Use "ndev" instead of "dev", as the rx queue back pointer
to a net_device struct, to avoid name clashing with a
"struct device" reference. This prepares the addition of a
"struct device" back pointer to the rx queue structure.
Remove duplicated rxq registration in the process.
Move napi_gro_receive() outside gfar_process_frame().
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are several (long standing) problems about how the status
field of the rx buffer descriptor (rxbd) is currently handled on
the error path:
- too many unnecessary 16bit reads of the two halves of the rxbd
status field (32bit), also resulting in overuse of endianness
convesion macros;
- "bdp->status = RXBD_LARGE" makes no sense, since the "large"
flag is read only (only eTSEC can write it), and trying to clear
the other status bits is also error prone in this context
(most of the rx status bits are read only anyway).
This is fixed with a single 32bit read of the "status" field,
and then the appropriate 16bit shifting is applied to access
the various status bits or the rx frame length. Also corrected
the use of the RXBD_LARGE flag.
Additional fix:
"rx_over_errors" stat is incremented instead of "rx_crc_errors"
in case of RXBD_OVERRUN occurrence.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use a more common consumer/ producer index design to improve
rx buffer allocation. Instead of allocating a single new buffer
(skb) on each iteration, bundle the allocation of several rx
buffers at a time. This also opens the path for further memory
optimizations.
Remove useless check of rxq->rfbptr, since this patch touches
rx pause frame handling code as well. rxq->rfbptr is always
initialized as part of Rx BD ring init.
Remove redundant (and misleading) 'amount_pull' parameter.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use device flag IFF_LIVE_ADDR_CHANGE to signal that
the device supports changing the hardware address when
the device is running.
This allows eth_mac_addr() to change the mac address
also when the network device's interface is open.
This capability is required by certain applications,
like bonding mode 6 (Adaptive Load Balancing).
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Handle TxFIFO underrun exceptions outside the fast path.
A controller reset is more reliable in this exceptional
case, as opposed to re-enabling on-the-fly the Tx DMA.
As the controller reset is handled outside the fast path
by the reset_gfar() workqueue handler, the locking
scheme on the Tx path is significantly simplified.
Because the Tx processing (xmit queues and tx napi) is
disabled during controller reset, tstat access from xmit
does not require locking. So the scope of the txlock on
the processing path is now reduced to num_txbdfree, which
is shared only between process context (xmit) and softirq
(clean_tx_ring). As a result, the txlock must not guard
against interrupt context, and the spin_lock_irqsave()
from xmit can be replaced by spin_lock_bh(). Likewise,
the locking has been downgraded for clean_tx_ring().
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
of_device_id is always used as const.
(See driver.of_match_table and open firmware functions)
Signed-off-by: Fabian Frederick <fabf@skynet.be>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use of_property_read*() to get arch endian consistent
property values. Do some refactoring in the process.
Signed-off-by: Jingchang Lu <jingchang.lu@freescale.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use conversion macros to correctly access the BE
fields of the Rx and Tx Frame Control Block on LE CPUs.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use conversion macros to correctly access the BE
fields of the Rx and Tx Buffer Descriptors on LE CPUs.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
eTSEC of-nodes may have children which are not queue-group nodes. For
example new-style fixed-phy declarations. These where incorrectly
assumed to be additional queue-groups.
Change the search to filter out any nodes which are not queue-groups,
or have been disabled.
Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 6ce29b0e2a ("gianfar: Avoid unnecessary reg accesses in adjust_link()")
eliminates unnecessary calls to adjust_link for phy devices which don't support
interrupts and need polling. As part of that work, the 'new_state' local flag,
which was used to reduce logging noise on the console, was eliminated.
Unfortunately, that means that a 'Link is Down' log message will now be
issued continuously if a link is configured as UP, the link state is down,
and the associated phy requires polling. This occurs because priv->oldduplex
is -1 in this case, which always differs from phydev->duplex. In addition,
phydev->speed may also differ from priv->oldspeed. gfar_update_link_state()
is therefore called each time a phy is polled, even if the link state did not
change.
Cc: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit cd1e65044d ("of/device: Don't register disabled
devices"), the disabled device will not be registered at all. So we
don't need to do the check again in the platform device driver.
Signed-off-by: Kevin Hao <haokexin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The same macros are used for rx as well. So rename it.
