 length
ipg: Remove ipg driver
dl2k: Add support for IP1000A-based cards
snmp: Remove duplicate OUTMCAST stat increment
net: thunder: Check for driver data in nicvf_remove()
...
Dan Carpenter [Wed, 4 Nov 2015 13:21:40 +0000 (16:21 +0300)]
drm/vc4: checking for NULL instead of IS_ERR
vc4_plane_init() returns an ERR_PTR on error, it doesn't return NULL.
This was obviously intended because the next lines call
PTR_ERR(primary_plane) already.
net/core: revert "net: fix __netdev_update_features return.." and add comment
This reverts commit 00ee59271777 ("net: fix __netdev_update_features return
on ndo_set_features failure")
and adds a comment explaining why it's okay to return a value other than
0 upon error. Some drivers might actually change flags and return an
error so it's better to fire a spurious notification rather than miss
these.
af_unix: take receive queue lock while appending new skb
While possibly in future we don't necessarily need to use
sk_buff_head.lock this is a rather larger change, as it affects the
af_unix fd garbage collector, diag and socket cleanups. This is too much
for a stable patch.
For the time being grab sk_buff_head.lock without disabling bh and irqs,
so don't use locked skb_queue_tail.
rtnetlink: fix frame size warning in rtnl_fill_ifinfo
Fix the following warning:
CC net/core/rtnetlink.o
net/core/rtnetlink.c: In function ‘rtnl_fill_ifinfo’:
net/core/rtnetlink.c:1308:1: warning: the frame size of 2864 bytes is larger than 2048 bytes [-Wframe-larger-than=]
}
^
by splitting up the huge rtnl_fill_ifinfo into some smaller ones, so we
don't have the huge frame allocations at the same time.
Andrew Lunn [Mon, 16 Nov 2015 22:36:46 +0000 (23:36 +0100)]
net: phy: Use interrupts when available in NOLINK state
The NOLINK state will poll the phy once a second to see if the link
has come up. If the phy has an interrupt line, this polling can be
skipped, since the phy should interrupt when the link returns.
The 88E1540 can be found embedded in the Marvell 88E6352 switch. It
is compatible with the 88E1510, so add support for it, using the 88E1510 specific functions.
Yang Shi [Mon, 16 Nov 2015 22:35:35 +0000 (14:35 -0800)]
arm64: bpf: make BPF prologue and epilogue align with ARM64 AAPCS
Save and restore FP/LR in BPF prog prologue and epilogue, save SP to FP
in prologue in order to get the correct stack backtrace.
However, ARM64 JIT used FP (x29) as eBPF fp register, FP is subjected to
change during function call so it may cause the BPF prog stack base address
change too.
Use x25 to replace FP as BPF stack base register (fp). Since x25 is callee
saved register, so it will keep intact during function call.
It is initialized in BPF prog prologue when BPF prog is started to run
everytime. Save and restore x25/x26 in BPF prologue and epilogue to keep
them intact for the outside of BPF. Actually, x26 is unnecessary, but SP
requires 16 bytes alignment.
Sabrina Dubroca [Mon, 16 Nov 2015 21:54:20 +0000 (22:54 +0100)]
macvlan: fix leak in macvlan_handle_frame
Reset pskb in macvlan_handle_frame in case skb_share_check returned a
clone.
Fixes: 8a4eb5734e8d ("net: introduce rx_handler results and logic around that") Signed-off-by: Sabrina Dubroca <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Sabrina Dubroca [Mon, 16 Nov 2015 21:44:53 +0000 (22:44 +0100)]
ipvlan: fix use after free of skb
ipvlan_handle_frame is a rx_handler, and when it returns a value other
than RX_HANDLER_CONSUMED (here, NET_RX_DROP aka RX_HANDLER_ANOTHER),
__netif_receive_skb_core expects that the skb still exists and will
process it further, but we just freed it.
Fixes: 2ad7bf363841 ("ipvlan: Initial check-in of the IPVLAN driver.") Signed-off-by: Sabrina Dubroca <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Sabrina Dubroca [Mon, 16 Nov 2015 21:34:26 +0000 (22:34 +0100)]
ipvlan: fix leak in ipvlan_rcv_frame
Pass a **skb to ipvlan_rcv_frame so that if skb_share_check returns a
new skb, we actually use it during further processing.
