- 16 Apr, 2009 7 commits
-
-
Michael Buesch authored
This patch adds poisoning and sanity checking to the RX DMA buffers. This is used for protection against buggy hardware/firmware that raises RX interrupts without doing an actual DMA transfer. This mechanism protects against rare "bad packets" (due to uninitialized skb data) and rare kernel crashes due to uninitialized RX headers. The poison is selected to not match on valid frames and to be cheap for checking. The poison check mechanism _might_ trigger incorrectly, if we are voluntarily receiving frames with bad PLCP headers. However, this is nonfatal, because the chance of such a match is basically zero and in case it happens it just results in dropping the packet. Bad-PLCP RX defaults to off, and you should leave it off unless you want to listen to the latest news broadcasted by your microwave oven. This patch also moves the initialization of the RX-header "length" field in front of the mapping of the DMA buffer. The CPU should not touch the buffer after we mapped it. Cc: stable@kernel.org Reported-by:
Francesco Gringoli <francesco.gringoli@ing.unibs.it> Signed-off-by:
Michael Buesch <mb@bu3sch.de> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Vasanthakumar Thiagarajan authored
Currently rx status for frames which are completed from reorder buffer is taken from it's cb area which is not always right, cb is not holding the rx status when driver uses mac80211's non-irq rx handler to pass it's received frames. This results in dropping almost all frames from reorder buffer when security is enabled by doing double decryption (first in hw, second in sw because of wrong rx status). This patch copies rx status into cb area before the frame is put into reorder buffer. After this patch, there is a significant improvement in throughput with ath9k + WPA2(AES). Signed-off-by:
Vasanthakumar Thiagarajan <vasanth@atheros.com> Acked-by:
Johannes Berg <johannes@sipsolutions.net> Cc: stable@kernel.org Signed-off-by:
-
Luis R. Rodriguez authored
We won't ever get here as regulatory_hint_core() can only fail on -ENOMEM and in that case we don't initialize cfg80211 but this is technically correct code. This is actually good for stable, where we don't check for -ENOMEM failure on __regulatory_hint()'s failure. Cc: stable@kernel.org Reported-by:
Quentin Armitage <Quentin@armitage.org.uk> Signed-off-by:
Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by:
-
Atsushi Nemoto authored
The commit a390d1f3 ("phylib: convert state_queue work to delayed_work") missed converting 'expires' value to 'delay' value. Signed-off-by:
Atsushi Nemoto <anemo@mba.ocn.ne.jp> Acked-by:
Marcin Slusarz <marcin.slusarz@gmail.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Rafael J. Wysocki authored
Prevent ixgbe from putting the adapter into D3 during shutdown except when we're going to power off the system, since doing that may generally cause problems with kexec to happen (such problems were observed for igb and forcedeth). For this purpose seperate ixgbe_shutdown() from ixgbe_suspend() and use the appropriate PCI PM callbacks in both of them. Signed-off-by:
"Rafael J. Wysocki" <rjw@sisk.pl> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
Rafael J. Wysocki authored
Prevent e1000e from putting the adapter into D3 during shutdown except when we're going to power off the system, since doing that may generally cause problems with kexec to happen (such problems were observed for igb and forcedeth). For this purpose seperate e1000e_shutdown() from e1000e_suspend() and use the appropriate PCI PM callbacks in both of them. Signed-off-by:
-
Rafael J. Wysocki authored
Prevent e1000 from putting the adapter into D3 during shutdown except when we're going to power off the system, since doing that may generally cause problems with kexec to happen (such problems were observed for igb and forcedeth). For this purpose seperate e1000_shutdown() from e1000_suspend() and use the appropriate PCI PM callbacks in both of them. Signed-off-by:
-
- 15 Apr, 2009 13 commits
-
-
Eric Dumazet authored
