-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=9CQy
-----END PGP SIGNATURE-----
Merge tag 'rxrpc-rewrite-20160304' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs
David Howells says:
====================
RxRPC: Rewrite part 1
Here's the first set of patches from my RxRPC rewrite, aimed at net-next.
These do some clean ups and bug fixes. Most of the changes are small, but
there are a couple of bigger changes:
(*) Convert call flag numbers and event numbers into enums. Then rename
the event numbers to all have _EV_ in their name to stop confusion.
Fix one instance of an event bit being used instead of a flag bit.
(*) A copy of the Rx protocol header is kept in the sk_buff private data.
Keep this in host byte order rather than network byte order as it
makes more sense. A number of other fields then get converted into
host byte order too.
Conversion between host and network byte order is then done at the
packet reception/generation stage.
This is based on net-next/master
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Nicolas Dichtel says:
====================
uapi: consolidate DIV_ROUND_UP definition
The inital goal was to consolidate ethtool.h uapi header. But I took the
opportunity to remove all duplicate definitions of DIV_ROUND_UP.
v3: add patch #2 and #3
v2: split the patch
define DIV_ROUND_UP in uapi
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
INT_MAX needs limits.h in userland.
When ethtool.h is included by a userland app, we got the following error:
.../usr/include/linux/ethtool.h: In function 'ethtool_validate_speed':
.../usr/include/linux/ethtool.h:1471:18: error: 'INT_MAX' undeclared (first use in this function)
return speed <= INT_MAX || speed == SPEED_UNKNOWN
^
Fixes: e02564ee33 ("ethtool: make validate_speed accept all speeds between 0 and INT_MAX")
CC: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Let's use __KERNEL_DIV_ROUND_UP, which is defined in uapi/linux/kernel.h.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
let's use the common definition to avoid the following warning during the
compilation:
drivers/scsi/cxgbi/cxgb4i/cxgb4i.c:161:0: warning: "DIV_ROUND_UP" redefined
#define DIV_ROUND_UP(n, d) (((n) + (d) - 1) / (d))
^
In file included from include/linux/list.h:8:0,
from include/linux/module.h:9,
from drivers/scsi/cxgbi/cxgb4i/cxgb4i.c:16:
include/linux/kernel.h:67:0: note: this is the location of the previous definition
#define DIV_ROUND_UP __KERNEL_DIV_ROUND_UP
^
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
DIV_ROUND_UP is defined in linux/kernel.h only for the kernel.
When ethtool.h is included by a userland app, we got the following error:
include/linux/ethtool.h:1218:8: error: variably modified 'queue_mask' at file scope
__u32 queue_mask[DIV_ROUND_UP(MAX_NUM_QUEUE, 32)];
^
Let's add a common definition in uapi and use it everywhere.
Fixes: ac2c7ad0e5 ("net/ethtool: introduce a new ioctl for per queue setting")
CC: Kan Liang <kan.liang@intel.com>
Suggested-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rolf is no longer in his previous role at Netronome and as far as I know no
longer working on the NFP driver. Thus it does not seem appropriate for him
to be a co-maintainer anymore.
Reviewed-by: Dinan Gunawardena <dinan.gunawardena@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the ICMP message processing code, don't try to map ICMP codes to UNIX
error codes as the caller (IPv4/IPv6) already did that for us (ee_errno).
Signed-off-by: David Howells <dhowells@redhat.com>
The version number rxkad places in the response should be network byte
order.
Whilst we're at it, rearrange the code to be more readable.
Signed-off-by: David Howells <dhowells@redhat.com>
Use ACCESS_ONCE() when accessing the other-end pointer into a circular
buffer as it's possible the other-end pointer might change whilst we're
doing this, and if we access it twice, we might get some weird things
happening.
Signed-off-by: David Howells <dhowells@redhat.com>
Currently, received RxRPC packets outside the range 1-13 are rejected.
There are, however, holes in the range that should also be rejected - plus
at least one type we don't yet support - so reject these also.
Signed-off-by: David Howells <dhowells@redhat.com>
The upper bound of the defined range for rx_mtu is being set in the same
member as the lower bound (extra1) rather than the correct place (extra2).
I'm not entirely sure why this compiles.
Signed-off-by: David Howells <dhowells@redhat.com>
Currently, a copy of the Rx packet header is copied into the the sk_buff
private data so that we can advance the pointer into the buffer,
potentially discarding the original. At the moment, this copy is held in
network byte order, but this means we're doing a lot of unnecessary
translations.
