Commit graph

155188 commits

Author SHA1 Message Date
Luciano Coelho
9f483dc3d1 wl12xx: pass the wake up condition when configuring the wake up event
Changed the function wl12xx_acx_wake_up_conditions() so that it receives an
argument with the actual wake up condition, instead of having
WAKE_UP_EVENT_DTIM_BITMAP hardcoded.  This is needed because we have to use
different conditions in 1271.

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:46 -04:00
Luciano Coelho
052a625a85 wl12xx: add support for fixed address in wl12xx_spi_read
In the wl1271 implementation, we need to read memory from the register
partition using fixed addresses.  This change adds the possibility to request
fixed address when calling wl12xx_spi_read() or wl12xx_spi_reg_read().

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:45 -04:00
Luciano Coelho
27797d68f7 wl12xx: add support for new WL1271 chip revision
This patch adds the code that recognizes the new WL1271 chip revision (PG 2.0).
Full support for this chip is not yet implemented and support for WL1271 PG 1.0
is not guaranteed anymore.

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:45 -04:00
Luciano Coelho
0d1c38398f wl12xx: moved firmware version reading routine to chip-specific functions
With WL1271, the firmware version can only be read right after booting the
chip.  To keep WL1251 aligned with this procedure, the code that reads the
firmware version initially has been moved to a common place where it can be
read from both chipsets.

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:45 -04:00
Luciano Coelho
a336e26664 wl12xx: add wl12xx_spi_reg_read() and wl12xx_spi_reg_write() functions
In some cases we need to read more than 32 bits from the register area.
These functions were added to support that, like the existing
wl12xx_spi_mem_read() and wl12xx_spi_mem_write() already do for large blocks
in the memory area.

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:45 -04:00
Juuso Oikarinen
0628817acc wl12xx: removed chipset interrupt source configuration from fw wakeup
The chipset source interrupt source configuration during fw wakeup was
blocking interrupts on the wl1271. The configuration is effectively
unused here as it is used to trigger an interrupt for the chipset
wake-up event, which is not handled, or waited for, in the fw wakeup
anyway on either wl1251 or wl1271.

Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:45 -04:00
Kalle Valo
4721213fdd wl12xx: fix rx descriptor use
Rx descriptor was incorrectly allocated from stack, use struct wl12xx
instead. Needed for DMA transfers.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:43 -04:00
Kalle Valo
53d65423ba wl12xx: use wl12xx_mem_read32() to read the rx counter
As a side effect the transfer is now DMA safe.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:43 -04:00
Kalle Valo
5262c12d16 wl12xx: allocate buffer the spi busy word from struct wl12xx
Needed for DMA transfers.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:43 -04:00
Kalle Valo
56343a3cfd wl12xx: allocate buffer spi read/write command buffer kzalloc()
Needed for DMA safe transfers.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:42 -04:00
Kalle Valo
8d47cdb617 wl12xx: reserve buffer for partition command in struct wl12xx
This is now DMA safe.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:42 -04:00
Kalle Valo
c4f5c85218 wl12xx: fix error handling in wl12xx_probe()
Resources were not freed properly in some cases.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:42 -04:00
Kalle Valo
1d3b813061 wl12xx: reserver buffer for read32()/write32() in struct wl12xx
The buffer is needed for DMA safe transfers.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:42 -04:00
Kalle Valo
ff25839bf0 wl12xx: cmd and acx interface rework
Rework cmd and acx interfaces, it was just too confusing earlier. Now
all commands need to contain all the needed headers, either just cmd
headers or both cmd and acx headers. This accomplish to remove the
extra copy done for each command. The interfaces are now properly
documented as well.

Also try to make all commands safe for DMA transfers. I might have missed
some, but most of them should be fixed now.

And this is not all! As a free bonus you will also get some cosmetic
cleanups and code reorganisation. Order today!

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:41 -04:00
Vasanthakumar Thiagarajan
c4f9f16b30 ath9k: Make sure we configure a non-zero beacon interval
This patch moves the sanity check on beacon interval
which was done only for mesh mode to all operating
modes just to be safe.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:41 -04:00
Andrey Yurovsky
e0d6133cba libertas: remove ps_supported flag, use fwcapinfo
Power save support depends on the firmware capabilities rather than the
card's hardware interface.  Use the FW_CAPINFO_PS bit in the firmware
capabilities mask throughout the driver in place of the redundant
ps_supported flag and don't make decisions about PS support in the
interface drivers (with the exception of a special case in the USB
driver).

