- 28 Apr, 2009 4 commits
-
-
Johannes Berg authored
Under certain circumstances iwlwifi can get stuck and will no longer accept scan requests, because the core code (cfg80211) thinks that it's still processing one. This fixes one of the points where it can happen, but I've still seen it (although only with my radio-off-when-idle patch). Signed-off-by:
Johannes Berg <johannes@sipsolutions.net> Acked-by:
Reinette Chatre <reinette.chatre@intel.com> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Jussi Kivilinna authored
rndis_wext_link_change() might be called from rndis_command() at initialization stage and priv->workqueue/priv->work have not been initialized yet. This causes invalid opcode at rndis_wext_bind on some brands of bcm4320. Fix by initializing workqueue/workers in rndis_wext_bind() before rndis_command is used. This bug has existed since 2.6.25, reported at: http://bugzilla.kernel.org/show_bug.cgi?id=12794Signed-off-by:
Jussi Kivilinna <jussi.kivilinna@mbnet.fi> Signed-off-by:
-
Stephen Rothwell authored
drivers/net/wireless/iwlwifi/iwl3945-base.c:1415: error: __ksymtab_iwl3945_rx_queue_reset causes a section type conflict I am pretty sure that this is a compiler bug, so not to worry. However, as far as I can see, iwl-3945.o (the only user) and iwl3945-base.o are always linked into the same module, so the EXPORT_SYMBOL (which causes the problem) should not be needed. Correct? Signed-off-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
-
Eric Dumazet authored
In 2.6.25 we added UDP mem accounting. This unfortunatly added a penalty when a frame is transmitted, since we have at TX completion time to call sock_wfree() to perform necessary memory accounting. This calls sock_def_write_space() and utimately scheduler if any thread is waiting on the socket. Thread(s) waiting for an incoming frame was scheduled, then had to sleep again as event was meaningless. (All threads waiting on a socket are using same sk_sleep anchor) This adds lot of extra wakeups and increases latencies, as noted by Christoph Lameter, and slows down softirq handler. Reference : http://marc.info/?l=linux-netdev&m=124060437012283&w=2 Fortunatly, Davide Libenzi recently added concept of keyed wakeups into kernel, and particularly for sockets (see commit 37e5540b epoll keyed wakeups: make sockets use keyed wakeups) Davide goal was to optimize epoll, but this new wakeup infrastructure can help non epoll users as well, if they care to setup an appropriate handler. This patch introduces new DEFINE_WAIT_FUNC() helper and uses it in wait_for_packet(), so that only relevant event can wakeup a thread blocked in this function. Trace of function calls from bnx2 TX completion bnx2_poll_work() is : __kfree_skb() skb_release_head_state() sock_wfree() sock_def_write_space() __wake_up_sync_key() __wake_up_common() receiver_wake_function() : Stops here since thread is waiting for an INPUT Reported-by:
Christoph Lameter <cl@linux.com> Signed-off-by:
Eric Dumazet <dada1@cosmosbay.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
- 27 Apr, 2009 12 commits
-
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <mike@compulab.co.il> Signed-off-by:
-
Anton Blanchard authored
Right now we have no upper limit on the size of the route cache hash table. On a 128GB POWER6 box it ends up as 32MB: IP route cache hash table entries: 4194304 (order: 9, 33554432 bytes) It would be nice to cap this for memory consumption reasons, but a massive hashtable also causes a significant spike when measuring OS jitter. With a 32MB hashtable and 4 million entries, rt_worker_func is taking 5 ms to complete. On another system with more memory it's taking 14 ms. Even though rt_worker_func does call cond_sched() to limit its impact, in an HPC environment we want to keep all sources of OS jitter to a minimum. With the patch applied we limit the number of entries to 512k which can still be overriden by using the rt_entries boot option: IP route cache hash table entries: 524288 (order: 6, 4194304 bytes) With this patch rt_worker_func now takes 0.460 ms on the same system. Signed-off-by:Anton Blanchard <anton@samba.org> Acked-by:
-
Karsten Keil authored
Move the entry about CAPI 2.0 to the beginning and add a URL. Incorporate changes suggested by Randy Dunlap, thanks for proofreading. Signed-off-by:
Karsten Keil <keil@b1-systems.de> Signed-off-by:
-
Tilman Schmidt authored
isdn: document Kernel CAPI driver interface Create a file Documentation/isdn/INTERFACE.CAPI describing the interface between the kernel CAPI subsystem and ISDN device drivers, analogous to the existing Documentation/isdn/INTERFACE for the old isdn4linux subsystem. Also add kerneldoc comments to the exported functions in drivers/isdn/capi/kcapi.c. Impact: Documentation Signed-off-by:
Tilman Schmidt <tilman@imap.cc> Signed-off-by:
-
Tilman Schmidt authored
After the merging of mISDN, state which files refer only to the old isdn4linux subsystem. Also add a few missing files. Signed-off-by:
-
Waskiewicz Jr, Peter P authored
The current code writes the PME enabled bit in PCI config space which is wrong. This was needed for pre-release hardware, and was not removed from the driver. Also, we need to clear the WUS (wake up status) after we resume. Otherwise we can't wake for the same event again since it's still asserted in the hardware. Plus, the multicast lists were being written improperly, causing multicast WoL to fail. Signed-off-by:
Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
Stephen Hemminger authored
From: Stephen Hemminger <shemminger@vyatta.com> The veth driver will oops if sysfs hooks are open while module is removed. The net device destructor can not point to code in a module; basically there are only two possible safe values: NULL - no destructor, or free_netdev - free on last use Signed-off-by: -
Nicolas Dichtel authored
When kernel inserts a temporary SA for IKE, it uses the wrong hash value for dst list. Two hash values were calcultated before: one with source address and one with a wildcard source address. Bug hinted by Junwei Zhang <junwei.zhang@6wind.com> Signed-off-by:
Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by:
-
Ayaz Abdulla authored
This patch fixes the tx_timeout() to properly handle the clean up of the tx ring. It also sets the tx put pointer back to the correct position to be in sync with HW. Signed-off-by:
Ayaz Abdulla <aabdulla@nvidia.com> Signed-off-by:
-
Adrian Bunk authored
Unless I miss anything this should fix a bug. Signed-off-by:
Adrian Bunk <bunk@kernel.org> Signed-off-by:
-
Yevgeny Petrilin authored
If we failed to allocate new fragments for receive buffer, the packet should be dropped and packets should be reused. Signed-off-by:
Yevgeny Petrilin <yevgenyp@mellanox.co.il> Signed-off-by:
-
Yevgeny Petrilin authored
In case of mlx4_en_activate_cq() failure, the cleanup code would go to rx_err and try to disable unactivated rings. Signed-off-by:
-
- 26 Apr, 2009 2 commits
-
-
Jay Vosburgh authored
Currently, the VLAN event handler does not adjust the VLAN device's carrier state when the real device or the VLAN device is set administratively up or down. The following patch adds a transfer of operating state from the real device to the VLAN device when the real device is administratively set up or down, and sets the carrier state up or down during init, open and close of the VLAN device. This permits observers above the VLAN device that care about the carrier state (bonding's link monitor, for example) to receive updates for administrative changes by more closely mimicing the behavior of real devices. Signed-off-by:
Jay Vosburgh <fubar@us.ibm.com> Signed-off-by:
Patrick McHardy <kaber@trash.net>
-
-
- 24 Apr, 2009 4 commits
-
-
Jan Engelhardt authored
Related-to: commit 325fb5b4 The compat path suffers from a similar problem. It only uses a __be32 when all of the recent code uses, and expects, an nf_inet_addr everywhere. As a result, addresses stored by xt_recents were filled with whatever other stuff was on the stack following the be32. Signed-off-by:
Jan Engelhardt <jengelh@medozas.de> With a minor compile fix from Roman. Reported-and-tested-by:
Roman Hoog Antink <rha@open.ch> Signed-off-by:
-
Pablo Neira Ayuso authored