The reasons it was done this way are that we need the values in network
byte order occasionally and we can use the copy, slightly modified, as part
of an iov array when sending an ack or an abort packet.
However, it seems more reasonable on review that it would be better kept in
host byte order and that we make up a new header when we want to send
another packet.
To this end, rename the original header struct to rxrpc_wire_header (with
BE fields) and institute a variant called rxrpc_host_header that has host
order fields. Change the struct in the sk_buff private data into an
rxrpc_host_header and translate the values when filling it in.
This further allows us to keep values kept in various structures in host
byte order rather than network byte order and allows removal of some fields
that are byteswapped duplicates.
Signed-off-by: David Howells <dhowells@redhat.com>
Convert call flag and event numbers into enums and move their definitions
outside of the struct.
Also move the call state enum outside of the struct and add an extra
element to count the number of states.
Signed-off-by: David Howells <dhowells@redhat.com>
Fix a case where RXRPC_CALL_RELEASE (an event) is being used to specify a
flag bit. RXRPC_CALL_RELEASED should be used instead.
Signed-off-by: David Howells <dhowells@redhat.com>
Some devices declare a high number of TX queues, then set a much
lower real_num_tx_queues
This cause setups using fq_codel, sfq or fq as the default qdisc to consume
more memory than really needed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Andrew and Ying Huang's test robot both reported usage count problems that
trace back to the 'keep address on ifdown' patch.
>From Andrew:
We execute CRIU test on linux-next. On the current linux-next kernel
they hangs on creating a network namespace.
The kernel log contains many massages like this:
[ 1036.122108] unregister_netdevice: waiting for lo to become free.
Usage count = 2
[ 1046.165156] unregister_netdevice: waiting for lo to become free.
Usage count = 2
[ 1056.210287] unregister_netdevice: waiting for lo to become free.
Usage count = 2
I tried to revert this patch and the bug disappeared.
Here is a set of commands to reproduce this bug:
[root@linux-next-test linux-next]# uname -a
Linux linux-next-test 4.5.0-rc6-next-20160301+ #3 SMP Wed Mar 2
17:32:18 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
[root@linux-next-test ~]# unshare -n
[root@linux-next-test ~]# ip link set up dev lo
[root@linux-next-test ~]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
group default qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
[root@linux-next-test ~]# logout
[root@linux-next-test ~]# unshare -n
-----
The problem is a change made to RTM_DELADDR case in __ipv6_ifa_notify that
was added in an early version of the offending patch and is no longer
needed.
Fixes: f1705ec197 ("net: ipv6: Make address flushing on ifdown optional")
Cc: Andrey Wagin <avagin@gmail.com>
Cc: Ying Huang <ying.huang@linux.intel.com>
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Tested-by: Jeremiah Mahler <jmmahler@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since ezchip network driver is written with big endian EZChip platform it
is necessary to add support for little endian architecture.
The first issue is that the order of the bits in a bit field is
implementation specific. So all the bit fields are removed.
Named constants are used to access necessary fields.
And the second one is that network byte order is big endian.
For example, data on ethernet is transmitted with most-significant
octet (byte) first. So in case of little endian architecture
it is important to swap data byte order when we read it from
register. In case of unaligned access we can use "get_unaligned_be32"
and in other case we can use function "ioread32_rep" which reads all
data from register and works either with little endian or big endian
architecture.
And then when we are going to write data to register we need to restore
byte order using the function "put_unaligned_be32" in case of
unaligned access and in other case "iowrite32_rep".
The last little fix is a space between type and pointer to observe
coding style.
Signed-off-by: Lada Trimasova <ltrimas@synopsys.com>
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Noam Camus <noamc@ezchip.com>
Cc: Tal Zilcer <talz@ezchip.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make use of ARCH_RENESAS in place of ARCH_SHMOBILE.
This is part of an ongoing process to migrate from ARCH_SHMOBILE to
ARCH_RENESAS the motivation for which being that RENESAS seems to be a more
appropriate name than SHMOBILE for the majority of Renesas ARM based SoCs.