It's safe to ignore the new skb in the ipvlan_xmit_* functions, because
they call ipvlan_rcv_frame with local == true, so that dev_forward_skb
is called and always takes ownership of the skb.
Fixes: 2ad7bf363841 ("ipvlan: Initial check-in of the IPVLAN driver.") Signed-off-by: Sabrina Dubroca <[email protected]> Signed-off-by: David S. Miller <[email protected]>
David S. Miller [Tue, 17 Nov 2015 19:38:36 +0000 (14:38 -0500)]
Merge branch 'vlan-reorder'
Vladislav Yasevich says:
====================
Fix issues with vlans without REORDER_HEADER
A while ago Phil Sutter brought up an issue with vlans without
REORDER_HEADER and bridges. The problem was that if a vlan
without REORDER_HEADER was a port in the bridge, the bridge ended
up forwarding corrupted packets that still contained the vlan header.
The same issue exists for bridge mode macvlan/macvtap devices.
An additional issue with vlans without REORDER_HEADER is that stacking
them also doesn't work. The reason here is that skb_reorder_vlan_header()
function assumes that it on ETH_HLEN bytes deep into the packet. That
is not the case, when you a vlan without REORRDER_HEADER flag set.
This series attempts to correct these 2 issues.
1) To solve the stacked vlans problem, the patch simply use
skb->mac_len as an offset to start copying mac addresses that
is part of header reordering.
2) To fix the issue with bridge/macvlan/macvtap, the second patch
simply doesn't write the vlan header back to the packet if the
vlan device is either a bridge or a macvlan port. This ends up
being the simplest and least performance intrussive solution.
I've considered extending patch 2 to all stacked devices (essentially
checked for the presense of rx_handler), but that feels like a broader
restriction and _may_ break existing uses.
====================
Vlad Yasevich [Mon, 16 Nov 2015 20:43:45 +0000 (15:43 -0500)]
vlan: Do not put vlan headers back on bridge and macvlan ports
When a vlan is configured with REORDER_HEADER set to 0, the vlan
header is put back into the packet and makes it appear that
the vlan header is still there even after it's been processed.
This posses a problem for bridge and macvlan ports. The packets
passed to those device may be forwarded and at the time of the
forward, vlan headers end up being unexpectedly present.
With the patch, we make sure that we do not put the vlan header
back (when REORDER_HEADER is 0) if a bridge or macvlan has
been configured on top of the vlan device.
Vlad Yasevich [Mon, 16 Nov 2015 20:43:44 +0000 (15:43 -0500)]
vlan: Fix untag operations of stacked vlans with REORDER_HEADER off
When we have multiple stacked vlan devices all of which have
turned off REORDER_HEADER flag, the untag operation does not
locate the ethernet addresses correctly for nested vlans.
The reason is that in case of REORDER_HEADER flag being off,
the outer vlan headers are put back and the mac_len is adjusted
to account for the presense of the header. Then, the subsequent
untag operation, for the next level vlan, always use VLAN_ETH_HLEN
to locate the begining of the ethernet header and that ends up
being a multiple of 4 bytes short of the actuall beginning
of the mac header (the multiple depending on the how many vlan
encapsulations ethere are).
As a reslult, if there are multiple levles of vlan devices
with REODER_HEADER being off, the recevied packets end up
being dropped.
To solve this, we use skb->mac_len as the offset. The value
is always set on receive path and starts out as a ETH_HLEN.
The value is also updated when the vlan header manupations occur
so we know it will be correct.
Timo Teräs [Mon, 16 Nov 2015 12:36:32 +0000 (14:36 +0200)]
via-velocity: unconditionally drop frames with bad l2 length
By default the driver allowed incorrect frames to be received. What is
worse the code does not handle very short frames correctly. The FCS
length is unconditionally subtracted, and the underflow can cause
skb_put to be called with large number after implicit cast to unsigned.
