platform/kernel/linux-rpi.git
4 years agonet: wireless: intel: iwlwifi: fix GRO_NORMAL packet stalling
Alexander Lobakin [Wed, 27 Nov 2019 09:41:23 +0000 (12:41 +0300)]
net: wireless: intel: iwlwifi: fix GRO_NORMAL packet stalling

[ Upstream commit b167191e2a851cb2e4c6ef8b91c83ff73ef41872 ]

Commit 6570bc79c0df ("net: core: use listified Rx for GRO_NORMAL in
napi_gro_receive()") has applied batched GRO_NORMAL packets processing
to all napi_gro_receive() users, including mac80211-based drivers.

However, this change has led to a regression in iwlwifi driver [1][2] as
it is required for NAPI users to call napi_complete_done() or
napi_complete() and the end of every polling iteration, whilst iwlwifi
doesn't use NAPI scheduling at all and just calls napi_gro_flush().
In that particular case, packets which have not been already flushed
from napi->rx_list stall in it until at least next Rx cycle.

Fix this by adding a manual flushing of the list to iwlwifi driver right
before napi_gro_flush() call to mimic napi_complete() logics.

I prefer to open-code gro_normal_list() rather than exporting it for 2
reasons:
* to prevent from using it and napi_gro_flush() in any new drivers,
  as it is the *really* bad way to use NAPI that should be avoided;
* to keep gro_normal_list() static and don't lose any CC optimizations.

I also don't add the "Fixes:" tag as the mentioned commit was only a
trigger that only exposed an improper usage of NAPI in this particular
driver.

[1] https://lore.kernel.org/netdev/PSXP216MB04388962C411CD0B17A86F47804A0@PSXP216MB0438.KORP216.PROD.OUTLOOK.COM
[2] https://bugzilla.kernel.org/show_bug.cgi?id=205647

Signed-off-by: Alexander Lobakin <alobakin@dlink.ru>
Acked-by: Luca Coelho <luciano.coelho@intel.com>
Reported-by: Nicholas Johnson <nicholas.johnson-opensource@outlook.com.au>
Tested-by: Nicholas Johnson <nicholas.johnson-opensource@outlook.com.au>
Reviewed-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoibmvnic: Fix completion structure initialization
Thomas Falcon [Mon, 25 Nov 2019 23:12:53 +0000 (17:12 -0600)]
ibmvnic: Fix completion structure initialization

[ Upstream commit 070eca955c4af1248cb78a216463ff757a5dc511 ]

Fix multiple calls to init_completion for device completion
structures. Instead, initialize them during device probe and
reinitialize them later as needed.

Signed-off-by: Thomas Falcon <tlfalcon@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoRDMA/bnxt_re: Fix chip number validation Broadcom's Gen P5 series
Luke Starrett [Thu, 21 Nov 2019 06:22:21 +0000 (01:22 -0500)]
RDMA/bnxt_re: Fix chip number validation Broadcom's Gen P5 series

[ Upstream commit e284b159c6881c8bec9713daba2653268f4c4948 ]

In the first version of Gen P5 ASIC, chip-id was always set to 0x1750 for
all adaptor port configurations. This has been fixed in the new chip rev.

Due to this missing fix users are not able to use adaptors based on latest
chip rev of Broadcom's Gen P5 adaptors.

Fixes: ae8637e13185 ("RDMA/bnxt_re: Add chip context to identify 57500 series")
Link: https://lore.kernel.org/r/1574317343-23300-2-git-send-email-devesh.sharma@broadcom.com
Signed-off-by: Naresh Kumar PBS <nareshkumar.pbs@broadcom.com>
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Luke Starrett <luke.starrett@broadcom.com>
Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobpf: Provide better register bounds after jmp32 instructions
Yonghong Song [Thu, 21 Nov 2019 17:06:50 +0000 (09:06 -0800)]
bpf: Provide better register bounds after jmp32 instructions

[ Upstream commit 581738a681b6faae5725c2555439189ca81c0f1f ]

With latest llvm (trunk https://github.com/llvm/llvm-project),
test_progs, which has +alu32 enabled, failed for strobemeta.o.
The verifier output looks like below with edit to replace large
decimal numbers with hex ones.
 193: (85) call bpf_probe_read_user_str#114
   R0=inv(id=0)
 194: (26) if w0 > 0x1 goto pc+4
   R0_w=inv(id=0,umax_value=0xffffffff00000001)
 195: (6b) *(u16 *)(r7 +80) = r0
 196: (bc) w6 = w0
   R6_w=inv(id=0,umax_value=0xffffffff,var_off=(0x0; 0xffffffff))
 197: (67) r6 <<= 32
   R6_w=inv(id=0,smax_value=0x7fffffff00000000,umax_value=0xffffffff00000000,
            var_off=(0x0; 0xffffffff00000000))
 198: (77) r6 >>= 32
   R6=inv(id=0,umax_value=0xffffffff,var_off=(0x0; 0xffffffff))
 ...
 201: (79) r8 = *(u64 *)(r10 -416)
   R8_w=map_value(id=0,off=40,ks=4,vs=13872,imm=0)
 202: (0f) r8 += r6
   R8_w=map_value(id=0,off=40,ks=4,vs=13872,umax_value=0xffffffff,var_off=(0x0; 0xffffffff))
 203: (07) r8 += 9696
   R8_w=map_value(id=0,off=9736,ks=4,vs=13872,umax_value=0xffffffff,var_off=(0x0; 0xffffffff))
 ...
 255: (bf) r1 = r8
   R1_w=map_value(id=0,off=9736,ks=4,vs=13872,umax_value=0xffffffff,var_off=(0x0; 0xffffffff))
 ...
 257: (85) call bpf_probe_read_user_str#114
 R1 unbounded memory access, make sure to bounds check any array access into a map

The value range for register r6 at insn 198 should be really just 0/1.
The umax_value=0xffffffff caused later verification failure.

After jmp instructions, the current verifier already tried to use just
obtained information to get better register range. The current mechanism is
for 64bit register only. This patch implemented to tighten the range
for 32bit sub-registers after jmp32 instructions.
With the patch, we have the below range ranges for the
above code sequence:
 193: (85) call bpf_probe_read_user_str#114
   R0=inv(id=0)
 194: (26) if w0 > 0x1 goto pc+4
   R0_w=inv(id=0,smax_value=0x7fffffff00000001,umax_value=0xffffffff00000001,
            var_off=(0x0; 0xffffffff00000001))
 195: (6b) *(u16 *)(r7 +80) = r0
 196: (bc) w6 = w0
   R6_w=inv(id=0,umax_value=0xffffffff,var_off=(0x0; 0x1))
 197: (67) r6 <<= 32
   R6_w=inv(id=0,umax_value=0x100000000,var_off=(0x0; 0x100000000))
 198: (77) r6 >>= 32
   R6=inv(id=0,umax_value=1,var_off=(0x0; 0x1))
 ...
 201: (79) r8 = *(u64 *)(r10 -416)
   R8_w=map_value(id=0,off=40,ks=4,vs=13872,imm=0)
 202: (0f) r8 += r6
   R8_w=map_value(id=0,off=40,ks=4,vs=13872,umax_value=1,var_off=(0x0; 0x1))
 203: (07) r8 += 9696
   R8_w=map_value(id=0,off=9736,ks=4,vs=13872,umax_value=1,var_off=(0x0; 0x1))
 ...
 255: (bf) r1 = r8
   R1_w=map_value(id=0,off=9736,ks=4,vs=13872,umax_value=1,var_off=(0x0; 0x1))
 ...
 257: (85) call bpf_probe_read_user_str#114
 ...

At insn 194, the register R0 has better var_off.mask and smax_value.
Especially, the var_off.mask ensures later lshift and rshift
maintains proper value range.

Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20191121170650.449030-1-yhs@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoRDMA/bnxt_re: Fix stat push into dma buffer on gen p5 devices
Devesh Sharma [Thu, 21 Nov 2019 06:22:22 +0000 (01:22 -0500)]
RDMA/bnxt_re: Fix stat push into dma buffer on gen p5 devices

[ Upstream commit 98998ffe5216c7fa2c0225bb5b049ca5cdf8d195 ]

Due to recent advances in the firmware for Broadcom's gen p5 series of
adaptors the driver code to report hardware counters has been broken
w.r.t. roce devices.

The new firmware command expects dma length to be specified during stat
dma buffer allocation.

Fixes: 2792b5b95ed5 ("bnxt_en: Update firmware interface spec. to 1.10.0.89.")
Link: https://lore.kernel.org/r/1574317343-23300-3-git-send-email-devesh.sharma@broadcom.com
Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoRDMA/bnxt_re: Fix missing le16_to_cpu
Devesh Sharma [Thu, 21 Nov 2019 06:22:23 +0000 (01:22 -0500)]
RDMA/bnxt_re: Fix missing le16_to_cpu

[ Upstream commit fca5b9dc0986aa49b3f0a7cfe24b6c82422ac1d7 ]

From sparse:

drivers/infiniband/hw/bnxt_re/main.c:1274:18: warning: cast from restricted __le16
drivers/infiniband/hw/bnxt_re/main.c:1275:18: warning: cast from restricted __le16
drivers/infiniband/hw/bnxt_re/main.c:1276:18: warning: cast from restricted __le16
drivers/infiniband/hw/bnxt_re/main.c:1277:21: warning: restricted __le16 degrades to integer

Fixes: 2b827ea1926b ("RDMA/bnxt_re: Query HWRM Interface version from FW")
Link: https://lore.kernel.org/r/1574317343-23300-4-git-send-email-devesh.sharma@broadcom.com
Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agotools, bpf: Fix build for 'make -s tools/bpf O=<dir>'
Quentin Monnet [Tue, 19 Nov 2019 10:56:26 +0000 (10:56 +0000)]
tools, bpf: Fix build for 'make -s tools/bpf O=<dir>'

[ Upstream commit a89b2cbf71d64b61e79bbe5cb7ff4664797eeaaf ]

Building selftests with 'make TARGETS=bpf kselftest' was fixed in commit
55d554f5d140 ("tools: bpf: Use !building_out_of_srctree to determine
srctree"). However, by updating $(srctree) in tools/bpf/Makefile for
in-tree builds only, we leave out the case where we pass an output
directory to build BPF tools, but $(srctree) is not set. This
typically happens for:

    $ make -s tools/bpf O=/tmp/foo
    Makefile:40: /tools/build/Makefile.feature: No such file or directory

Fix it by updating $(srctree) in the Makefile not only for out-of-tree
builds, but also if $(srctree) is empty.

Detected with test_bpftool_build.sh.

Fixes: 55d554f5d140 ("tools: bpf: Use !building_out_of_srctree to determine srctree")
Signed-off-by: Quentin Monnet <quentin.monnet@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Link: https://lore.kernel.org/bpf/20191119105626.21453-1-quentin.monnet@netronome.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: phy: initialise phydev speed and duplex sanely
Russell King [Fri, 22 Nov 2019 15:23:23 +0000 (15:23 +0000)]
net: phy: initialise phydev speed and duplex sanely

[ Upstream commit a5d66f810061e2dd70fb7a108dcd14e535bc639f ]

When a phydev is created, the speed and duplex are set to zero and
-1 respectively, rather than using the predefined SPEED_UNKNOWN and
DUPLEX_UNKNOWN constants.

There is a window at initialisation time where we may report link
down using the 0/-1 values.  Tidy this up and use the predefined
constants, so debug doesn't complain with:

"Unsupported (update phy-core.c)/Unsupported (update phy-core.c)"

when the speed and duplex settings are printed.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoice: Fix setting coalesce to handle DCB configuration
Brett Creeley [Fri, 8 Nov 2019 14:23:23 +0000 (06:23 -0800)]
ice: Fix setting coalesce to handle DCB configuration

[ Upstream commit e25f9152bc07de534b2b590ce6c052ea25dd8900 ]

Currently there can be a case where a DCB map is applied and there are
more interrupt vectors (vsi->num_q_vectors) than Rx queues (vsi->num_rxq)
and Tx queues (vsi->num_txq). If we try to set coalesce settings in this
case it will report a false failure. Fix this by checking if vector index
is valid with respect to the number of Tx and Rx queues configured.

Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoice: Only disable VF state when freeing each VF resources
Akeem G Abodunrin [Fri, 8 Nov 2019 14:23:22 +0000 (06:23 -0800)]
ice: Only disable VF state when freeing each VF resources

[ Upstream commit 1f9639d2fb9188a59acafae9dea626391c442a8d ]

It is wrong to set PF disable state flag for all VFs when freeing VF
resources - Instead, we should set VF disable state flag for each VF with
its resources being returned to the device. Right now, all VF opcodes,
mailbox communication to clear its resources as well fails - since we
already indicate that PF is in disable state, with all VFs not active. In
addition, we don't need to notify VF that PF is intending to reset it, if
it is already in disabled state.

Signed-off-by: Akeem G Abodunrin <akeem.g.abodunrin@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agodrm/amdgpu: fix bad DMA from INTERRUPT_CNTL2
Sam Bobroff [Sun, 17 Nov 2019 23:53:54 +0000 (10:53 +1100)]
drm/amdgpu: fix bad DMA from INTERRUPT_CNTL2

[ Upstream commit 3d0e3ce52ce3eb4b9de3caf9c38dbb5a4d3e13c3 ]

The INTERRUPT_CNTL2 register expects a valid DMA address, but is
currently set with a GPU MC address.  This can cause problems on
systems that detect the resulting DMA read from an invalid address
(found on a Power8 guest).

Instead, use the DMA address of the dummy page because it will always
be safe.

Fixes: 27ae10641e9c ("drm/amdgpu: add interupt handler implementation for si v3")
Signed-off-by: Sam Bobroff <sbobroff@linux.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomips: fix build when "48 bits virtual memory" is enabled
Mike Rapoport [Thu, 21 Nov 2019 16:21:31 +0000 (18:21 +0200)]
mips: fix build when "48 bits virtual memory" is enabled

[ Upstream commit 3ed6751bb8fa89c3014399bb0414348499ee202a ]

With CONFIG_MIPS_VA_BITS_48=y the build fails miserably:

  CC      arch/mips/kernel/asm-offsets.s
In file included from arch/mips/include/asm/pgtable.h:644,
                 from include/linux/mm.h:99,
                 from arch/mips/kernel/asm-offsets.c:15:
include/asm-generic/pgtable.h:16:2: error: #error CONFIG_PGTABLE_LEVELS is not consistent with __PAGETABLE_{P4D,PUD,PMD}_FOLDED
 #error CONFIG_PGTABLE_LEVELS is not consistent with __PAGETABLE_{P4D,PUD,PMD}_FOLDED
  ^~~~~
include/asm-generic/pgtable.h:390:28: error: unknown type name 'p4d_t'; did you mean 'pmd_t'?
 static inline int p4d_same(p4d_t p4d_a, p4d_t p4d_b)
                            ^~~~~
                            pmd_t

[ ... more such errors ... ]

scripts/Makefile.build:99: recipe for target 'arch/mips/kernel/asm-offsets.s' failed
make[2]: *** [arch/mips/kernel/asm-offsets.s] Error 1

This happens because when CONFIG_MIPS_VA_BITS_48 enables 4th level of the
page tables, but neither pgtable-nop4d.h nor 5level-fixup.h are included to
cope with the 5th level.

Replace #ifdef conditions around includes of the pgtable-nop{m,u}d.h with
explicit CONFIG_PGTABLE_LEVELS and add include of 5level-fixup.h for the
case when CONFIG_PGTABLE_LEVELS==4

Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Paul Burton <paulburton@kernel.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <jhogan@kernel.org>
Cc: linux-mips@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: linux-mm@kvack.org
Cc: Mike Rapoport <rppt@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agolibtraceevent: Fix memory leakage in copy_filter_type
Hewenliang [Tue, 19 Nov 2019 01:44:15 +0000 (20:44 -0500)]
libtraceevent: Fix memory leakage in copy_filter_type

[ Upstream commit 10992af6bf46a2048ad964985a5b77464e5563b1 ]

It is necessary to free the memory that we have allocated when error occurs.