Signed-off-by: Simon Horman <horms+renesas@verge.net.au>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new NET_DEVLINK infrastructure can be a loadable module, but the drivers
using it might be built-in, which causes link errors like:
drivers/net/built-in.o: In function `mlx4_load_one':
:(.text+0x2fbfda): undefined reference to `devlink_port_register'
:(.text+0x2fc084): undefined reference to `devlink_port_unregister'
drivers/net/built-in.o: In function `mlxsw_sx_port_remove':
:(.text+0x33a03a): undefined reference to `devlink_port_type_clear'
:(.text+0x33a04e): undefined reference to `devlink_port_unregister'
There are multiple ways to avoid this:
a) add 'depends on NET_DEVLINK || !NET_DEVLINK' dependencies
for each user
b) use 'select NET_DEVLINK' from each driver that uses it
and hide the symbol in Kconfig.
c) make NET_DEVLINK a 'bool' option so we don't have to
list it as a dependency, and rely on the APIs to be
stubbed out when it is disabled
d) use IS_REACHABLE() rather than IS_ENABLED() to check for
NET_DEVLINK in include/net/devlink.h
This implements a variation of approach a) by adding an
intermediate symbol that drivers can depend on, and changes
the three drivers using it.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 09d4d087cd ("mlx4: Implement devlink interface")
Fixes: c4745500e9 ("mlxsw: Implement devlink interface")
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Johan Hedberg says:
====================
pull request: bluetooth-next 2016-03-01
Here's our main set of Bluetooth & 802.15.4 patches for the 4.6 kernel.
- New Bluetooth HCI driver for Intel/AG6xx controllers
- New Broadcom ACPI IDs
- LED trigger support for indicating Bluetooth powered state
- Various fixes in mac802154, 6lowpan and related drivers
- New USB IDs for AR3012 Bluetooth controllers
Please let me know if there are any issues pulling. Thanks.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
I added this check in setup_tc to multiple drivers,
if (handle != TC_H_ROOT || tc->type != TC_SETUP_MQPRIO)
Unfortunately restricting to TC_H_ROOT like this breaks the old
instantiation of mqprio to setup a hardware qdisc. This patch
relaxes the test to only check the type to make it equivalent
to the check before I broke it. With this the old instantiation
continues to work.
A good smoke test is to setup mqprio with,
# tc qdisc add dev eth4 root mqprio num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 0@0 1@1 2@2 3@3 4@4 5@5 6@6 7@7
Fixes: e4c6734eaa ("net: rework ndo tc op to consume additional qdisc handle paramete")
Reported-by: Singh Krishneil <krishneil.k.singh@intel.com>
Reported-by: Jake Keller <jacob.e.keller@intel.com>
CC: Murali Karicheri <m-karicheri2@ti.com>
CC: Shradha Shah <sshah@solarflare.com>
CC: Or Gerlitz <ogerlitz@mellanox.com>
CC: Ariel Elior <ariel.elior@qlogic.com>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
CC: Bruce Allan <bruce.w.allan@intel.com>
CC: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Left over from c24588afc5
("atl1c: using fixed TXQ configuration for l2cb and l1c")
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8cc785f6f4 ("net: ipv4: make the ping
/proc code AF-independent") removed the code using it, but renamed this
variable instead of removing it.
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3b766cd832 ("net/core: Add reading VF
statistics through the PF netdevice") added that variable but it's never
been used.
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Hariprasad Shenai says:
====================
cxgb4/cxgb4vf: Cleanup and minor fixes
This series sets FBMIN to 64 bytes for Chelsio's T6 series of adapters,
check to replenish fl is revised, some code cleanup in cxgb4vf sge
initialization code and removes dead code.
This patch series has been created against net-next tree and includes
patches on cxgb4 and cxgb4vf driver.
We have included all the maintainers of respective drivers. Kindly review
the change and let us know in case of any review comments.
====================
Function t4vf_wait_dev_ready() is already called in t4vf_prep_adapter(),
no need to call it again in adap_init0().
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adds a new function t4vf_fl_pkt_align() and use the same in SGE
initialization code to find out freelist packet alignment
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
T4 and T5 hardware will not coalesce Free List PCI-E Fetch Requests if
the Host Driver provides more Free List Pointers than the Fetch Burst
Minimum value. So if we set FBMIN to 64 bytes and the Host Driver
supplies 128 bytes of Free List Pointer data, the hardware will issue two
64-byte PCI-E Fetch Requests rather than a single coallesced 128-byte
Fetch Request. T6 fixes this. So, for T4/T5 we set the FBMIN value to 128
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use freelist capacity instead of freelist size while checking, if
freelist needs to be refilled
Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Alexandre TORGUE says:
====================
stmmac: enhance driver performances and update the version
According to Giuseppe, I send the v3 series.
This is a subset of patches to rework the driver in order to improve its
performances and make it more robust under stress conditions.
All patches have been ported on STi mainstream kernel branch and
tested on ARM STiH4xx platforms and newer ones.
This series also updates the driver version and prepares it
to include further development to support new chips.