This patch adds missing role attribute to the DCCP type, otherwise the creation of entries is not of any use. The attribute added is CTA_PROTOINFO_DCCP_ROLE which contains the role of the conntrack original tuple. Signed-off-by:
Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by:
-
Laszlo Attila Toth authored
Signed-off-by:
Laszlo Attila Toth <panther@balabit.hu> Signed-off-by:
-
Patrick McHardy authored
Commit d0dba725 (netfilter: ctnetlink: add callbacks to the per-proto nlattrs) changed the protocol registration function to abort if the to-be registered protocol doesn't provide a new callback function. The DCCP and UDP-Lite IPv6 protocols were missed in this conversion, add the required callback pointer. Reported-and-tested-by:
Steven Jan Springl <steven@springl.ukfsn.org> Signed-off-by:
-
- 22 Apr, 2009 10 commits
-
-
Pablo Neira Ayuso authored
This patch fixes a (bogus?) gcc warning during compilation: net/netfilter/nf_conntrack_netlink.c:1234: warning: 'helpname' may be used uninitialized in this function net/netfilter/nf_conntrack_netlink.c:991: warning: 'helpname' may be used uninitialized in this function In fact, helpname is initialized by ctnetlink_parse_help() so I cannot see a way to use it without being initialized. Signed-off-by:
-
Jean Delvare authored
Patch "af_rose/x25: Sanity check the maximum user frame size" (commit 83e0bbcb) from Alan Cox got locking wrong. If we bail out due to user frame size being too large, we must unlock the socket beforehand. Signed-off-by:
Jean Delvare <jdelvare@suse.de> Signed-off-by:
-
Paul Moore authored
The NetLabel address selector mechanism has a problem where it can get mistakenly remove the wrong selector when similar addresses are used. The problem is caused when multiple addresses are configured that have different netmasks but the same address, e.g. 127.0.0.0/8 and 127.0.0.0/24. This patch fixes the problem. Reported-by:
Etienne Basset <etienne.basset@numericable.fr> Signed-off-by:
Paul Moore <paul.moore@hp.com> Acked-by:
James Morris <jmorris@namei.org> Tested-by:
-
Heiko Schocher authored
If using the UCC on a MPC8360 in RMII mode, don;t set UCC_GETH_UPSMR_RPM bit in the upsmr register. Signed-off-by:
Heiko Schocher <hs@denx.de> Acked-by:
Li Yang <leoli@freescale.com> Signed-off-by:
-
Jianjun kong authored
While ifconfig eth0 up kernel calls open() of 8139 driver(8139too.c). In rtl8139_hw_start() of rtl8139_open(), 8139 driver enable RX before setting up the DMA buffer address. In this interval where RX was enabled and DMA buffer address is not yet set up, any incoming broadcast packet would be send to a strange physical address: 0x003e8800 which is the default value of DMA buffer address. Unfortunately, this address is used by Linux kernel. So kernel panics. This patch fix it by setting up DMA buffer address before RX enabled and everything is fine even under broadcast packets attack. Signed-off-by:
Jonathan Lin <jon.lin@vatics.com> Signed-off-by:
Amos Kong <jianjun@zeuux.org> Signed-off-by:
-
Hendrik Brueckner authored
AF_IUCV runs into a race when queuing incoming iucv messages and receiving the resulting backlog. If the Linux system is under pressure (high load or steal time), the message queue grows up, but messages are not received and queued onto the backlog queue. In that case, applications do not receive any data with recvmsg() even if AF_IUCV puts incoming messages onto the message queue. The race can be avoided if the message queue spinlock in the message_pending callback is spreaded across the entire callback function. Signed-off-by:
Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Signed-off-by:
Ursula Braun <ursula.braun@de.ibm.com> Signed-off-by:
-
Hendrik Brueckner authored
Add few more sk states in iucv_sock_shutdown(). Signed-off-by:
-
Hendrik Brueckner authored
Reject incoming iucv messages if the receive direction has been shut down. It avoids that the queue of outstanding messages increases and exceeds the message limit of the iucv communication path. Signed-off-by:
-
Hendrik Brueckner authored