V2: put the USB special case in the right place.

Signed-off-by: Andrey Yurovsky <andrey@cozybit.com>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:41 -04:00
Bob Copeland
f0f3d388ba ath5k: enable hardware LEDs
Cardbus and some PCI cards use hardware LEDs rather than software GPIOs.
Program them with the proper blink patterns when idle, scanning or
associated.  Fixes http://bugzilla.kernel.org/show_bug.cgi?id=13288.

Tested-by: Frans Pop <elendil@planet.nl>
Tested-by: Mark Hindley <mark@hindley.org.uk>
Signed-off-by: Bob Copeland <me@bobcopeland.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:41 -04:00
Bob Copeland
46026e8f48 ath5k: cleanup ath5k_hw struct
ah_gpios array isn't used, and ah_current_channel can be a pointer
instead of an embedded struct.  Removing these and some other
write-only variables, and moving some things around for better
packing and cache utilization saves 116 bytes.

   text	   data	    bss	    dec	    hex	filename
 121762	    472	     64	 122298	  1ddba	ath5k_before.ko
 121646	    472	     64	 122182	  1dd46	ath5k.ko

Signed-off-by: Bob Copeland <me@bobcopeland.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:41 -04:00
Vasanthakumar Thiagarajan
8a46097a6c ath9k: downgrade ASSERT() in ath_clone_txbuf()
We can easily run out of tx buf if there is any stuck in
transmission, so downgrade it to WARN_ON().

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:39 -04:00
Vasanthakumar Thiagarajan
c3d8f02ed9 ath9k: Remove unnecessary count for addba attempt
mac80211 already has one to keep track of number of failure
addba attempts.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:39 -04:00
Vasanthakumar Thiagarajan
089e698d23 ath9k: Nuke unneccesary helper function to see if aggr is active
IEEE80211_TX_CTL_AMPDU in tx control flags should be enough
to confirm BA is negotiated for that tid.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:38 -04:00
Johannes Berg
e36d56b648 cfg80211: pass netdev to change_virtual_intf
If there was a reason I'm passing the ifidx I cannot
remember it any more and don't see one now, so let's
just pass the pointer itself.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-10 14:57:38 -04:00
David S. Miller
e5a8a896f5 Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6 2009-07-09 20:18:24 -07:00
Roland Dreier
e594e96e8a cxgb3: Fix crash caused by stashing wrong netdev_queue
Commit c3a8c5b6 ("cxgb3: move away from LLTX") exposed a bug in how
cxgb3 looks up the netdev_queue it stashes away in a qset during
initialization.  For multiport devices, the TX queue index it uses is
offset by the first_qset index of each port.  This leads to a crash
once LLTX is removed, since hard_start_xmit is called with one TX
queue lock held, while the TX reclaim timer task grabs a different
(wrong) TX queue lock when it frees skbs.

Fix this by removing the first_qset offset used to look up the TX
queue passed into t3_sge_alloc_qset() from setup_sge_qsets().

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Acked-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-09 17:16:06 -07:00
Yi Zou
8faa2a786a ixgbe: Fix coexistence of FCoE and Flow Director in 82599
Fix coexistence of Fiber Channel over Ethernet (FCoE) and Flow Director (FDIR)
in 82599 and remove the disabling of FDIR when FCoE is enabled.

Currently, FDIR is turned off when FCoE is enabled under the assumption that
FCoE is always enabled with DCB being turned on. However, FDIR does not have
to be turned off all the time when FCoE is enabled since FCoE can be enabled
without DCB being turned on, e.g., use link pause only. This patch makes sure
that when DCB is turned on or off, FDIR is turned on or off correspondingly;
and when FCoE is enabled, it does not disable FDIR, rather, it will have FDIR
set up properly so FCoE and FDIR can coexist regardless of DCB being on or off.

Signed-off-by: Yi Zou <yi.zou@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-09 17:16:05 -07:00
Jiri Olsa
ad46276952 memory barrier: adding smp_mb__after_lock
Adding smp_mb__after_lock define to be used as a smp_mb call after
a lock.