Fixes: ef3072cd1d5c ("tools lib traceevent: Get rid of die in add_filter_type()")
Signed-off-by: Hewenliang <hewenliang4@huawei.com>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tzvetomir Stoyanov <tstoyanov@vmware.com>
Link: http://lore.kernel.org/lkml/20191119014415.57210-1-hewenliang4@huawei.com
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocrypto: vmx - Avoid weird build failures
Michael Ellerman [Wed, 20 Nov 2019 11:27:38 +0000 (22:27 +1100)]
crypto: vmx - Avoid weird build failures

[ Upstream commit 4ee812f6143d78d8ba1399671d78c8d78bf2817c ]

In the vmx crypto Makefile we assign to a variable called TARGET and
pass that to the aesp8-ppc.pl and ghashp8-ppc.pl scripts.

The variable is meant to describe what flavour of powerpc we're
building for, eg. either 32 or 64-bit, and big or little endian.

Unfortunately TARGET is a fairly common name for a make variable, and
if it happens that TARGET is specified as a command line parameter to
make, the value specified on the command line will override our value.

In particular this can happen if the kernel Makefile is driven by an
external Makefile that uses TARGET for something.

This leads to weird build failures, eg:
  nonsense  at /build/linux/drivers/crypto/vmx/ghashp8-ppc.pl line 45.
  /linux/drivers/crypto/vmx/Makefile:20: recipe for target 'drivers/crypto/vmx/ghashp8-ppc.S' failed

Which shows that we passed an empty value for $(TARGET) to the perl
script, confirmed with make V=1:

  perl /linux/drivers/crypto/vmx/ghashp8-ppc.pl  > drivers/crypto/vmx/ghashp8-ppc.S

We can avoid this confusion by using override, to tell make that we
don't want anything to override our variable, even a value specified
on the command line. We can also use a less common name, given the
script calls it "flavour", let's use that.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomac80211: consider QoS Null frames for STA_NULLFUNC_ACKED
Thomas Pedersen [Tue, 19 Nov 2019 05:35:38 +0000 (21:35 -0800)]
mac80211: consider QoS Null frames for STA_NULLFUNC_ACKED

[ Upstream commit 08a5bdde3812993cb8eb7aa9124703df0de28e4b ]

Commit 7b6ddeaf27ec ("mac80211: use QoS NDP for AP probing")
let STAs send QoS Null frames as PS triggers if the AP was
a QoS STA.  However, the mac80211 PS stack relies on an
interface flag IEEE80211_STA_NULLFUNC_ACKED for
determining trigger frame ACK, which was not being set for
acked non-QoS Null frames. The effect is an inability to
trigger hardware sleep via IEEE80211_CONF_PS since the QoS
Null frame was seemingly never acked.

This bug only applies to drivers which set both
IEEE80211_HW_REPORTS_TX_ACK_STATUS and
IEEE80211_HW_PS_NULLFUNC_STACK.

Detect the acked QoS Null frame to restore STA power save.

Fixes: 7b6ddeaf27ec ("mac80211: use QoS NDP for AP probing")
Signed-off-by: Thomas Pedersen <thomas@adapt-ip.com>
Link: https://lore.kernel.org/r/20191119053538.25979-4-thomas@adapt-ip.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocrypto: sun4i-ss - Fix 64-bit size_t warnings on sun4i-ss-hash.c
Corentin Labbe [Thu, 14 Nov 2019 10:49:06 +0000 (11:49 +0100)]
crypto: sun4i-ss - Fix 64-bit size_t warnings on sun4i-ss-hash.c

[ Upstream commit a7126603d46fe8f01aeedf589e071c6aaa6c6c39 ]

If you try to compile this driver on a 64-bit platform then you
will get warnings because it mixes size_t with unsigned int which
only works on 32-bit.

This patch fixes all of the warnings on sun4i-ss-hash.c.
Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocrypto: sun4i-ss - Fix 64-bit size_t warnings
Herbert Xu [Tue, 12 Nov 2019 02:38:34 +0000 (10:38 +0800)]
crypto: sun4i-ss - Fix 64-bit size_t warnings

[ Upstream commit d6e9da21ee8246b5e556b3b153401ab045adb986 ]

If you try to compile this driver on a 64-bit platform then you
will get warnings because it mixes size_t with unsigned int which
only works on 32-bit.

This patch fixes all of the warnings.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agos390/cpumf: Adjust registration of s390 PMU device drivers
Thomas Richter [Fri, 15 Nov 2019 11:50:46 +0000 (12:50 +0100)]
s390/cpumf: Adjust registration of s390 PMU device drivers

[ Upstream commit 6a82e23f45fe0aa821e7a935e39d0acb20c275c0 ]

Linux-next commit titled "perf/core: Optimize perf_init_event()"
changed the semantics of PMU device driver registration.
It was done to speed up the lookup/handling of PMU device driver
specific events. It also enforces that only one PMU device
driver will be registered of type PERF_EVENT_RAW.

This change added these line in function perf_pmu_register():

  ...
  +       ret = idr_alloc(&pmu_idr, pmu, max, 0, GFP_KERNEL);
  +       if (ret < 0)
                goto free_pdc;
  +
  +       WARN_ON(type >= 0 && ret != type);

The warn_on generates a message. We have 3 PMU device drivers,
each registered as type PERF_TYPE_RAW.
The cf_diag device driver (arch/s390/kernel/perf_cpumf_cf_diag.c)
always hits the WARN_ON because it is the second PMU device driver
(after sampling device driver arch/s390/kernel/perf_cpumf_sf.c)
which is registered as type 4 (PERF_TYPE_RAW).
So when the sampling device driver is registered, ret has value 4.
When cf_diag device driver is registered with type 4,
ret has value of 5 and WARN_ON fires.

Adjust the PMU device drivers for s390 to support the new
semantics required by perf_pmu_register().

Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomt76: fix possible out-of-bound access in mt7615_fill_txs/mt7603_fill_txs
Lorenzo Bianconi [Wed, 6 Nov 2019 23:01:58 +0000 (01:01 +0200)]
mt76: fix possible out-of-bound access in mt7615_fill_txs/mt7603_fill_txs

[ Upstream commit e8b970c8e367e85fab9b8ac4f36080e5d653c38e ]

Fix possible out-of-bound access of status rates array in
mt7615_fill_txs/mt7603_fill_txs routines

Fixes: c5211e997eca ("mt76: mt7603: rework and fix tx status reporting")
Fixes: 4af81f02b49c ("mt76: mt7615: sync with mt7603 rate control changes")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: ethernet: ti: ale: clean ale tbl on init and intf restart
Grygorii Strashko [Tue, 19 Nov 2019 22:19:13 +0000 (00:19 +0200)]
net: ethernet: ti: ale: clean ale tbl on init and intf restart

[ Upstream commit 7fe579dfb90fcdf0c7722f33c772d5f0d1bc7cb6 ]

Clean CPSW ALE on init and intf restart (up/down) to avoid reading obsolete
or garbage entries from ALE table.

Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: soc-pcm: check symmetry before hw_params
Shengjiu Wang [Tue, 12 Nov 2019 10:46:42 +0000 (18:46 +0800)]
ASoC: soc-pcm: check symmetry before hw_params

[ Upstream commit 5cca59516de5df9de6bdecb328dd55fb5bcccb41 ]

This reverts commit 957ce0c6b8a1f (ASoC: soc-pcm: check symmetry after
hw_params).

That commit cause soc_pcm_params_symmetry can't take effect.
cpu_dai->rate, cpu_dai->channels and cpu_dai->sample_bits
are updated in the middle of soc_pcm_hw_params, so move
soc_pcm_params_symmetry to the end of soc_pcm_hw_params is
not a good solution, for judgement of symmetry in the function
is always true.

FIXME:
According to the comments of that commit, I think the case
described in the commit should disable symmetric_rates
in Back-End, rather than changing the position of
soc_pcm_params_symmetry.

Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Link: https://lore.kernel.org/r/1573555602-5403-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agofbtft: Make sure string is NULL terminated
Andy Shevchenko [Wed, 20 Nov 2019 09:57:12 +0000 (11:57 +0200)]
fbtft: Make sure string is NULL terminated

[ Upstream commit 21f585480deb4bcf0d92b08879c35d066dfee030 ]

New GCC warns about inappropriate use of strncpy():

drivers/staging/fbtft/fbtft-core.c: In function ‘fbtft_framebuffer_alloc’:
drivers/staging/fbtft/fbtft-core.c:665:2: warning: ‘strncpy’ specified bound 16 equals destination size [-Wstringop-truncation]
  665 |  strncpy(info->fix.id, dev->driver->name, 16);
      |  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Later on the copy is being used with the assumption to be NULL terminated.
Make sure string is NULL terminated by switching to snprintf().

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20191120095716.26628-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoiwlwifi: check kasprintf() return value
Johannes Berg [Tue, 5 Nov 2019 13:50:32 +0000 (14:50 +0100)]
iwlwifi: check kasprintf() return value

[ Upstream commit 5974fbb5e10b018fdbe3c3b81cb4cc54e1105ab9 ]

kasprintf() can fail, we should check the return value.

Fixes: 5ed540aecc2a ("iwlwifi: use mac80211 throughput trigger")
Fixes: 8ca151b568b6 ("iwlwifi: add the MVM driver")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobnxt_en: Improve RX buffer error handling.
Michael Chan [Mon, 18 Nov 2019 08:56:36 +0000 (03:56 -0500)]
bnxt_en: Improve RX buffer error handling.

[ Upstream commit 19b3751ffa713d04290effb26fe01009010f2206 ]

When hardware reports RX buffer errors, the latest 57500 chips do not
require reset.  The packet is discarded by the hardware and the
ring will continue to operate.

Also, add an rx_buf_errors counter for this type of error.  It can help
the user to identify if the aggregation ring is too small.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agos390/kasan: support memcpy_real with TRACE_IRQFLAGS
Vasily Gorbik [Tue, 5 Nov 2019 16:33:20 +0000 (17:33 +0100)]
s390/kasan: support memcpy_real with TRACE_IRQFLAGS

[ Upstream commit 13f9bae579c6bd051e58f326913dd09af1291208 ]

Currently if the kernel is built with CONFIG_TRACE_IRQFLAGS and KASAN
and used as crash kernel it crashes itself due to
trace_hardirqs_off/trace_hardirqs_on being called with DAT off. This
happens because trace_hardirqs_off/trace_hardirqs_on are instrumented and
kasan code tries to perform access to shadow memory to validate memory
accesses. Kasan shadow memory is populated with vmemmap, so all accesses
require DAT on.

memcpy_real could be called with DAT on or off (with kasan enabled DAT
is set even before early code is executed).

Make sure that trace_hardirqs_off/trace_hardirqs_on are called with DAT
on and only actual __memcpy_real is called with DAT off.

Also annotate __memcpy_real and _memcpy_real with __no_sanitize_address
to avoid further problems due to switching DAT off.

Reviewed-by: Philipp Rudo <prudo@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agos390/crypto: Fix unsigned variable compared with zero
YueHaibing [Thu, 14 Nov 2019 07:30:05 +0000 (15:30 +0800)]
s390/crypto: Fix unsigned variable compared with zero

[ Upstream commit 0398d4ab1677f7d8cd43aac2aa29a93dfcf9e2e3 ]

s390_crypto_shash_parmsize() return type is int, it
should not be stored in a unsigned variable, which
compared with zero.

Reported-by: Hulk Robot <hulkci@huawei.com>
Fixes: 3c2eb6b76cab ("s390/crypto: Support for SHA3 via CPACF (MSA6)")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Joerg Schmidbauer <jschmidb@linux.vnet.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Fix to show function entry line as probe-able
Masami Hiramatsu [Thu, 24 Oct 2019 09:12:54 +0000 (18:12 +0900)]
perf probe: Fix to show function entry line as probe-able

commit 91e2f539eeda26ab00bd03fae8dc434c128c85ed upstream.

Fix die_walk_lines() to list the function entry line correctly.  Since
the dwarf_entrypc() does not return the entry pc if the DIE has only
range attribute, __die_walk_funclines() fails to list the declaration
line (entry line) in that case.

To solve this issue, this introduces die_entrypc() which correctly
returns the entry PC (the first address range) even if the DIE has only
range attribute. With this fix die_walk_lines() shows the function entry
line is able to probe correctly.

Fixes: 4cc9cec636e7 ("perf probe: Introduce lines walker interface")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lore.kernel.org/lkml/157190837419.1859.4619125803596816752.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Backlund <tmb@mageia.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
4 years agoperf session: Fix decompression of PERF_RECORD_COMPRESSED records
Alexey Budankov [Mon, 18 Nov 2019 14:21:03 +0000 (17:21 +0300)]
perf session: Fix decompression of PERF_RECORD_COMPRESSED records

[ Upstream commit bb1835a3b86c73aa534ef6430ad40223728dfbc0 ]

Avoid termination of trace loading in case the last record in the
decompressed buffer partly resides in the following mmaped
PERF_RECORD_COMPRESSED record.

In this case NULL value returned by fetch_mmaped_event() means to
proceed to the next mmaped record then decompress it and load compressed
events.

The issue can be reproduced like this:

  $ perf record -z -- some_long_running_workload
  $ perf report --stdio -vv
  decomp (B): 44519 to 163000
  decomp (B): 48119 to 174800
  decomp (B): 65527 to 131072
  fetch_mmaped_event: head=0x1ffe0 event->header_size=0x28, mmap_size=0x20000: fuzzed perf.data?
  Error:
  failed to process sample
  ...

Testing:

  71: Zstd perf.data compression/decompression              : Ok

  $ tools/perf/perf report -vv --stdio
  decomp (B): 59593 to 262160
  decomp (B): 4438 to 16512
  decomp (B): 285 to 880
  Looking at the vmlinux_path (8 entries long)
  Using vmlinux for symbols
  decomp (B): 57474 to 261248
  prefetch_event: head=0x3fc78 event->header_size=0x28, mmap_size=0x3fc80: fuzzed or compressed perf.data?
  decomp (B): 25 to 32
  decomp (B): 52 to 120
  ...

Fixes: 57fc032ad643 ("perf session: Avoid infinite loop when seeing invalid header.size")
Link: https://marc.info/?l=linux-kernel&m=156580812427554&w=2
Co-developed-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/cf782c34-f3f8-2f9f-d6ab-145cee0d5322@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobrcmfmac: remove monitor interface when detaching
Rafał Miłecki [Mon, 18 Nov 2019 12:38:55 +0000 (13:38 +0100)]
brcmfmac: remove monitor interface when detaching

[ Upstream commit 4f61563da075bc8faefddfd5f8fc0cc14c49650a ]

This fixes a minor WARNING in the cfg80211:
[  130.658034] ------------[ cut here ]------------
[  130.662805] WARNING: CPU: 1 PID: 610 at net/wireless/core.c:954 wiphy_unregister+0xb4/0x198 [cfg80211]

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet-af_xdp: Use correct number of channels from ethtool
Luigi Rizzo [Tue, 19 Nov 2019 00:19:51 +0000 (16:19 -0800)]
net-af_xdp: Use correct number of channels from ethtool

[ Upstream commit 3de88c9113f88c04abda339f1aa629397bf89e02 ]

Drivers use different fields to report the number of channels, so take
the maximum of all data channels (rx, tx, combined) when determining the
size of the xsk map. The current code used only 'combined' which was set
to 0 in some drivers e.g. mlx4.