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix the following spare warning:
drivers/net/ethernet/freescale/gianfar.c:3521:60: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/freescale/gianfar.c:3521:60: expected unsigned int [noderef] <asn:2>*addr
drivers/net/ethernet/freescale/gianfar.c:3521:60: got unsigned int [usertype] *rfbptr
drivers/net/ethernet/freescale/gianfar.c:205:16: warning: incorrect type in assignment (different address spaces)
drivers/net/ethernet/freescale/gianfar.c:205:16: expected unsigned int [usertype] *rfbptr
drivers/net/ethernet/freescale/gianfar.c:205:16: got unsigned int [noderef] <asn:2>*<noident>
drivers/net/ethernet/freescale/gianfar.c:2918:44: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/freescale/gianfar.c:2918:44: expected unsigned int [noderef] <asn:2>*addr
drivers/net/ethernet/freescale/gianfar.c:2918:44: got unsigned int [usertype] *rfbptr
Signed-off-by: Kevin Hao <haokexin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Here's the set of driver core patches for 3.19-rc1.
They are dominated by the removal of the .owner field in platform
drivers. They touch a lot of files, but they are "simple" changes, just
removing a line in a structure.
Other than that, a few minor driver core and debugfs changes. There are
some ath9k patches coming in through this tree that have been acked by
the wireless maintainers as they relied on the debugfs changes.
Everything has been in linux-next for a while.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iEYEABECAAYFAlSOD20ACgkQMUfUDdst+ylLPACg2QrW1oHhdTMT9WI8jihlHVRM
53kAoLeteByQ3iVwWurwwseRPiWa8+MI
=OVRS
-----END PGP SIGNATURE-----
Merge tag 'driver-core-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
Pull driver core update from Greg KH:
"Here's the set of driver core patches for 3.19-rc1.
They are dominated by the removal of the .owner field in platform
drivers. They touch a lot of files, but they are "simple" changes,
just removing a line in a structure.
Other than that, a few minor driver core and debugfs changes. There
are some ath9k patches coming in through this tree that have been
acked by the wireless maintainers as they relied on the debugfs
changes.
Everything has been in linux-next for a while"
* tag 'driver-core-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core: (324 commits)
Revert "ath: ath9k: use debugfs_create_devm_seqfile() helper for seq_file entries"
fs: debugfs: add forward declaration for struct device type
firmware class: Deletion of an unnecessary check before the function call "vunmap"
firmware loader: fix hung task warning dump
devcoredump: provide a one-way disable function
device: Add dev_<level>_once variants
ath: ath9k: use debugfs_create_devm_seqfile() helper for seq_file entries
ath: use seq_file api for ath9k debugfs files
debugfs: add helper function to create device related seq_file
drivers/base: cacheinfo: remove noisy error boot message
Revert "core: platform: add warning if driver has no owner"
drivers: base: support cpu cache information interface to userspace via sysfs
drivers: base: add cpu_device_create to support per-cpu devices
topology: replace custom attribute macros with standard DEVICE_ATTR*
cpumask: factor out show_cpumap into separate helper function
driver core: Fix unbalanced device reference in drivers_probe
driver core: fix race with userland in device_add()
sysfs/kernfs: make read requests on pre-alloc files use the buffer.
sysfs/kernfs: allow attributes to request write buffer be pre-allocated.
fs: sysfs: return EGBIG on write if offset is larger than file size
...