Making it nop for x86, since {read|write|spin}_lock() on x86 are
full memory barriers.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-09 17:06:58 -07:00
Jiri Olsa
a57de0b433 net: adding memory barrier to the poll and receive callbacks
Adding memory barrier after the poll_wait function, paired with
receive callbacks. Adding fuctions sock_poll_wait and sk_has_sleeper
to wrap the memory barrier.

Without the memory barrier, following race can happen.
The race fires, when following code paths meet, and the tp->rcv_nxt
and __add_wait_queue updates stay in CPU caches.

CPU1                         CPU2

sys_select                   receive packet
  ...                        ...
  __add_wait_queue           update tp->rcv_nxt
  ...                        ...
  tp->rcv_nxt check          sock_def_readable
  ...                        {
  schedule                      ...
                                if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
                                        wake_up_interruptible(sk->sk_sleep)
                                ...
                             }

If there was no cache the code would work ok, since the wait_queue and
rcv_nxt are opposit to each other.

Meaning that once tp->rcv_nxt is updated by CPU2, the CPU1 either already
passed the tp->rcv_nxt check and sleeps, or will get the new value for
tp->rcv_nxt and will return with new data mask.
In both cases the process (CPU1) is being added to the wait queue, so the
waitqueue_active (CPU2) call cannot miss and will wake up CPU1.

The bad case is when the __add_wait_queue changes done by CPU1 stay in its
cache, and so does the tp->rcv_nxt update on CPU2 side.  The CPU1 will then
endup calling schedule and sleep forever if there are no more data on the
socket.

Calls to poll_wait in following modules were ommited:
	net/bluetooth/af_bluetooth.c
	net/irda/af_irda.c
	net/irda/irnet/irnet_ppp.c
	net/mac80211/rc80211_pid_debugfs.c
	net/phonet/socket.c
	net/rds/af_rds.c
	net/rfkill/core.c
	net/sunrpc/cache.c
	net/sunrpc/rpc_pipe.c
	net/tipc/socket.c

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-09 17:06:57 -07:00
Anton Vorontsov
1b614fb9a0 netpoll: Fix carrier detection for drivers that are using phylib
Using early netconsole and gianfar driver this error pops up:

  netconsole: timeout waiting for carrier

It appears that net/core/netpoll.c:netpoll_setup() is using
cond_resched() in a loop waiting for a carrier.

The thing is that cond_resched() is a no-op when system_state !=
SYSTEM_RUNNING, and so drivers/net/phy/phy.c's state_queue is never
scheduled, therefore link detection doesn't work.

I belive that the main problem is in cond_resched()[1], but despite
how the cond_resched() story ends, it might be a good idea to call
msleep(1) instead of cond_resched(), as suggested by Andrew Morton.

[1] http://lkml.org/lkml/2009/7/7/463

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 20:09:44 -07:00
David S. Miller
d2daeabf62 Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-2.6 2009-07-08 18:13:13 -07:00
Jaswinder Singh Rajput
1ce822fa04 includecheck fix: include/linux, rfkill.h
fix the following 'make includecheck' warning:

  include/linux/rfkill.h: linux/types.h is included more than once.

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:32 -04:00
Larry Finger
47ab3840a3 p54: tx refused but queue active
In the mainline kernel, p54usb will fail because the TX queue length can
become < 0. This problem has been reported as Bugzilla #13725. The failure
is expressed by the following message in the logs:

WARNING: at net/mac80211/tx.c:1325 ieee80211_tx+0x23c/0x298 [mac80211]()
Hardware name: HP Pavilion dv2700 Notebook PC
tx refused but queue active

This problem has been recently observed in the wireless-testing tree, where
a full solution is being tested. That fix is too invasive for 2.6.31-rcX,
but the simple change supplied here will prevent the failure.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:30 -04:00
Jay Sternberg
804ef71ee1 Atheros Kconfig needs to be dependent on WLAN_80211
Atheros top level menu needs a "depends WLAN_80211" to properly indent
within menuconfig and xconfig interfaces.