Tested: compiled and run xdpsock -q 3 -r -S on mlx4

Signed-off-by: Luigi Rizzo <lrizzo@google.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20191119001951.92930-1-lrizzo@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agox86/insn: Add some Intel instructions to the opcode map
Adrian Hunter [Fri, 15 Nov 2019 13:54:47 +0000 (15:54 +0200)]
x86/insn: Add some Intel instructions to the opcode map

[ Upstream commit b980be189c9badba50634671e2303e92bf28e35a ]

Add to the opcode map the following instructions:
        cldemote
        tpause
        umonitor
        umwait
        movdiri
        movdir64b
        enqcmd
        enqcmds
        encls
        enclu
        enclv
        pconfig
        wbnoinvd

For information about the instructions, refer Intel SDM May 2019
(325462-070US) and Intel Architecture Instruction Set Extensions
May 2019 (319433-037).

The instruction decoding can be tested using the perf tools'
"x86 instruction decoder - new instructions" test as folllows:

  $ perf test -v "new " 2>&1 | grep -i cldemote
  Decoded ok: 0f 1c 00                    cldemote (%eax)
  Decoded ok: 0f 1c 05 78 56 34 12        cldemote 0x12345678
  Decoded ok: 0f 1c 84 c8 78 56 34 12     cldemote 0x12345678(%eax,%ecx,8)
  Decoded ok: 0f 1c 00                    cldemote (%rax)
  Decoded ok: 41 0f 1c 00                 cldemote (%r8)
  Decoded ok: 0f 1c 04 25 78 56 34 12     cldemote 0x12345678
  Decoded ok: 0f 1c 84 c8 78 56 34 12     cldemote 0x12345678(%rax,%rcx,8)
  Decoded ok: 41 0f 1c 84 c8 78 56 34 12  cldemote 0x12345678(%r8,%rcx,8)
  $ perf test -v "new " 2>&1 | grep -i tpause
  Decoded ok: 66 0f ae f3                 tpause %ebx
  Decoded ok: 66 0f ae f3                 tpause %ebx
  Decoded ok: 66 41 0f ae f0              tpause %r8d
  $ perf test -v "new " 2>&1 | grep -i umonitor
  Decoded ok: 67 f3 0f ae f0              umonitor %ax
  Decoded ok: f3 0f ae f0                 umonitor %eax
  Decoded ok: 67 f3 0f ae f0              umonitor %eax
  Decoded ok: f3 0f ae f0                 umonitor %rax
  Decoded ok: 67 f3 41 0f ae f0           umonitor %r8d
  $ perf test -v "new " 2>&1 | grep -i umwait
  Decoded ok: f2 0f ae f0                 umwait %eax
  Decoded ok: f2 0f ae f0                 umwait %eax
  Decoded ok: f2 41 0f ae f0              umwait %r8d
  $ perf test -v "new " 2>&1 | grep -i movdiri
  Decoded ok: 0f 38 f9 03                 movdiri %eax,(%ebx)
  Decoded ok: 0f 38 f9 88 78 56 34 12     movdiri %ecx,0x12345678(%eax)
  Decoded ok: 48 0f 38 f9 03              movdiri %rax,(%rbx)
  Decoded ok: 48 0f 38 f9 88 78 56 34 12  movdiri %rcx,0x12345678(%rax)
  $ perf test -v "new " 2>&1 | grep -i movdir64b
  Decoded ok: 66 0f 38 f8 18              movdir64b (%eax),%ebx
  Decoded ok: 66 0f 38 f8 88 78 56 34 12  movdir64b 0x12345678(%eax),%ecx
  Decoded ok: 67 66 0f 38 f8 1c           movdir64b (%si),%bx
  Decoded ok: 67 66 0f 38 f8 8c 34 12     movdir64b 0x1234(%si),%cx
  Decoded ok: 66 0f 38 f8 18              movdir64b (%rax),%rbx
  Decoded ok: 66 0f 38 f8 88 78 56 34 12  movdir64b 0x12345678(%rax),%rcx
  Decoded ok: 67 66 0f 38 f8 18           movdir64b (%eax),%ebx
  Decoded ok: 67 66 0f 38 f8 88 78 56 34 12       movdir64b 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i enqcmd
  Decoded ok: f2 0f 38 f8 18              enqcmd (%eax),%ebx
  Decoded ok: f2 0f 38 f8 88 78 56 34 12  enqcmd 0x12345678(%eax),%ecx
  Decoded ok: 67 f2 0f 38 f8 1c           enqcmd (%si),%bx
  Decoded ok: 67 f2 0f 38 f8 8c 34 12     enqcmd 0x1234(%si),%cx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%eax),%ebx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%eax),%ecx
  Decoded ok: 67 f3 0f 38 f8 1c           enqcmds (%si),%bx
  Decoded ok: 67 f3 0f 38 f8 8c 34 12     enqcmds 0x1234(%si),%cx
  Decoded ok: f2 0f 38 f8 18              enqcmd (%rax),%rbx
  Decoded ok: f2 0f 38 f8 88 78 56 34 12  enqcmd 0x12345678(%rax),%rcx
  Decoded ok: 67 f2 0f 38 f8 18           enqcmd (%eax),%ebx
  Decoded ok: 67 f2 0f 38 f8 88 78 56 34 12       enqcmd 0x12345678(%eax),%ecx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%rax),%rbx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%rax),%rcx
  Decoded ok: 67 f3 0f 38 f8 18           enqcmds (%eax),%ebx
  Decoded ok: 67 f3 0f 38 f8 88 78 56 34 12       enqcmds 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i enqcmds
  Decoded ok: f3 0f 38 f8 18              enqcmds (%eax),%ebx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%eax),%ecx
  Decoded ok: 67 f3 0f 38 f8 1c           enqcmds (%si),%bx
  Decoded ok: 67 f3 0f 38 f8 8c 34 12     enqcmds 0x1234(%si),%cx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%rax),%rbx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%rax),%rcx
  Decoded ok: 67 f3 0f 38 f8 18           enqcmds (%eax),%ebx
  Decoded ok: 67 f3 0f 38 f8 88 78 56 34 12       enqcmds 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i encls
  Decoded ok: 0f 01 cf                    encls
  Decoded ok: 0f 01 cf                    encls
  $ perf test -v "new " 2>&1 | grep -i enclu
  Decoded ok: 0f 01 d7                    enclu
  Decoded ok: 0f 01 d7                    enclu
  $ perf test -v "new " 2>&1 | grep -i enclv
  Decoded ok: 0f 01 c0                    enclv
  Decoded ok: 0f 01 c0                    enclv
  $ perf test -v "new " 2>&1 | grep -i pconfig
  Decoded ok: 0f 01 c5                    pconfig
  Decoded ok: 0f 01 c5                    pconfig
  $ perf test -v "new " 2>&1 | grep -i wbnoinvd
  Decoded ok: f3 0f 09                    wbnoinvd
  Decoded ok: f3 0f 09                    wbnoinvd

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: x86@kernel.org
Link: http://lore.kernel.org/lkml/20191115135447.6519-3-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: Intel: bytcr_rt5640: Update quirk for Acer Switch 10 SW5-012 2-in-1
Hans de Goede [Tue, 19 Nov 2019 14:51:38 +0000 (15:51 +0100)]
ASoC: Intel: bytcr_rt5640: Update quirk for Acer Switch 10 SW5-012 2-in-1

[ Upstream commit 0bb887709eb16bdc4b5baddd8337abf3de72917f ]

When the Acer Switch 10 SW5-012 quirk was added we did not have
jack-detection support yet; and the builtin microphone selection of
the original quirk is wrong too.

Fix the microphone-input quirk and add jack-detection info so that the
internal-microphone and headphone/set jack on the Switch 10 work properly.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20191119145138.59162-1-hdegoede@redhat.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agofirmware_loader: Fix labels with comma for builtin firmware
Linus Walleij [Fri, 15 Nov 2019 22:59:11 +0000 (23:59 +0100)]
firmware_loader: Fix labels with comma for builtin firmware

[ Upstream commit 553671b7685972ca671da5f71cf6414b54376e13 ]

Some firmware images contain a comma, such as:
EXTRA_FIRMWARE "brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt"
as Broadcom firmware simply tags the device tree compatible
string at the end of the firmware parameter file. And the
compatible string contains a comma.

This doesn't play well with gas:

drivers/base/firmware_loader/builtin/brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt.gen.S: Assembler messages:
drivers/base/firmware_loader/builtin/brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt.gen.S:4: Error: bad instruction `_fw_brcm_brcmfmac4334_sdio_samsung,gt_s7710_txt_bin:'
drivers/base/firmware_loader/builtin/brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt.gen.S:9: Error: bad instruction `_fw_brcm_brcmfmac4334_sdio_samsung,gt_s7710_txt_name:'
drivers/base/firmware_loader/builtin/brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt.gen.S:15: Error: can't resolve `.rodata' {.rodata section} - `_fw_brcm_brcmfmac4334_sdio_samsung' {*UND* section}
make[6]: *** [../scripts/Makefile.build:357: drivers/base/firmware_loader/builtin/brcm/brcmfmac4334-sdio.samsung,gt-s7710.txt.gen.o] Error 1

We need to get rid of the comma from the labels used by the
assembly stub generator.

Replacing a comma using GNU Make subst requires a helper
variable.

Cc: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20191115225911.3260-1-linus.walleij@linaro.org
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: phy: avoid matching all-ones clause 45 PHY IDs
Russell King [Fri, 15 Nov 2019 20:08:37 +0000 (20:08 +0000)]
net: phy: avoid matching all-ones clause 45 PHY IDs

[ Upstream commit b95e86d846b63b02ecdc94802ddbeaf9005fb6d9 ]

We currently match clause 45 PHYs using any ID read from a MMD marked
as present in the "Devices in package" registers 5 and 6.  However,
this is incorrect.  45.2 says:

  "The definition of the term package is vendor specific and could be
   a chip, module, or other similar entity."

so a package could be more or less than the whole PHY - a PHY could be
made up of several modules instantiated onto a single chip such as the
Marvell 88x3310, or some of the MMDs could be disabled according to
chip configuration, such as the Broadcom 84881.

In the case of Broadcom 84881, the "Devices in package" registers
contain 0xc000009b, meaning that there is a PHYXS present in the
package, but all registers in MMD 4 return 0xffff.  This leads to our
matching code incorrectly binding this PHY to one of our generic PHY
drivers.

This patch changes the way we determine whether to attempt to match a
MMD identifier, or use it to request a module - if the identifier is
all-ones, then we skip over it. When reading the identifiers, we
initialise phydev->c45_ids.device_ids to all-ones, only reading the
device ID if the "Devices in package" registers indicates we should.

This avoids the generic drivers incorrectly matching on a PHY ID of
0xffffffff.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobnxt_en: Return proper error code for non-existent NVM variable
Vasundhara Volam [Mon, 18 Nov 2019 08:56:41 +0000 (03:56 -0500)]
bnxt_en: Return proper error code for non-existent NVM variable

[ Upstream commit 05069dd4c577f9b143dfd243d55834333c4470c5 ]

For NVM params that are not supported in the current NVM
configuration, return the error as -EOPNOTSUPP.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoselftests, bpf: Workaround an alu32 sub-register spilling issue
Yonghong Song [Sun, 17 Nov 2019 21:40:36 +0000 (13:40 -0800)]
selftests, bpf: Workaround an alu32 sub-register spilling issue

[ Upstream commit 2ea2612b987ad703235c92be21d4e98ee9c2c67c ]

Currently, with latest llvm trunk, selftest test_progs failed obj
file test_seg6_loop.o with the following error in verifier:

  infinite loop detected at insn 76

The byte code sequence looks like below, and noted that alu32 has been
turned off by default for better generated codes in general:

      48:       w3 = 100
      49:       *(u32 *)(r10 - 68) = r3
      ...
  ;             if (tlv.type == SR6_TLV_PADDING) {
      76:       if w3 == 5 goto -18 <LBB0_19>
      ...
      85:       r1 = *(u32 *)(r10 - 68)
  ;     for (int i = 0; i < 100; i++) {
      86:       w1 += -1
      87:       if w1 == 0 goto +5 <LBB0_20>
      88:       *(u32 *)(r10 - 68) = r1

The main reason for verification failure is due to partial spills at
r10 - 68 for induction variable "i".

Current verifier only handles spills with 8-byte values. The above 4-byte
value spill to stack is treated to STACK_MISC and its content is not
saved. For the above example:

    w3 = 100
      R3_w=inv100 fp-64_w=inv1086626730498
    *(u32 *)(r10 - 68) = r3
      R3_w=inv100 fp-64_w=inv1086626730498
    ...
    r1 = *(u32 *)(r10 - 68)
      R1_w=inv(id=0,umax_value=4294967295,var_off=(0x0; 0xffffffff))
      fp-64=inv1086626730498

To resolve this issue, verifier needs to be extended to track sub-registers
in spilling, or llvm needs to enhanced to prevent sub-register spilling
in register allocation phase. The former will increase verifier complexity
and the latter will need some llvm "hacking".

Let us workaround this issue by declaring the induction variable as "long"
type so spilling will happen at non sub-register level. We can revisit this
later if sub-register spilling causes similar or other verification issues.

Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20191117214036.1309510-1-yhs@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoselftests, bpf: Fix test_tc_tunnel hanging
Jiri Benc [Fri, 15 Nov 2019 12:43:23 +0000 (13:43 +0100)]
selftests, bpf: Fix test_tc_tunnel hanging

[ Upstream commit 3b054b7133b4ad93671c82e8d6185258e3f1a7a5 ]

When run_kselftests.sh is run, it hangs after test_tc_tunnel.sh. The reason
is test_tc_tunnel.sh ensures the server ('nc -l') is run all the time,
starting it again every time it is expected to terminate. The exception is
the final client_connect: the server is not started anymore, which ensures
no process is kept running after the test is finished.

For a sit test, though, the script is terminated prematurely without the
final client_connect and the 'nc' process keeps running. This in turn causes
the run_one function in kselftest/runner.sh to hang forever, waiting for the
runaway process to finish.

Ensure a remaining server is terminated on cleanup.

Fixes: f6ad6accaa99 ("selftests/bpf: expand test_tc_tunnel with SIT encap")
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/bpf/60919291657a9ee89c708d8aababc28ebe1420be.1573821780.git.jbenc@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf/core: Fix the mlock accounting, again
Alexander Shishkin [Fri, 15 Nov 2019 16:08:18 +0000 (18:08 +0200)]
perf/core: Fix the mlock accounting, again

[ Upstream commit 36b3db03b4741b8935b68fffc7e69951d8d70a89 ]

Commit:

  5e6c3c7b1ec2 ("perf/aux: Fix tracking of auxiliary trace buffer allocation")

tried to guess the correct combination of arithmetic operations that would
undo the AUX buffer's mlock accounting, and failed, leaking the bottom part
when an allocation needs to be charged partially to both user->locked_vm
and mm->pinned_vm, eventually leaving the user with no locked bonus:

  $ perf record -e intel_pt//u -m1,128 uname
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.061 MB perf.data ]

  $ perf record -e intel_pt//u -m1,128 uname
  Permission error mapping pages.
  Consider increasing /proc/sys/kernel/perf_event_mlock_kb,
  or try again with a smaller value of -m/--mmap_pages.
  (current value: 1,128)

Fix this by subtracting both locked and pinned counts when AUX buffer is
unmapped.

