remove unneeded includes; ok mpi@

Fix up flags passed to usbd_open_pipe_intr() - interrupt pipes are always
opened exclusively, and the flags argument is passed to usbd_setup_xfer().
USBD_EXCLUSIVE_USE has the same value as USBD_NO_COPY, which means that
data transferred would not be copied to the buffer where the driver expects
it.

None of these drivers actually used the interrupt data for anything, and
in some the interrupt pipe code was #if 0'd out, so this doesn't change
anything, but fixing this up prevents unwary developers like me from
copying code that doesn't work.

ok mglocker@

Nuke all occurrences of usbd_abort_pipe() if it gets called right
before usbd_close_pipe(), since usbd_close_pipe() already takes care
about aborting non-empty pipes.

As investigated by gerhard@ usbdi.c rev. 1.57 did add usbd_abort_pipe()
to usbd_close_pipe(), but the drivers didn't get cleaned up afterwards.

ok gerhard@

Change users of IFQ_SET_MAXLEN() and IFQ_IS_EMPTY() to use the "new" API.

ok dlg@ tobhe@

use the UE_GET_XFERTYPE macro where applicable

ok mpi@

remove duplicate device ids

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

Use LIST_FOREACH construct instead of homebrew loop.

ok bluhm@, mpi@

move counting if_opackets next to counting if_obytes in if_enqueue.

this means packets are consistently counted in one place, unlike the
many and various ways that drivers thought they should do it.

ok mpi@ deraadt@

Avoid calling usbd_set_config_no() in *_attach() and let the stack do
it instead.

If anything bad happen due to a malformed descriptor it makes no sense
to try to attach a driver, and bail before probing.

This is similar to the change to avoid calling usbd_set_config_index().

G/C IFQ_SET_READY().

replace IFF_OACTIVE manipulation with mpsafe operations.

there are two things shared between the network stack and drivers
in the send path: the send queue and the IFF_OACTIVE flag. the send
queue is now protected by a mutex. this diff makes the oactive
functionality mpsafe too.

IFF_OACTIVE is part of if_flags. there are two problems with that.
firstly, if_flags is a short and we dont have any MI atomic operations
to manipulate a short. secondly, while we could make the IFF_OACTIVE
operates mpsafe, all changes to other flags would have to be made
safe at the same time, otherwise a read-modify-write cycle on their
updates could clobber the oactive change.

instead, this moves the oactive mark into struct ifqueue and provides
an API for changing it. there's ifq_set_oactive, ifq_clr_oactive,
and ifq_is_oactive. these are modelled on ifsq_set_oactive,
ifsq_clr_oactive, and ifsq_is_oactive in dragonflybsd.

this diff includes changes to all the drivers manipulating IFF_OACTIVE
to now use the ifsq_{set,clr_is}_oactive API too.

ok kettenis@ mpi@ jmatthew@ deraadt@

You only need <net/if_dl.h> if you're using LLADDR() or a sockaddr_dl.

shuffle struct ifqueue so in flight mbufs are protected by a mutex.

the code is refactored so the IFQ macros call newly implemented ifq
functions. the ifq code is split so each discipline (priq and hfsc
in our case) is an opaque set of operations that the common ifq
code can call. the common code does the locking, accounting (ifq_len
manipulation), and freeing of the mbuf if the disciplines enqueue
function rejects it. theyre kind of like bufqs in the block layer
with their fifo and nscan disciplines.

the new api also supports atomic switching of disciplines at runtime.
the hfsc setup in pf_ioctl.c has been tweaked to build a complete
hfsc_if structure which it attaches to the send queue in a single
operation, rather than attaching to the interface up front and
building up a list of queues.

the send queue is now mutexed, which raises the expectation that
packets can be enqueued or purged on one cpu while another cpu is
dequeueing them in a driver for transmission. a lot of drivers use
IFQ_POLL to peek at an mbuf and attempt to fit it on the ring before
committing to it with a later IFQ_DEQUEUE operation. if the mbuf
gets freed in between the POLL and DEQUEUE operations, fireworks
will ensue.

to avoid this, the ifq api introduces ifq_deq_begin, ifq_deq_rollback,
and ifq_deq_commit. ifq_deq_begin allows a driver to take the ifq
mutex and get a reference to the mbuf they wish to try and tx. if
there's space, they can ifq_deq_commit it to remove the mbuf and
release the mutex. if there's no space, ifq_deq_rollback simply
releases the mutex. this api was developed to make updating the
drivers using IFQ_POLL easy, instead of having to do significant
semantic changes to avoid POLL that we cannot test on all the
hardware.

the common code has been tested pretty hard, and all the driver
modifications are straightforward except for de(4). if that breaks
it can be dealt with later.

ok mpi@ jmatthew@

arp_ifinit() is no longer needed.