This is purely a visual issue but it effects all subsequent drivers.
The issue is the top level menu does not include a dependency on
WLAN_80211 so within the tree structure, Atheros is at the same level as
WLAN_80211 but when WLAN_80211 collapsed, the menu disappears along with
all subsequent drives, so it is really a subordinate.

Signed-off-by: Jay Sternberg <jay.e.sternberg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:29 -04:00
Johannes Berg
b9744d19e3 mac80211: fix docbook
These two functions no longer exist in mac80211,
so trying to insert them generates warnings in
the document.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:28 -04:00
Johannes Berg
4ff176674e mac80211_hwsim: avoid NULL access
There's a race condition -- started can be set to true
before channel is set due to the way mac80211 callbacks
currently work (->start should probably pass the channel
we would like to have initially). For now simply add a
check to hwsim to avoid dereferencing the NULL channel
pointer.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:26 -04:00
Clyde McPherson
2fbddeb5c4 ssb: Add support for 4318E
Added support for the Broadcom 4318E chipset on PCMCIA/CF cards. The
4318E can do 802.11A/B/G, only B and G mode are supported in b43.

Signed-off-by: Clyde McPherson <ccmcphe@verizon.net>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:25 -04:00
Clyde McPherson
cff782cd94 b43: Add support for 4318E
Added support for the Broadcom 4318E chipset on PCMCIA/CF cards. The
4318E can do 802.11A/B/G, only B and G mode are supported in b43.

Signed-off-by: Clyde McPherson <ccmcphe@verizon.net>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:24 -04:00
Hin-Tak Leung
8b339d0580 zd1211rw: adding SONY IFU-WLM2 (054c:0257) as a zd1211b device
Yevgen Kotikov reported success on the sourceforge zd1211-devs list
with the following details:

Brand/retail: SONY IFU-WLM2
USB-IDs: Vendor: 0x054C Device: 0x0257
chip ID: zd1211b chip 054c:0257 v4802 high 00-0b-6b AL2230_RF pa0 -----
FCC ID: unknown

Signed-off-by: Hin-Tak Leung <htl10@users.sourceforge.net>
Tested-by: Yevgen Kotikov <yevgen.kotikov@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:23 -04:00
Pascal Terjan
b4b223cdd5 zd1211rw: 07b8:6001 is a ZD1211B
On a shuttle machine here we got 07b8:6001 device, handled by zd1211rw, which does not work.
Scanning is OK but association does not work, we get "direct probe to AP xxx timed out"

It appears that this simple patch makes the device work perfectly.

This id was already there in initial import of the driver so I don't know if it has ever been
working as ZD1211 (which would mean they changed it and kept the id :( ).

Signed-off-by: Pascal Terjan <pterjan@mandriva.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
2009-07-08 15:24:08 -04:00
Anton Vorontsov
bff38771e1 netpoll: Introduce netpoll_carrier_timeout kernel option
Some PHYs require longer timeouts for carrier detection, and
auto-negotiation process may take indefinite amount of time.

It may be inconvenient to force longer timeouts for sane PHYs,
so let's introduce a kernel command line option.

Since we're using module_param(), the option also can be
changed in runtime.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 11:10:56 -07:00
Divy Le Ray
34701fde8f cxgb3: Drain Mac Tx fifo when the port goes down.
Drain the MAC TX fifos when a port goes down.
Back pressure might otherwise occur, leading to both
ports of the same adapter to hang.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:27 -07:00
Divy Le Ray
88045b3cf0 cxgb3: fix mac index mapping
Override the mac index computation for the gen2 adapter,
as each port is expected to use index 0.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:26 -07:00
Divy Le Ray
dce7d1d031 cxgb3: Fix mss table initialization
Fix the HW SMT table initialization to avoid random
mss miscomputations for offload connections.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:25 -07:00
Divy Le Ray
2e8c07c35b cxgb3: use request_firmware() for the EDC registers setup
use request_firmware() to load the phy's EDC programmation