Reported-by: Thomas Richter <tmricht@linux.ibm.com>
Tested-by: Thomas Richter <tmricht@linux.ibm.com>
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: wm5100: add missed pm_runtime_disable
Chuhong Yuan [Mon, 18 Nov 2019 07:37:07 +0000 (15:37 +0800)]
ASoC: wm5100: add missed pm_runtime_disable

[ Upstream commit b1176bbb70866f24099cd2720283c7219fb4a81c ]

The driver forgets to call pm_runtime_disable in remove and
probe failure.
Add the calls to fix it.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20191118073707.28298-1-hslester96@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agospi: st-ssc4: add missed pm_runtime_disable
Chuhong Yuan [Mon, 18 Nov 2019 02:48:48 +0000 (10:48 +0800)]
spi: st-ssc4: add missed pm_runtime_disable

[ Upstream commit cd050abeba2a95fe5374eec28ad2244617bcbab6 ]

The driver forgets to call pm_runtime_disable in probe failure
and remove.
Add the missed calls to fix it.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Link: https://lore.kernel.org/r/20191118024848.21645-1-hslester96@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: wm2200: add missed operations in remove and probe failure
Chuhong Yuan [Mon, 18 Nov 2019 07:36:33 +0000 (15:36 +0800)]
ASoC: wm2200: add missed operations in remove and probe failure

[ Upstream commit 2dab09be49a1e7a4dd13cb47d3a1441a2ef33a87 ]

This driver misses calls to pm_runtime_disable and regulator_bulk_disable
in remove and a call to free_irq in probe failure.
Add the calls to fix it.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Link: https://lore.kernel.org/r/20191118073633.28237-1-hslester96@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agosched/uclamp: Fix overzealous type replacement
Valentin Schneider [Fri, 15 Nov 2019 10:39:08 +0000 (10:39 +0000)]
sched/uclamp: Fix overzealous type replacement

[ Upstream commit 7763baace1b738d65efa46d68326c9406311c6bf ]

Some uclamp helpers had their return type changed from 'unsigned int' to
'enum uclamp_id' by commit

  0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")

but it happens that some do return a value in the [0, SCHED_CAPACITY_SCALE]
range, which should really be unsigned int. The affected helpers are
uclamp_none(), uclamp_rq_max_value() and uclamp_eff_value(). Fix those up.

Note that this doesn't lead to any obj diff using a relatively recent
aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
properly returns an 11 bit value (bits_per(1024)) and doesn't seem to do
anything funny. I'm still marking this as fixing the above commit to be on
the safe side.

Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Reviewed-by: Qais Yousef <qais.yousef@arm.com>
Acked-by: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Dietmar.Eggemann@arm.com
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: patrick.bellasi@matbug.net
Cc: qperret@google.com
Cc: surenb@google.com
Cc: tj@kernel.org
Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
Link: https://lkml.kernel.org/r/20191115103908.27610-1-valentin.schneider@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobtrfs: don't prematurely free work in run_ordered_work()
Omar Sandoval [Mon, 16 Sep 2019 18:30:53 +0000 (11:30 -0700)]
btrfs: don't prematurely free work in run_ordered_work()

[ Upstream commit c495dcd6fbe1dce51811a76bb85b4675f6494938 ]

We hit the following very strange deadlock on a system with Btrfs on a
loop device backed by another Btrfs filesystem:

1. The top (loop device) filesystem queues an async_cow work item from
   cow_file_range_async(). We'll call this work X.
2. Worker thread A starts work X (normal_work_helper()).
3. Worker thread A executes the ordered work for the top filesystem
   (run_ordered_work()).
4. Worker thread A finishes the ordered work for work X and frees X
   (work->ordered_free()).
5. Worker thread A executes another ordered work and gets blocked on I/O
   to the bottom filesystem (still in run_ordered_work()).
6. Meanwhile, the bottom filesystem allocates and queues an async_cow
   work item which happens to be the recently-freed X.
7. The workqueue code sees that X is already being executed by worker
   thread A, so it schedules X to be executed _after_ worker thread A
   finishes (see the find_worker_executing_work() call in
   process_one_work()).

Now, the top filesystem is waiting for I/O on the bottom filesystem, but
the bottom filesystem is waiting for the top filesystem to finish, so we
deadlock.

This happens because we are breaking the workqueue assumption that a
work item cannot be recycled while it still depends on other work. Fix
it by waiting to free the work item until we are done with all of the
related ordered work.

P.S.:

One might ask why the workqueue code doesn't try to detect a recycled
work item. It actually does try by checking whether the work item has
the same work function (find_worker_executing_work()), but in our case
the function is the same. This is the only key that the workqueue code
has available to compare, short of adding an additional, layer-violating
"custom key". Considering that we're the only ones that have ever hit
this, we should just play by the rules.

Unfortunately, we haven't been able to create a minimal reproducer other
than our full container setup using a compress-force=zstd filesystem on
top of another compress-force=zstd filesystem.

Suggested-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobtrfs: don't prematurely free work in end_workqueue_fn()
Omar Sandoval [Mon, 16 Sep 2019 18:30:54 +0000 (11:30 -0700)]
btrfs: don't prematurely free work in end_workqueue_fn()

[ Upstream commit 9be490f1e15c34193b1aae17da58e14dd9f55a95 ]

Currently, end_workqueue_fn() frees the end_io_wq entry (which embeds
the work item) and then calls bio_endio(). This is another potential
instance of the bug in "btrfs: don't prematurely free work in
run_ordered_work()".

In particular, the endio call may depend on other work items. For
example, btrfs_end_dio_bio() can call btrfs_subio_endio_read() ->
__btrfs_correct_data_nocsum() -> dio_read_error() ->
submit_dio_repair_bio(), which submits a bio that is also completed
through a end_workqueue_fn() work item. However,
__btrfs_correct_data_nocsum() waits for the newly submitted bio to
complete, thus it depends on another work item.

This example currently usually works because we use different workqueue
helper functions for BTRFS_WQ_ENDIO_DATA and BTRFS_WQ_ENDIO_DIO_REPAIR.
However, it may deadlock with stacked filesystems and is fragile
overall. The proper fix is to free the work item at the very end of the
work function, so let's do that.

Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agommc: tmio: Add MMC_CAP_ERASE to allow erase/discard/trim requests
Eugeniu Rosca [Fri, 15 Nov 2019 13:44:30 +0000 (14:44 +0100)]
mmc: tmio: Add MMC_CAP_ERASE to allow erase/discard/trim requests

[ Upstream commit c91843463e9e821dc3b48fe37e3155fa38299f6e ]

Isolated initially to renesas_sdhi_internal_dmac [1], Ulf suggested
adding MMC_CAP_ERASE to the TMIO mmc core:

On Fri, Nov 15, 2019 at 10:27:25AM +0100, Ulf Hansson wrote:
 -- snip --
 This test and due to the discussions with Wolfram and you in this
 thread, I would actually suggest that you enable MMC_CAP_ERASE for all
 tmio variants, rather than just for this particular one.

 In other words, set the cap in tmio_mmc_host_probe() should be fine,
 as it seems none of the tmio variants supports HW busy detection at
 this point.
 -- snip --

Testing on R-Car H3ULCB-KF doesn't reveal any issues (v5.4-rc7):

root@rcar-gen3:~# lsblk
NAME         MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
mmcblk0      179:0    0 59.2G  0 disk  <--- eMMC
mmcblk0boot0 179:8    0    4M  1 disk
mmcblk0boot1 179:16   0    4M  1 disk
mmcblk1      179:24   0   30G  0 disk  <--- SD card

root@rcar-gen3:~# time blkdiscard /dev/mmcblk0
real    0m8.659s
user    0m0.001s
sys     0m1.920s

root@rcar-gen3:~# time blkdiscard /dev/mmcblk1
real    0m1.176s
user    0m0.001s
sys     0m0.124s

[1] https://lore.kernel.org/linux-renesas-soc/20191112134808.23546-1-erosca@de.adit-jv.com/

Cc: Wolfram Sang <wsa+renesas@sang-engineering.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Andrew Gabbasov <andrew_gabbasov@mentor.com>
Originally-by: Harish Jenny K N <harish_kandiga@mentor.com>
Suggested-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocrypto: virtio - deal with unsupported input sizes
Ard Biesheuvel [Sat, 9 Nov 2019 17:09:27 +0000 (18:09 +0100)]
crypto: virtio - deal with unsupported input sizes

[ Upstream commit 19c5da7d4a2662e85ea67d2d81df57e038fde3ab ]

Return -EINVAL for input sizes that are not a multiple of the AES
block size, since they are not supported by our CBC chaining mode.

While at it, remove the pr_err() that reports unsupported key sizes
being used: we shouldn't spam the kernel log with that.

Fixes: dbaf0624ffa5 ("crypto: add virtio-crypto driver")
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Gonglei <arei.gonglei@huawei.com>
Cc: virtualization@lists.linux-foundation.org
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoxhci-pci: Allow host runtime PM as default also for Intel Ice Lake xHCI
Mika Westerberg [Fri, 15 Nov 2019 16:50:03 +0000 (18:50 +0200)]
xhci-pci: Allow host runtime PM as default also for Intel Ice Lake xHCI

[ Upstream commit 07a594f353655b1628f598add352e7e754f44869 ]

Intel Ice Lake has two xHCI controllers one on PCH and the other as part
of the CPU itself. The latter is also part of the so called Type C
Subsystem (TCSS) sharing ACPI power resources with the PCIe root ports
and the Thunderbolt controllers. In order to put the whole TCSS block
into D3cold the xHCI needs to be runtime suspended as well when idle.

For this reason allow runtime PM as default for Ice Lake TCSS xHCI
controller.

Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/1573836603-10871-5-git-send-email-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agotun: fix data-race in gro_normal_list()
Petar Penkov [Thu, 14 Nov 2019 17:52:09 +0000 (09:52 -0800)]
tun: fix data-race in gro_normal_list()

[ Upstream commit c39e342a050a4425348e6fe7f75827c0a1a7ebc5 ]

There is a race in the TUN driver between napi_busy_loop and
napi_gro_frags. This commit resolves the race by adding the NAPI struct
via netif_tx_napi_add, instead of netif_napi_add, which disables polling
for the NAPI struct.

KCSAN reported:
BUG: KCSAN: data-race in gro_normal_list.part.0 / napi_busy_loop

write to 0xffff8880b5d474b0 of 4 bytes by task 11205 on cpu 0:
 gro_normal_list.part.0+0x77/0xb0 net/core/dev.c:5682
 gro_normal_list net/core/dev.c:5678 [inline]
 gro_normal_one net/core/dev.c:5692 [inline]
 napi_frags_finish net/core/dev.c:5705 [inline]
 napi_gro_frags+0x625/0x770 net/core/dev.c:5778
 tun_get_user+0x2150/0x26a0 drivers/net/tun.c:1976
 tun_chr_write_iter+0x79/0xd0 drivers/net/tun.c:2022
 call_write_iter include/linux/fs.h:1895 [inline]
 do_iter_readv_writev+0x487/0x5b0 fs/read_write.c:693
 do_iter_write fs/read_write.c:970 [inline]
 do_iter_write+0x13b/0x3c0 fs/read_write.c:951
 vfs_writev+0x118/0x1c0 fs/read_write.c:1015
 do_writev+0xe3/0x250 fs/read_write.c:1058
 __do_sys_writev fs/read_write.c:1131 [inline]
 __se_sys_writev fs/read_write.c:1128 [inline]
 __x64_sys_writev+0x4e/0x60 fs/read_write.c:1128
 do_syscall_64+0xcc/0x370 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

read to 0xffff8880b5d474b0 of 4 bytes by task 11168 on cpu 1:
 gro_normal_list net/core/dev.c:5678 [inline]
 napi_busy_loop+0xda/0x4f0 net/core/dev.c:6126
 sk_busy_loop include/net/busy_poll.h:108 [inline]
 __skb_recv_udp+0x4ad/0x560 net/ipv4/udp.c:1689
 udpv6_recvmsg+0x29e/0xe90 net/ipv6/udp.c:288
 inet6_recvmsg+0xbb/0x240 net/ipv6/af_inet6.c:592
 sock_recvmsg_nosec net/socket.c:871 [inline]
 sock_recvmsg net/socket.c:889 [inline]
 sock_recvmsg+0x92/0xb0 net/socket.c:885
 sock_read_iter+0x15f/0x1e0 net/socket.c:967
 call_read_iter include/linux/fs.h:1889 [inline]
 new_sync_read+0x389/0x4f0 fs/read_write.c:414
 __vfs_read+0xb1/0xc0 fs/read_write.c:427
 vfs_read fs/read_write.c:461 [inline]
 vfs_read+0x143/0x2c0 fs/read_write.c:446
 ksys_read+0xd5/0x1b0 fs/read_write.c:587
 __do_sys_read fs/read_write.c:597 [inline]
 __se_sys_read fs/read_write.c:595 [inline]
 __x64_sys_read+0x4c/0x60 fs/read_write.c:595
 do_syscall_64+0xcc/0x370 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 11168 Comm: syz-executor.0 Not tainted 5.4.0-rc6+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011

Fixes: 943170998b20 ("tun: enable NAPI for TUN/TAP driver")
Signed-off-by: Petar Penkov <ppenkov@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agospi: tegra20-slink: add missed clk_unprepare
Chuhong Yuan [Fri, 15 Nov 2019 08:31:22 +0000 (16:31 +0800)]
spi: tegra20-slink: add missed clk_unprepare

[ Upstream commit 04358e40ba96d687c0811c21d9dede73f5244a98 ]

The driver misses calling clk_unprepare in probe failure and remove.
Add the calls to fix it.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Link: https://lore.kernel.org/r/20191115083122.12278-1-hslester96@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoregulator: core: Let boot-on regulators be powered off
Pascal Paillet [Wed, 13 Nov 2019 10:27:37 +0000 (11:27 +0100)]
regulator: core: Let boot-on regulators be powered off

[ Upstream commit 089b3f61ecfc43ca4ea26d595e1d31ead6de3f7b ]

Boot-on regulators are always kept on because their use_count value
is now incremented at boot time and never cleaned.

Only increment count value for alway-on regulators.
regulator_late_cleanup() is now able to power off boot-on regulators
when unused.

Fixes: 05f224ca6693 ("regulator: core: Clean enabling always-on regulators + their supplies")
Signed-off-by: Pascal Paillet <p.paillet@st.com>
Link: https://lore.kernel.org/r/20191113102737.27831-1-p.paillet@st.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: wm8904: fix regcache handling
Michael Walle [Tue, 12 Nov 2019 22:36:29 +0000 (23:36 +0100)]
ASoC: wm8904: fix regcache handling

[ Upstream commit e9149b8c00d25dbaef1aa174fc604bed207e576d ]

The current code assumes that the power is turned off in
SND_SOC_BIAS_OFF. If there are no actual regulator the codec isn't
turned off and the registers are not reset to their default values but
the regcache is still marked as dirty. Thus a value might not be written
to the hardware if it is set to the default value. Do a software reset
before turning off the power to make sure the registers are always reset
to their default states.

Signed-off-by: Michael Walle <michael@walle.cc>
Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20191112223629.21867-1-michael@walle.cc
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoiwlwifi: mvm: fix unaligned read of rx_pkt_status
Wang Xuerui [Fri, 15 Nov 2019 07:28:02 +0000 (09:28 +0200)]
iwlwifi: mvm: fix unaligned read of rx_pkt_status

[ Upstream commit c5aaa8be29b25dfe1731e9a8b19fd91b7b789ee3 ]

This is present since the introduction of iwlmvm.
Example stack trace on MIPS:

[<ffffffffc0789328>] iwl_mvm_rx_rx_mpdu+0xa8/0xb88 [iwlmvm]
[<ffffffffc0632b40>] iwl_pcie_rx_handle+0x420/0xc48 [iwlwifi]

Tested with a Wireless AC 7265 for ~6 months, confirmed to fix the
problem. No other unaligned accesses are spotted yet.