We need to use dma_mapping_error() to check the dma address returned
by dma_map_single/page(). Otherwise we would get warning like this:
WARNING: at lib/dma-debug.c:1140
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.18.0-rc2-next-20141029 #196
task: c0834300 ti: effe6000 task.ti: c0874000
NIP: c02b2c98 LR: c02b2c98 CTR: c030abc4
REGS: effe7d70 TRAP: 0700 Not tainted (3.18.0-rc2-next-20141029)
MSR: 00021000 <CE,ME> CR: 22044022 XER: 20000000
GPR00: c02b2c98 effe7e20 c0834300 00000098 00021000 00000000 c030b898 00000003
GPR08: 00000001 00000000 00000001 749eec9d 22044022 1001abe0 00000020 ef278678
GPR16: ef278670 ef278668 ef278660 070a8040 c087f99c c08cdc60 00029000 c0840d44
GPR24: c08be6e8 c0840000 effe7e78 ef041340 00000600 ef114e10 00000000 c08be6e0
NIP [c02b2c98] check_unmap+0x51c/0x9e4
LR [c02b2c98] check_unmap+0x51c/0x9e4
Call Trace:
[effe7e20] [c02b2c98] check_unmap+0x51c/0x9e4 (unreliable)
[effe7e70] [c02b31d8] debug_dma_unmap_page+0x78/0x8c
[effe7ed0] [c03d1640] gfar_clean_rx_ring+0x208/0x488
[effe7f40] [c03d1a9c] gfar_poll_rx_sq+0x3c/0xa8
[effe7f60] [c04f8714] net_rx_action+0xc0/0x178
[effe7f90] [c00435a0] __do_softirq+0x100/0x1fc
[effe7fe0] [c0043958] irq_exit+0xa4/0xc8
[effe7ff0] [c000d14c] call_do_irq+0x24/0x3c
[c0875e90] [c00048a0] do_IRQ+0x8c/0xf8
[c0875eb0] [c000ed10] ret_from_except+0x0/0x18
For TX, we need to unmap the pages which has already been mapped and
free the skb before return.
For RX, move the dma mapping and error check to gfar_new_skb(). We
would reuse the original skb in the rx ring when either allocating
skb failure or dma mapping error.
Signed-off-by: Kevin Hao <haokexin@gmail.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The hardware can automatically generate pause frames when the number
of free buffers drops under a certain threshold, but in order to do this,
the address of the last free buffer needs to be written to a specific
register for each RX queue.
This has to be done in 'gfar_clean_rx_ring' which is called for each
RX queue. In order not to impact performance, by adding a register write
for each incoming packet, this operation is done only when the PAUSE frame
transmission is enabled.
Whenever the link is readjusted, this capability is turned on or off.
Signed-off-by: Matei Pavaluca <matei.pavaluca@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Local flow control options needed in order to resolve the negotiation
are incorrectly calculated.
Previously 'mii_advertise_flowctrl' was called to determine the local advertising
options, but these were determined based on FLOW_CTRL_RX/TX flags which are
never set through ethtool.
The patch simply translates from ethtool flow options to mii flow options.
Signed-off-by: Pavaluca Matei <matei.pavaluca@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The phy device supports 802.3x flow control, but the specific flags are not set
in the phy initialisation code. Flow control flags need to be added to the
supported capabilities of the phydev by the driver.
This is needed in order for ethtool to work ('ethtool -A' code checks for these
flags)
Signed-off-by: Pavaluca Matei <matei.pavaluca@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For each Rx frame the eTSEC writes its FCS (Frame Check Sequence)
to the Rx buffer.
The eTSEC h/w manual states in the "Receive Buffer Descriptor Field
Descriptions" table:
"Data length is the number of octets written by the eTSEC into this BD's
data buffer if L is cleared (the value is equal to MRBLR), or, if L is
set, the length of the frame including *CRC*, FCB (if RCTRL[PRSDEP > 00),
preamble (if MACCFG2[PreAmRxEn]=1), time stamp (if RCTRL[TS] = 1) and
any padding (RCTRL[PAL])."
Though the FCS bytes are removed by the driver before passing the skb
to the net stack, the Rx buffer size computation does not currently
take into account the FCS bytes (4 bytes).
Because the Rx buffer size is multiple of 512 bytes, leaving out the
FCS is not a problem for the default MTU of 1500, as the Rx buffer size
is 1536 in this case. However, for custom MTUs, where the difference
between the MTU size and the Rx buffer size is less, this can be a
problem as the computed Rx buffer size won't be enough to accomodate
the FCS for a received frame that is big enough (close to MTU size).
In such case the received frame is considered to be incomplete (L flag
not set in the RxBD status) and silently dropped.
Note that the driver does not currently support S/G on Rx, so it has to
compute its Rx buffer size based on the MTU of the device.
Reported-by: Kristian Otnes <kotnes@cisco.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace PPC specific eieio() with arch independent wmb()
for other architectures, i.e. ARM.
The eieio() macro is not defined on ARM and generates
build error.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use arch independent code to replace the powerpc dependent
spin_event_timeout() from gfar_halt_nodisable().
