Drivers do this to try to break out of the ->poll()'ing loop
when the device is being brought administratively down.

Now that we have a napi_disable() "pending" state we are going
to solve that problem generically.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch fixes the parsing of the RX data header channel field.

The current code parses the header incorrectly and passes a wrong
channel number and frequency for each frame to mac80211.
The FIXMEs added by this patch don't matter for now as the code
where they live won't get executed anyway. They will be fixed later.
Signed-off-by: NMichael Buesch <mb@bu3sch.de>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This should fix the kernel warn/oops reported while routing.

The tulip driver has a fencepost bug with new NAPI in 2.6.24
It has an off by one bug if a full quantum is reached.
Signed-off-by: NStephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

meth didn't set a valid mac address during probing, but later during
open. Newer kernel refuse to open device with 00:00:00:00:00:00 as mac
address -> dead ethernet. This patch sets the mac address in the probe
function and uses only the mac address from the netdevice struct when
setting up the hardware.
Signed-off-by: NThomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This was missed when commit e2ac455a
fixed the compile errors in drivers/net/netx-eth.c caused by
commit 09f75cd7.
Signed-off-by: NAdrian Bunk <adrian.bunk@movial.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It is possible for the TX ring to have packets sit in it for unbounded
amounts of time.

The only way to defer TX interrupts in the chip is to periodically set
"mark" bits, when processing of a TX descriptor with the mark bit set
is complete it triggers the interrupt for the TX queue's LDG.

A consequence of this kind of scheme is that if packet flow suddenly
stops, the remaining TX packets will just sit there.

If this happens, since those packets could be charged to TCP socket
send queues, such sockets could get stuck.

The simplest solution is to divorce the socket ownership of the packet
once the device takes the SKB, by using skb_orphan() in
niu_start_xmit().

In hindsight, it would have been much nicer if the chip provided two
interrupt sources for TX (like basically every other ethernet chip
does).  Namely, keep the "mark" bit, but also signal the LDG when the
TX queue becomes completely empty.  That way there is no need to have
a deadlock breaker like this.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Noticed by Paul Lodridge.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

niu_slowpath_interrupt() expects values to be setup in lp->{v0,v1,v2}
but they aren't.  That's only done by niu_schedule_napi() which is
done later in the interrupt path.

If niu_rx_error() returns zero, and v0 is clear, hit the
RX_DMA_CTL_STATE register with a RX_DMA_CTL_STAT_MEX.

Only emit verbose RX error logs if a fatal channel or port error is
signalled.  Other cases will be recorded into statistics by
niu_log_rxchan_errors().
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

1) close should do napi_disable() not napi_enable
2) remove unused local var 'todo'
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

skb->truesize was not being incremented at all to
reflect the page based data added to RX SKBs.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts changeset fa4f0774
([CASSINI]: dont touch page_count) because it breaks the driver.

The local page counting added by this changeset did not account
for the asynchronous page count changes done by kfree_skb()
and friends.

The change adds extra atomics and on top of it all appears to be
totally unnecessary as well.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NNick Piggin <npiggin@suse.de>

Here's proposed fix for RX checksum handling in cassini; it affects
little-endian working with half-duplex gigabit, but obviously needs
testing on big-endian too.

The problem is, we need to convert checksum to fixed-endian *before*
correcting for (unstripped) FCS.  On big-endian it won't matter
(conversion is no-op), on little-endian it will, but only if FCS is
not stripped by hardware; i.e. in half-duplex gigabit mode when
->crc_size is set.

cassini.c part is that fix, cassini.h one consists of trivial
endianness annotations.  With that applied the sucker is endian-clean,
according to sparse.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Move veth.h from net/ to linux/ since it is a user api, and add it to
user header processing Kbuild.

[ Use header-y as suggested by Sam Ravnborg.  -DaveM ]
Signed-off-by: NStephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is a trivial fix of debug message.
When a persist flag is set, the message should say "enabled".
Signed-off-by: NToyo Abe <tabe@miraclelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Initially transmit buffer pointers were only reset. But buffer
descriptors were possibly still set as ready, and buffer in upper
layer was not freed. This caused driver hang under big load.  Now
reset clean properly the buffer descriptor and freed upper layer.
Signed-off-by: NGregory CLEMENT <gclement00@gmail.com>
Signed-off-by: NHaavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Same story as with olympic - htons(readw()) when swab16(readw()) is needed,
missing conversions to le32 when dealing with shared descriptors, etc.
Olympic got those fixes in 2.4.0-test2, 3c359 didn't.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