Signed-off-by: Wang Xuerui <wangxuerui@qiniu.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobcache: fix deadlock in bcache_allocator
Andrea Righi [Wed, 13 Nov 2019 08:03:21 +0000 (16:03 +0800)]
bcache: fix deadlock in bcache_allocator

[ Upstream commit 84c529aea182939e68f618ed9813740c9165c7eb ]

bcache_allocator can call the following:

 bch_allocator_thread()
  -> bch_prio_write()
     -> bch_bucket_alloc()
        -> wait on &ca->set->bucket_wait

But the wake up event on bucket_wait is supposed to come from
bch_allocator_thread() itself => deadlock:

[ 1158.490744] INFO: task bcache_allocato:15861 blocked for more than 10 seconds.
[ 1158.495929]       Not tainted 5.3.0-050300rc3-generic #201908042232
[ 1158.500653] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 1158.504413] bcache_allocato D    0 15861      2 0x80004000
[ 1158.504419] Call Trace:
[ 1158.504429]  __schedule+0x2a8/0x670
[ 1158.504432]  schedule+0x2d/0x90
[ 1158.504448]  bch_bucket_alloc+0xe5/0x370 [bcache]
[ 1158.504453]  ? wait_woken+0x80/0x80
[ 1158.504466]  bch_prio_write+0x1dc/0x390 [bcache]
[ 1158.504476]  bch_allocator_thread+0x233/0x490 [bcache]
[ 1158.504491]  kthread+0x121/0x140
[ 1158.504503]  ? invalidate_buckets+0x890/0x890 [bcache]
[ 1158.504506]  ? kthread_park+0xb0/0xb0
[ 1158.504510]  ret_from_fork+0x35/0x40

Fix by making the call to bch_prio_write() non-blocking, so that
bch_allocator_thread() never waits on itself.

Moreover, make sure to wake up the garbage collector thread when
bch_prio_write() is failing to allocate buckets.

BugLink: https://bugs.launchpad.net/bugs/1784665
BugLink: https://bugs.launchpad.net/bugs/1796292
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
Signed-off-by: Coly Li <colyli@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agotracing/kprobe: Check whether the non-suffixed symbol is notrace
Masami Hiramatsu [Tue, 29 Oct 2019 08:31:44 +0000 (17:31 +0900)]
tracing/kprobe: Check whether the non-suffixed symbol is notrace

[ Upstream commit c7411a1a126f649be71526a36d4afac9e5aefa13 ]

Check whether the non-suffixed symbol is notrace, since suffixed
symbols are generated by the compilers for optimization. Based on
these suffixed symbols, notrace check might not work because
some of them are just a partial code of the original function.
(e.g. cold-cache (unlikely) code is separated from original
 function as FUNCTION.cold.XX)

For example, without this fix,
  # echo p device_add.cold.67 > /sys/kernel/debug/tracing/kprobe_events
  sh: write error: Invalid argument

  # cat /sys/kernel/debug/tracing/error_log
  [  135.491035] trace_kprobe: error: Failed to register probe event
    Command: p device_add.cold.67
               ^
  # dmesg | tail -n 1
  [  135.488599] trace_kprobe: Could not probe notrace function device_add.cold.67

With this,
  # echo p device_add.cold.66 > /sys/kernel/debug/tracing/kprobe_events
  # cat /sys/kernel/debug/kprobes/list
  ffffffff81599de9  k  device_add.cold.66+0x0    [DISABLED]

Actually, kprobe blacklist already did similar thing,
see within_kprobe_blacklist().

Link: http://lkml.kernel.org/r/157233790394.6706.18243942030937189679.stgit@devnote2
Fixes: 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoMIPS: ralink: enable PCI support only if driver for mt7621 SoC is selected
Sergio Paracuellos [Sat, 19 Oct 2019 08:12:33 +0000 (10:12 +0200)]
MIPS: ralink: enable PCI support only if driver for mt7621 SoC is selected

[ Upstream commit 3b2fa0c92686562ac0b8cf00c0326a45814f8e18 ]

Some versions of SoC MT7621 have three PCI express hosts. Some boards
make use of those PCI through the staging driver mt7621-pci. Recently
PCI support has been removed from MT7621 Soc kernel configuration due
to a build error. This makes imposible to compile staging driver and
produces a regression for gnubee based boards. Enable support for PCI
again but enable it only if staging mt7621-pci driver is selected.

Fixes: c4d48cf5e2f0 ("MIPS: ralink: deactivate PCI support for SOC_MT7621")
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Cc: ralf@linux-mips.org
Cc: jhogan@kernel.org
Cc: john@phrozen.org
Cc: NeilBrown <neil@brown.name>
Cc: linux-mips@vger.kernel.org
Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20191019081233.7337-1-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agotracing: use kvcalloc for tgid_map array allocation
Yuming Han [Thu, 24 Oct 2019 03:34:30 +0000 (11:34 +0800)]
tracing: use kvcalloc for tgid_map array allocation

[ Upstream commit 6ee40511cb838f9ced002dff7131bca87e3ccbdd ]

Fail to allocate memory for tgid_map, because it requires order-6 page.
detail as:

c3 sh: page allocation failure: order:6,
   mode:0x140c0c0(GFP_KERNEL), nodemask=(null)
c3 sh cpuset=/ mems_allowed=0
c3 CPU: 3 PID: 5632 Comm: sh Tainted: G        W  O    4.14.133+ #10
c3 Hardware name: Generic DT based system
c3 Backtrace:
c3 [<c010bdbc>] (dump_backtrace) from [<c010c08c>](show_stack+0x18/0x1c)
c3 [<c010c074>] (show_stack) from [<c0993c54>](dump_stack+0x84/0xa4)
c3 [<c0993bd0>] (dump_stack) from [<c0229858>](warn_alloc+0xc4/0x19c)
c3 [<c0229798>] (warn_alloc) from [<c022a6e4>](__alloc_pages_nodemask+0xd18/0xf28)
c3 [<c02299cc>] (__alloc_pages_nodemask) from [<c0248344>](kmalloc_order+0x20/0x38)
c3 [<c0248324>] (kmalloc_order) from [<c0248380>](kmalloc_order_trace+0x24/0x108)
c3 [<c024835c>] (kmalloc_order_trace) from [<c01e6078>](set_tracer_flag+0xb0/0x158)
c3 [<c01e5fc8>] (set_tracer_flag) from [<c01e6404>](trace_options_core_write+0x7c/0xcc)
c3 [<c01e6388>] (trace_options_core_write) from [<c0278b1c>](__vfs_write+0x40/0x14c)
c3 [<c0278adc>] (__vfs_write) from [<c0278e10>](vfs_write+0xc4/0x198)
c3 [<c0278d4c>] (vfs_write) from [<c027906c>](SyS_write+0x6c/0xd0)
c3 [<c0279000>] (SyS_write) from [<c01079a0>](ret_fast_syscall+0x0/0x54)

Switch to use kvcalloc to avoid unexpected allocation failures.

Link: http://lkml.kernel.org/r/1571888070-24425-1-git-send-email-chunyan.zhang@unisoc.com
Signed-off-by: Yuming Han <yuming.han@unisoc.com>
Signed-off-by: Chunyan Zhang <chunyan.zhang@unisoc.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoRDMA/efa: Clear the admin command buffer prior to its submission
Gal Pressman [Tue, 12 Nov 2019 09:26:08 +0000 (11:26 +0200)]
RDMA/efa: Clear the admin command buffer prior to its submission

[ Upstream commit 64c264872b8879e2ab9017eefe9514d4c045c60e ]

We cannot rely on the entry memcpy as we only copy the actual size of the
command, the rest of the bytes must be memset to zero.

Currently providing non-zero memory will not have any user visible impact.
However, since admin commands are extendable (in a backwards compatible
way) everything beyond the size of the command must be cleared to prevent
issues in the future.

Fixes: 0420e542569b ("RDMA/efa: Implement functions that submit and complete admin commands")
Link: https://lore.kernel.org/r/20191112092608.46964-1-galpress@amazon.com
Reviewed-by: Daniel Kranzdorf <dkkranzd@amazon.com>
Reviewed-by: Firas JahJah <firasj@amazon.com>
Signed-off-by: Gal Pressman <galpress@amazon.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoqtnfmac: fix using skb after free
Sergey Matyukevich [Wed, 13 Nov 2019 11:06:47 +0000 (11:06 +0000)]
qtnfmac: fix using skb after free

[ Upstream commit 4a33f21cef84b1b933958c99ed5dac1726214b35 ]

KASAN reported use-after-free error:

[  995.220767] BUG: KASAN: use-after-free in qtnf_cmd_send_with_reply+0x169/0x3e0 [qtnfmac]
[  995.221098] Read of size 2 at addr ffff888213d1ded0 by task kworker/1:1/71

The issue in qtnf_cmd_send_with_reply impacts all the commands that do
not need response other then return code. For such commands, consume_skb
is used for response skb and right after that return code in response
skb is accessed.

Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agox86/crash: Add a forward declaration of struct kimage
Lianbo Jiang [Fri, 8 Nov 2019 09:00:27 +0000 (17:00 +0800)]
x86/crash: Add a forward declaration of struct kimage

[ Upstream commit 112eee5d06007dae561f14458bde7f2a4879ef4e ]

Add a forward declaration of struct kimage to the crash.h header because
future changes will invoke a crash-specific function from the realmode
init path and the compiler will complain otherwise like this:

  In file included from arch/x86/realmode/init.c:11:
  ./arch/x86/include/asm/crash.h:5:32: warning: ‘struct kimage’ declared inside\
   parameter list will not be visible outside of this definition or declaration
      5 | int crash_load_segments(struct kimage *image);
        |                                ^~~~~~
  ./arch/x86/include/asm/crash.h:6:37: warning: ‘struct kimage’ declared inside\
   parameter list will not be visible outside of this definition or declaration
      6 | int crash_copy_backup_region(struct kimage *image);
        |                                     ^~~~~~
  ./arch/x86/include/asm/crash.h:7:39: warning: ‘struct kimage’ declared inside\
   parameter list will not be visible outside of this definition or declaration
      7 | int crash_setup_memmap_entries(struct kimage *image,
        |

 [ bp: Rewrite the commit message. ]

Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Lianbo Jiang <lijiang@redhat.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: bhe@redhat.com
Cc: d.hatayama@fujitsu.com
Cc: dhowells@redhat.com
Cc: dyoung@redhat.com
Cc: ebiederm@xmission.com
Cc: horms@verge.net.au
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jürgen Gross <jgross@suse.com>
Cc: kexec@lists.infradead.org
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: vgoyal@redhat.com
Cc: x86-ml <x86@kernel.org>
Link: https://lkml.kernel.org/r/20191108090027.11082-4-lijiang@redhat.com
Link: https://lkml.kernel.org/r/201910310233.EJRtTMWP%25lkp@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoqtnfmac: fix invalid channel information output
Sergey Matyukevich [Wed, 13 Nov 2019 11:06:51 +0000 (11:06 +0000)]
qtnfmac: fix invalid channel information output

[ Upstream commit 24227a9e956a7c9913a7e6e7199a9ae3f540fe88 ]

Do not attempt to print frequency for an invalid channel
provided by firmware. That channel may simply not exist.

Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoqtnfmac: fix debugfs support for multiple cards
Sergey Matyukevich [Wed, 13 Nov 2019 11:06:49 +0000 (11:06 +0000)]
qtnfmac: fix debugfs support for multiple cards

[ Upstream commit dd4c2260dab04f5ae7bdb79b9470e7da56f48145 ]

Fix merge artifact for commit 0b68fe10b8e8 ("qtnfmac: modify debugfs
to support multiple cards") and finally add debugfs support
for multiple qtnfmac wireless cards.

Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocpufreq: Register drivers only after CPU devices have been registered
Viresh Kumar [Thu, 14 Nov 2019 03:36:17 +0000 (09:06 +0530)]
cpufreq: Register drivers only after CPU devices have been registered

[ Upstream commit 46770be0cf94149ca48be87719bda1d951066644 ]

The cpufreq core heavily depends on the availability of the struct
device for CPUs and if they aren't available at the time cpufreq driver
is registered, we will never succeed in making cpufreq work.

This happens due to following sequence of events:

- cpufreq_register_driver()
  - subsys_interface_register()
  - return 0; //successful registration of driver

... at a later point of time

- register_cpu();
  - device_register();
    - bus_probe_device();
      - sif->add_dev();
- cpufreq_add_dev();
  - get_cpu_device(); //FAILS
  - per_cpu(cpu_sys_devices, num) = &cpu->dev; //used by get_cpu_device()
  - return 0; //CPU registered successfully

Because the per-cpu variable cpu_sys_devices is set only after the CPU
device is regsitered, cpufreq will never be able to get it when
cpufreq_add_dev() is called.

This patch avoids this failure by making sure device structure of at
least CPU0 is available when the cpufreq driver is registered, else
return -EPROBE_DEFER.

Reported-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Co-developed-by: Amit Kucheria <amit.kucheria@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Tested-by: Amit Kucheria <amit.kucheria@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobcache: fix static checker warning in bcache_device_free()
Coly Li [Wed, 13 Nov 2019 08:03:17 +0000 (16:03 +0800)]
bcache: fix static checker warning in bcache_device_free()

[ Upstream commit 2d8869518a525c9bce5f5268419df9dfbe3dfdeb ]

Commit cafe56359144 ("bcache: A block layer cache") leads to the
following static checker warning:

    ./drivers/md/bcache/super.c:770 bcache_device_free()
    warn: variable dereferenced before check 'd->disk' (see line 766)

drivers/md/bcache/super.c
   762  static void bcache_device_free(struct bcache_device *d)
   763  {
   764          lockdep_assert_held(&bch_register_lock);
   765
   766          pr_info("%s stopped", d->disk->disk_name);
                                      ^^^^^^^^^
Unchecked dereference.

   767
   768          if (d->c)
   769                  bcache_device_detach(d);
   770          if (d->disk && d->disk->flags & GENHD_FL_UP)
                    ^^^^^^^
Check too late.

   771                  del_gendisk(d->disk);
   772          if (d->disk && d->disk->queue)
   773                  blk_cleanup_queue(d->disk->queue);
   774          if (d->disk) {
   775                  ida_simple_remove(&bcache_device_idx,
   776                                    first_minor_to_idx(d->disk->first_minor));
   777                  put_disk(d->disk);
   778          }
   779

It is not 100% sure that the gendisk struct of bcache device will always
be there, the warning makes sense when there is problem in block core.

This patch tries to remove the static checking warning by checking
d->disk to avoid NULL pointer deferences.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Coly Li <colyli@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoparport: load lowlevel driver if ports not found
Sudip Mukherjee [Wed, 16 Oct 2019 14:45:39 +0000 (15:45 +0100)]
parport: load lowlevel driver if ports not found

[ Upstream commit 231ec2f24dad18d021b361045bbd618ba62a274e ]

Usually all the distro will load the parport low level driver as part
of their initialization. But we can get into a situation where all the
parallel port drivers are built as module and we unload all the modules
at a later time. Then if we just do "modprobe parport" it will only
load the parport module and will not load the low level driver which
will actually register the ports. So, check the bus if there is any
parport registered, if not, load the low level driver.

We can get into the above situation with all distro but only Suse has
setup the alias for "parport_lowlevel" and so it only works in Suse.
Users of Debian based distro will need to load the lowlevel module
manually.

Signed-off-by: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Link: https://lore.kernel.org/r/20191016144540.18810-3-sudipm.mukherjee@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonvme: Discard workaround for non-conformant devices
Eduard Hasenleithner [Tue, 12 Nov 2019 20:55:01 +0000 (21:55 +0100)]
nvme: Discard workaround for non-conformant devices

[ Upstream commit 530436c45ef2e446c12538a400e465929a0b3ade ]

Users observe IOMMU related errors when performing discard on nvme from
non-compliant nvme devices reading beyond the end of the DMA mapped
ranges to discard.

Two different variants of this behavior have been observed: SM22XX
controllers round up the read size to a multiple of 512 bytes, and Phison
E12 unconditionally reads the maximum discard size allowed by the spec
(256 segments or 4kB).