Added GRS/GTS read accessors to clean-up the implementation
of gfar_halt_nodisable().
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix the 32-bit memory access that is not endian safe,
i.e. not giving the desired byte layout for a LE CPU:
tempval = *((u32 *) (tmpbuf + 4)), where 'char tmpbuf[]'.
Get rid of rendundant local vars (tmpbuf[] and idx) and
forced casts. Cleanup comments.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This excludes the PPC specific instructions for PPC based SoC
(MPC85xx family) version identification from ARM builds.
The PPC specific macro mfspr() from asm/reg.h is not defined
by the ARM architecture.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Include linux/of_address.h for of_iomap() and linux/of_irq.h
for irq_of_parse_and_map().
This wasn't an issue for PPC, because these were implicitly
included from asm/prom.h (via linux/of.h) for PPC builds only.
ARM builds need these includes explicitly.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The line before the changed if condition is:
priv->phy_node = of_parse_phandle(np, "phy-handle", 0);
. If this call succeeds priv->phy_node must not be overwritten in the if
block; otherwise the reference to the node returned by of_parse_phandle
is lost. So add a check that the if block isn't executed in this case.
Furthermore in the fixed phy case no reference is aquired for phy_node
resulting in an of_node_put without holding a reference. To fix that,
get a reference on the MAC dt node.
Fixes: be40364544 ("gianfar: use the new fixed PHY helpers")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
of_node_put is a noop when being called with NULL.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking updates from David Miller:
1) Seccomp BPF filters can now be JIT'd, from Alexei Starovoitov.
2) Multiqueue support in xen-netback and xen-netfront, from Andrew J
Benniston.
3) Allow tweaking of aggregation settings in cdc_ncm driver, from Bjørn
Mork.
4) BPF now has a "random" opcode, from Chema Gonzalez.
5) Add more BPF documentation and improve test framework, from Daniel
Borkmann.
6) Support TCP fastopen over ipv6, from Daniel Lee.
7) Add software TSO helper functions and use them to support software
TSO in mvneta and mv643xx_eth drivers. From Ezequiel Garcia.
8) Support software TSO in fec driver too, from Nimrod Andy.
9) Add Broadcom SYSTEMPORT driver, from Florian Fainelli.
10) Handle broadcasts more gracefully over macvlan when there are large
numbers of interfaces configured, from Herbert Xu.
11) Allow more control over fwmark used for non-socket based responses,
from Lorenzo Colitti.
12) Do TCP congestion window limiting based upon measurements, from Neal
Cardwell.
13) Support busy polling in SCTP, from Neal Horman.
14) Allow RSS key to be configured via ethtool, from Venkata Duvvuru.
15) Bridge promisc mode handling improvements from Vlad Yasevich.
16) Don't use inetpeer entries to implement ID generation any more, it
performs poorly, from Eric Dumazet.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1522 commits)
rtnetlink: fix userspace API breakage for iproute2 < v3.9.0
tcp: fixing TLP's FIN recovery
net: fec: Add software TSO support
net: fec: Add Scatter/gather support
net: fec: Increase buffer descriptor entry number
net: fec: Factorize feature setting
net: fec: Enable IP header hardware checksum
net: fec: Factorize the .xmit transmit function
bridge: fix compile error when compiling without IPv6 support
bridge: fix smatch warning / potential null pointer dereference
via-rhine: fix full-duplex with autoneg disable
bnx2x: Enlarge the dorq threshold for VFs
bnx2x: Check for UNDI in uncommon branch
bnx2x: Fix 1G-baseT link
bnx2x: Fix link for KR with swapped polarity lane
sctp: Fix sk_ack_backlog wrap-around problem
net/core: Add VF link state control policy
net/fsl: xgmac_mdio is dependent on OF_MDIO
net/fsl: Make xgmac_mdio read error message useful
net_sched: drr: warn when qdisc is not work conserving
...
Conflicts:
drivers/net/xen-netback/netback.c
net/core/filter.c
A filter bug fix overlapped some cleanups and a conversion
over to some new insn generation macros.
A xen-netback bug fix overlapped the addition of multi-queue
support.