And indeed, an skb_over_panic() was observed with via-velocity.
This removes the module parameter as it does not work in it's
current state, and should be implemented via NETIF_F_RXALL if needed.
Linus Torvalds [Tue, 17 Nov 2015 18:11:08 +0000 (10:11 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull vfs fixes from Al Viro:
"A fs-cache regression fix, and adding a warning about obnoxiou^W
moderation of list given in MAINTAINERS"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
MAINTAINERS: [email protected] is moderated for non-subscribers
FS-Cache: Add missing initialization of ret in cachefiles_write_page()
Following changes that appeared in lk 4.0.0, the gadget udc driver for
some ARM based Atmel SoCs (e.g. at91sam9x5 and sama5d3 families)
incorrectly deduced full-speed USB link speed even when the hardware
had negotiated a high-speed link. The fix is to make sure that the
UDPHS Interrupt Enable Register value does not mask the SPEED bit
in the Interrupt Status Register.
For a mass storage gadget this problem lead to failures when the host
had a USB 3 port with the xhci_hcd driver. If the host was a USB 2
port using the ehci_hcd driver then the mass storage gadget worked
(but probably at a lower speed than it should have).
Ben McCauley [Mon, 16 Nov 2015 16:47:24 +0000 (10:47 -0600)]
usb: dwc3: gadget: let us set lower max_speed
In some SoCs, dwc3 is implemented as a USB2.0 only
core, meaning that it can't ever achieve SuperSpeed.
Currect driver always sets gadget.max_speed to
USB_SPEED_SUPER unconditionally. This can causes
issues to some Host stacks where the host will issue
a GetBOS() request and we will reply with a BOS
containing Superspeed Capability Descriptor.
At least Windows seems to be upset by this fact and
prints a warning that we should connect $this device
to another port.
[ [email protected] : rewrote entire commit, including
source code comment to make a lot clearer what the
problem is ]
Bin Liu [Fri, 6 Nov 2015 18:08:56 +0000 (12:08 -0600)]
usb: musb: fix tx fifo flush handling
Here are a few changes in musb_h_tx_flush_fifo().
- It has been observed that sometimes (if not always) musb is unable
to flush tx fifo during urb dequeue when disconnect a device. But
it seems to be harmless, since the tx fifo flush is done again in
musb_ep_program() when re-use the hw_ep.
But the WARN() floods the console in the case when multiple tx urbs
are queued, so change it to dev_WARN_ONCE().
- applications could queue up many tx urbs, then the 1ms delay could
causes minutes of delay in device disconnect. So remove it to get
better user experience. The 1ms delay does not help the flushing
anyway.
- cleanup the debug code - related to lastcsr.
----
Note: The tx fifo flush issue has been observed during device disconnect
on AM335x.
To reproduce the issue, ensure tx urb(s) are queued when unplug the usb
device which is connected to AM335x usb host port.
I found using a usb-ethernet device and running iperf (client on AM335x)
has very high chance to trigger the problem.
Better to turn on dev_dbg() in musb_cleanup_urb() with CPPI enabled to
see the issue when aborting the tx channel.
Peter Chen [Tue, 3 Nov 2015 07:57:38 +0000 (15:57 +0800)]
usb: gadget: f_loopback: fix the warning during the enumeration
The current code tries to allocate memory with GFP_KERNEL at
interrupt context, it would show below warning during the enumeration
when I test it with chipidea hardware, change GFP flag as GFP_ATOMIC
can fix this issue.
Douglas Anderson [Thu, 22 Oct 2015 20:05:03 +0000 (13:05 -0700)]
usb: dwc2: host: Fix remote wakeup when not in DWC2_L2
In commit 734643dfbdde ("usb: dwc2: host: add flag to reflect bus
state") we changed dwc2_port_suspend() not to set the lx_state
anymore (instead it sets the new bus_suspended variable). This
introduced a bug where we would fail to detect device insertions if:
1. Plug empty hub into dwc2
2. Plug USB flash drive into the empty hub.
3. Wait a few seconds
4. Unplug USB flash drive
5. Less than 2 seconds after step 4, plug the USB flash drive in again.
The dwc2_hcd_rem_wakeup() function should have been changed to look at
the new bus_suspended variable.