Make nvme_setup_discard unconditionally allocate the maximum DSM buffer
so the driver DMA maps a memory range that will always succeed.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=202665
Signed-off-by: Eduard Hasenleithner <eduard@hasenleithner.at>
[changelog, use existing define, kernel coding style]
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: ethernet: ti: Add dependency for TI_DAVINCI_EMAC
Mao Wenan [Tue, 12 Nov 2019 06:33:58 +0000 (14:33 +0800)]
net: ethernet: ti: Add dependency for TI_DAVINCI_EMAC

[ Upstream commit b2ef81dcdf3835bd55e5f97ff30131bb327be7fa ]

If TI_DAVINCI_EMAC=y and GENERIC_ALLOCATOR is not set,
below erros can be seen:
drivers/net/ethernet/ti/davinci_cpdma.o: In function `cpdma_desc_pool_destroy.isra.14':
davinci_cpdma.c:(.text+0x359): undefined reference to `gen_pool_size'
davinci_cpdma.c:(.text+0x365): undefined reference to `gen_pool_avail'
davinci_cpdma.c:(.text+0x373): undefined reference to `gen_pool_avail'
davinci_cpdma.c:(.text+0x37f): undefined reference to `gen_pool_size'
drivers/net/ethernet/ti/davinci_cpdma.o: In function `__cpdma_chan_free':
davinci_cpdma.c:(.text+0x4a2): undefined reference to `gen_pool_free_owner'
drivers/net/ethernet/ti/davinci_cpdma.o: In function `cpdma_chan_submit_si':
davinci_cpdma.c:(.text+0x66c): undefined reference to `gen_pool_alloc_algo_owner'
davinci_cpdma.c:(.text+0x805): undefined reference to `gen_pool_free_owner'
drivers/net/ethernet/ti/davinci_cpdma.o: In function `cpdma_ctlr_create':
davinci_cpdma.c:(.text+0xabd): undefined reference to `devm_gen_pool_create'
davinci_cpdma.c:(.text+0xb79): undefined reference to `gen_pool_add_owner'
drivers/net/ethernet/ti/davinci_cpdma.o: In function `cpdma_check_free_tx_desc':
davinci_cpdma.c:(.text+0x16c6): undefined reference to `gen_pool_avail'

This patch mades TI_DAVINCI_EMAC select GENERIC_ALLOCATOR.

Fixes: 99f629718272 ("net: ethernet: ti: cpsw: drop TI_DAVINCI_CPDMA config option")
Signed-off-by: Mao Wenan <maowenan@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agos390/disassembler: don't hide instruction addresses
Ilya Leoshkevich [Thu, 31 Oct 2019 17:25:16 +0000 (18:25 +0100)]
s390/disassembler: don't hide instruction addresses

[ Upstream commit 544f1d62e3e6c6e6d17a5e56f6139208acb5ff46 ]

Due to kptr_restrict, JITted BPF code is now displayed like this:

000000000b6ed1b2ebdff0800024  stmg    %r13,%r15,128(%r15)
000000004cde2ba041d0f040      la      %r13,64(%r15)
00000000fbad41b0a7fbffa0      aghi    %r15,-96

Leaking kernel addresses to dmesg is not a concern in this case, because
this happens only when JIT debugging is explicitly activated, which only
root can do.

Use %px in this particular instance, and also to print an instruction
address in show_code and PCREL (e.g. brasl) arguments in print_insn.
While at present functionally equivalent to %016lx, %px is recommended
by Documentation/core-api/printk-formats.rst for such cases.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agor8169: respect EEE user setting when restarting network
Heiner Kallweit [Sun, 10 Nov 2019 13:44:54 +0000 (14:44 +0100)]
r8169: respect EEE user setting when restarting network

[ Upstream commit 7ec3f872bc85ada93db34448d73bb399d6b82c2c ]

Currently, if network is re-started, we advertise all supported EEE
modes, thus potentially overriding a manual adjustment the user made
e.g. via ethtool. Be friendly to the user and preserve a manual
setting on network re-start.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: dsa: sja1105: Disallow management xmit during switch reset
Vladimir Oltean [Sat, 9 Nov 2019 11:32:24 +0000 (13:32 +0200)]
net: dsa: sja1105: Disallow management xmit during switch reset

[ Upstream commit af580ae2dcb250719857b4b7024bd4bb0c2e05fb ]

The purpose here is to avoid ptp4l fail due to this condition:

  timed out while polling for tx timestamp
  increasing tx_timestamp_timeout may correct this issue, but it is likely caused by a driver bug
  port 1: send peer delay request failed

So either reset the switch before the management frame was sent, or
after it was timestamped as well, but not in the middle.

The condition may arise either due to a true timeout (i.e. because
re-uploading the static config takes time), or due to the TX timestamp
actually getting lost due to reset. For the former we can increase
tx_timestamp_timeout in userspace, for the latter we need this patch.

Locking all traffic during switch reset does not make sense at all,
though. Forcing all CPU-originated traffic to potentially block waiting
for a sleepable context to send > 800 bytes over SPI is not a good idea.
Flows that are autonomously forwarded by the switch will get dropped
anyway during switch reset no matter what. So just let all other
CPU-originated traffic be dropped as well.

Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: Intel: kbl_rt5663_rt5514_max98927: Add dmic format constraint
Yu-Hsuan Hsu [Mon, 23 Sep 2019 16:29:40 +0000 (00:29 +0800)]
ASoC: Intel: kbl_rt5663_rt5514_max98927: Add dmic format constraint

[ Upstream commit e2db787bdcb4f2722ecf410168f0583764634e45 ]

On KBL platform, the microphone is attached to external codec(rt5514)
instead of PCH. However, TDM slot between PCH and codec is 16 bits only.
In order to avoid setting wrong format, we should add a constraint to
force to use 16 bits format forever.

Signed-off-by: Yu-Hsuan Hsu <yuhsuan@chromium.org>
Acked-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190923162940.199580-1-yuhsuan@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agobpf, testing: Workaround a verifier failure for test_progs
Yonghong Song [Thu, 7 Nov 2019 17:00:45 +0000 (09:00 -0800)]
bpf, testing: Workaround a verifier failure for test_progs

[ Upstream commit b7a0d65d80a0c5034b366392624397a0915b7556 ]

With latest llvm compiler, running test_progs will have the following
verifier failure for test_sysctl_loop1.o:

  libbpf: load bpf program failed: Permission denied
  libbpf: -- BEGIN DUMP LOG ---
  libbpf:
  invalid indirect read from stack var_off (0x0; 0xff)+196 size 7
  ...
  libbpf: -- END LOG --
  libbpf: failed to load program 'cgroup/sysctl'
  libbpf: failed to load object 'test_sysctl_loop1.o'

The related bytecode looks as below:

  0000000000000308 LBB0_8:
      97:       r4 = r10
      98:       r4 += -288
      99:       r4 += r7
     100:       w8 &= 255
     101:       r1 = r10
     102:       r1 += -488
     103:       r1 += r8
     104:       r2 = 7
     105:       r3 = 0
     106:       call 106
     107:       w1 = w0
     108:       w1 += -1
     109:       if w1 > 6 goto -24 <LBB0_5>
     110:       w0 += w8
     111:       r7 += 8
     112:       w8 = w0
     113:       if r7 != 224 goto -17 <LBB0_8>

And source code:

     for (i = 0; i < ARRAY_SIZE(tcp_mem); ++i) {
             ret = bpf_strtoul(value + off, MAX_ULONG_STR_LEN, 0,
                               tcp_mem + i);
             if (ret <= 0 || ret > MAX_ULONG_STR_LEN)
                     return 0;
             off += ret & MAX_ULONG_STR_LEN;
     }

Current verifier is not able to conclude that register w0 before '+'
at insn 110 has a range of 1 to 7 and thinks it is from 0 - 255. This
leads to more conservative range for w8 at insn 112, and later verifier
complaint.

Let us workaround this issue until we found a compiler and/or verifier
solution. The workaround in this patch is to make variable 'ret' volatile,
which will force a reload and then '&' operation to ensure better value
range. With this patch, I got the below byte code for the loop:

  0000000000000328 LBB0_9:
     101:       r4 = r10
     102:       r4 += -288
     103:       r4 += r7
     104:       w8 &= 255
     105:       r1 = r10
     106:       r1 += -488
     107:       r1 += r8
     108:       r2 = 7
     109:       r3 = 0
     110:       call 106
     111:       *(u32 *)(r10 - 64) = r0
     112:       r1 = *(u32 *)(r10 - 64)
     113:       if w1 s< 1 goto -28 <LBB0_5>
     114:       r1 = *(u32 *)(r10 - 64)
     115:       if w1 s> 7 goto -30 <LBB0_5>
     116:       r1 = *(u32 *)(r10 - 64)
     117:       w1 &= 7
     118:       w1 += w8
     119:       r7 += 8
     120:       w8 = w1
     121:       if r7 != 224 goto -21 <LBB0_9>

Insn 117 did the '&' operation and we got more precise value range
for 'w8' at insn 120. The test is happy then:

  #3/17 test_sysctl_loop1.o:OK

Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20191107170045.2503480-1-yhs@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoiio: dac: ad5446: Add support for new AD5600 DAC
Stefan Popa [Wed, 6 Nov 2019 09:47:21 +0000 (11:47 +0200)]
iio: dac: ad5446: Add support for new AD5600 DAC

[ Upstream commit 6376cbe549fffb378403cee78efd26b8a2c8e450 ]

The AD5600 is a single channel, 16-bit resolution, voltage output digital
to analog converter (DAC). The AD5600 uses a 3-wire SPI interface. It is
part of the AD5541 family of DACs.

The ad5446 IIO driver implements support for some of these DACs (in the
AD5441 family), so the change is a simple entry in this driver.

Link: https://www.analog.com/media/en/technical-documentation/data-sheets/AD5600.pdf
Signed-off-by: Stefan Popa <stefan.popa@analog.com>
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoASoC: rt5677: Mark reg RT5677_PWR_ANLG2 as volatile
Ben Zhang [Wed, 6 Nov 2019 01:13:30 +0000 (17:13 -0800)]
ASoC: rt5677: Mark reg RT5677_PWR_ANLG2 as volatile

[ Upstream commit eabf424f7b60246c76dcb0ea6f1e83ef9abbeaa6 ]

The codec dies when RT5677_PWR_ANLG2(MX-64h) is set to 0xACE1
while it's streaming audio over SPI. The DSP firmware turns
on PLL2 (MX-64 bit 8) when SPI streaming starts.  However regmap
does not believe that register can change by itself. When
BST1 (bit 15) is turned on with regmap_update_bits(), it doesn't
read the register first before write, so PLL2 power bit is
cleared by accident.

Marking MX-64h as volatile in regmap solved the issue.

Signed-off-by: Ben Zhang <benzh@chromium.org>
Signed-off-by: Curtis Malainey <cujomalainey@chromium.org>
Link: https://lore.kernel.org/r/20191106011335.223061-6-cujomalainey@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agospi: pxa2xx: Add missed security checks
Chuhong Yuan [Sat, 9 Nov 2019 08:09:43 +0000 (16:09 +0800)]
spi: pxa2xx: Add missed security checks

[ Upstream commit 5eb263ef08b5014cfc2539a838f39d2fd3531423 ]

pxa2xx_spi_init_pdata misses checks for devm_clk_get and
platform_get_irq.
Add checks for them to fix the bugs.

Since ssp->clk and ssp->irq are used in probe, they are mandatory here.
So we cannot use _optional() for devm_clk_get and platform_get_irq.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Link: https://lore.kernel.org/r/20191109080943.30428-1-hslester96@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: vim2m: media_device_cleanup was called too early
Hans Verkuil [Sat, 9 Nov 2019 13:03:08 +0000 (14:03 +0100)]
media: vim2m: media_device_cleanup was called too early

[ Upstream commit 9f22e88a4bba270d3427684cee84dfbf67489e86 ]

Running the contrib/test/test-media script in v4l-utils with the vim2m argument
will cause this kernel warning:

[  554.430157] ------------[ cut here ]------------
[  554.433034] DEBUG_LOCKS_WARN_ON(lock->magic != lock)
[  554.433064] WARNING: CPU: 0 PID: 616 at kernel/locking/mutex.c:938 __mutex_lock+0xd7a/0x1380
[  554.439736] Modules linked in: vim2m v4l2_mem2mem vivid rc_cec videobuf2_dma_contig v4l2_dv_timings cec videobuf2_vmalloc videobuf2_memops v4l2_tpg videobuf2_v4l2 videobuf2_common videodev mc rc_core [last unloaded: vivid]
[  554.445794] CPU: 0 PID: 616 Comm: sleep Not tainted 5.4.0-rc1-virtme #1
[  554.448481] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-0-ga698c8995f-prebuilt.qemu.org 04/01/2014
[  554.453088] RIP: 0010:__mutex_lock+0xd7a/0x1380
[  554.454955] Code: d2 0f 85 de 05 00 00 44 8b 05 82 d9 f7 00 45 85 c0 0f 85 bf f3 ff ff 48 c7 c6 e0 30 a6 b7 48 c7 c7 e0 2e a6 b7 e8 5c 76 36 fe <0f> 0b e9 a5 f3 ff ff 65 48 8b 1c 25 80 ef 01 00 be 08 00 00 00 48
[  554.462836] RSP: 0018:ffff88803a4cfad0 EFLAGS: 00010282
[  554.465129] RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffffb5a3d24f
[  554.468143] RDX: 0000000000000000 RSI: 0000000000000004 RDI: ffffffffb85273f4
[  554.471000] RBP: ffff88803a4cfc50 R08: fffffbfff701e681 R09: fffffbfff701e681
[  554.473990] R10: fffffbfff701e680 R11: ffffffffb80f3403 R12: 0000000000000000
[  554.476831] R13: dffffc0000000000 R14: ffffffffb9714f00 R15: ffff888053103fc8
[  554.479622] FS:  00007fac6358a540(0000) GS:ffff88805d000000(0000) knlGS:0000000000000000
[  554.482673] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  554.484949] CR2: 00007fac6343faf0 CR3: 0000000036c22000 CR4: 00000000003406f0
[  554.487811] Call Trace:
[  554.488860]  ? v4l2_release+0x1b8/0x390 [videodev]
[  554.490818]  ? do_exit+0x946/0x2980
[  554.492269]  ? mutex_lock_io_nested+0x1250/0x1250
[  554.494128]  ? __lock_acquire+0xe90/0x3c30
[  554.495774]  ? fsnotify_first_mark+0x120/0x120
[  554.497487]  ? vim2m_device_release+0x50/0x50 [vim2m]
[  554.499469]  ? v4l2_release+0x1b8/0x390 [videodev]
[  554.501493]  v4l2_release+0x1b8/0x390 [videodev]
[  554.503430]  __fput+0x256/0x790
[  554.504711]  task_work_run+0x109/0x190
[  554.506145]  do_exit+0x95e/0x2980
[  554.507421]  ? vfs_lock_file+0x21/0xf0
[  554.509013]  ? find_held_lock+0x33/0x1c0
[  554.510382]  ? __close_fd+0xee/0x190
[  554.511862]  ? release_task.part.21+0x1310/0x1310
[  554.513701]  ? lock_downgrade+0x6d0/0x6d0
[  554.515299]  do_group_exit+0xeb/0x2d0
[  554.516862]  __x64_sys_exit_group+0x35/0x40
[  554.518610]  do_syscall_64+0x90/0x450
[  554.520142]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
[  554.522289] RIP: 0033:0x7fac6348ecf6
[  554.523876] Code: Bad RIP value.
[  554.525294] RSP: 002b:00007ffe6373dc58 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
[  554.528555] RAX: ffffffffffffffda RBX: 00007fac6357f760 RCX: 00007fac6348ecf6
[  554.531537] RDX: 0000000000000000 RSI: 000000000000003c RDI: 0000000000000000
[  554.534709] RBP: 0000000000000000 R08: 00000000000000e7 R09: ffffffffffffff80
[  554.536752] R10: 00007ffe6373db24 R11: 0000000000000246 R12: 00007fac6357f760
[  554.538643] R13: 0000000000000002 R14: 00007fac63588428 R15: 0000000000000000
[  554.540634] irq event stamp: 21731
[  554.541618] hardirqs last  enabled at (21731): [<ffffffffb75b3cd4>] _raw_spin_unlock_irq+0x24/0x30
[  554.544145] hardirqs last disabled at (21730): [<ffffffffb75b3ada>] _raw_spin_lock_irq+0xa/0x40
[  554.547027] softirqs last  enabled at (20148): [<ffffffffb780064d>] __do_softirq+0x64d/0x906
[  554.550385] softirqs last disabled at (19857): [<ffffffffb5926bd5>] irq_exit+0x175/0x1a0
[  554.553668] ---[ end trace a389c80c2ca84244 ]---

This is caused by media_device_cleanup() which destroys
v4l2_dev->mdev->req_queue_mutex. But v4l2_release() tries to lock
that mutex after media_device_cleanup() is called.

