When filling netlink info, dport is being returned as flags. Fix
instances to return correct value.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Let the tasklet only be enabled after open(), and be disabled for
the other situation. The tasklet is only necessary after open() for
tx/rx, so it could be disabled by default.
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the required pieces to 8250-OMAP UART driver for DMA
support. The TX burst size is set to 1 so we can send an arbitrary
amount of bytes.
The RX burst is currently set to 48 which means we receive an DMA
interrupt every 48 bytes and have to reprogram everything. Less bytes in
the RX-FIFO mean that no DMA transfer will happen and the UART will send a
RX-timeout _or_ RDI event at which point the FIFO will be manually purged.
There is a workaround for TX-DMA on AM33xx where we put the first byte
into the FIFO to kick start the DMA process. Haven't seen this problem on
OMAP36xx (beagle board xm) or DRA7xx.
On AM375x there is "Usage Note 2.7: UART: Cannot Acknowledge Idle
Requests in Smartidle Mode When Configured for DMA Operations" in the
errata document. This problem persists even after disabling DMA in the
UART and will be addressed in the HWMOD.
v10:
- delay update_registers() from set_termios() until TX-DMA is
done. It has been reported / proved that invoking
update_registers() while TX-DMA is in progress may stall the
DMA operation and it won't finish.
- use the new omap DMA-TX-RX hooks and DMA only interrupt
routine.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We have (or will have) custom DMA callbacks in the omap driver due to
the different behaviour in the RX and TX case. To make this work
we need a few changes in the IRQ handler to invoke the rx_handler again
after the "manual" mode or retry the tx_handler again before falling
back to the manual mode.
Heikki didn't want to see the extra hacks in the generic / default irq
handler and Peter wasn't too happy about an OMAP-only IRQ handler. The
way I planned it is to use this extra IRQ routine only in DMA case. If
Peter dislike this approach then I hope Heikki doesn't block changes in
the default IRQ handler :)
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It has been reported that generating an MLD listener report on
devices with large MTUs (e.g. 9000) and a high number of IPv6
addresses can trigger a skb_over_panic():
skbuff: skb_over_panic: text:ffffffff80612a5d len:3776 put:20
head:ffff88046d751000 data:ffff88046d751010 tail:0xed0 end:0xec0
dev:port1
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:100!
invalid opcode: 0000 [#1] SMP
Modules linked in: ixgbe(O)
CPU: 3 PID: 0 Comm: swapper/3 Tainted: G O 3.14.23+ #4
[...]
Call Trace:
<IRQ>
[<ffffffff80578226>] ? skb_put+0x3a/0x3b
[<ffffffff80612a5d>] ? add_grhead+0x45/0x8e
[<ffffffff80612e3a>] ? add_grec+0x394/0x3d4
[<ffffffff80613222>] ? mld_ifc_timer_expire+0x195/0x20d
[<ffffffff8061308d>] ? mld_dad_timer_expire+0x45/0x45
[<ffffffff80255b5d>] ? call_timer_fn.isra.29+0x12/0x68
[<ffffffff80255d16>] ? run_timer_softirq+0x163/0x182
[<ffffffff80250e6f>] ? __do_softirq+0xe0/0x21d
[<ffffffff8025112b>] ? irq_exit+0x4e/0xd3
[<ffffffff802214bb>] ? smp_apic_timer_interrupt+0x3b/0x46
[<ffffffff8063f10a>] ? apic_timer_interrupt+0x6a/0x70
mld_newpack() skb allocations are usually requested with dev->mtu
in size, since commit 72e09ad107 ("ipv6: avoid high order allocations")
we have changed the limit in order to be less likely to fail.
However, in MLD/IGMP code, we have some rather ugly AVAILABLE(skb)
macros, which determine if we may end up doing an skb_put() for
adding another record. To avoid possible fragmentation, we check
the skb's tailroom as skb->dev->mtu - skb->len, which is a wrong
assumption as the actual max allocation size can be much smaller.
The IGMP case doesn't have this issue as commit 57e1ab6ead
("igmp: refine skb allocations") stores the allocation size in
the cb[].
