This change was applied to 2.4, but not to 2.5.

Currently, incoming packets are reassembled in a sk_buff which is then
sent to the upper layers.  The sk_buff needs to be big enough to hold the
worst case packet since the size cannot be known in advance.  This would
not be so bad if the ATM layer paid attention to the mtu (usually 1500 bytes),
but for some reason it blithely ignores it and typically passes 9188 bytes
as the maximum size.  This means that at best 5/6 of the space in every
sk_buff is wasted.  So instead let's just allocate an assembly buffer (sarb)
which gets reused for every packet, and copy each assembled packet into
a minimally sized sk_buff before sending.

Reduce constant string space by reusing __FILE__

I saw that you also reported problems with USB/uhci-hcd on your systems. Can you test
the following patch and see if it works now?

This patch for 2.6 fixes freecom.c -- an error in return-code
interpretation was introduced somewhere during 2.5.

This also adds a few more checks to try to catch commands that appear to
want to move data, but of length 0.  This is likely a bug in some higher
layer -- I'll bring it up on linux-scsi.

into kroah.com:/home/linux/BK/gregkh-2.6

into home.osdl.org:/home/torvalds/v2.5/linux

into nuts.ninka.net:/disk1/davem/BK/net-2.5

The patch below adds the ebt_limit match, which is the equivalent of
the iptables limit match. This enables limiting the number of frames
processed per time unit.

The patch is from Tom Marshall <tommy_at_home.tig-grr.com>.

them instead of direct references to low2highfoo.

Parts of this come from a janitor named Domen, unfortunately,
I forgot his e-mail... Anyhow, it's obvious that janitors
did not try to compile result of their efforts, this version
at least builds.

into home.osdl.org:/home/torvalds/v2.5/linux

- common umc8672_init() for built-in and module
- release hwif only if hwif->chipset == ide_umc8672
- mark exit functions with __exit
- do not use ide_hwifs[] directly
- minor cleanups

- common qd65xx_init() for built-in and module
- release hwif only if hwif->chipset == ide_qd65xx
- mark exit functions with __exit
- do not use ide_hwifs[] directly
- minor cleanups

- common ht6560b_init() for built-in and module
- release hwif only if hwif->chipset == ide_ht6560b
- when releasing hwif, restore hwif->channel to the default value
- mark exit functions with __exit
- do not use ide_hwifs[] directly
- minor cleanups

- common dtc2278_init() for built-in and module
- touch hwifs only if hwif->chipset == ide_unknown for both ports,
  so we don't thrash already used hwifs when loading module
- release hwif only if hwif->chipset == ide_dtc2278
- mark exit functions with __exit
- do not use ide_hwifs[] directly
- minor cleanups

- common ali14xx_init() for built-in and module
- do not call findPort() twice
- touch hwifs only if chipset was found and initialized
- release hwif only if hwif->chipset == ide_ali14xx
- when releasing hwif, restore hwif->channel to the default value
- mark exit functions with __exit
- do not use ide_hwifs[] directly
- minor cleanups

Blind strncpy()->strlcpy() conversion in 2.5.70
made it impossible to change any setting, revert it.

into home.osdl.org:/home/torvalds/v2.5/linux

Finally admit defeat and change the maintainers entry for 8250/16x50
serial.

This removes "do_machine_check()" in favour of just calling the
appropriate machine check handler directly from the lowlevel exception
code.