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:24 -07:00
Divy Le Ray
5e65951556 cxgb3: AEL2020 phy support update
We don't always see the link status update interrupt
when we come out of reset and the peer is up.
Check and report the link status right before enabling interrupts.
Also fix LED settings, to get a consistent link status.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:23 -07:00
Divy Le Ray
cfe2462c6a cxgb3: Fix T3C MAC max packet size access
Max packet size is not the only field in T3C's High Water Mark register.
Mask the register to access this field.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:22 -07:00
Divy Le Ray
619f05cf69 cxgb3: fix phy power down
2 phys are were not getting the Global Tx disable bit set
when powered down, leading to an inconsistent link state
on peer.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:20 -07:00
Divy Le Ray
2c3d50f7db cxgb3: AQ100X phy support update
Add missing SUPPORTED_TP flag.
Update FW version checking.
Do the full initialization even if the FW version is unknown,
it might help catching further issues.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:54:18 -07:00
Florian Fainelli
c3b8542307 r6040: bump driver version to 0.24 and date to 08 July 2009
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:49:28 -07:00
Florian Fainelli
35976d4d55 r6040: restore MIER register correctly when IRQ line is shared
When the r6040 device IRQ line is shared we will enter the driver
interrupt service routine, mask off the device interrupt enable
register (MIER) and return with IRQ_NONE, we would then leave the
device with interrupts disabled, this patch fixes that issue.

Reported-by: Steve Holland <sdh4@iastate.edu>
Signed-off-by: Joe Chou <joe.chou@rdc.com.tw>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:49:26 -07:00
Jarek Poplawski
345aa03120 ipv4: Fix fib_trie rebalancing, part 4 (root thresholds)
Pawel Staszewski wrote:
<blockquote>
Some time ago i report this:
http://bugzilla.kernel.org/show_bug.cgi?id=6648

and now with 2.6.29 / 2.6.29.1 / 2.6.29.3 and 2.6.30 it back
dmesg output:
oprofile: using NMI interrupt.
Fix inflate_threshold_root. Now=15 size=11 bits
...
Fix inflate_threshold_root. Now=15 size=11 bits

cat /proc/net/fib_triestat
Basic info: size of leaf: 40 bytes, size of tnode: 56 bytes.
Main:
        Aver depth:     2.28
        Max depth:      6
        Leaves:         276539
        Prefixes:       289922
        Internal nodes: 66762
          1: 35046  2: 13824  3: 9508  4: 4897  5: 2331  6: 1149  7: 5
9: 1  18: 1
        Pointers: 691228
Null ptrs: 347928
Total size: 35709  kB
</blockquote>

It seems, the current threshold for root resizing is too aggressive,
and it causes misleading warnings during big updates, but it might be
also responsible for memory problems, especially with non-preempt
configs, when RCU freeing is delayed long after call_rcu.

It should be also mentioned that because of non-atomic changes during
resizing/rebalancing the current lookup algorithm can miss valid leaves
so it's additional argument to shorten these activities even at a cost
of a minimally longer searching.

This patch restores values before the patch "[IPV4]: fib_trie root
node settings", commit: 965ffea43d from
v2.6.22.

Pawel's report:
<blockquote>
I dont see any big change of (cpu load or faster/slower
routing/propagating routes from bgpd or something else) - in avg there
is from 2% to 3% more of CPU load i dont know why but it is - i change
from "preempt" to "no preempt" 3 times and check this my "mpstat -P ALL
1 30"
always avg cpu load was from 2 to 3% more compared to "no preempt"
[...]
cat /proc/net/fib_triestat
Basic info: size of leaf: 20 bytes, size of tnode: 36 bytes.
Main:
        Aver depth:     2.44
        Max depth:      6
        Leaves:         277814
        Prefixes:       291306
        Internal nodes: 66420
          1: 32737  2: 14850  3: 10332  4: 4871  5: 2313  6: 942  7: 371  8: 3  17: 1
        Pointers: 599098
Null ptrs: 254865
Total size: 18067  kB
</blockquote>

According to this and other similar reports average depth is slightly
increased (~0.2), and root nodes are shorter (log 17 vs. 18), but
there is no visible performance decrease. So, until memory handling is
improved or added parameters for changing this individually, this
patch resets to safer defaults.

Reported-by: Pawel Staszewski <pstaszewski@itcare.pl>
Reported-by: Jorge Boncompte [DTI2] <jorge@dti2.net>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Tested-by: Pawel Staszewski <pstaszewski@itcare.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-08 10:46:45 -07:00