If you need to find a difference between addresses of two
struct members, subtract offsetof() or cast addresses to
char * and subtract those if you prefer it that way.  Doing
that same with s/char */u32/, OTOH...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Both store MAC address in CIS; there's no decoder for that
type (0x88) so the drivers work with raw data.  It is
byteswapped, so ntohs() works for little-endian, but for
big-endian it's wrong.  ntohs(le16_to_cpu()) does the
right thing on both (and always expands to swab16()).
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

missing conversions in a couple of places
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

* shift before cpu_to_le64(), not after it
* writel() converts to l-e itself
* misc missing conversions
* in set_multicast() hash_table[] is host-endian; we feed it to card
  via writel() and populate it as host-endian, so we'd better put the
  first element into it also in host-endian
* pci_unmap_single() et.al. expect host-endian, not little-endian
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

pci_unmap_single() and friends getting a little-endian address...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

* usb_control_message() to/from stack (breaks on e.g. arm); some
  places did kmalloc() for buffer, some just worked from stack.
  Added kmalloc()/memcpy()/kfree() in asix_read_cmd()/asix_write_cmd(),
  removed that crap from callers.
* Fixed a leak in ax88172_bind() - on success it forgot to kfree() the
  buffer.
* Endianness bug in ax88178_bind() - we read a word from eeprom and work with
  it without converting to host-endian
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

skb->protocol is net-endian, TYVM...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

txlo_dma_addr should be host-endian; we pass it to typhoon_tso_fill(),
which does arithmetics on it, converts to l-e and passes it to card.
Unfortunately, we forgot le32_to_cpu() when initializing it from
face->txLoAddr, which sits in shared memory and is little-endian.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

rxBuffCleared is little-endian; we miss le32_to_cpu() in checks for
rx ring overruns.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

One cpu_to_le16() too many when passing argument for TYPHOON_CMD_XCVR_SELECT;
we end up passing host-endian while the hardware expects little-endian.  The
other place doing that (typhoon_start_runtime()) does the right thing, so the
card will recover at the next ifconfig up/tx timeout/resume, which limits the
amount of mess, but still, WTF?
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

in typhoon_get_drvinfo() .parm2 is little-endian; not critical
since we just get the firmware id flipped in get_drvinfo output
on big-endian boxen, but...
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

txBytes and rxBytesGood are both 64bit; using le32_to_cpu() won't work
on big-endian for obvious reasons.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

if log_len is larger than 4K then we are killing the stack.
allocate on heap instead and limit size to what practically can
be used (PAGE_SIZE)
Signed-off-by: NReinette Chatre <reinette.chatre@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

This patch moves _cancel_deferred_work out of mutex protection and removes
unnecessary mutex in pci_suspend and pci_resume.

Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: NZhu Yi <yi.zhu@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Some people would like to know what p54 is.
Signed-off-by: NMichael Wu <flamingice@sourmilk.net>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Thanks to Matthias Mueller for reporting this device.
Signed-off-by: NMichael Wu <flamingice@sourmilk.net>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

tg3_nvram_write_block_unbuffered() is reading data from nvram into
allocated buffer before overwriting a part of it with user-supplied
data.  Then it feeds the entire page back to nvram.  It should be
storing the words it had read as little-endian, not as host-endian.
Note that tg3_set_eeprom() does exactly that for padding the same
data to full words before it gets passed down to tg3_nvram_write_block()
and then to tg3_nvram_write_block_unbuffered().

Moreover, when we get to sending the entire thing back to nvram, we
go through it word-by-word, doing essentially
	writel(swab32(le32_to_cpu(word)), ...)
so if we want them to reach the card in host-independent endianness,
we'd better really have all that buffer filled with fixed-endian.
For user-supplied part we obviously do have that (it's an array of
octets memcpy'd in), ditto for padding of user-supplied part to word
boundaries (taken care of in tg3_set_eeprom()).  The rest of the
buffer gets filled by tg3_nvram_write_block_unbuffered() and it would
damn better be consistent with that (and with tg3_get_eeprom(), while
we are at it - there we also convert the words read from nvram to
little-endian before returning the buffer to user).

The bug should get triggered on big-endian boxen when set_eeprom is done
for less than entire page.  Then the words that should've been unaffected
at all will actually get byteswapped in place in nvram.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixed misannotations, introduced a new helper - tg3_nvram_read_le().
It gets __le32 * instead of u32 * and puts there the value converted
to little-endian.  A lot of callers of tg3_nvram_read() were doing
that; converted them to tg3_nvram_read_le().

At that point the driver is practically endian-clean; the only remaining
place is an actual bug, AFAICS; will be dealt with in the next patch.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The patch fixes the STATUS_RF_KILL_HW state is not cleared problem if the
device goes to suspend when the rf_kill switch is enabled. The bug causes
the driver always thinks the rf_kill switch is enabled (although it is
disabled) after resume.
Signed-off-by: NZhu Yi <yi.zhu@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>