Set a reserved_tailroom to make it fit into the MTU and use
skb_availroom() helper instead. This also allows to get rid of
igmp_skb_size().
Reported-by: Wei Liu <lw1a2.jing@gmail.com>
Fixes: 72e09ad107 ("ipv6: avoid high order allocations")
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Cc: David L Stevens <david.stevens@oracle.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The omap needs a DMA request pending right away. If it is
enqueued once the bytes are in the FIFO then nothing will happen
and the FIFO will be later purged via RX-timeout interrupt.
This patch enqueues RX-DMA request on completion but not if it
was aborted on error. The first enqueue will happen in the driver
in startup.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch provides mostly a copy of serial8250_tx_dma() +
__dma_tx_complete() with the following extensions:
- DMA bug
At least on AM335x the following problem exists: Even if the TX FIFO is
empty and a TX transfer is programmed (and started) the UART does not
trigger the DMA transfer.
After $TRESHOLD number of bytes have been written to the FIFO manually the
UART reevaluates the whole situation and decides that now there is enough
room in the FIFO and so the transfer begins.
This problem has not been seen on DRA7 or beagle board xm (OMAP3). I am not
sure if this is UART-IP core specific or DMA engine.
The workaround is to use a threshold of one byte, program the DMA
transfer minus one byte and then to put the first byte into the FIFO to
kick start the transfer.
- support for runtime PM
RPM is enabled on start_tx(). We can't disable RPM on DMA complete callback
because there is still data in the FIFO which is being sent. We have to wait
until the FIFO is empty before we disable it.
For this to happen we fake a TX sent error and enable THRI. Once the
FIFO is empty we receive an interrupt and since the TTY-buffer is still
empty we "put RPM" via __stop_tx(). Should it been filed then in the
start_tx() path we should program the DMA transfer and remove the error
flag and the THRI bit.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The OMAP has a few corner cases where it needs a share of kindness of
affection to do the right thing. Heikki Krogerus suggested that instead
adding the quirks into the default DMA implementation, OMAP could get
its own copy of the function. And Alan suggested the same thing so here
we go.
This patch provides callbacks for custom TX/RX DMA implementation. If
there are not setup / used, then the default (current) implementation is
used.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After dmaengine_terminate_all() has been invoked then both DMA drivers
(edma and omap-dma) do not invoke dma_cookie_complete() to mark the
transfer as complete. This dma_cookie_complete() is performed by the
Synopsys DesignWare driver which is probably the only one that is used
by omap8250-dma and hence don't see following problem…
…which is that once a RX transfer has been terminated then following
query of channel status reports DMA_IN_PROGRESS (again: the actual
transfer has been canceled, there is nothing going on anymore).
This means that serial8250_rx_dma() never enqueues another DMA transfer
because it (wrongly) assumes that there is a transer already pending.
Vinod Koul refuses to accept a patch which adds this
dma_cookie_complete() to both drivers and so dmaengine_tx_status() would
report DMA_COMPLETE instead (and behave like the Synopsys DesignWare
driver already does). He argues that I am not allowed to use the cookie
to query the status and that the driver already cleaned everything up after
the invokation of dmaengine_terminate_all().
To end this I add a bookkeeping whether or not a RX-transfer has been
started to the 8250-dma code. It has already been done for the TX side.
*Now* we learn about the RX status based on our bookkeeping and don't
need dmaengine_tx_status() for this anymore.
Cc: vinod.koul@intel.com
Reviewed-by: Tony Lindgren <tony@atomide.com>
Tested-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Right now it is possible that serial8250_tx_dma() fails and returns
-EBUSY. The caller (serial8250_start_tx()) will then enable
UART_IER_THRI which will generate an interrupt once the TX FIFO is
empty.
In serial8250_handle_irq() nothing will happen because up->dma is set
and so serial8250_tx_chars() won't be invoked. We end up with plenty of
interrupts and some "too much work for irq" output.
This patch introduces dma_tx_err in struct uart_8250_port to signal that
the last invocation of serial8250_tx_dma() failed so we can fill the TX
FIFO manually. Should the next invocation of serial8250_start_tx()
succeed then the dma_tx_err flag along with the THRI bit is removed and
DMA only usage may continue.
Reviewed-by: Tony Lindgren <tony@atomide.com>
Tested-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch provides a 8250-core based UART driver for the internal OMAP
UART. The long term goal is to provide the same functionality as the
current OMAP uart driver and DMA support.
I tried to merge omap-serial code together with the 8250-core code.
There should should be hardly a noticable difference. The trigger levels
are different compared to omap-serial:
- omap serial
TX: Interrupt comes after TX FIFO has room for 16 bytes.
TX of 4096 bytes in one go results in 256 interrupts
RX: Interrupt comes after there is on byte in the FIFO.
RX of 4096 bytes results in 4096 interrupts.
- this driver
TX: Interrupt comes once the TX FIFO is empty.
TX of 4096 bytes results in 65 interrupts. That means there will
be gaps on the line while the driver reloads the FIFO.
RX: Interrupt comes once there are 48 bytes in the FIFO or less over
"longer" time frame. We have
1 / 11520 * 10^3 * 16 => 1.38… ms
1.38ms to react and purge the FIFO on 115200,8N1. Since the other
driver fired after each byte it had ~5.47ms time to react. This
_may_ cause problems if one relies on no missing bytes and has no
flow control. On the other hand we get only 85 interrupts for the
same amount of data.
It has been only tested as console UART on am335x-evm, dra7-evm and
beagle bone. I also did some longer raw-transfers to meassure the load.
The device name is ttyS based instead of ttyO. If a ttyO based node name
is required please ask udev for it. If both driver are activated (this
and omap-serial) then this serial driver will take control over the
device due to the link order
v9…v10:
- Tony noticed that omap3 won't show anything after waking up
from core off. In v9 I reworked the register restore and set
IER to 0 by accident. This went unnoticed because start_tx
usually sets ier (either due to DMA bug or due to TX-complete
IRQ).
- dropped EFR and SLEEP from capabilities. We do have both but
nobody should touch it. We already handle SLEEP ourself.
- make the private copy of the registers (like EFR) u8 instead
u32
- drop MDR1 & DL[ML] reset in restore registers. Does not look
required it is set to the required value later.
- update MDR1 & SCR only if changed.
- set MDR1 as the last thing. The errata says that we should
setup everything before MDR1 set.
- avoid div by 0 in omap_8250_get_divisor() if baud rate gets
very large (Frans Klaver fixed the same thing omap-serial)
- drop "is in early stage" from Kconfig.
v8…v9:
- less on a file seems to hang the am335x after a while. I
believe I introduce this bug a while ago since I can reproduce
this prior to v8. Fixed by redoing the omap8250_restore_regs()
v7…v8:
- redo the register write. There is now one function for that
which is used from set_termios() and runtime-resume.
- drop PORT_OMAP_16750 and move the setup to the omap file. We
have our own set termios function anyway (Heikki Krogerus)
- use MEM instead of MEM32. TRM of AM/DM37x says that 32bit
access on THR might result in data abort. We only need 32bit
access in the errata function which is before we use 8250's
read function so it doesn't matter.
v4…v7:
- change trigger levels after some tests with raw transfers.
v3…v4:
- drop RS485 support
- wire up ->throttle / ->unthrottle
v2…v3:
- wire up startup & shutdown for wakeup-irq handling.
- RS485 handling (well the core does).
v1…v2:
- added runtime PM. Could somebody could please double check
this?
- added omap_8250_set_termios()
Reviewed-by: Tony Lindgren <tony@atomide.com>
Tested-by: Tony Lindgren <tony@atomide.com>
Tested-by: Frans Klaver <frans.klaver@xsens.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Using a single fixed string is smaller code size than using
a format and many string arguments.
Reduces overall code size a little.
$ size net/ipv4/igmp.o* net/ipv6/mcast.o* net/ipv6/ip6_flowlabel.o*
text data bss dec hex filename
34269 7012 14824 56105 db29 net/ipv4/igmp.o.new
34315 7012 14824 56151 db57 net/ipv4/igmp.o.old
30078 7869 13200 51147 c7cb net/ipv6/mcast.o.new
30105 7869 13200 51174 c7e6 net/ipv6/mcast.o.old
11434 3748 8580 23762 5cd2 net/ipv6/ip6_flowlabel.o.new
11491 3748 8580 23819 5d0b net/ipv6/ip6_flowlabel.o.old
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
if I boot with console=ttyS0 and the omap driver is module I end up with
| console [ttyS0] disabled
| omap8250 44e09000.serial: ttyS0 at MMIO 0x44e09000 (irq = 88, base_baud = 3000000) is a 8250
| Unable to handle kernel paging request at virtual address c07a9de0
| Modules linked in: 8250_omap(+)
| CPU: 0 PID: 908 Comm: modprobe Not tainted 3.17.0-rc5+ #1593
| PC is at serial8250_console_setup+0x0/0xc8
| LR is at register_console+0x13c/0x3a4
| [<c0078788>] (register_console) from [<c02d0340>] (uart_add_one_port+0x3cc/0x420)
| [<c02d0340>] (uart_add_one_port) from [<c02d38a4>] (serial8250_register_8250_port+0x298/0x39c)
| [<c02d38a4>] (serial8250_register_8250_port) from [<bf006274>] (omap8250_probe+0x218/0x3dc [8250_omap])
| [<bf006274>] (omap8250_probe [8250_omap]) from [<c02e3424>] (platform_drv_probe+0x2c/0x5c)
| [<c02e3424>] (platform_drv_probe) from [<c02e1eac>] (driver_probe_device+0x104/0x228)
…
| [<c009fa48>] (SyS_init_module) from [<c000e6e0>] (ret_fast_syscall+0x0/0x30)
| Code: 7823603b f8314620 051b3013 491ed416 (44792204)
because serial8250_console_setup() is already gone.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Frans reworded the two comments with better English for better
understanding. His review hit the mailing list after the patch got
applied so here is an incremental update.
Reported-by: Frans Klaver <frans.klaver@xsens.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
By default the arch_fast_hash hashing function pointers are initialized
to jhash(2). If during boot-up a CPU with SSE4.2 is detected they get
updated to the CRC32 ones. This dispatching scheme incurs a function
pointer lookup and indirect call for every hashing operation.
rhashtable as a user of arch_fast_hash e.g. stores pointers to hashing
functions in its structure, too, causing two indirect branches per
hashing operation.
Using alternative_call we can get away with one of those indirect branches.
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Cc: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When using mxs-auart based console, sometime we need the sysrq function
to help debugging kernel. The sysrq code is basically there,
this patch just simply enable it.
Signed-off-by: Janusz Uzycki <j.uzycki@elproma.com.pl>
Reviewed-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This error message is not necessary. The driver core code will print all
probe error messages. It also resolves some error codes to proper error
messages. For example -EPROBE_DEFER will only be printed as an info message.
This patch removes the error message as the core prints the same
information.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A port count mismatch occurs if mutex_lock_interruptible()
exits uart_open() and the port has already been opened. This may
prematurely close a port on an open tty. Since uart_close() is _always_
called if uart_open() fails, the port count must be corrected if errors
occur.
Always increment the port count in uart_open(), regardless of errors;
always decrement the port count in uart_close(). Note that
tty_port_close_start() decrements the port count when uart_open()
was successful.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
uart_start() only claims the port->lock to call __uart_start(),
which does the actual processing. Eliminate the extra acquire/release
in uart_write(); call __uart_start() directly with port->lock already
held.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The key function of uart_add_one_port() is to cross-reference the
UART driver's port structure with the serial core's state table;
keep the assignments together and document this crucial association.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_port_init() initializes close_delay and closing_wait to these
same values; remove.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Note the serial_struct flags for which the kernel ignores and performs
no action. The flags cannot be removed since they form part of the
userspace interface via the TIOCSSERIAL/TIOCGSERIAL ioctls.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The userspace-defined ASYNC_* flags in include/uapi/linux/tty_flags.h
are the authoritative bit definitions for the serial_struct flags,
and thus for any derivative values or fields.
Although the serial core provides the TIOCSSERIAL and TIOCGSERIAL
ioctls to set and retrieve these flags from userspace, it defines these
bits independently, as UPF_* macros.
Define the UPF_* macros which are userspace-modifiable directly from
the ASYNC_* symbolic constants. Add compile-time test to ensure the
bits changeable by TIOCSSERIAL match the defined range in the uapi
header.
Add ASYNCB_MAGIC_MULTIPLIER to the uapi header since this bit is
programmable by userspace.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The low-level uart driver may modify termios settings to override
settings that are not compatible with the uart, such as CRTSCTS.
Thus, callers of the low-level uart driver's set_termios() method must
hold termios_rwsem write lock to prevent concurrent access to termios,
in case such override occurs.
The termios_rwsem lock requirement does not extend to console setup
(ie., uart_set_options), as console setup cannot race with tty
operations. Nor does this lock requirement extend to functions which
cannot be concurrent with tty ioctls (ie., uart_port_startup() and
uart_resume_port()).
Further, always claim the port mutex to protect hardware
re-reprogramming in the set_termios() uart driver method. Note this
is unnecessary for console initialization in uart_set_options()
which cannot be concurrent with other uart operations.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The tty buffers (and any line discipline buffers) must be flushed after
the UART hardware has shutdown; otherwise, a racing open on the same
tty may receive data from the previous session, which is a security
hazard. However, holding the port mutex while flushing the line
discipline buffers creates a lock inversion if the set_termios()
handler takes the port mutex (as it does in the followup patch,
'serial: Fix locking for uart driver set_termios method'.
Flush the ldisc buffers after dropping the port mutex; the tty lock
is still held which prevents a concurrent open() from advancing while
flushing. Since no new rx data is possible after uart_shutdown() until
a new open reinitializes the port, the later flush has no impact on
what data is being discarded.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the context of the final tty & port close, flushing the tx
ring buffer after the hardware has already been shutdown and
the ring buffer freed is neither required nor desirable.
uart_flush_buffer() performs 3 operations:
1. Resets tx ring buffer indices, but the tx ring buffer has
already been freed and the indices are reset if the port is
re-opened.
2. Calls uart driver's flush_buffer() method
5 in-tree uart drivers define flush_buffer() methods:
amba-pl011, atmel-serial, imx, serial-tegra, timbuart
These have been refactored into the shutdown() method, if
required.
3. Kicks the ldisc for more writing, but this is undesirable.
The file handle is being released; any waiting writer will
will be kicked out by tty_release() with a warning. Further,
the N_TTY ldisc may generate SIGIO for a file handle which
is no longer valid.
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tom Lendacky says:
====================
amd-xgbe: AMD XGBE driver updates 2014-11-04
The following series of patches includes functional updates to the
driver as well as some trivial changes for function renaming and
spelling fixes.
- Move channel and ring structure allocation into the device open path
- Rename the pre_xmit function to dev_xmit
- Explicitly use the u32 data type for the device descriptors
- Use page allocation for the receive buffers
- Add support for split header/payload receive
- Add support for per DMA channel interrupts
- Add support for receive side scaling (RSS)
- Add support for ethtool receive side scaling commands
- Fix the spelling of descriptors
- After a PCS reset, sync the PCS and PHY modes
- Add dependency on HAS_IOMEM to both the amd-xgbe and amd-xgbe-phy
drivers
This patch series is based on net-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
tty_ldisc_flush() first clears the line discipline input buffer,
then clears the tty flip buffers. However, this allows for existing
data in the tty flip buffers to be added after the ldisc input
buffer has been cleared, but before the flip buffers have been cleared.
Add an optional ldisc parameter to tty_buffer_flush() to allow
tty_ldisc_flush() to pass the ldisc to clear.
NB: Initially, the plan was to do this automatically in
tty_buffer_flush(). However, an audit of the behavior of existing
line disciplines showed that performing a ldisc buffer flush on
ioctl(TCFLSH) was not always the outcome. For example, some line
disciplines have flush_buffer() methods but not ioctl() methods,
so a ->flush_buffer() command would be unexpected.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When changing the ldisc on one end of a pty pair, there may be
waiting readers/writers on the other end which may not exit from
the ldisc i/o loop, preventing tty_ldisc_lock_pair_timeout() from
acquiring the other side's ldisc lock.
Only acquire this side's ldisc lock; although this will no longer
prevent the other side from writing new input, that input will not
be processed until after the ldisc change completes. This has no
effect on normal ttys; new input from the driver was never disabled.
Remove tty_ldisc_enable_pair().
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When hanging up one end of a pty pair, there may be waiting
readers/writers on the other end which may not exit, preventing
tty_ldisc_lock_pair() from acquiring the other side's ldisc lock.
Only acquire this side's ldisc lock; although this will no longer
prevent the other side from writing new input, that input will not
be processing until after the ldisc hangup is complete.
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tty_ldisc_lock(), tty_ldisc_unlock(), and tty_ldisc_lock_nested()
are low-level aliases for the underlying lock mechanism. Rename
with double underscore to allow for new, higher level functions
with those names.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When releasing one end of a pty pair, that end may just have written
to the other, which the input processing worker, flush_to_ldisc(), is
still working on but has not completed the copy to the other end's
read buffer. So input may not appear to be available to a waiting
reader but yet TTY_OTHER_CLOSED is now observed. The n_tty line
discipline has worked around this by waiting for input processing
to complete and then re-checking if input is available before
exiting with -EIO.
Since the tty/ldisc lock reordering, the wait for input processing
to complete can now occur during final close before setting
TTY_OTHER_CLOSED. In this way, a waiting reader is guaranteed to
see input available (if any) before observing TTY_OTHER_CLOSED.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With the revised tty lock order and lockdep annotation, claiming
the pty slave lock is now safe while still holding the pty master lock.
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Eliminate the requirement of specifying the tty lock nesting at
lock time; instead, set the lock subclass for slave ptys at pty
install (normal ttys and master ptys use subclass 0).
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When releasing the master pty, the slave pty also needs to be locked
to prevent concurrent tty count changes for the slave pty and to
ensure that only one parallel master and slave release observe the
final close, and proceed to destruct the pty pair. Conversely, when
releasing the slave pty, locking the master pty is not necessary
(since the master's state can be inferred by the slave tty count).
Introduce tty_lock_slave()/tty_unlock_slave() which acquires/releases
the tty lock of the slave pty. Remove tty_lock_pair()/tty_unlock_pair().
Dropping the tty_lock is no longer required to re-establish a stable
lock order.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The amd-xgbe-phy driver needs to perform ioremap calls, so add HAS_IOMEM
to its build dependency.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The amd-xgbe driver needs to perform ioremap calls, so add HAS_IOMEM
to its build dependency.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support to sync the states of the PCS and the PHY
after a reset is performed. If the PCS and the PHY are not in the
same state after reset an extra mode change would be performed. This
extra mode change might not be needed if the PCS and the PHY are
synced up after reset.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes the spelling of the word "descriptor" in a couple
of locations.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for ethtool receive side scaling (RSS) commands.
Support is added to get/set the RSS hash key and the RSS lookup table.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch provides support for receive side scaling (RSS). RSS allows
for spreading incoming network packets across the Rx queues. When used
in conjunction with the per DMA channel interrupt support, this allows
the receive processing to be spread across multiple processors.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch provides support for interrupts that are generated by the
Tx/Rx DMA channel pairs of the device. This allows for Tx and Rx
processing to run across multiple processsors.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Provide support for splitting IP packets so that the header and
payload can be sent to different DMA addresses. This will allow
the IP header to be put into the linear part of the skb while the
payload can be added as frags.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use page allocations for Rx buffers instead of pre-allocating skbs
of a set size.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The Tx and Rx descriptors are unsigned 32 bit values. Use the u32
type, rather than unsigned int, to map these descriptors.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The pre_xmit function name implies that it performs operations prior
to transmitting the packet when in fact it is responsible for setting
up the descriptors and initiating the transmit. Rename this to
function from pre_xmit to dev_xmit, which is consistent with the name
used during receive processing - dev_read.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move the channel and ring tracking structures allocation to device
open. This will allow for future support to vary the number of Tx/Rx
queues without unloading the module.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The local o_tty variable in tty_release() is now accessed only
when closing the pty master.
Set o_tty to slave pty when closing pty master, otherwise NULL;
use o_tty != NULL as replacement for pty_master.
Reviewed-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