By moving media_device_cleanup() to the video_device's release function it is
guaranteed that the mutex is valid whenever v4l2_release is called.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: vicodec: media_device_cleanup was called too early
Hans Verkuil [Sat, 9 Nov 2019 14:06:18 +0000 (15:06 +0100)]
media: vicodec: media_device_cleanup was called too early

[ Upstream commit 693c5f144aeb9636ae161a3c61a838c50b2ae41c ]

Running the contrib/test/test-media script in v4l-utils with the vicodec argument
will cause this kernel warning:

[  372.298824] ------------[ cut here ]------------
[  372.298848] DEBUG_LOCKS_WARN_ON(lock->magic != lock)
[  372.298896] WARNING: CPU: 11 PID: 2220 at kernel/locking/mutex.c:938 __mutex_lock+0x919/0xc10
[  372.298907] Modules linked in: vicodec v4l2_mem2mem vivid rc_cec v4l2_tpg videobuf2_dma_contig cec rc_core v4l2_dv_timings videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 videobuf2_common videodev mc vmw_balloon vmw_vmci button vmwgfx [last unloaded: vimc]
[  372.298961] CPU: 11 PID: 2220 Comm: sleep Not tainted 5.4.0-rc1-test-no #150
[  372.298970] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/29/2019
[  372.298983] RIP: 0010:__mutex_lock+0x919/0xc10
[  372.298995] Code: 59 83 e8 9a fc 16 ff 44 8b 05 23 61 38 01 45 85 c0 0f 85 ef f7 ff ff 48 c7 c6 a0 1f 87 82 48 c7 c7 a0 1e 87 82 e8 cd bb f7 fe <0f> 0b e9 d5 f7 ff ff f6 c3 04 0f 84 3b fd ff ff 49 89 df 41 83 e7
[  372.299004] RSP: 0018:ffff8881b400fb80 EFLAGS: 00010286
[  372.299014] RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
[  372.299022] RDX: 0000000000000003 RSI: 0000000000000004 RDI: ffffed1036801f62
[  372.299030] RBP: ffff8881b400fcf0 R08: ffffffff81217c91 R09: fffffbfff061c271
[  372.299038] R10: fffffbfff061c270 R11: ffffffff830e1383 R12: ffff88814761dc80
[  372.299046] R13: 0000000000000000 R14: ffff88814761cbf0 R15: ffff88814761d030
[  372.299055] FS:  0000000000000000(0000) GS:ffff8881b68c0000(0000) knlGS:0000000000000000
[  372.299063] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  372.299071] CR2: 00007f606d78aa20 CR3: 0000000003013002 CR4: 00000000001606e0
[  372.299153] Call Trace:
[  372.299176]  ? __kasan_slab_free+0x12f/0x180
[  372.299187]  ? kmem_cache_free+0x9b/0x250
[  372.299200]  ? do_exit+0xcdf/0x1200
[  372.299210]  ? do_group_exit+0x85/0x130
[  372.299220]  ? __x64_sys_exit_group+0x23/0x30
[  372.299231]  ? do_syscall_64+0x5e/0x1c0
[  372.299241]  ? entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  372.299295]  ? v4l2_release+0xed/0x190 [videodev]
[  372.299309]  ? mutex_lock_io_nested+0xb80/0xb80
[  372.299323]  ? find_held_lock+0x85/0xa0
[  372.299335]  ? fsnotify+0x5b0/0x600
[  372.299351]  ? locks_remove_file+0x78/0x2b0
[  372.299363]  ? __fsnotify_update_child_dentry_flags.part.0+0x170/0x170
[  372.299383]  ? vidioc_querycap+0x50/0x50 [vicodec]
[  372.299426]  ? v4l2_release+0xed/0x190 [videodev]
[  372.299467]  v4l2_release+0xed/0x190 [videodev]
[  372.299484]  __fput+0x15a/0x390
[  372.299499]  task_work_run+0xb2/0xe0
[  372.299512]  do_exit+0x4d0/0x1200
[  372.299528]  ? do_user_addr_fault+0x367/0x610
[  372.299538]  ? release_task+0x990/0x990
[  372.299552]  ? rwsem_spin_on_owner+0x170/0x170
[  372.299567]  ? vmacache_find+0xb2/0x100
[  372.299580]  do_group_exit+0x85/0x130
[  372.299592]  __x64_sys_exit_group+0x23/0x30
[  372.299602]  do_syscall_64+0x5e/0x1c0
[  372.299614]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[  372.299624] RIP: 0033:0x7f606d74a9d6
[  372.299640] Code: Bad RIP value.
[  372.299648] RSP: 002b:00007fff65364468 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
[  372.299658] RAX: ffffffffffffffda RBX: 00007f606d83b760 RCX: 00007f606d74a9d6
[  372.299666] RDX: 0000000000000000 RSI: 000000000000003c RDI: 0000000000000000
[  372.299673] RBP: 0000000000000000 R08: 00000000000000e7 R09: ffffffffffffff80
[  372.299681] R10: 00007fff65364334 R11: 0000000000000246 R12: 00007f606d83b760
[  372.299689] R13: 0000000000000002 R14: 00007f606d844428 R15: 0000000000000000
[  372.299704] ---[ end trace add7d62ca4bc65e3 ]---

This is caused by media_device_cleanup() which destroys
v4l2_dev->mdev->req_queue_mutex. But v4l2_release() tries to lock
that mutex after media_device_cleanup() is called.

By moving media_device_cleanup() to the v4l2_device's release function it is
guaranteed that the mutex is valid whenever v4l2_release is called.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoEDAC/ghes: Fix grain calculation
Robert Richter [Wed, 6 Nov 2019 09:33:23 +0000 (09:33 +0000)]
EDAC/ghes: Fix grain calculation

[ Upstream commit 7088e29e0423d3195e09079b4f849ec4837e5a75 ]

The current code to convert a physical address mask to a grain
(defined as granularity in bytes) is:

e->grain = ~(mem_err->physical_addr_mask & ~PAGE_MASK);

This is broken in several ways:

1) It calculates to wrong grain values. E.g., a physical address mask
of ~0xfff should give a grain of 0x1000. Without considering
PAGE_MASK, there is an off-by-one. Things are worse when also
filtering it with ~PAGE_MASK. This will calculate to a grain with the
upper bits set. In the example it even calculates to ~0.

2) The grain does not depend on and is unrelated to the kernel's
page-size. The page-size only matters when unmapping memory in
memory_failure(). Smaller grains are wrongly rounded up to the
page-size, on architectures with a configurable page-size (e.g. arm64)
this could round up to the even bigger page-size of the hypervisor.

Fix this with:

e->grain = ~mem_err->physical_addr_mask + 1;

The grain_bits are defined as:

grain = 1 << grain_bits;

Change also the grain_bits calculation accordingly, it is the same
formula as in edac_mc.c now and the code can be unified.

The value in ->physical_addr_mask coming from firmware is assumed to
be contiguous, but this is not sanity-checked. However, in case the
mask is non-contiguous, a conversion to grain_bits effectively
converts the grain bit mask to a power of 2 by rounding it up.

Suggested-by: James Morse <james.morse@arm.com>
Signed-off-by: Robert Richter <rrichter@marvell.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Cc: "linux-edac@vger.kernel.org" <linux-edac@vger.kernel.org>
Cc: Tony Luck <tony.luck@intel.com>
Link: https://lkml.kernel.org/r/20191106093239.25517-11-rrichter@marvell.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoiio: cros_ec_baro: set info_mask_shared_by_all_available field
Gwendal Grignou [Wed, 6 Nov 2019 17:55:33 +0000 (09:55 -0800)]
iio: cros_ec_baro: set info_mask_shared_by_all_available field

[ Upstream commit e9a4cbcaaa391ef44d623d548ee715e77265030c ]

Field was already set for light/proximity and
accelerometer/gyroscope/magnetometer sensors.

Fixes: ae7b02ad2f32 ("iio: common: cros_ec_sensors: Expose cros_ec_sensors frequency range via iio sysfs")
Signed-off-by: Gwendal Grignou <gwendal@chromium.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: v4l2-ctrl: Lock main_hdl on operations of requests_queued.
Pi-Hsun Shih [Sun, 10 Nov 2019 06:29:10 +0000 (07:29 +0100)]
media: v4l2-ctrl: Lock main_hdl on operations of requests_queued.

[ Upstream commit df4a3e7f88e3b0d7ae46d70b9ff8e3c0ea730785 ]

There's a race condition between the list_del_init in the
v4l2_ctrl_request_complete, and the list_add_tail in the
v4l2_ctrl_request_queue, since they can be called in different thread
and the requests_queued list is not protected by a lock. This can lead
to that the v4l2_ctrl_handler is still in the requests_queued list while
the request_is_queued is already set to false, which would cause
use-after-free if the v4l2_ctrl_handler is later released.

Fix this by locking the ->lock of main_hdl (which is the owner of the
requests_queued list) when doing list operations on the
->requests_queued list.

Signed-off-by: Pi-Hsun Shih <pihsun@chromium.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: cedrus: Use helpers to access capture queue
Jernej Skrabec [Sat, 26 Oct 2019 07:27:52 +0000 (09:27 +0200)]
media: cedrus: Use helpers to access capture queue

[ Upstream commit 1fd50a2c294457508f06b8b631d01a58de81cdd2 ]

Accessing capture queue structue directly is not safe. Use helpers for
that.

Acked-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: si470x-i2c: add missed operations in remove
Chuhong Yuan [Sun, 10 Nov 2019 06:28:15 +0000 (07:28 +0100)]
media: si470x-i2c: add missed operations in remove

[ Upstream commit 2df200ab234a86836a8879a05a8007d6b884eb14 ]

The driver misses calling v4l2_ctrl_handler_free and
v4l2_device_unregister in remove like what is done in probe failure.
Add the calls to fix it.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoice: delay less
Mitch Williams [Wed, 6 Nov 2019 10:05:36 +0000 (02:05 -0800)]
ice: delay less

[ Upstream commit 88bb432a55de8ae62106305083a8bfbb23b01ad2 ]

Shorten the delay for SQ responses, but increase the number of loops.
Max delay time is unchanged, but some operations complete much more
quickly.

In the process, add a new define to make the delay count and delay time
more explicit. Add comments to make things more explicit.

This fixes a problem with VF resets failing on with many VFs.

Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocrypto: atmel - Fix authenc support when it is set to m
Herbert Xu [Mon, 28 Oct 2019 07:39:07 +0000 (15:39 +0800)]
crypto: atmel - Fix authenc support when it is set to m

[ Upstream commit 1520c72596dde7f22b8bd6bed3ef7df2b8b7ef39 ]

As it is if CONFIG_CRYPTO_DEV_ATMEL_AUTHENC is set to m it is in
effect disabled.  This patch fixes it by using IS_ENABLED instead
of ifdef.

Fixes: 89a82ef87e01 ("crypto: atmel-authenc - add support to...")
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Reviewed-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agosoundwire: intel: fix PDI/stream mapping for Bulk
Pierre-Louis Bossart [Tue, 22 Oct 2019 23:29:48 +0000 (18:29 -0500)]
soundwire: intel: fix PDI/stream mapping for Bulk

[ Upstream commit c134f914e9f55b7817e2bae625ec0e5f1379f7cd ]

The previous formula is incorrect for PDI0/1, the mapping is not
linear but has a discontinuity between PDI1 and PDI2.

This change has no effect on PCM PDIs (same mapping).

Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20191022232948.17156-1-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: pvrusb2: Fix oops on tear-down when radio support is not present
Mike Isely [Wed, 6 Nov 2019 11:11:14 +0000 (12:11 +0100)]
media: pvrusb2: Fix oops on tear-down when radio support is not present

[ Upstream commit 7f404ae9cf2a285f73b3c18ab9303d54b7a3d8e1 ]

In some device configurations there's no radio or radio support in the
driver.  That's OK, as the driver sets itself up accordingly.  However
on tear-down in these caes it's still trying to tear down radio
related context when there isn't anything there, leading to
dereferences through a null pointer and chaos follows.

How this bug survived unfixed for 11 years in the pvrusb2 driver is a
mystery to me.

[hverkuil: fix two checkpatch warnings]

Signed-off-by: Mike Isely <isely@pobox.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoselftests: net: Fix printf format warnings on arm
Masami Hiramatsu [Wed, 23 Oct 2019 04:58:07 +0000 (13:58 +0900)]
selftests: net: Fix printf format warnings on arm

[ Upstream commit 670cd6849ea36ea4df2f2941cf4717dff8755abe ]

Fix printf format warnings on arm (and other 32bit arch).

 - udpgso.c and udpgso_bench_tx use %lu for size_t but it
   should be unsigned long long on 32bit arch.

 - so_txtime.c uses %ld for int64_t, but it should be
   unsigned long long on 32bit arch.

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Willem de Bruijn <willemb@google.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agofsi: core: Fix small accesses and unaligned offsets via sysfs
Andrew Jeffery [Fri, 8 Nov 2019 05:19:39 +0000 (15:49 +1030)]
fsi: core: Fix small accesses and unaligned offsets via sysfs

[ Upstream commit 9f4c2b516b4f031e3cd0e45957f4150b3c1a083d ]

Subtracting the offset delta from four-byte alignment lead to wrapping
of the requested length where `count` is less than `off`. Generalise the
length handling to enable and optimise aligned access sizes for all
offset and size combinations. The new formula produces the following
results for given offset and count values:

    offset  count | length
    --------------+-------
    0       1     | 1
    0       2     | 2
    0       3     | 2
    0       4     | 4
    0       5     | 4
    1       1     | 1
    1       2     | 1
    1       3     | 1
    1       4     | 1
    1       5     | 1
    2       1     | 1
    2       2     | 2
    2       3     | 2
    2       4     | 2
    2       5     | 2
    3       1     | 1
    3       2     | 1
    3       3     | 1
    3       4     | 1
    3       5     | 1

We might need something like this for the cfam chardevs as well, for
example we don't currently implement any alignment restrictions /
handling in the hardware master driver.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20191108051945.7109-6-joel@jms.id.au
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoath10k: fix get invalid tx rate for Mesh metric
Miaoqing Pan [Wed, 6 Nov 2019 18:04:37 +0000 (20:04 +0200)]
ath10k: fix get invalid tx rate for Mesh metric

[ Upstream commit 05a11003a56507023f18d3249a4d4d119c0a3e9c ]

ath10k does not provide transmit rate info per MSDU
in tx completion, mark that as -1 so mac80211
will ignore the rates. This fixes mac80211 update Mesh
link metric with invalid transmit rate info.

Tested HW: QCA9984
Tested FW: 10.4-3.9.0.2-00035

Signed-off-by: Hou Bao Hou <houbao@codeaurora.org>
Signed-off-by: Anilkumar Kolli <akolli@codeaurora.org>
Signed-off-by: Miaoqing Pan <miaoqing@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agomedia: exynos4-is: fix wrong mdev and v4l2 dev order in error path
Seung-Woo Kim [Mon, 4 Nov 2019 09:46:32 +0000 (10:46 +0100)]
media: exynos4-is: fix wrong mdev and v4l2 dev order in error path

[ Upstream commit 4d741cbd58bf889c8a68cf6e592a7892b5c2802e ]

When driver is built as module and probe during insmod is deferred
because of sensor subdevs, there is NULL pointer deference because
mdev is cleaned up and then access it from v4l2_device_unregister().
Fix the wrong mdev and v4l2 dev order in error path of probe.