Increment if_ipackets in if_input().

Note that pseudo-drivers not using if_input() are not affected by this
conversion.

ok mikeb@, kettenis@, claudio@, dlg@

simply do if_input() between splnet() and splx(), like other drivers
(such as if_udav.c).

ok by mpi@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Convert to if_input().

ok dlg@

unifdef INET

No need for <netinet/in_systm.h> here.

Kill a bunch of DVACT_DEACTIVATE handlers whose only purpose were to
set the 'dying' flag of a device. Such handlers are useless now that
usbd_detach() already set this flag.

Even if the purpose of this flag is questionnable on OpenBSD because
DVACT_DEACTIVATE acts as a pre-detach event, this is just a first small
step towards the simplification of the autoconf(9) device states.

This cleaning is now possible thanks to the work of pirofti@ to convert
all the USB drivers to properly use usbd_is_dying().

Discussed many times with deraadt@

Clean-up usbd_abort_pipe() usage.


This function never fails. So change it's return type to void and
adjust the copy-pasted callers that were checking the return.

"If it compiles, ok" mpi@

Most network drivers include netinet/in_var.h, but apparently they
don't have to. Just remove these include lines.
Compiled on amd64 i386 sparc64; OK henning@ mikeb@

Get rid of various 'typedef struct' definitions and use plain structure
definitions instead. We don't change usb.h for now to stay compatible
with userland.

Tested by mpi@ on macppc and myself on i386.

ok mpi@

proc.h is way too much header for usb to handle.

Rewrite the receive filter handling code and cleanup the ioctl bits.

ok sthen@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

a bit more ansi; ok tedu

garbage collect "usb events". without /dev/usb there is no way to access
them from userland, and nothing in the kernel uses them.

ok krw@, miod@

* replace per-driver dying and/or other state variables with use of
usbd_deactivete() and usbd_is_dying()
* use usbd_deactivate() in activate()/DEACTIVATE
* convert a few more direct checks of the associated bus' dying flag
with usbd_is_dying()

don't free network related resources if they were not allocated

check that a timeout(9) has been initialized before deleting it

ok phessler

create another kthread to run xfer abort tasks. xfer abort tasks cannot
be run from the generic task kthread, because xfers that need to be
aborted block newly queued tasks from running (i.e. the xfer to be aborted
blocks the abort of that task). as there are now three types of usb
tasks, add an argument to usb_init_task() and another member to struct
usb_task to specify the task type.

fixes boot hangs that are showing up because we now use usb tasks to
attach/detach usb devices.

Delete comments about these drivers needing powerhook support; they don't.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Eliminate the redundant bits of code for MTU and multicast handling
from the individual drivers now that ether_ioctl() handles this.

Shrinks the i386 kernels by..
RAMDISK - 2176 bytes
RAMDISKB - 1504 bytes
RAMDISKC - 736 bytes

Tested by naddy@/okan@/sthen@/brad@/todd@/jmc@ and lots of users.
Build tested on almost all archs by todd@/brad@

ok naddy@

Set the IFCAP_VLAN_MTU capabilities flag so these interfaces are allowed
to transmit full sized VLAN tagged frames.

Use ether_ioctl().

Convert timeout_add() calls using multiples of hz to timeout_add_sec()

Really just the low-hanging fruit of (hopefully) forthcoming timeout
conversions.

ok art@, krw@

Sanitize the use of timeouts:

Instead of calling timeout_set(..., NULL, NULL) in attach routines and
later
timeout_del(...)
timeout_set(..., func, arg)
timeout_add(..., time)
set the function and argument in the initial timeout_set() call and only
use timeout_add(..., time) later.

ok dlg, fgsch, krw, winiger

treat usb vendor/product names as a locator, and have usbd_print handle it,
so that it shows up before the :. as a result, all the usb devices do not
need to have name printing code anymore. all this now works and prints
nicely because usbd_probe_and_attach() is serialized. ok kettenis

Remove the definition and usage of the USB_DECLARE_DRIVER_CLASS and
USB_DECLARE_DRIVER macros.