Latest tcpdump/libpcap triggers annoying messages because of high order page allocation failures (when lowmem exhausted or fragmented) These allocation errors are correctly handled so could be silent. [22660.208901] tcpdump: page allocation failure. order:5, mode:0xc0d0 [22660.208921] Pid: 13866, comm: tcpdump Not tainted 2.6.30-rc2 #170 [22660.208936] Call Trace: [22660.208950] [<c04e2b46>] ? printk+0x18/0x1a [22660.208965] [<c02760f7>] __alloc_pages_internal+0x357/0x460 [22660.208980] [<c0276251>] __get_free_pages+0x21/0x40 [22660.208995] [<c04cc835>] packet_set_ring+0x105/0x3d0 [22660.209009] [<c04ccd1d>] packet_setsockopt+0x21d/0x4d0 [22660.209025] [<c0270400>] ? filemap_fault+0x0/0x450 [22660.209040] [<c0449e34>] sys_setsockopt+0x54/0xa0 [22660.209053] [<c044b97f>] sys_socketcall+0xef/0x270 [22660.209067] [<c0202e34>] sysenter_do_call+0x12/0x26 Signed-off-by:
Eric Dumazet <dada1@cosmosbay.com> Signed-off-by:
-
Markus Brunner authored
After a transmit timed out, the reset task will be called, which will free the allocated resources(stop_gfar). If gfar_poll will be called before the resources get allocated again gfar_clean_tx_ring will call dev_kfree_skb_any(NULL). Example crash: ops: Kernel access of bad area, sig: 11 [#1] PREEMPT RSBBA100 Modules linked in: NIP: c01a10c4 LR: c013b254 CTR: c013c038 REGS: c02e7d20 TRAP: 0300 Not tainted (2.6.27.20) MSR: 00001032 <ME,IR,DR> CR: 24000082 XER: 20000000 DAR: 000000a0, DSISR: 20000000 TASK = c02ce578[0] 'swapper' THREAD: c02e6000 GPR00: 000000a0 c02e7dd0 c02ce578 00000000 00000040 00000001 c02ec1c0 00001032 GPR08: c080d1e0 df9ea800 00000000 00000000 24000082 ffffffff 0404f000 00000000 GPR16: ffffffbf ffffffff ffffffff ffdff7ff ffffffff c02d0fd4 00100100 00200200 GPR24: c031220c 00000001 00000001 00000000 00000000 df849800 ff109000 df849b80 NIP [c01a10c4] dev_kfree_skb_irq+0x18/0x70 LR [c013b254] gfar_clean_tx_ring+0x70/0x11c Call Trace: [c02e7dd0] [c003e978] update_wall_time+0x730/0x744 (unreliable) [c02e7df0] [c013b254] gfar_clean_tx_ring+0x70/0x11c [c02e7e10] [c013c07c] gfar_poll+0x44/0x150 [c02e7e30] [c01a064c] net_rx_action+0xa8/0x19c [c02e7e70] [c00251d4] __do_softirq+0x64/0xc0 [c02e7e90] [c0006384] do_softirq+0x40/0x58 [c02e7ea0] [c00250a8] irq_exit+0x40/0x9c [c02e7eb0] [c000642c] do_IRQ+0x90/0xac [c02e7ec0] [c0010ab4] ret_from_except+0x0/0x14 --- Exception: 501 at cpu_idle+0x9c/0xf8 LR = cpu_idle+0x9c/0xf8 [c02e7f80] [c0009820] cpu_idle+0x58/0xf8 (unreliable) [c02e7fa0] [c01fb8c8] __got2_end+0x7c/0x90 [c02e7fc0] [c026c794] start_kernel+0x2c0/0x2d4 [c02e7ff0] [00003438] 0x3438 Instruction dump: 7fa00124 80010024 bba10014 38210020 7c0803a6 4e800020 9421ffe0 7c0802a6 7c6b1b78 90010024 380300a0 bfa10014 <7d200028> 3129ffff 7d20012d 40a2fff4 Kernel panic - not syncing: Fatal exception in interrupt This Patch calls netif_stop_queue before calling stop_gfar. Signed-off-by:Markus Brunner <super.firetwister@gmail.com> Signed-off-by:
-
Stanislaw Gruszka authored
Add LRO alignment initially committed in 621544eb ("[LRO]: fix lro_gen_skb() alignment") and removed in 0dcffac1 ("myri10ge: add multislices support") during conversion to multi-slice. Signed-off-by:
Stanislaw Gruszka <sgruszka@redhat.com> Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
Alexander Beregalov <a.beregalov@gmail.com> Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Wu Fengguang authored
Move the update of real_num_tx_queues from ixgbe_acquire_msix_vectors() to ixgbe_set_num_queues(), to ensure it be always in sync with num_tx_queues. Signed-off-by:
Wu Fengguang <fengguang.wu@intel.com> Signed-off-by:
-
Wu Fengguang authored
Don't do the num_tx_queues based masking on calculating tx queue index. 1) num_tx_queues is not always power-of-2, because it also depends on the online cpu numbers. So the masking could be a performance bug on a 6 cpu system. 2) queue_mapping will be limited by real_num_tx_queues=num_tx_queues in the generic netdev function set_cur_queue_map(). So the bound limiting here is not necessary. Signed-off-by: -
David S. Miller authored
This reverts commit 244f46ae. Alan Cox did the research, and just like the other radio protocols zero-length frames have meaning because at the top level ROSE is X.25 PLP. So this zero-length filtering is invalid. Signed-off-by:
-
Ben Hutchings authored
falcon_sim_phy_event() used EFX_OWORD_FIELD, which operates on bitfields in 128-bit values, on an event, which is a 64-bit value. This should be harmless - these macros always use little-endian ordering, so it would read and write back the following 8 bytes unchanged - but it is obviously wrong. Signed-off-by:
Ben Hutchings <bhutchings@solarflare.com> Signed-off-by:
-
Ben Hutchings authored
sfc could call netif_napi_add() multiple times for the same napi_struct, corrupting the list of napi_structs for the associated device and leading to a busy-loop on device removal. Move the call to netif_napi_add() and add a call to netif_napi_del() in the obvious places. Signed-off-by:
-
- 14 Apr, 2009 9 commits
-
-
Jay Vosburgh authored
Remove debug printk I accidently left in as part of commit: commit 6146b1a4 Author: Jay Vosburgh <fubar@us.ibm.com> Date: Tue Nov 4 17:51:15 2008 -0800 bonding: Fix ALB mode to balance traffic on VLANs Reported by Duncan Gibb <duncan.gibb@siriusit.co.uk> Signed-off-by:
Jay Vosburgh <fubar@us.ibm.com> Signed-off-by:
-
Jesse Brandeburg authored
e1000/e1000e compile report a possible unused variable, fix that for now. Shortly after this a small refactor and bug fix will follow in the same code. Signed-off-by:
Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by:
-
Alexander Beregalov authored
Reported-by:
Subrata Modak <subrata@linux.vnet.ibm.com> Signed-off-by:
-
Michal Schmidt authored
The BUG_ON(skge->tx_ring.to_use != skge->tx_ring.to_clean) in skge_up() was sometimes observed when setting MTU. skge_down() disables the TX queue, but then reenables it by mistake via skge_tx_clean(). Fix it by moving the waking of the queue from skge_tx_clean() to the other caller. And to make sure start_xmit is not in progress on another CPU, skge_down() should call netif_tx_disable(). The bug was reported to me by Jiri Jilek whose Debian system sometimes failed to boot. He tested the patch and the bug did not happen anymore. Signed-off-by:
Michal Schmidt <mschmidt@redhat.com> Acked-by:
Stephen Hemminger <shemminger@vyatta.com> Signed-off-by:
-
Herbert Xu authored
Since everybody has been focusing on baremetal GRO performance no one noticed when I added a bug that zapped gso_size for all GRO packets. This only gets picked up when you forward the skb out of an interface. Thanks to Mark Wagner for noticing this bug when testing kvm. Reported-by:
Mark Wagner <mwagner@redhat.com> Signed-off-by:
Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Yang Hongyang authored
After switch (rthdr->type) {...},the check below is completely useless.Because: if the type is 2,then hdrlen must be 2 and segments_left must be 1,clearly the check is redundant;if the type is not 2,then goto sticky_done,the check is useless too. Signed-off-by:Yang Hongyang <yanghy@cn.fujitsu.com> Reviewed-by:
Shan Wei <shanwei@cn.fujitsu.com> Signed-off-by:
-
Herbert Xu authored
When I made the tun driver use non-linear packets as the preferred option, it broke non-GSO users because they would end up allocating a completely non-linear packet, which triggers a crash when we call eth_type_trans. This patch reverts non-GSO users to using linear packets and adds a check to ensure that GSO users can't cause crashes in the same way. Signed-off-by:
-
Ilpo Järvinen authored
A long-standing feature in tcp_init_metrics() is such that any of its goto reset prevents call to tcp_init_cwnd(). Signed-off-by:
Ilpo Järvinen <ilpo.jarvinen@helsinki.fi> Signed-off-by:
-
Stephen Hemminger authored
When vlan acceleration is used on receive, the vlan tag is maintained outside of the skb data. The existing vlan tag match only works on TX path because it uses vlan_get_tag which tests for VLAN_HW_TX_ACCEL. Signed-off-by:
-
- 13 Apr, 2009 11 commits
-
-
Herbert Xu authored
Hi: gro: Normalise skb before bypassing GRO on netpoll VLAN path When we detect netpoll RX on the GRO VLAN path we bail out and call the normal VLAN receive handler. However, the packet needs to be normalised by calling eth_type_trans since that's what the normal path expects (normally the GRO path does the fixup). This patch adds the necessary call to vlan_gro_frags. Signed-off-by:
-
Ajit Khaparde authored
This patch fixes the default value of pause auto-negotiation supported by PCS. Signed-off-by:
Ajit Khaparde <ajitk@serverengines.com> Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
Alexander Beregalov authored
Signed-off-by:
-
