- 10 Mar, 2005 40 commits
-
-
Patrick McHardy authored
This check is wrong, gcc optimizes it away: if ((len -= sizeof(pi)) > len) return -EINVAL; This could be responsible for the BUG. If len is 2 or 3 and TUN_NO_PI isn't set it underflows. alloc_skb() allocates len + 2, which is 0 or 1 byte. skb_reserve tries to reserve 2 bytes and things explode in skb_put. Signed-off-by:Patrick McHardy <kaber@trash.net> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
David S. Miller authored
Otherwise declarations for disable_irq()/enable_irq() will not be obtained on platforms not doing CONFIG_GENERIC_HARDIRQS. Signed-off-by: -
Robert Olsson authored
Signed-off-by: -
Patrick McHardy authored
Cacheing the current position reduces complexity from O(n^2) to O(n). Signed-off-by:
-
Stephen Hemminger authored
Low level optimization of the comparison with zero address. On most cpu's faster to use unrolled or rather than a loop and memcmp. This is in the fastpath of bridge forwarding. Signed-off-by:
Stephen Hemminger <shemminger@osdl.org> Signed-off-by:
-
Stephen Hemminger authored
Optimize bridge forwarding table for the fastpath of updating an existing entry. Use RCU to find the entry and change it's time. Fallback to normal locking for insert. This gives about a 1/3 improvement in packets forwarding per second on SMP. Signed-off-by:
-
Stephen Hemminger authored
For 2.6, I changed the bridge forwarding table timeout code to keep a threaded list of in order entries. Well, it turns out that this is a performance hit because we end up constantly moving entries around in the list. Later patch changes this to be in place update with RCU. This version just uses a 100ms garbage collection timer. Signed-off-by:
-
Stephen Hemminger authored
Get rid of unneeded include. Signed-off-by:
-
Stephen Hemminger authored
Replace the existing mac hash in the bridge code with the nice inline jenkins hash. This should provide better distribution across hash buckets and compiles to code that is similar in complexity. Signed-off-by:
-
Hideaki Yoshifuji authored
Signed-off-by:
Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org> Signed-off-by:
-
Stephen Hemminger authored
The tc_u32_mark structure is used as part of the netlink message from the user API to the kernel, so it needs to be moved to include/linux/pkt_cls.h and have types changed from u32 to __u32. Also, the definition of u32 performance counters doesn't need to depend on the config option. The definition can exist even if the code isn't enabled. Signed-off-by:
-
Stephen Hemminger authored
Trivial reordering of netdevice structure to save four bytes. Signed-off-by:
-
Patrick McHardy authored
Signed-off-by: -
Neil Horman authored
Signed-off-by:
Neil Horman <nhorman@redhat.com> Signed-off-by:
-
Domen Puncer authored
Signed-off-by:
Scott Feldman <sfeldma@pobox.com> Signed-off-by:
Maximilian Attems <janitor@sternwelten.at> Signed-off-by:
Domen Puncer <domen@coderock.org> Signed-off-by:
-
Patrick McHardy authored
Signed-off-by:
-
Jeff Garzik authored
Never called AFAICS, and only assigned -- to a no-op stub -- in one driver. Signed-off-by:
Jeff Garzik <jgarzik@pobox.com> Signed-off-by:
-
Herbert Xu authored
As promised here is the patch that moves dst_release out of dst->ops->check. It bloats sk_dst_check/__sk_dst_check slightly but they're only used in a handful of places so it isn't too bad. I actually counted, it's about a few hundred bytes. Signed-off-by:
Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by:
-
Patrick McHardy authored
Tunnel mode packets are rerouted if the tunnel destination address is different from the original destination address, otherwise the old route is used. This is inconsistent, the old route might have been selected for a given output device or using routing by tos/fwmark. Always choose a new route in tunnel mode. Signed-off-by:
-
Jean Tourrilhes authored
<Patch from Adrian Bunk> make needlessly global code static remove unneeded EXPORT_SYMBOL's from irport.c Signed-off-by:
Adrian Bunk <bunk@stusta.de> Signed-off-by:
Jean Tourrilhes <jt@hpl.hp.com> Signed-off-by:
-
Jean Tourrilhes authored
cleanup some construct obsoleted by Linus's patch Signed-off-by:
-
Jean Tourrilhes authored
<Original patch from Steffen Pingel> support NSC PC8738x chipset (IBM x40 & ...) Signed-off-by:
-
Jean Tourrilhes authored
<Original patch from Iavor Fetvadjie> allow IrDA socket to connect on arbitrary LSAPs Signed-off-by:
-
Jean Tourrilhes authored
<Patch from Randy Dunlap> Mark exit code properly in VIA driver Signed-off-by:
Randy Dunlap <rddunlap@osdl.org> Signed-off-by:
-
Jean Tourrilhes authored
<Patch from John K. Luebs> Proper turnaround computations in the stir4200 driver Take care of Tx packet without IrDA metadata (speed) Signed-off-by:
John K. Luebs <jkluebs@lu...> Signed-off-by:
-
Jean Tourrilhes authored
Forgot to convert a few usb_unlink_urb() in usb_kill_urb() <Patch from John K. Luebs> Proper sysfs support Signed-off-by:
-
Jean Tourrilhes authored
poll would improperly exit when the discovery log was empty Signed-off-by:
-
David S. Miller authored
into sunset.davemloft.net:/home/davem/src/BK/net-2.6
-
Matthew Wilcox authored
Use CONFIG_64BIT everywhere instead. Signed-off-by:
Matthew Wilcox <willy@parisc-linux.org> Signed-off-by:
Linus Torvalds <torvalds@osdl.org>
-
Matthew Wilcox authored
Indentation and header file ordering Signed-off-by:
-
Matthew Wilcox authored
The EISA EEPROM driver was being initialised too early, before miscdevs can be registered. But EISA needs to be registered early -- before PCI. So stash the EEPROM address in a global variable at EISA init time, then pick it up later in the EISA EEPROM driver. Also ioremap() it, use readb instead of gsc_readb and rename the miscdev to not have a space in the name. Signed-off-by:
-
Matthew Wilcox authored
There can be only one Asp in the machine, so allocate its controller structure statically. Signed-off-by:
-
Matthew Wilcox authored
Ditch ccio_mem_ratio. "Ported" from similar change to sba driver. From: Joel Soete <soete.joel@tiscali.be> Signed-off-by:
Grant Grundler <grundler@parisc-linux.org> Signed-off-by:
-
Matthew Wilcox authored
- PCI_DMA_BUS_IS_PHYS is only true if we don't have an IOMMU. Getting this wrong results in system panics when we run out of bounce buffers. - parisc_vmerge_boundary and parisc_vmerge_max_size need to be available on all machines, not just ones with GSC. Signed-off-by:
-
Matthew Wilcox authored
_dbg_write_trylock() needs to return an int. Also update Thibaut's email address Signed-off-by:
-
Matthew Wilcox authored
Make access_ok an inline function to get rid of an unused variable warning Signed-off-by:
-
Matthew Wilcox authored
- Match other architectures by initialising preempt_count to 1. - PREEMPT_ACTIVE needs to not overlap with HARDIRQ_MASK Signed-off-by:
-
Matthew Wilcox authored
The extraneous semicolon in mb() breaks the SMP build in the new skbuff.h From: Grant Grundler <grundler@parisc-linux.org> Signed-off-by:
-
Matthew Wilcox authored
The result from __raw_readX() can never be const so use __fswabXX() instead of cpu_to_leXX() Signed-off-by:
-
Matthew Wilcox authored
We no longer need to define our own HARDIRQ_BITS Signed-off-by:
-