No binary change.

ok dlg.

Remove the definition and usage of the IF_INPUT macro which was defined
as ether_input_mbuf which is itself a macro for ether_input.

No binary change.

ok dlg.

Remove the definition and use of the device_ptr_t which was a struct device *.

No binary change.

ok mk.

Remove the definition and usage of the USBGETSOFTC macro, which was really
only a cast to (void *).

ok jsg.

Remove the definition and use of the USBDEVNAME macro.

(This might look easy, but it was a big diff. Thanks to dlg and especially jsg
for looking over it; we found at least four mistakes in the initial diff.)

ok jsg.

Remove the definition and use of the USBDEV macro. It only created confusion
and the address of it's argument: USBDEV(sc->sc_dev) yields &sc->sc_dev.

No binary changes.

ok jsg.

Remove definitions and usage of usb_callout and related macros. These macros
were used as a layer of confusion^Wabstraction around the timeout(9) API.

No binary change.

ok jsg.

Remove the definition and use of if_deactivate(). It was defined empty and
thus produced no code at all.

No binary change.

ok jsg.

Remove the "Static" declaration of many functions. It was defined to be empty
and it was not consistently used. It was confusing as it suggested these
functions were static, which they were not.

discussed with dlg and jsg, ok jsg.

Last part of FreeBSD/NetBSD sepcific code removal.

ok jsg@

Farewell USB_{ATTACH,MATCH,DETACH}* you will not be missed.
ok deraadt@ krw@ mbalmer@

Remove logprintf macro

Remove Ether_ifattach macro

More lockmgr -> rwlock low hanging fruit.

ok dlg@

Remove _KERNEL_OPT leftovers.

add ZyXEL USB id, remove useless comments and reorder USB devices.

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

allow bpf(4) to ignore packets based on their direction (inbound or
outbound), using a new BIOCSDIRFILT ioctl;
guidance, feedback and ok canacar@

Remove last NRND NetBSDisms from tree.

ok deraadt@ brad@

typos from alexey dobriyan;

Use usbd_clear_endpoint_stall_async() when clearing endpoint stalls in
an interrupt context.

From NetBSD

ok dlg@

{TrendNet,Trendware} -> TRENDnet, ok jsg@

Don't keep the devinfo string on the stack, instead use malloc/free.
This should cure some rare stack overflows.

From augustss NetBSD

ok dlg@ pascoe@

Update for GREENHOUSE2 -> REALTEK USB vid change.

clear IFF_RUNNING & IFF_OACTIVE in foo_stop() before de-allocating resources.

don't call mii_pollstat() right after mii_tick() in foo_tick_task()

remove netns crud.
some drivers actually had hooks for SIOCSIFADDR, most just useless includes
"looks good" deraadt miod brad

ethernet on OQO model 01 device cable. dlg ok

- make sure int is in running state before touching the multicast filters
- call foo_setmulti only instead of init'ing the chip
- don't overwrite potential error return with success when calling
ether_addmulti/ether_delmulti

ok dlg@

Three mode devices found in the wild, I think

strncpy->strlcpy
ok henning@

Add Longshine LCS-8138TX and Micronet SP128AR found in linux driver
ok dlg@

from netbsd: if_url.c 1.8, if_urlreg.h 1.2, urlphy.c 1.9 and urlphyreg.h 1.2
via Jonathan Gray
log message:

Remove advertising clauses from code written by Shingo WATANABE.
He has kindly consented it.

don't need to set ifp->if_mtu or ifp->if_output in each driver,
{ether,atm,fddi}_ifattach already does this.

ok mcbride@ markus@ henning@

put "do { } while (0)" wrappers on all the debug maroc functions

init ether address correctly, since our Ether_ifattach macro does not do
that job; 3612, rspmn@arcor.de