This fixes below null pointer deference:
   Unable to handle kernel NULL pointer dereference at virtual address 00000000
   pgd = ca026f68
   [00000000] *pgd=00000000
   Internal error: Oops: 5 [#1] PREEMPT SMP ARM
   [...]
   Hardware name: SAMSUNG EXYNOS (Flattened Device Tree)
   PC is at ida_free+0x7c/0x160
   LR is at xas_start+0x44/0x204
   [...]
   [<c0dafd60>] (ida_free) from [<c083c20c>] (__media_device_unregister_entity+0x18/0xc0)
   [<c083c20c>] (__media_device_unregister_entity) from [<c083c2e0>] (media_device_unregister_entity+0x2c/0x38)
   [<c083c2e0>] (media_device_unregister_entity) from [<c0843404>] (v4l2_device_release+0xd0/0x104)
   [<c0843404>] (v4l2_device_release) from [<c0632558>] (device_release+0x28/0x98)
   [<c0632558>] (device_release) from [<c0db1204>] (kobject_put+0xa4/0x208)
   [<c0db1204>] (kct_put) from [<bf00bac4>] (fimc_capture_subdev_unregistered+0x58/0x6c [s5p_fimc])
   [<bf00bac4>] (fimc_capture_subdev_unregistered [s5p_fimc]) from [<c084a1cc>] (v4l2_device_unregister_subdev+0x6c/0xa8)
   [<c084a1cc>] (v4l2_device_unregister_subdev) from [<c084a350>] (v4l2_device_unregister+0x64/0x94)
   [<c084a350>] (v4l2_device_unregister) from [<bf0101ac>] (fimc_md_probe+0x4ec/0xaf8 [s5p_fimc])
   [...]

Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Fixes: 9832e155f1ed ("[media] media-device: split media initialization and registration")
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agodrm/amdgpu: Avoid accidental thread reactivation.
Andrey Grodzovsky [Wed, 6 Nov 2019 17:36:29 +0000 (12:36 -0500)]
drm/amdgpu: Avoid accidental thread reactivation.

[ Upstream commit a28fda312a9fabdf0e5f5652449d6197c9fb0a90 ]

Problem:
During GPU reset we call the GPU scheduler to suspend it's
thread, those two functions in amdgpu also suspend and resume
the sceduler for their needs but this can collide with GPU
reset in progress and accidently restart a suspended thread
before time.

Fix:
Serialize with GPU reset.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoselftests: proc: Make va_max 1MB
Masami Hiramatsu [Wed, 23 Oct 2019 04:57:40 +0000 (13:57 +0900)]
selftests: proc: Make va_max 1MB

[ Upstream commit 2f3571ea71311bbb2cbb9c3bbefc9c1969a3e889 ]

Currently proc-self-map-files-002.c sets va_max (max test address
of user virtual address) to 4GB, but it is too big for 32bit
arch and 1UL << 32 is overflow on 32bit long.
Also since this value should be enough bigger than vm.mmap_min_addr
(64KB or 32KB by default), 1MB should be enough.

Make va_max 1MB unconditionally.

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agocgroup: freezer: don't change task and cgroups status unnecessarily
Honglei Wang [Wed, 30 Oct 2019 08:18:10 +0000 (16:18 +0800)]
cgroup: freezer: don't change task and cgroups status unnecessarily

[ Upstream commit 742e8cd3e1ba6f19cad6d912f8d469df5557d0fd ]

It's not necessary to adjust the task state and revisit the state
of source and destination cgroups if the cgroups are not in freeze
state and the task itself is not frozen.

And in this scenario, it wakes up the task who's not supposed to be
ready to run.

Don't do the unnecessary task state adjustment can help stop waking
up the task without a reason.

Signed-off-by: Honglei Wang <honglei.wang@oracle.com>
Acked-by: Roman Gushchin <guro@fb.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agos390/bpf: Use kvcalloc for addrs array
Ilya Leoshkevich [Thu, 7 Nov 2019 14:18:38 +0000 (15:18 +0100)]
s390/bpf: Use kvcalloc for addrs array

[ Upstream commit 166f11d11f6f70439830d09bfa5552ec1b368494 ]

A BPF program may consist of 1m instructions, which means JIT
instruction-address mapping can be as large as 4m. s390 has
FORCE_MAX_ZONEORDER=9 (for memory hotplug reasons), which means maximum
kmalloc size is 1m. This makes it impossible to JIT programs with more
than 256k instructions.

Fix by using kvcalloc, which falls back to vmalloc for larger
allocations. An alternative would be to use a radix tree, but that is
not supported by bpf_prog_fill_jited_linfo.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20191107141838.92202-1-iii@linux.ibm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agolibbpf: Fix negative FD close() in xsk_setup_xdp_prog()
Andrii Nakryiko [Thu, 7 Nov 2019 05:40:59 +0000 (21:40 -0800)]
libbpf: Fix negative FD close() in xsk_setup_xdp_prog()

[ Upstream commit 9656b346b280c3e49c8a116c3a715f966633b161 ]

Fix issue reported by static analysis (Coverity). If bpf_prog_get_fd_by_id()
fails, xsk_lookup_bpf_maps() will fail as well and clean-up code will attempt
close() with fd=-1. Fix by checking bpf_prog_get_fd_by_id() return result and
exiting early.

Fixes: 10a13bb40e54 ("libbpf: remove qidconf and better support external bpf programs.")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20191107054059.313884-1-andriin@fb.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Filter out instances except for inlined subroutine and subprogram
Masami Hiramatsu [Wed, 30 Oct 2019 07:09:30 +0000 (16:09 +0900)]
perf probe: Filter out instances except for inlined subroutine and subprogram

[ Upstream commit da6cb952a89efe24bb76c4971370d485737a2d85 ]

Filter out instances except for inlined_subroutine and subprogram DIE in
die_walk_instances() and die_is_func_instance().

This fixes an issue that perf probe sets some probes on calling address
instead of a target function itself.

When perf probe walks on instances of an abstruct origin (a kind of
function prototype of inlined function), die_walk_instances() can also
pass a GNU_call_site (a GNU extension for call site) to callback. Since
it is not an inlined instance of target function, we have to filter out
when searching a probe point.

Without this patch, perf probe sets probes on call site address too.This
can happen on some function which is marked "inlined", but has actual
symbol. (I'm not sure why GCC mark it "inlined"):

  # perf probe -D vfs_read
  p:probe/vfs_read _text+2500017
  p:probe/vfs_read_1 _text+2499468
  p:probe/vfs_read_2 _text+2499563
  p:probe/vfs_read_3 _text+2498876
  p:probe/vfs_read_4 _text+2498512
  p:probe/vfs_read_5 _text+2498627

With this patch:

Slightly different results, similar tho:

  # perf probe -D vfs_read
  p:probe/vfs_read _text+2498512

Committer testing:

  # uname -a
  Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux

Before:

  # perf probe -D vfs_read
  p:probe/vfs_read _text+3131557
  p:probe/vfs_read_1 _text+3130975
  p:probe/vfs_read_2 _text+3131047
  p:probe/vfs_read_3 _text+3130380
  p:probe/vfs_read_4 _text+3130000
  # uname -a
  Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux
  #

After:

  # perf probe -D vfs_read
  p:probe/vfs_read _text+3130000
  #

Fixes: db0d2c6420ee ("perf probe: Search concrete out-of-line instances")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lore.kernel.org/lkml/157241937063.32002.11024544873990816590.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Skip end-of-sequence and non statement lines
Masami Hiramatsu [Wed, 30 Oct 2019 07:09:21 +0000 (16:09 +0900)]
perf probe: Skip end-of-sequence and non statement lines

[ Upstream commit f4d99bdfd124823a81878b44b5e8750b97f73902 ]

Skip end-of-sequence and non-statement lines while walking through lines
list.

The "end-of-sequence" line information means:

 "the current address is that of the first byte after the
  end of a sequence of target machine instructions."
 (DWARF version 4 spec 6.2.2)

This actually means out of scope and we can not probe on it.

On the other hand, the statement lines (is_stmt) means:

 "the current instruction is a recommended breakpoint location.
  A recommended breakpoint location is intended to “represent”
  a line, a statement and/or a semantically distinct subpart
  of a statement."

 (DWARF version 4 spec 6.2.2)

So, non-statement line info also should be skipped.

These can reduce unneeded probe points and also avoid an error.

E.g. without this patch:

  # perf probe -a "clear_tasks_mm_cpumask:1"
  Added new events:
    probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_1 (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_2 (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_3 (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_4 (on clear_tasks_mm_cpumask:1)

  You can now use it in all perf tools, such as:

   perf record -e probe:clear_tasks_mm_cpumask_4 -aR sleep 1

  #

This puts 5 probes on one line, but acutally it's not inlined function.
This is because there are many non statement instructions at the
function prologue.

With this patch:

  # perf probe -a "clear_tasks_mm_cpumask:1"
  Added new event:
    probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1)

  You can now use it in all perf tools, such as:

   perf record -e probe:clear_tasks_mm_cpumask -aR sleep 1

  #

Now perf-probe skips unneeded addresses.

Committer testing:

Slightly different results, but similar:

Before:

  # uname -a
  Linux quaco 5.3.8-200.fc30.x86_64 #1 SMP Tue Oct 29 14:46:22 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux
  #
  # perf probe -a "clear_tasks_mm_cpumask:1"
  Added new events:
    probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_1 (on clear_tasks_mm_cpumask:1)
    probe:clear_tasks_mm_cpumask_2 (on clear_tasks_mm_cpumask:1)

  You can now use it in all perf tools, such as:

   perf record -e probe:clear_tasks_mm_cpumask_2 -aR sleep 1

  #

After:

  # perf probe -a "clear_tasks_mm_cpumask:1"
  Added new event:
    probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask:1)

  You can now use it in all perf tools, such as:

   perf record -e probe:clear_tasks_mm_cpumask -aR sleep 1

  # perf probe -l
    probe:clear_tasks_mm_cpumask (on clear_tasks_mm_cpumask@kernel/cpu.c)
  #

Fixes: 4cc9cec636e7 ("perf probe: Introduce lines walker interface")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lore.kernel.org/lkml/157241936090.32002.12156347518596111660.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Fix to show calling lines of inlined functions
Masami Hiramatsu [Wed, 30 Oct 2019 07:09:40 +0000 (16:09 +0900)]
perf probe: Fix to show calling lines of inlined functions

[ Upstream commit 86c0bf8539e7f46d91bd105e55eda96e0064caef ]

Fix to show calling lines of inlined functions (where an inline function
is called).

die_walk_lines() filtered out the lines inside inlined functions based
on the address. However this also filtered out the lines which call
those inlined functions from the target function.

To solve this issue, check the call_file and call_line attributes and do
not filter out if it matches to the line information.

Without this fix, perf probe -L doesn't show some lines correctly.
(don't see the lines after 17)

  # perf probe -L vfs_read
  <vfs_read@/home/mhiramat/ksrc/linux/fs/read_write.c:0>
        0  ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
        1  {
        2         ssize_t ret;

        4         if (!(file->f_mode & FMODE_READ))
                          return -EBADF;
        6         if (!(file->f_mode & FMODE_CAN_READ))
                          return -EINVAL;
        8         if (unlikely(!access_ok(buf, count)))
                          return -EFAULT;

       11         ret = rw_verify_area(READ, file, pos, count);
       12         if (!ret) {
       13                 if (count > MAX_RW_COUNT)
                                  count =  MAX_RW_COUNT;
       15                 ret = __vfs_read(file, buf, count, pos);
       16                 if (ret > 0) {
                                  fsnotify_access(file);
                                  add_rchar(current, ret);
                          }

With this fix:

  # perf probe -L vfs_read
  <vfs_read@/home/mhiramat/ksrc/linux/fs/read_write.c:0>
        0  ssize_t vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
        1  {
        2         ssize_t ret;

        4         if (!(file->f_mode & FMODE_READ))
                          return -EBADF;
        6         if (!(file->f_mode & FMODE_CAN_READ))
                          return -EINVAL;
        8         if (unlikely(!access_ok(buf, count)))
                          return -EFAULT;

       11         ret = rw_verify_area(READ, file, pos, count);
       12         if (!ret) {
       13                 if (count > MAX_RW_COUNT)
                                  count =  MAX_RW_COUNT;
       15                 ret = __vfs_read(file, buf, count, pos);
       16                 if (ret > 0) {
       17                         fsnotify_access(file);
       18                         add_rchar(current, ret);
                          }
       20                 inc_syscr(current);
                  }

Fixes: 4cc9cec636e7 ("perf probe: Introduce lines walker interface")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lore.kernel.org/lkml/157241937995.32002.17899884017011512577.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Return a better scope DIE if there is no best scope
Masami Hiramatsu [Tue, 5 Nov 2019 00:16:49 +0000 (09:16 +0900)]
perf probe: Return a better scope DIE if there is no best scope

[ Upstream commit c701636aeec4c173208697d68da6e4271125564b ]

Make find_best_scope() returns innermost DIE at given address if there
is no best matched scope DIE. Since Gcc sometimes generates intuitively
strange line info which is out of inlined function address range, we
need this fixup.

Without this, sometimes perf probe failed to probe on a line inside an
inlined function:

  # perf probe -D ksys_open:3
  Failed to find scope of probe point.
    Error: Failed to add events.

With this fix, 'perf probe' can probe it:

  # perf probe -D ksys_open:3
  p:probe/ksys_open _text+25707308
  p:probe/ksys_open_1 _text+25710596
  p:probe/ksys_open_2 _text+25711114
  p:probe/ksys_open_3 _text+25711343
  p:probe/ksys_open_4 _text+25714058
  p:probe/ksys_open_5 _text+2819653
  p:probe/ksys_open_6 _text+2819701

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tom Zanussi <tom.zanussi@linux.intel.com>
Link: http://lore.kernel.org/lkml/157291300887.19771.14936015360963292236.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agonet: avoid potential false sharing in neighbor related code
Eric Dumazet [Tue, 5 Nov 2019 22:11:51 +0000 (14:11 -0800)]
net: avoid potential false sharing in neighbor related code

[ Upstream commit 25c7a6d1f90e208ec27ca854b1381ed39842ec57 ]

There are common instances of the following construct :

if (n->confirmed != now)
n->confirmed = now;

A C compiler could legally remove the conditional.

Use READ_ONCE()/WRITE_ONCE() to avoid this problem.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
4 years agoperf probe: Skip overlapped location on searching variables
Masami Hiramatsu [Wed, 30 Oct 2019 07:09:49 +0000 (16:09 +0900)]
perf probe: Skip overlapped location on searching variables

[ Upstream commit dee36a2abb67c175265d49b9a8c7dfa564463d9a ]

Since debuginfo__find_probes() callback function can be called with  the
location which already passed, the callback function must filter out
such overlapped locations.

add_probe_trace_event() has already done it by commit 1a375ae7659a
("perf probe: Skip same probe address for a given line"), but
add_available_vars() doesn't. Thus perf probe -v shows same address
repeatedly as below:

  # perf probe -V vfs_read:18
  Available variables at vfs_read:18
          @<vfs_read+217>
                  char*   buf
                  loff_t* pos
                  ssize_t ret
                  struct file*    file
          @<vfs_read+217>
                  char*   buf
                  loff_t* pos
                  ssize_t ret
                  struct file*    file
          @<vfs_read+226>
                  char*   buf
                  loff_t* pos
                  ssize_t ret
                  struct file*    file

With this fix, perf probe -V shows it correctly:

  # perf probe -V vfs_read:18
  Available variables at vfs_read:18
          @<vfs_read+217>
                  char*   buf
                  loff_t* pos
                  ssize_t ret
                  struct file*    file
          @<vfs_read+226>
                  char*   buf
                  loff_t* pos
                  ssize_t ret
                  struct file*    file

Fixes: cf6eb489e5c0 ("perf probe: Show accessible local variables")
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lore.kernel.org/lkml/157241938927.32002.4026859017790562751.stgit@devnote2
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>