In detail, these patches are:
o to rework and improve the internal DMA bus settings
Fine tuning is mandatory on some platforms for both
performance and stability issues.
o to rework and optimize the descriptor management.
This will help a lot on performance side and preparing
the inclusion on the GMAC4.x.
o to add a set of optimizations for both xmit and rx functions.
These will help a lot on performance side and making the driver
more robust in case of low memory conditions and under some
stress test, performed for example on IP-STB.
Below some throughput figures obtained on some boxes before and after
the patches.
nuttcp (mbps) iperf (Mbps)
------------------------------------------------------------------
tcp udp tcp udp
tx rx tx rx tx rx tx rx
------------------------------------------
old 680 800 480 506 760 800 600 700
new 830 880 540 630 840 880 700 800
V2: - rx_copybreak is now managed by using ethtool.
V3: - improve comments on PCIe detailing that there are no regressions
- rework some APIs to properly define some params as bool as expected
- rework the formula to get the element inside the ring. Comparing V2,
patches 4 and 13 have been merged because the same formula have been
used. After this rework, no evident benefit has been noticed in terms
of performances so the table above is still valid. Disassembling the
code for SH4 and ARM, with the new formula just an instr is saved
(depending on compiler flags) and this gives us not so relevanti gain,
for example, on SH4 where some instr are executed in the same pipeline
stage.
Ring sizes are now fixed and maybe they can be reworked to be tuned
w/o using stmmaceth= cmdline option. Indeed, nobody change these sizes
and indeed the numbers selected by default respect the budget and
avoid to pass invalid setup. These are the best driver default sizes
for ring and chain.
====================
This patch just updates the driver to the version fully
tested on STi platforms. This version is Oct_2015.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is a threshold now used to also limit the skb allocation
when use zero-copy. This is to avoid that there are incoherence
in the ring due to a failure on skb allocation under very
aggressive testing and under low memory conditions.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to allow this driver to copy tiny frames during the reception
process. This is giving more stability while stressing the driver on STi
embedded systems.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
phy_bus_name can be NULL when "fixed-link" property isn't used.
Then, since "stmmac: do not poll phy handler when attach a switch",
phy_bus_name ptr needs to be checked before strcmp is called.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch avoids to call the stmmac_adjust_link when
the driver is connected to a switch by using the FIXED_PHY
support. Prior this patch the phydev->irq was set as PHY_POLL
so periodically the phy handler was invoked spending useless
time because the link cannot actually change.
Note that the stmmac_adjust_link will be called just one
time and this guarantees that the ST glue logic will be
setup according to the mode and speed fixed.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to fill the first descriptor just before granting
the DMA engine so at the end of the xmit.
The patch takes care about the algorithm adopted to mitigate the
interrupts, then it fixes the last segment in case of no fragments.
Moreover, this new implementation does not pass any "ter" field when
prepare the descriptors because this is not necessary.
The patch also details the memory barrier in the xmit.
As final results, this patch guarantees the same performances
but fixing a case if small datagram are sent. In fact, this
kind of test is impacted if no coalesce is done.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The dirty index can be updated out of the loop where all the
tx resources are claimed. This will help on performances too.
Also a useless debug printk has been removed from the main loop.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch "inline" get_tx_owner and get_ls routines. It Results in a
unique read to tdes0, instead of three, to check TX_OWN and LS bits,
and other status bits.
It helps improve driver TX path by removing two uncached read/writes
inside TX clean loop for enhanced descriptors but not for normal ones
because the des1 must be read in any case.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Acked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to optimize the way to manage the TDES inside the
xmit function. When prepare the frame, some settings (e.g. OWN
bit) can be merged. This has been reworked to improve the tx
performances.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The RDES0 register can be read several times while doing RX of a
packet.
This patch slightly improves RX path performance by reading rdes0
once for two operation: check rx owner, get rx status bits.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Acked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Optimize tx_clean by avoiding a des3 read in stmmac_clean_desc3().
In ring mode, TX, des3 seems only used when xmit a jumbo frame.
In case of normal descriptors, it may also be used for time
stamping.
Clean it in the above two case, without reading it.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
last_segment field is read twice from dma descriptors in stmmac_clean().
Add last_segment to dma data so that this flag is from priv
structure in cache instead of memory.
It avoids reading twice from memory for each loop in stmmac_clean().
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the code pulls out the length field when
unmapping a buffer directly from the descriptor. This will result
in an uncached read to a dma_alloc_coherent() region. There is no
need to do this, so this patch simply puts the value directly into
a data structure which will hit the cache.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Alexandre TORGUE <alexandre.torgue@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller