The SBCF firmware format has been extended to support extra headers
after the main payload. These extra headers are used to sign the
firmware code with more than one certificate. This eases up
distributing single code images that work in more than one SKU of the
device.
The changes to support this feature will be spread in a series of
commits. This one just adds the support to parse the extra headers and
store them in i2400m->fw_hdrs. Coming changes to the loader code will
use that to determine which header to upload to the device.
The i2400m_fw_check() function now iterates over all the headers and
for each, calls i2400m_fw_hdr_check(), which does some basic checks on
each header. It then stores the headers for the bootloader code to use.
The i2400m_dev_bootstrap() function has been modified to cleanup
i2400m->fw_hdrs when done.
wimax/i2400m: verify firmware format version is known
Make sure the bootloading code checks that the format of the file is
understood (major version match). This also fixes a dumb typo in
extracting the major version field.
The i2400m based devices can get in a sort of a deadlock some times;
when they boot, they send a reboot "barker" (a magic number) and then
the driver has to echo that same barker to ack reception
(echo/ack). Then the device does a final ack by sending an ACK barker.
The first time this happens, we don't know ahead of time with barker
the device is going to send, as different device models and SKUs will
send different barker depending on the EEPROM programming.
If the device has sent the barker before the driver has been able to
read it, the driver looses, as it doesn't know which barker it has to
echo/ack back. With older devices, we tried a couple of combinations
and that always worked; but now, with adding support for more, in
which we have an unlimited number of new barkers, that is not an
option.
So we rework said case so that when the device gets stuck, we just
cycle through all the known types until one forces the device to send
an ack. Otherwise, the driver gives up and aborts.
wimax/i2400m: retry loading firmware files in sequence
The i2400m firmware loader is given a list of firmware files to try to
load by the probe() function (which can be different based on the
device's model / generation).
Current code didn't attempt to load, check and try to boot with each
file, but just to try to load if off disk. This is limiting in some
cases, where we might want to try to load a firmware and if it fails
to load onto the device, just fall back to another one.
This changes the behaviour so all files are tried for being loaded
from disk, checked and uploaded to the device until one suceeds in
bringing the device up.
wimax/i2400m: rework bootrom initialization to be more flexible
This modifies the bootrom initialization code of the i2400m driver so
it can more easily support upcoming hardware.
Currently, the code detects two types of barkers (magic numbers) sent
by the device to indicate the types of firmware it would take (signed
vs non-signed).
This schema is extended so that multiple reboot barkers are
recognized; upcoming hw will expose more types barkers which will have
to match a header in the firmware image before we can load it.
For that, a barker database is introduced; the first time the device
sends a barker, it is matched in the database. That gives the driver
the information needed to decide how to upload the firmware and which
types of firmware to use. The database can be populated from module
parameters.
The execution flow is not altered; a new function
(i2400m_is_boot_barker) is introduced to determine in the RX path if
the device has sent a boot barker. This function is becoming heavier,
so it is put away from the hot reception path [this is why there is
some reorganization in sdio-rx.c:i2400ms_rx and
usb-notifc.c:i2400mu_notification_grok()].
The documentation on the process has also been updated.
All these modifications are heavily based on previous work by Dirk
Brandewie <dirk.brandewie@intel.com>.
wimax/i2400m: decide properly if using signed vs non-signed firmware loading
The i2400m based devices can boot two main types of firmware images:
signed and non-signed. Signed images have signature data included that
must match that of a certificate stored in the device.
Currently the code is making the decission on what type of firmware
load (signed vs non-signed) is going to be loaded based on a hardcoded
decission in __i2400m_ack_verify(), based on the barker the device
sent upon boot.
This is not flexible enough as future hardware will emit more barkers;
thus the bit has to be set in a place where there is better knowledge
of what is going on. This will be done in follow-up commits -- however
this patch paves the way for it.
So the querying of the mode is packed into i2400m_boot_is_signed();
the main changes are just using i2400m_boot_is_signed() to determine
the method to follow and setting i2400m->sboot in
i2400m_is_boot_barker(). The modifications in i2400m_dnload_init() and
i2400m_dnload_finalize() are just reorganizing the order of the if
blocks and thus look larger than they really are.
wimax: allow specifying debug levels as command line option
Add "debug" module options to all the wimax modules (including
drivers) so that the debug levels can be set upon kernel boot or
module load time.
This is needed as currently there was a limitation where the debug
levels could only be set when a device was succesfully
enumerated. This made it difficult to debug issues that made a device
not probe properly.
wimax/i2400m: during probe, call sdio_disable at most once
In the Intel Wireless Multicomm 3200, the initialization is
orchestrated by a component called Top. This component also monitors
how many times a function is reset (via sdio_disable) to detect
possible issues and will reset the whole multifunction device if any
function triggers a maximum reset level.
During WiMAX's probe, the driver needs to wait for Top to come up
before it can enable the WiMAX function. If it cannot, it will return
-ENODEV and the Top driver will rescan the SDIO bus once done
loading.
Currently, the WiMAX SDIO probe routine was trying a few times before
returning -ENODEV, and this was triggering Top's too-many-resets
detector. This is, in any case, unnecessary because the Top driver will
force the bus rescan when the functions can be probed successfully.
Added then a maxtries argument to i2400ms_enable_func() and set it to
1 when calling from probe. We want to reuse this function instead of
flat calling out sdio_enable_func() to take advantage of hardware
quirk workarounds.
Reported-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Cindy H Kao [Wed, 19 Aug 2009 02:27:18 +0000 (19:27 -0700)]
wimax/i2400m: don't write to memory allocated by request_firmware()
In kernel 2.6.31, the firmware requested to ram could be marked
with read only attribute, and we can't write any thing directly
to the memory when setting up the last JUMP brh cmd.
Changed so that the scratch buffer is used.
Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
wimax/i2400m: be smarter about copying command buffer to bm_cmd_buf
Because some underlying bus APIs (like USB) don't like data buffers in
the stack or vmalloced areas, the i2400m driver provides a scratch
buffer (i2400m->bm_cmd_buf) for said low-level drivers to copy command
data to before passing it to said API. This is only used during boot
mode.
However, at some the code was copying the buffer even when the command
was already specified in said buffer. This is ok, but it needs to be
more careful. As thus, change so that:
(a) the copy happens only if command buffer is not the scratch buffer
Dirk Brandewie [Wed, 12 Aug 2009 18:29:46 +0000 (11:29 -0700)]
wimax/i2400m: USB driver uses a configurable endpoint map
Newer generations of the i2400m USB WiMAX device use a different
endpoint map; in order to make it easy to support it, we make the
endpoint-to-function mapeable instead of static.
Cindy H Kao [Tue, 18 Aug 2009 02:39:12 +0000 (19:39 -0700)]
wimax/iwmc3200: increase wait time before enable retry
When trying to enable the iwmc3200 WiMAX SDIO function, we loop
waiting for the top controller to be up and running (at which point we
can succesfully enable the function). Between each try we wait for
I2400MS_INIT_SLEEP_INTERVAL ms.
Integration tests have found the current value of 10ms to be too
short; it was upped to 100ms to give more time to the top controller
to be ready.
Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Cindy H Kao [Tue, 18 Aug 2009 02:39:12 +0000 (19:39 -0700)]
wimax/iwmc3200: don't disable the SDIO function if enable fails
In the iwmc3200, disabling the WiMAX SDIO function when enable fails
would possibly result in a device reset triggered by the iwmc3200's
top controller since it monitors the bus reset activities from each
SDIO function. In any case, the disable makes no sense; if the enable
fails, it should not be disabled.
Thus we remove the unecessary sdio_disable_func() in
i2400ms_enable_function().
Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Cindy H Kao [Tue, 18 Aug 2009 02:17:58 +0000 (19:17 -0700)]
wimax/iwmc3200: overwrite SDIO IOR timeout value to avoid platform hang
The default SDIO IOE wait timeout returned from iwmc3200-wimax's CCCR
is not efficient. This inefficiency will actually cause problems on
Moorestown platforms (system hang).
This is a sillicon bug that requires a software patch to by
overwritting func->enable_timeout. The new value I2400MS_IOR_TIMEOUT
is recommended and verified from the system integration results.
Future sillicon releases will have this default value corrected in the
future.
Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Dirk Brandewie [Thu, 13 Aug 2009 20:48:29 +0000 (13:48 -0700)]
wimax/i2400m: Make boot retries a BUS-specific parameter
In i2400m-based devices, the driver's bootloader will retry to load
the firmware when things go wrong. The driver currently has a constant
(I2400M_BOOT_RETRIES) which governs the max number of tries.
However, different SKUs of the same hardware may admit or require
different numbers of retries due to it's particulars, so it is made a
BUS specific parameter and different values are assigned for 5x50
devices versus the 3200 ones.
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com> Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Dirk Brandewie [Tue, 18 Aug 2009 15:51:52 +0000 (08:51 -0700)]
wimax/i2400m: Ensure boot mode cmd and ack buffers are alloc'd before first message
The change to the SDIO boot mode RX chain could try to use the cmd and
ack buffers befor they were allocated. USB does not have the problem
but both were changed for consistency's sake.
wimax: indicate initial SW rfkill state is "blocked"
The WiMAX stack assumes that all WiMAX devices are SW OFF when they
are initialized. The recent changes in the RFKILL stack thus cause an
initial call after rfkill_register(), because by default, rfkill
considers devices to be SW ON upon registration.
So call rfkill_init_sw_state() to set it to SW OFF so
rfkill_register() doesn't do that unnecessary step.
Cindy H Kao [Thu, 27 Aug 2009 22:25:12 +0000 (15:25 -0700)]
wimax/i2400m: fix the bootmode RX deadlock in SDIO driver
i2400ms_bus_bm_wait_for_ack() causes a race condition. It happens
because this function clears i2400ms->bm_ack_size before waiting for
an interrupt, which is set by the interrupt service routine i2400ms_rx()
to indicate reception and size of received data; thus, if the interrupt
came right before the clearing/waiting, it is lost.
The fix is clear the bm_ack_size to -EINPROGRESS before we are enabling
the RX interrupt configuration in i2400ms_rx_setup(). Then everytime
when the interrupt service routine i2400ms_rx() is invoked during bootmode,
bm_ack_size is updated with the actual rx_size and it is cleared to
-EINPROGRESS again after the RX data is handled.
Signed-off-by: Cindy H Kao <cindy.h.kao@intel.com> Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Eric Dumazet [Thu, 15 Oct 2009 06:30:45 +0000 (06:30 +0000)]
inet: rename some inet_sock fields
In order to have better cache layouts of struct sock (separate zones
for rx/tx paths), we need this preliminary patch.
Goal is to transfert fields used at lookup time in the first
read-mostly cache line (inside struct sock_common) and move sk_refcnt
to a separate cache line (only written by rx path)
This patch adds inet_ prefix to daddr, rcv_saddr, dport, num, saddr,
sport and id fields. This allows a future patch to define these
fields as macros, like sk_refcnt, without name clashes.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Krishna Kumar [Wed, 14 Oct 2009 19:55:07 +0000 (19:55 +0000)]
genetlink: Optimize and one bug fix in genl_generate_id()
1. GENL_MIN_ID is a valid id -> no need to start at
GENL_MIN_ID + 1.
2. Avoid going through the ids two times: If we start at
GENL_MIN_ID+1 (*or bigger*) and all ids are over!, the
code iterates through the list twice (*or lesser*).
3. Simplify code - no need to start at idx=0 which gets
reset to GENL_MIN_ID.
Patch on net-next-2.6. Reboot test shows that first id
passed to genl_register_family was 16, next two were
GENL_ID_GENERATE and genl_generate_id returned 17 & 18
(user level testing of same code shows expected values
across entire range of MIN/MAX).
Signed-off-by: Krishna Kumar <krkumar2@in.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Einar Lueck [Wed, 14 Oct 2009 22:54:59 +0000 (22:54 +0000)]
ctcm rollback in case of errors
Group device now cleanly reacts to failures during channel start and
implements a clean rollback.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com> Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
lcs: Recognize return codes of ccw_device_set_online().
The creation of a new lcs device requires a call to the function
ccw_device_set_online() for the read and the write channel. Failure
of either call should terminate the lcs device creation immediately
with return code -ENODEV.
Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com> Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
lcs: ODEBUG: object is on stack, but not annotated.
Timer_list structure in lcs_send_lancmd() is allocated on stack.
Initialization with init_timer() leads to above ODEBUG message.
Instead use init_timer_on_stack() which prevents above msg.
Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com> Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
af_iucv: use sk functions to modify sk->sk_ack_backlog
Instead of modifying sk->sk_ack_backlog directly, use respective
socket functions.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Narender Kumar [Fri, 16 Oct 2009 15:50:11 +0000 (15:50 +0000)]
netxen: sysfs control for auto firmware recovery
Firmware hang detection and recovery (reset) need to
be disabled for diagnostic tools, which can run
some disruptive tests.
This adds a driver level control to turn off this
feature by diag tools.
Signed-off-by: Narender Kumar <narender.kumar@qlogic.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for different windowing scheme for on chip
memory in future chip revisions. This is required by
diagnostic tools.
Signed-off-by: Amit Kumar Salecha <amit@netxen.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Future revisions need different chip reset sequence
and firmware initialization.
Also clean up some never used debug code.
Signed-off-by: Amit Kumar Salecha <amit@netxen.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Future revisions of the chip have 128 bit memory
transactions. Require drivers to implement rmw
in case of sub-128 bit accesses by driver. This
is mostly used by diagnostic tools.
Signed-off-by: Amit Kumar Salecha <amit@netxen.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Amit Kumar Salecha <amit@netxen.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Phonet: hold socket before giving it to sk_deliver_skb()
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eilon Greenstein [Thu, 15 Oct 2009 07:18:47 +0000 (00:18 -0700)]
bnx2x: Do not call load/unload functionality from DCC
There is really no need to clear the MAC or the FW filtering rules - it was
added for completion, but caused race conditions with load/unload. Removing this
redundant code
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eilon Greenstein [Thu, 15 Oct 2009 07:18:08 +0000 (00:18 -0700)]
bnx2x: Changing the Disabled state to a flag
When working with DCC, a function can be disabled or enabled (virtual link down
or up). Using the function state introduced some race conditions with the
load/unload flow.
Using a separate flag to indicate that the function is disabled.
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eilon Greenstein [Thu, 15 Oct 2009 07:17:20 +0000 (00:17 -0700)]
bnx2x: Allowing 0 as initial fairness value
Value of zero was used to disable the fairness mechanism. Though the code
(driver and FW) allowed changing the value at run time, it did not allow to do
that if the mechanism was disabled to begin with.
Fixed the FW to allow turning on and off the mechanism at run time. Fixed the
code to read the value from the chip at the right sequence.
Without this fix, if the initial value was set to zero, traffic could not run on
the interface.
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Thu, 15 Oct 2009 03:40:11 +0000 (20:40 -0700)]
net: sk_drops consolidation
sock_queue_rcv_skb() can update sk_drops itself, removing need for
callers to take care of it. This is more consistent since
sock_queue_rcv_skb() also reads sk_drops when queueing a skb.
This adds sk_drops managment to many protocols that not cared yet.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Reported-by: "Tony Luck" <tony.luck@intel.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Acked-by: Tony Luck <tony.luck@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Thomas Gleixner [Wed, 14 Oct 2009 08:19:46 +0000 (01:19 -0700)]
net: Remove BKL from tun
The lock_kernel/unlock_kernel() in cycle_kernel_lock() which is called
in tun_chr_open() is not serializing against anything and safe to
remove.
tun_chr_fasync() is serialized by get/put_tun() and fasync_helper()
has no dependency on BKL. The modification of tun->flags is racy with
and without the BKL so removing it does not make it worse.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Sriram [Wed, 7 Oct 2009 02:44:30 +0000 (02:44 +0000)]
TI DaVinci EMAC: Clear statistics register properly.
The mechanism to clear the statistics register is dependent
on the status of GMIIEN bit in MAC control register. If the
GMIIEN bit is set, the stats registers are write to decrement.
If the GMIIEN bit is cleared, the stats registers are plain
read/write registers. The stats register clearing operation
must take into account the current state of GMIIEN as it
can be cleared when the interface is brought down.
With existing implementation logic, querying for interface stats
when the interface is down, can corrupt the statistics counters.
This patch examines the GMIIEN bit status in MAC_CONTROL
register before choosing an appropriate mask for clearing stats
registers.
Signed-off-by: Sriramakrishnan <srk@ti.com> Acked-by: Chaithrika U S <chaithrika@ti.com> Signed-off-by: David S. Miller <davem@davemloft.net>
françois romieu [Wed, 7 Oct 2009 12:44:20 +0000 (12:44 +0000)]
r8169: partial support and phy init for the 8168d
Extracted from Realtek's 8.012.00 r8168 driver.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com> Tested-by: Simon Farnsworth <simon.farnsworth@onelan.com> Cc: Edward Hsu <edward_hsu@realtek.com.tw> Signed-off-by: David S. Miller <davem@davemloft.net>
Ajit Khaparde [Tue, 13 Oct 2009 01:46:56 +0000 (01:46 +0000)]
ixgb: Fix erroneous display of stats by ethtool -S
Commit 5675f221 overlooked the way offsets for netdev stats were considered.
Because of this some of the stats shown by ethtool -S were wrong.
This patch fixes it.
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ajit Khaparde [Tue, 13 Oct 2009 01:46:29 +0000 (01:46 +0000)]
igb: Fix erroneous display of stats by ethtool -S
Commit 337e067d overlooked the way offsets for netdev stats were considered.
Because of this some of the stats shown by ethtool -S were wrong.
This patch fixes it.
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ajit Khaparde [Tue, 13 Oct 2009 01:45:48 +0000 (01:45 +0000)]
e1000: Fix erroneous display of stats by ethtool -S
Commit 23d26497 overlooked the way offsets for netdev stats were considered.
Because of this some of the stats shown by ethtool -S were wrong.
This patch fixes it.
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ajit Khaparde [Tue, 13 Oct 2009 01:45:09 +0000 (01:45 +0000)]
e1000e: Fix erroneous display of stats by ethtool -S
Commit fd8235bb overlooked the way offsets for netdev stats were considered.
Because of this some of the stats shown by ethtool -S were wrong.
This patch fixes it.
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Ajit Khaparde [Tue, 13 Oct 2009 01:47:33 +0000 (01:47 +0000)]
ixgbe: Fix erroneous display of stats by ethtool -S
Commit 59aa3cc4 overlooked the way offsets for netdev stats were considered.
Because of this some of the stats shown by ethtool -S were wrong.
This patch fixes it.
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dhananjay Phadke [Tue, 13 Oct 2009 05:31:45 +0000 (05:31 +0000)]
netxen: add sysfs entries for diag tools
Add sysfs entries to enable register and memory access
for diagnostic tools.
Signed-off-by: Narender Kumar <narender.kumar@qlogic.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Sub 64-bit / unaligned access to oncard memory was only used
by old diagnostic tools, causes some intermittent issues when
memory controller agent is used. The new access method was
added by commit ea6828b8aa3a8ebae8d7740f32f212ba1d2f0742
("netxen: improve pci memory access"). Firmware init anyway
uses 8-byte strides.
This also fixes address/offset calculation for NX2031 context
memory (SIU). For NX3031, SIU uses same register offsets
as packet memory (MIU).
Signed-off-by: Amit Kumar Salecha <amit@netxen.com> Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Karen Xie [Thu, 8 Oct 2009 09:11:05 +0000 (09:11 +0000)]
cxgb3: Added private MAC address and provisioning packet handler for iSCSI
This patch added support of private MAC address per port and provisioning
packet handler for iSCSI traffic only.
The above changes are isolated to the cxgb3 driver, independent of any scsi or iscsi driver changes.
Acked-by: Karen Xie <kxie@chelsio.com> Acked-by: Divy Le Ray <divy@chelsio.com> Signed-off-by: Rakesh Ranjan <rakesh@chelsio.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Magnus Damm [Fri, 9 Oct 2009 00:20:04 +0000 (00:20 +0000)]
net: add Runtime PM to the sh_eth driver
Add Runtime PM support to the sh_eth driver.
The clock to the ethernet hardware block will be
enabled as long as the network device is up.
Signed-off-by: Magnus Damm <damm@opensource.se> Tested-by: Kuninori Morimoto <morimoto.kuninori@renesas.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Magnus Damm [Fri, 9 Oct 2009 00:17:14 +0000 (00:17 +0000)]
net: allow sh_eth to get mac address through platform data
Extend the sh_eth driver to allow passing the mac address
using the platform data structure. This to simplify board
setup code.
Signed-off-by: Magnus Damm <damm@opensource.se> Tested-by: Kuninori Morimoto <morimoto.kuninori@renesas.com> Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows the CAN controller driver to define the number of echo
skb's used for the local loopback (echo), as suggested by Kurt Van
Dijck, with the function:
struct net_device *alloc_candev(int sizeof_priv,
unsigned int echo_skb_max);
The CAN drivers have been adapted accordingly. For the ems_usb driver,
as suggested by Sebastian Haas, the number of echo skb's has been
increased to 10, which improves the transmission performance a lot.
Signed-off-by: Wolfgang Grandegger <wg@grandegger.com> Signed-off-by: Kurt Van Dijck <kurt.van.dijck@eia.be> Signed-off-by: David S. Miller <davem@davemloft.net>
Cisco HDLC uses keepalive packets and sequence numbers to determine link
state. In rare cases both ends could transmit keepalive packets at the same
time, causing the received sequence numbers to be treated as incorrect.
Now we accept our current sequence number as well as the previous one.
Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl> Signed-off-by: David S. Miller <davem@davemloft.net>
Willy Tarreau [Tue, 13 Oct 2009 07:27:40 +0000 (00:27 -0700)]
tcp: fix tcp_defer_accept to consider the timeout
I was trying to use TCP_DEFER_ACCEPT and noticed that if the
client does not talk, the connection is never accepted and
remains in SYN_RECV state until the retransmits expire, where
it finally is deleted. This is bad when some firewall such as
netfilter sits between the client and the server because the
firewall sees the connection in ESTABLISHED state while the
server will finally silently drop it without sending an RST.
This behaviour contradicts the man page which says it should
wait only for some time :
TCP_DEFER_ACCEPT (since Linux 2.4)
Allows a listener to be awakened only when data arrives
on the socket. Takes an integer value (seconds), this
can bound the maximum number of attempts TCP will
make to complete the connection. This option should not
be used in code intended to be portable.
Also, looking at ipv4/tcp.c, a retransmit counter is correctly
computed :
case TCP_DEFER_ACCEPT:
icsk->icsk_accept_queue.rskq_defer_accept = 0;
if (val > 0) {
/* Translate value in seconds to number of
* retransmits */
while (icsk->icsk_accept_queue.rskq_defer_accept < 32 &&
val > ((TCP_TIMEOUT_INIT / HZ) <<
icsk->icsk_accept_queue.rskq_defer_accept))
icsk->icsk_accept_queue.rskq_defer_accept++;
icsk->icsk_accept_queue.rskq_defer_accept++;
}
break;
==> rskq_defer_accept is used as a counter of retransmits.
But in tcp_minisocks.c, this counter is only checked. And in
fact, I have found no location which updates it. So I think
that what was intended was to decrease it in tcp_minisocks
whenever it is checked, which the trivial patch below does.
Signed-off-by: Willy Tarreau <w@1wt.eu> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / ~andy / linux / log