Signed-off-by: David S. Miller <davem@davemloft.net>
Quoting David Miller:
"At the moment you call register_netdev() the device is visible, notifications
are sent to userspace, and userland tools can try to bring the interface up
and see the incorrect link state, before you do the netif_carrier_off().
Said another way, between the register_netdev() and netif_carrier_off() call,
userspace can see the device in an inconsistent state."
So call netif_carrier_off() prior to register_netdev().
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Building with CONFIG_DEBUG_SECTION_MISMATCH enabled, the following
WARNING is occured:
LD drivers/net/built-in.o
WARNING: drivers/net/built-in.o(.text+0xcd4c): Section mismatch in
reference from the function gfar_probe() to the function
.init.text:gfar_init_addr_hash_table()
The function gfar_probe() references
the function __init gfar_init_addr_hash_table().
This is often because gfar_probe lacks a __init
annotation or the annotation of gfar_init_addr_hash_table is wrong.
Signed-off-by: Xiubo Li <Li.Xiubo@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
of_phy_connect_fixed_link() is becoming obsolete, and also required
platform code to register the fixed PHYs at the specified addresses for
those to be usable. Get rid of it and use the new of_phy_is_fixed_link()
plus of_phy_register_fixed_link() helpers to transition over the new
scheme.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For phy devices that don't issue interrupts upon link
state changes, phylib polls the link state resulting in
repeated calls to adjust_link(), even if the link state
didn't change. As a result, some mac registers are
repeatedly read and written with the same values, which
is not ok.
To fix this, adjust_link() has been refactored to check
first whether the link state has changed and to take action
only if needed, updating mac registers and local state
variables. The 'new_state' local flag, set if one of the
link params changed (link, speed or duplex), has been
rendered useless and removed by this refactoring.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The P1010 device tree restricts the number of
supported interrupt groups to 1, although the eth
controller can support 2 interrupt groups and the
driver assumes the Multi-Group mode ("fsl,etsec2" model).
So, in this case the assumption that the Multi-Group
mode (MQ_MG_MODE) devices always support 2 interrupt
groups is false. To fix this, a check for the actual
number of interrupt groups enabled in the board's
device tree has been added in gfar_probe for the
"fsl,etsec2" devices.
Without this fix, P1010 based boards claim support for
2 Tx queues to the net stack but only one is actually
allocated, leading to NULL access in xmit. This issue
was introduced by enabling Single-Queue polling for
the P1010 devices.
(71ff9e3 gianfar: Use Single-Queue polling for
"fsl,etsec2")
Fixes: 71ff9e3df7
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
netpoll can call functions in hard irq context that are ordinarily
called in lesser contexts. For those functions use dev_kfree_skb_any
and dev_consume_skb_any so skbs are freed safely from hard irq
context.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
priv is not instantiated at gfar_of_init() time, when
parsing the DT for info on supported HW queues. Before
the netdev can be allocated, the number of supported
queues must be known. Because the number of supported
queues depends on device type, move the compatibility
checks before netdev allocation. Local vars are used
to hold the operation mode info before netdev allocation.
This fixes the null accesses for priv->.., in gfar_of_init.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the "fsl,etsec2" compatible models the driver currently
supports 8 Tx and Rx DMA rings (aka HW queues). However, there
are only 2 pairs of Rx/Tx interrupt lines, as these controllers
are integrated in low power SoCs with 2 CPUs at most. As a result,
there are at most 2 NAPI instances that have to service multiple
Tx and Rx queues for these devices. This complicates the NAPI
polling routine having to iterate over the mutiple Rx/Tx queues
hooked to the same interrupt lines. And there's also an overhead
at HW level, as the controller needs to service all the 8 Tx rings
in a round robin manner. The combined overhead shows up for multi
parallel Tx flows transmitted by the kernel stack, when the driver
usually starts returning NETDEV_TX_BUSY leading to NETDEV WATCHDOG
Tx timeout triggering if the Tx path is congested for too long.
As an alternative, this patch makes the driver support only one
Tx/Rx DMA ring per NAPI instance (per interrupt group or pair
of Tx/Rx interrupt lines) by default. The simplified single queue
polling routine (gfar_poll_sq) will be the default napi poll routine
for the etsec2 devices too. Some adjustments needed to be made to
link the Tx/Rx HW queues with each NAPI instance (2 in this case).