Let's fix it. Since commit b46146d59fda ("usb: dwc2: host: resume root
hub on remote wakeup") talks about needing the root hub resumed if the
bus was suspended, we'll include it in our test.
It appears that the "port_l1_change" should only be set to 1 if we were
in DWC2_L1 (the driver currently never sets this), so we'll update the
former "else" case based on this test.
Douglas Anderson [Tue, 20 Oct 2015 23:33:53 +0000 (16:33 -0700)]
usb: dwc2: host: Fix ahbcfg for rk3066
The comment for ahbcfg for rk3066 parameters (also used for rk3288)
claimed that ahbcfg was INCR16, but it wasn't. Since the bits weren't
shifted properly, the 0x7 ended up being masked and we ended up
programming 0x3 for the HBstLen. Let's set it to INCR16 properly.
As per Wu Liang Feng at Rockchip this may increase transmission
efficiency. I did blackbox tests with writing 0s to a USB-based SD
reader (forcefully capping CPU Freq to try to measure efficiency):
cd /sys/devices/system/cpu/cpu0/cpufreq
echo userspace > scaling_governor
echo 126000 > scaling_setspeed
for i in $(seq 10); do
dd if=/dev/zero of=/dev/sdb bs=1M count=750
done
With the above tests I found that speeds went from ~15MB/s to ~18MB/s.
Note that most other tests I did (including reading from the same USB
reader) didn't show any difference in performance.
arm64: kernel: pause/unpause function graph tracer in cpu_suspend()
The function graph tracer adds instrumentation that is required to trace
both entry and exit of a function. In particular the function graph
tracer updates the "return address" of a function in order to insert
a trace callback on function exit.
Kernel power management functions like cpu_suspend() are called
upon power down entry with functions called "finishers" that are in turn
called to trigger the power down sequence but they may not return to the
kernel through the normal return path.
When the core resumes from low-power it returns to the cpu_suspend()
function through the cpu_resume path, which leaves the trace stack frame
set-up by the function tracer in an incosistent state upon return to the
kernel when tracing is enabled.
This patch fixes the issue by pausing/resuming the function graph
tracer on the thread executing cpu_suspend() (ie the function call that
subsequently triggers the "suspend finishers"), so that the function graph
tracer state is kept consistent across functions that enter power down
states and never return by effectively disabling graph tracer while they
are executing.
Ville Syrjälä [Wed, 11 Nov 2015 17:11:28 +0000 (19:11 +0200)]
drm/i915: Don't clobber the addfb2 ioctl params
We try to convert the old way of of specifying fb tiling (obj->tiling)
into the new fb modifiers. We store the result in the passed in mode_cmd
structure. But that structure comes directly from the addfb2 ioctl, and
gets copied back out to userspace, which means we're clobbering the
modifiers that the user provided (all 0 since the DRM_MODE_FB_MODIFIERS
flag wasn't even set by the user). Hence if the user reuses the struct
for another addfb2, the ioctl will be rejected since it's now asking for
some modifiers w/o the flag set.
Fix the problem by making a copy of the user provided structure. We can
play any games we want with the copy.
Ville Syrjälä [Mon, 16 Nov 2015 15:02:34 +0000 (17:02 +0200)]
drm: Fix primary plane size for stereo doubled modes for legacy setcrtc
Properly double the hdisplay/vdisplay timings that we use as the primary
plane size with stereo doubled modes. Otherwise the modeset gets
rejected on machines where the primary plane must be fullscreen, and on
the rest only the first eye would get a visible plane.
drm/i915: Clear intel_crtc->atomic before updating it.
If an atomic update fails intel_crtc->atomic may have have some values left
from the last atomic check. One example is atomic->wait_for_vblank,
which results in spurious errors in kms_flip.
Marc Zyngier [Mon, 16 Nov 2015 19:13:28 +0000 (19:13 +0000)]
irqchip/gic: Add save/restore of the active state
When using EOImode==1, we may mark interrupts as being forwarded
to a virtual machine. In that case, the interrupt is left active
while being passed to the VM.