If iucv_sock_recvmsg() is called with MSG_PEEK flag set, the skb is enqueued twice. If the socket is then closed, the pointer to the skb is freed twice. Remove the skb_queue_head() call for MSG_PEEK, because the skb_recv_datagram() function already handles MSG_PEEK (does not dequeue the skb). Signed-off-by:
-
Ursula Braun authored
Make sure a second invocation of iucv_sock_close() guarantees proper freeing of an iucv path. Signed-off-by:
-
- 21 Apr, 2009 8 commits
-
-
Reinette Chatre authored
A few issues wrt DMA were uncovered when using the driver with swiotlb. - driver should not use memory after it has been mapped - iwl3945's RX queue management cannot use all of iwlagn because the size of the RX buffer is different. Revert back to using iwl3945 specific routines that map/unmap memory. - no need to "dma_syn_single_range_for_cpu" followed by pci_unmap_single, we can just call pci_unmap_single initially - only map the memory area that will be used by device. this is especially relevant to the mapping of iwl_cmd. we should not map the entire structure because the meta data at the beginning of structure contains the address to be used later for unmapping. If the address to be used for unmapping is stored in mapped data it creates a problem. - ensure that _if_ memory needs to be modified after it is mapped that we call _sync_single_for_cpu first, and then release it back to device with _sync_single_for_device - we mapped the wrong length of data for host commands, with mapped length differing with length provided to device, fix that. Thanks to Jason Andryuk <jandryuk@gmail.com> for significant bisecting help to find these issues. This fixes http://www.intellinuxwireless.org/bugzilla/show_bug.cgi?id=1964Signed-off-by:
Jason Andryuk <jandryuk@gmail.com> Tested-by:
Ben Gamari <bgamari@gmail.com> Signed-off-by:
-
Reinette Chatre authored
When debugging TX issues it is helpful to know the seq nr of the frame being transmitted. The seq nr is printed as part of ucode's log informing us which frame is being processed. Having this information printed in driver log makes it easy to match activities between driver and firmware. Also make possible to print TX flags directly. These are already printed as part of entire TX command, but having it printed directly in cpu format makes it easier to look at. Signed-off-by:
-
Rami Rosen authored
This patch fixes a build warning in mwl8.c. (Marvell TOPDOG wireless driver) The warning it fixes is: "large integer implicitly truncated to unsigned type." The rx_ctrl member of the mwl8k_rx_desc struct is 8 bit (__u8 ), whereas trying to assign it a 32 bit value (which is returned from cpu_to_le32()) causes the compiler to issue a truncation warning. Signed-off-by:
Rami Rosen <ramirose@gmail.com> Signed-off-by:
-
Johannes Berg authored
When checking whether or not a given frame needs to be moved to be properly aligned to a 4-byte boundary, we use & 4 which wasn't intended, this code should check the lowest two bits. Signed-off-by:
-
Reinette Chatre authored
It is expected that config interface will always succeed as mac80211 will only request what driver supports. The exception here is when a device has rfkill enabled. At this time the rfkill state is unknown to mac80211 and config interface can fail. When this happens we deal with this error instead of printing a WARN. Signed-off-by:
-
Reinette Chatre authored
Users reported lockup with work still trying to run after module has been unloaded. http://thread.gmane.org/gmane.linux.kernel.wireless.general/30594/focus=30601Signed-off-by:
TJ <ubuntu@tjworld.net> Reported-by:
Huaxu Wan <huaxu.wan@linux.intel.com> Signed-off-by:
-
Jay Sternberg authored
Fix the bug where some revisions of 6000 series hardware cannot be used. Later versions of 6000 series have the EEPROM replaced by OTP. For these devices to be used we need to expand valid EEPROM mask. Signed-off-by:
Jay Sternberg <jay.e.sternberg@linux.intel.com> Signed-off-by:
-
Johannes Berg authored
sparse says: drivers/net/wireless/atmel.c:1501:3: warning: Initializer entry defined twice drivers/net/wireless/atmel.c:1505:3: also defined here and it's correct; atmel has its own ndo_change_mtu and shouldn't use eth_change_mtu. Signed-off-by:
-