The gfar_poll_sq() is already successfully used by older SQ_SG_MODE
(single interrupt group) controllers.
This patch fixes Tx timeout triggering under heavy Tx traffic load
(i.e. iperf -c -P 8) for the "fsl,etsec2" (currently the only
MQ_MG_MODE devices). There's also a significant memory footprint
reduction by supporting 2 Rx/Tx DMA rings (at most), instead of 8,
for these devices.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are some concurrency issues on devices w/ 2 CPUs related
to the handling of Rx and Tx interrupts. eTSEC has separate
interrupt lines for Rx and Tx but a single imask register
to mask these interrupts and a single NAPI instance to handle
both Rx and Tx work. As a result, the Rx and Tx ISRs are
identical, both are invoking gfar_schedule_cleanup(), however
both handlers can be entered at the same time when the Rx and
Tx interrupts are taken by different CPUs. In this case
spurrious interrupts (SPU) show up (in /proc/interrupts)
indicating a concurrency issue. Also, Tx overruns followed
by Tx timeout have been observed under heavy Tx traffic load.
To address these issues, the schedule cleanup ISR part has
been changed to handle the Rx and Tx interrupts independently.
The patch adds a separate NAPI poll routine for Tx cleanup to
be triggerred independently by the Tx confirmation interrupts
only. Existing poll functions are modified to handle only
the Rx path processing. The Tx poll routine does not need a
budget, since Tx processing doesn't consume NAPI budget, and
hence it is registered with minimum NAPI weight.
NAPI scheduling does not require locking since there are
different NAPI instances between the Rx and Tx confirmation
paths now.
So, the patch fixes the occurence of spurrious Rx/Tx interrupts.
Tx overruns also occur less frequently now.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Programming the interrupt coalescing (IC) registers while
the controller/DMA is on may incur the loss of one Tx
confirmation interrupt, under certain conditions. This is
a subtle hw race because it does not occur during a burst
of Tx packets. It has been observed on p2020 devices that,
if just one packet is being xmit'ed, the Tx confirmation
doesn't trigger and BQL evetually blocks the Tx queues,
followed by Tx timeout and an un-responsive device.
This issue was not apparent prior to introducing BQL
support, as a late Tx confirmation was not an issue back then
and the next burst of Tx frames would have triggered the
Tx confirmation/ Tx ring cleanup anyway.
Bottom line, the hw specifications state that the IC registers
should not be programmed while the Rx/Tx blocks (the DMA) are
enabled. Further more, these registers are currently re-written
with the same values on the processing path, over and over again.
To fix this, rewriting the IC registers has been removed from
the processing path (napi poll). A complete MAC reset procedure
has been implemented for the ethtool -c option instead, to
reliably update these registers while the controller is stopped.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The device reset procedure, stop_gfar()/startup_gfar(), has
concurrency issues.
"Kernel access of bad area" oopses show up during Tx timeout
device reset or other reset cases (like changing MTU) that
happen while the interface still has traffic. The oopses
happen in start_xmit and clean_tx_ring when accessing tx_queue->
tx_skbuff which is NULL. The race comes from de-allocating the
tx_skbuff while transmission and napi processing are still
active. Though the Tx queues get temoprarily stopped when Tx
timeout occurs, they get re-enabled as a result of Tx congestion
handling inside the napi context (see clean_tx_ring()). Not
disabling the napi during reset is also a bug, because
clean_tx_ring() will try to access tx_skbuff while it is being
de-alloc'ed and re-alloc'ed.
To fix this, stop_gfar() needs to disable napi processing
after stopping the Tx queues. However, in order to prevent
clean_tx_ring() to re-enable the Tx queue before the napi
gets disabled, the device state DOWN has been introduced.
It prevents the Tx congestion management from re-enabling the
de-congested Tx queue while the device is brought down.
An additional locking state, RESETTING, has been introduced
to prevent simultaneous resets or to prevent configuring the
device while it is resetting.
The bogus 'rxlock's (for each Rx queue) have been removed since
their purpose is not justified, as they don't prevent nor are
suited to prevent device reset/reconfig races (such as this one).