If we suspend the system before the VM has deactivated the interrupt,
the active state will be lost (which may be very annoying, as this
may result in spurious interrupts and a confused guest).
To avoid this, save and restore the active state together with the
rest of the GIC registers.
Marc Zyngier [Mon, 16 Nov 2015 19:13:27 +0000 (19:13 +0000)]
irqchip/gic: Clear enable bits before restoring them
When restoring the GIC state (after a suspend/resume cycle,
for example), the driver directly writes the 'enabled' state
it has saved by accessing GICD_ISENABLERn, which performs
an OR operation between the value present in the register
and the value we write.
If whatever code that has run before we reentered the kernel
has enabled an interrupt that was previously disabled, we won't
restore that disabled state.
Making sure we first clear the register (by writting to
GICD_ICENABLERn) before restoring the enabled state.
Marc Zyngier [Mon, 16 Nov 2015 19:13:26 +0000 (19:13 +0000)]
irqchip/gic: Make sure all interrupts are deactivated at boot
When booting a GIC/GICv3 based system, we have no idea what
state the firmware (or previous kernel in the case of kexec)
has left the GIC, and some interrupts may still be active.
In order to garantee that we have a clean state, make sure
the active bits are cleared at init time.
This uses a slightly different way to express current_pt_regs()
that avoids the use of the header and gets away with the already
included asm/ptrace.h.
However, it looks like the dependency should not even there as
I do not see why __generic_dma_ops() cares about whether we have
an ACPI based system or not.
The current behavior is to fall back to the global dma_ops when
a device has not set its own dma_ops, but only for DT based systems.
This seems dangerous, as a random device might have different
requirements regarding IOMMU or coherency, so we should really
never have that fallback and just forbid DMA when we have not
initialized DMA for a device.
This removes the global dma_ops variable and the special-casing
for ACPI, and just returns the dma ops that got set for the
device, or the dummy_dma_ops if none were present.
The original code has apparently been copied from arm32 where we
rely on it for ISA devices things like the floppy controller, but
we should have no such devices on ARM64.
Ard Biesheuvel [Mon, 16 Nov 2015 10:18:14 +0000 (11:18 +0100)]
arm64: mm: use correct mapping granularity under DEBUG_RODATA
When booting a 64k pages kernel that is built with CONFIG_DEBUG_RODATA
and resides at an offset that is not a multiple of 512 MB, the rounding
that occurs in __map_memblock() and fixup_executable() results in
incorrect regions being mapped.
The following snippet from /sys/kernel/debug/kernel_page_tables shows
how, when the kernel is loaded 2 MB above the base of DRAM at 0x40000000,
the first 2 MB of memory (which may be inaccessible from non-secure EL1
or just reserved by the firmware) is inadvertently mapped into the end of
the module region.
The same issue is likely to occur on 16k pages kernels whose load
address is not a multiple of 32 MB (i.e., SECTION_SIZE). So round to
SWAPPER_BLOCK_SIZE instead of SECTION_SIZE.
drm/core: Fix old_fb handling in drm_mode_atomic_ioctl.
plane_mask should be cleared inside the retry loop, because it gets
reset on every retry. Without this fix the plane->fb refcounting might
get out of sync on retries, resulting in either leaked memory or
use-after-free.
drm/core: Set legacy_cursor_update in drm_atomic_helper_disable_plane.
legacy_cursor_update was being set in restore_fbdev_mode_atomic which was
probably unintended. Fix this by only setting it in the function that needs it.
All of these paths are interpreting a dax pmd mapping as a transparent
huge page and making the assumption that the pfn is covered by the
memmap, i.e. that the pfn has an associated struct page. PTE mappings
do not suffer the same fate since they have the _PAGE_SPECIAL flag to
cause the gup path to fault. We can do something similar for the PMD
path, or otherwise defer pmd support for cases where a struct page is
available. For now, 4.4-rc and -stable need to disable dax pmd support
by default.
For development the "depends on BROKEN" line can be removed from
CONFIG_FS_DAX_PMD.