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Resetting the device (stop_gfar()/startup_gfar()) should
be fast and to the point, in order to timely recover
from an error condition (like Tx timeout) or during
device reconfig. The irq free/ request routines are just
redundant here, and they should be part of the device
close/ open routines instead.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The RCTRL and TCTRL registers should not be changed
on-the-fly, while the controller is running, otherwise
unexpected behaviour occurs. But that's exactly what
gfar_vlan_mode() does, updating the VLAN acceleration
bits inside RCTRL/TCTRL. The attempt to lock these
operations doesn't help, but only adds to the confusion.
There's also a dependency for Rx FCB insertion (activating
/de-activating the TOE offload block on Rx) which might
change the required rx buffer size. This makes matters
worse as gfar_vlan_mode() ends up calling gfar_change_mtu(),
though the MTU size remains the same. Note that there are
other situations that may affect the required rx buffer size,
like changing RXCSUM or rx hw timestamping, but errorneously
the rx buffer size is not recomputed/ updated in the process.
To fix this, do the vlan updates properly inside the MAC
reset and reconfiguration procedure, which takes care of
the rx buffer size dependecy and the rx TOE block (PRSDEP)
activation/deactivation as well (in the correct order).
As a consequence, MTU/ rx buff size updates are done now
by the same MAC reset and reconfig procedure, so that out
of context updates to MAXFRM, MRBLR, and MACCFG inside
change_mtu() are no longer needed. The rx buffer size
dependecy to Rx FCB is now handled for the other cases too
(RXCSUM and rx hw timestamping).
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The main MAC config registers like: RCTRL/TCTRL, MRBLR,
MAXFRM, RXIC/TXIC, most fields of MACCFG1/2, should not
be changed on-the-fly, but at least after stopping the
DMA and disabling the Rx/Tx blocks and, for increased
reliability, after a MAC soft reset.
Impelement a complete MAC soft reset and reconfig procedure
following the latest HW advisories - gfar_mac_reset() - to
replace gfar_mac_init() and (the confusing) init_registers()
functions.
Factor out separate config functions for RCTRL and TCTRL,
insure programming order of the relevant config regs after
MAC soft reset.
Split gfar_hw_init() into gfar_mac_reset() and the remaining
global regs that don't need to be reconfigured after MAC soft
reset (FIFOCFG, ATTRELI, HW counters a.s.o).
As gfar_hw_init() now makes all the register writes @probe()
time, based on all the device flags and config options, it
must be moved further down, just before register_netdev(),
as the last config step when the config values are comitted
to HW. Also, move netif_carrier_off() after register_netdev(),
because it has no effect if called before.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
gfar_halt() and gfar_start() are responsible for stopping
and starting the DMA and the Rx/Tx hw rings. They implement
the support for the "graceful Rx/Tx stop/start" hw procedure,
and also disable/enable eTSEC's hw interrupts in the process.
The GRS/GTS procedure requires however to have the RQUEUE/TQUEUE
registers cleared first and to wait for a period of time for the
current frame to pass through the interface (around ~10ms for a
jumbo frame). Only then may the GTS and GRS bits from DMACTRL be
set to shut down the DMA, and finally the Tx_EN and Rx_EN bits in
MACCFG1 may be cleared to disable the Tx/Rx blocks.
The same register programming order applies to start the Rx/Tx:
enabling the RQUEUE/TQUEUE *before* clearing the GRS/GTS bits.
This is a HW recommendation in order to avoid a possible
controller "lock up" during graceful reset.
Cleanup the gfar_halt()/start() prototypes, to take priv instead
of ndev as their purpose is to operate on HW. Enabling the
RQUEUE/TQUEUE in the hw_init() is not needed anymore since
that's the job of gfar_start().
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Throughout the code there are places where the controller's
hw interrupt sources need to get disabled/enabled (masked/
un-masked) all at once. The recommendation for disabling
the interrupts is to clear the ievent first then the imask
register (not the other way around).
Use the gfar_ints_enable/disable() helpers to make these
operations consistent.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The RCTRL updates of the FSL_GIANFAR_DEV_HAS_PADDING device
flag get overriden by the FSL_GIANFAR_DEV_HAS_TIMER flag
settings, which impose a Rx padding alignment of 8 bytes.
As all the eTSEC devices that set HAS_PADDING also set the
HAS_TIMER flag, the HAS_PADDING flag is now obsolete.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Claudiu Manoil