FS-Cache: Add missing initialization of ret in cachefiles_write_page()
fs/cachefiles/rdwr.c: In function ‘cachefiles_write_page’:
fs/cachefiles/rdwr.c:882: warning: ‘ret’ may be used uninitialized in
this function
If the jump to label "error" is taken, "ret" will indeed be
uninitialized, and random stack data may be printed by the debug code.
Fixes: 102f4d900c9c8f5e ("FS-Cache: Handle a write to the page immediately beyond the EOF marker") Signed-off-by: Geert Uytterhoeven <[email protected]> Signed-off-by: David Howells <[email protected]> Signed-off-by: Al Viro <[email protected]>
Chris Bainbridge [Thu, 12 Nov 2015 19:24:47 +0000 (19:24 +0000)]
Revert "ACPI / SBS: Add 5 us delay to fix SBS hangs on MacBook"
Revert commit 3349fb64b292 (ACPI / SBS: Add 5 us delay to fix SBS
hangs on MacBook), since the delay introduced by it is not necessary
any more after commit add68d6aa9e2 (ACPI / SMBus: Fix boot stalls /
high CPU caused by reentrant code).
Chris Bainbridge [Thu, 12 Nov 2015 18:05:37 +0000 (18:05 +0000)]
ACPI / SMBus: Fix boot stalls / high CPU caused by reentrant code
In the SBS initialisation, a reentrant call to wait_event_timeout()
causes an intermittent boot stall of several minutes usually following
the "Switching to clocksource tsc" message. Another symptom of this bug
is high CPU usage from programs (Firefox, upowerd) querying the battery
state. This is caused by:
if (wait_event_timeout(hc->wait, smb_check_done(hc),
msecs_to_jiffies(timeout)))
2. ___wait_event sets task state to uninterruptible
3. ___wait_event calls the "condition" smb_check_done()
4. smb_check_done (sbshc.c) calls through to ec_read() in
drivers/acpi/ec.c
5. ec_guard() is reached which calls wait_event_timeout()
if (wait_event_timeout(ec->wait,
ec_transaction_completed(ec),
guard))
ie. wait_event_timeout() is being called again inside evaluation of
the previous wait_event_timeout() condition
5. The EC IRQ handler calls wake_up() and wakes up the sleeping task in
ec_guard()
6. The task is now in state running even though the wait "condition" is
still being evaluated
7. The "condition" check returns false so ___wait_event calls
schedule_timeout()
8. Since the task state is running, the scheduler immediately schedules
it again
9. This loop usually repeats for around 250 seconds even though the
original wait_event_timeout was only 1000ms.
The timeout is incorrect because each call to schedule_timeout()
usually returns immediately, taking less than 1ms, so the jiffies
timeout counter is not decremented. The task is now stuck in a
running state, and so is highly likely to be immediately
rescheduled, which takes less than a jiffy. The loop will never exit
if all schedule_timeout() calls take less than a jiffy.
Fix this by replacing SMBus reads in the wait_event_timeout condition
with checks of a boolean value that is updated by the EC query handler.
Ondrej Zary [Sun, 15 Nov 2015 21:36:11 +0000 (22:36 +0100)]
dl2k: Add support for IP1000A-based cards
Add support for IP1000A chips to dl2k driver.
IP1000A chip looks like a TC9020 with integrated PHY.
This allows IP1000A chips to work reliably because the ipg driver is
buggy - it loses packets under load and then completely stops
transmitting data.
Tested with Asus NX1101 v2.0 at 10, 100 and 1000Mbps:
vendor=0x13f0 device=0x1023 (rev 0x41)
subsystem vendor=0x1043 device=0x8180
MAC address registers access needed to be changed from 8-bit to 16-bit
because 8-bit does not work on IP1000A. 8-bit access is not even
allowed in the TC9020 datasheet (although it worked). 16-bit access
works on both.
Tested that it does not break D-Link DGE-550T (DL-2000 chip, probably
a rebranded TC9020):
vendor=0x1186 device=0x4000 (rev 0x0c)
subsystem vendor=0x1186 device=0x4000
projects / linux.git / log