Set devclass to DV_IFNET for all USB network drivers.
ok deraadt@ nate@

string cleanup; nate ok

branches: 1.10.4;
pad minimum frames with 0's; based on netbsd

sync with NetBSD

Plug memory leak; nyar-junk@int0rnet.net

get rid of trailing whitespace

shorten address printout

Kill __FUNCTION__.

branches: 1.4.2;
Fix ioctl handling
Add default MTU
fix typo

errors noted by Jason Ackley <jason@ackley.net>

Add a driver for Realtek RTL8150L based ethernet devices
From NetBSD

Meant to commit only ulpt and committed everything. Most things weren't ready

Sync ulpt driver with NetBSD

Fix up flags passed to usbd_open_pipe_intr() - interrupt pipes are always
opened exclusively, and the flags argument is passed to usbd_setup_xfer().
USBD_EXCLUSIVE_USE has the same value as USBD_NO_COPY, which means that
data transferred would not be copied to the buffer where the driver expects
it.

None of these drivers actually used the interrupt data for anything, and
in some the interrupt pipe code was #if 0'd out, so this doesn't change
anything, but fixing this up prevents unwary developers like me from
copying code that doesn't work.

ok mglocker@

Nuke all occurrences of usbd_abort_pipe() if it gets called right
before usbd_close_pipe(), since usbd_close_pipe() already takes care
about aborting non-empty pipes.

As investigated by gerhard@ usbdi.c rev. 1.57 did add usbd_abort_pipe()
to usbd_close_pipe(), but the drivers didn't get cleaned up afterwards.

ok gerhard@

Change users of IFQ_SET_MAXLEN() and IFQ_IS_EMPTY() to use the "new" API.

ok dlg@ tobhe@

use the UE_GET_XFERTYPE macro where applicable

ok mpi@

remove duplicate device ids

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

Use LIST_FOREACH construct instead of homebrew loop.

ok bluhm@, mpi@

move counting if_opackets next to counting if_obytes in if_enqueue.

this means packets are consistently counted in one place, unlike the
many and various ways that drivers thought they should do it.

ok mpi@ deraadt@

Avoid calling usbd_set_config_no() in *_attach() and let the stack do
it instead.

If anything bad happen due to a malformed descriptor it makes no sense
to try to attach a driver, and bail before probing.

This is similar to the change to avoid calling usbd_set_config_index().

G/C IFQ_SET_READY().

replace IFF_OACTIVE manipulation with mpsafe operations.

there are two things shared between the network stack and drivers
in the send path: the send queue and the IFF_OACTIVE flag. the send
queue is now protected by a mutex. this diff makes the oactive
functionality mpsafe too.

IFF_OACTIVE is part of if_flags. there are two problems with that.
firstly, if_flags is a short and we dont have any MI atomic operations
to manipulate a short. secondly, while we could make the IFF_OACTIVE
operates mpsafe, all changes to other flags would have to be made
safe at the same time, otherwise a read-modify-write cycle on their
updates could clobber the oactive change.

instead, this moves the oactive mark into struct ifqueue and provides
an API for changing it. there's ifq_set_oactive, ifq_clr_oactive,
and ifq_is_oactive. these are modelled on ifsq_set_oactive,
ifsq_clr_oactive, and ifsq_is_oactive in dragonflybsd.

this diff includes changes to all the drivers manipulating IFF_OACTIVE
to now use the ifsq_{set,clr_is}_oactive API too.

ok kettenis@ mpi@ jmatthew@ deraadt@

You only need <net/if_dl.h> if you're using LLADDR() or a sockaddr_dl.

shuffle struct ifqueue so in flight mbufs are protected by a mutex.

the code is refactored so the IFQ macros call newly implemented ifq
functions. the ifq code is split so each discipline (priq and hfsc
in our case) is an opaque set of operations that the common ifq
code can call. the common code does the locking, accounting (ifq_len
manipulation), and freeing of the mbuf if the disciplines enqueue
function rejects it. theyre kind of like bufqs in the block layer
with their fifo and nscan disciplines.

the new api also supports atomic switching of disciplines at runtime.
the hfsc setup in pf_ioctl.c has been tweaked to build a complete
hfsc_if structure which it attaches to the send queue in a single
operation, rather than attaching to the interface up front and
building up a list of queues.

the send queue is now mutexed, which raises the expectation that
packets can be enqueued or purged on one cpu while another cpu is
dequeueing them in a driver for transmission. a lot of drivers use
IFQ_POLL to peek at an mbuf and attempt to fit it on the ring before
committing to it with a later IFQ_DEQUEUE operation. if the mbuf
gets freed in between the POLL and DEQUEUE operations, fireworks
will ensue.

to avoid this, the ifq api introduces ifq_deq_begin, ifq_deq_rollback,
and ifq_deq_commit. ifq_deq_begin allows a driver to take the ifq
mutex and get a reference to the mbuf they wish to try and tx. if
there's space, they can ifq_deq_commit it to remove the mbuf and
release the mutex. if there's no space, ifq_deq_rollback simply
releases the mutex. this api was developed to make updating the
drivers using IFQ_POLL easy, instead of having to do significant
semantic changes to avoid POLL that we cannot test on all the
hardware.

the common code has been tested pretty hard, and all the driver
modifications are straightforward except for de(4). if that breaks
it can be dealt with later.

ok mpi@ jmatthew@

arp_ifinit() is no longer needed.

Increment if_ipackets in if_input().

Note that pseudo-drivers not using if_input() are not affected by this
conversion.

ok mikeb@, kettenis@, claudio@, dlg@

simply do if_input() between splnet() and splx(), like other drivers
(such as if_udav.c).

ok by mpi@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Convert to if_input().

ok dlg@

unifdef INET

No need for <netinet/in_systm.h> here.

Kill a bunch of DVACT_DEACTIVATE handlers whose only purpose were to
set the 'dying' flag of a device. Such handlers are useless now that
usbd_detach() already set this flag.

Even if the purpose of this flag is questionnable on OpenBSD because
DVACT_DEACTIVATE acts as a pre-detach event, this is just a first small
step towards the simplification of the autoconf(9) device states.

This cleaning is now possible thanks to the work of pirofti@ to convert
all the USB drivers to properly use usbd_is_dying().

Discussed many times with deraadt@

Clean-up usbd_abort_pipe() usage.


This function never fails. So change it's return type to void and
adjust the copy-pasted callers that were checking the return.

"If it compiles, ok" mpi@

Most network drivers include netinet/in_var.h, but apparently they
don't have to. Just remove these include lines.
Compiled on amd64 i386 sparc64; OK henning@ mikeb@

Get rid of various 'typedef struct' definitions and use plain structure
definitions instead. We don't change usb.h for now to stay compatible
with userland.

Tested by mpi@ on macppc and myself on i386.

ok mpi@

proc.h is way too much header for usb to handle.

Rewrite the receive filter handling code and cleanup the ioctl bits.

ok sthen@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

a bit more ansi; ok tedu

garbage collect "usb events". without /dev/usb there is no way to access
them from userland, and nothing in the kernel uses them.

ok krw@, miod@

* replace per-driver dying and/or other state variables with use of
usbd_deactivete() and usbd_is_dying()
* use usbd_deactivate() in activate()/DEACTIVATE
* convert a few more direct checks of the associated bus' dying flag
with usbd_is_dying()

don't free network related resources if they were not allocated

check that a timeout(9) has been initialized before deleting it

ok phessler

create another kthread to run xfer abort tasks. xfer abort tasks cannot
be run from the generic task kthread, because xfers that need to be
aborted block newly queued tasks from running (i.e. the xfer to be aborted
blocks the abort of that task). as there are now three types of usb
tasks, add an argument to usb_init_task() and another member to struct
usb_task to specify the task type.

fixes boot hangs that are showing up because we now use usb tasks to
attach/detach usb devices.

Delete comments about these drivers needing powerhook support; they don't.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Eliminate the redundant bits of code for MTU and multicast handling
from the individual drivers now that ether_ioctl() handles this.

Shrinks the i386 kernels by..
RAMDISK - 2176 bytes
RAMDISKB - 1504 bytes
RAMDISKC - 736 bytes

Tested by naddy@/okan@/sthen@/brad@/todd@/jmc@ and lots of users.
Build tested on almost all archs by todd@/brad@

ok naddy@

Set the IFCAP_VLAN_MTU capabilities flag so these interfaces are allowed
to transmit full sized VLAN tagged frames.

Use ether_ioctl().

Convert timeout_add() calls using multiples of hz to timeout_add_sec()

Really just the low-hanging fruit of (hopefully) forthcoming timeout
conversions.

ok art@, krw@

Sanitize the use of timeouts:

Instead of calling timeout_set(..., NULL, NULL) in attach routines and
later
timeout_del(...)
timeout_set(..., func, arg)
timeout_add(..., time)
set the function and argument in the initial timeout_set() call and only
use timeout_add(..., time) later.

ok dlg, fgsch, krw, winiger

treat usb vendor/product names as a locator, and have usbd_print handle it,
so that it shows up before the :. as a result, all the usb devices do not
need to have name printing code anymore. all this now works and prints
nicely because usbd_probe_and_attach() is serialized. ok kettenis

Remove the definition and usage of the USB_DECLARE_DRIVER_CLASS and
USB_DECLARE_DRIVER macros.

No binary change.

ok dlg.

Remove the definition and usage of the IF_INPUT macro which was defined
as ether_input_mbuf which is itself a macro for ether_input.

No binary change.

ok dlg.

Remove the definition and use of the device_ptr_t which was a struct device *.

No binary change.

ok mk.

Remove the definition and usage of the USBGETSOFTC macro, which was really
only a cast to (void *).

ok jsg.

Remove the definition and use of the USBDEVNAME macro.

(This might look easy, but it was a big diff. Thanks to dlg and especially jsg
for looking over it; we found at least four mistakes in the initial diff.)

ok jsg.

Remove the definition and use of the USBDEV macro. It only created confusion
and the address of it's argument: USBDEV(sc->sc_dev) yields &sc->sc_dev.

No binary changes.

ok jsg.

Remove definitions and usage of usb_callout and related macros. These macros
were used as a layer of confusion^Wabstraction around the timeout(9) API.

No binary change.

ok jsg.

Remove the definition and use of if_deactivate(). It was defined empty and
thus produced no code at all.

No binary change.

ok jsg.

Remove the "Static" declaration of many functions. It was defined to be empty
and it was not consistently used. It was confusing as it suggested these
functions were static, which they were not.

discussed with dlg and jsg, ok jsg.

Last part of FreeBSD/NetBSD sepcific code removal.

ok jsg@

Farewell USB_{ATTACH,MATCH,DETACH}* you will not be missed.
ok deraadt@ krw@ mbalmer@

Remove logprintf macro

Remove Ether_ifattach macro

More lockmgr -> rwlock low hanging fruit.

ok dlg@

Remove _KERNEL_OPT leftovers.

add ZyXEL USB id, remove useless comments and reorder USB devices.

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

allow bpf(4) to ignore packets based on their direction (inbound or
outbound), using a new BIOCSDIRFILT ioctl;
guidance, feedback and ok canacar@

Remove last NRND NetBSDisms from tree.

ok deraadt@ brad@

typos from alexey dobriyan;

Use usbd_clear_endpoint_stall_async() when clearing endpoint stalls in
an interrupt context.

From NetBSD

ok dlg@

{TrendNet,Trendware} -> TRENDnet, ok jsg@

Don't keep the devinfo string on the stack, instead use malloc/free.
This should cure some rare stack overflows.

From augustss NetBSD

ok dlg@ pascoe@

Update for GREENHOUSE2 -> REALTEK USB vid change.

clear IFF_RUNNING & IFF_OACTIVE in foo_stop() before de-allocating resources.

don't call mii_pollstat() right after mii_tick() in foo_tick_task()

remove netns crud.
some drivers actually had hooks for SIOCSIFADDR, most just useless includes
"looks good" deraadt miod brad

ethernet on OQO model 01 device cable. dlg ok

- make sure int is in running state before touching the multicast filters
- call foo_setmulti only instead of init'ing the chip
- don't overwrite potential error return with success when calling
ether_addmulti/ether_delmulti

ok dlg@

Three mode devices found in the wild, I think

strncpy->strlcpy
ok henning@

Add Longshine LCS-8138TX and Micronet SP128AR found in linux driver
ok dlg@

from netbsd: if_url.c 1.8, if_urlreg.h 1.2, urlphy.c 1.9 and urlphyreg.h 1.2
via Jonathan Gray
log message:

Remove advertising clauses from code written by Shingo WATANABE.
He has kindly consented it.

don't need to set ifp->if_mtu or ifp->if_output in each driver,
{ether,atm,fddi}_ifattach already does this.

ok mcbride@ markus@ henning@

put "do { } while (0)" wrappers on all the debug maroc functions

init ether address correctly, since our Ether_ifattach macro does not do
that job; 3612, rspmn@arcor.de

Set devclass to DV_IFNET for all USB network drivers.
ok deraadt@ nate@

string cleanup; nate ok

branches: 1.10.4;
pad minimum frames with 0's; based on netbsd

sync with NetBSD

Plug memory leak; nyar-junk@int0rnet.net

get rid of trailing whitespace

shorten address printout

Kill __FUNCTION__.

branches: 1.4.2;
Fix ioctl handling
Add default MTU
fix typo

errors noted by Jason Ackley <jason@ackley.net>

Add a driver for Realtek RTL8150L based ethernet devices
From NetBSD

Meant to commit only ulpt and committed everything. Most things weren't ready

Sync ulpt driver with NetBSD

Nuke all occurrences of usbd_abort_pipe() if it gets called right
before usbd_close_pipe(), since usbd_close_pipe() already takes care
about aborting non-empty pipes.

As investigated by gerhard@ usbdi.c rev. 1.57 did add usbd_abort_pipe()
to usbd_close_pipe(), but the drivers didn't get cleaned up afterwards.

ok gerhard@

Change users of IFQ_SET_MAXLEN() and IFQ_IS_EMPTY() to use the "new" API.

ok dlg@ tobhe@

use the UE_GET_XFERTYPE macro where applicable

ok mpi@

remove duplicate device ids

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

Use LIST_FOREACH construct instead of homebrew loop.

ok bluhm@, mpi@

move counting if_opackets next to counting if_obytes in if_enqueue.

this means packets are consistently counted in one place, unlike the
many and various ways that drivers thought they should do it.

ok mpi@ deraadt@

Avoid calling usbd_set_config_no() in *_attach() and let the stack do
it instead.

If anything bad happen due to a malformed descriptor it makes no sense
to try to attach a driver, and bail before probing.

This is similar to the change to avoid calling usbd_set_config_index().

G/C IFQ_SET_READY().

replace IFF_OACTIVE manipulation with mpsafe operations.

there are two things shared between the network stack and drivers
in the send path: the send queue and the IFF_OACTIVE flag. the send
queue is now protected by a mutex. this diff makes the oactive
functionality mpsafe too.

IFF_OACTIVE is part of if_flags. there are two problems with that.
firstly, if_flags is a short and we dont have any MI atomic operations
to manipulate a short. secondly, while we could make the IFF_OACTIVE
operates mpsafe, all changes to other flags would have to be made
safe at the same time, otherwise a read-modify-write cycle on their
updates could clobber the oactive change.

instead, this moves the oactive mark into struct ifqueue and provides
an API for changing it. there's ifq_set_oactive, ifq_clr_oactive,
and ifq_is_oactive. these are modelled on ifsq_set_oactive,
ifsq_clr_oactive, and ifsq_is_oactive in dragonflybsd.

this diff includes changes to all the drivers manipulating IFF_OACTIVE
to now use the ifsq_{set,clr_is}_oactive API too.

ok kettenis@ mpi@ jmatthew@ deraadt@

You only need <net/if_dl.h> if you're using LLADDR() or a sockaddr_dl.

shuffle struct ifqueue so in flight mbufs are protected by a mutex.

the code is refactored so the IFQ macros call newly implemented ifq
functions. the ifq code is split so each discipline (priq and hfsc
in our case) is an opaque set of operations that the common ifq
code can call. the common code does the locking, accounting (ifq_len
manipulation), and freeing of the mbuf if the disciplines enqueue
function rejects it. theyre kind of like bufqs in the block layer
with their fifo and nscan disciplines.

the new api also supports atomic switching of disciplines at runtime.
the hfsc setup in pf_ioctl.c has been tweaked to build a complete
hfsc_if structure which it attaches to the send queue in a single
operation, rather than attaching to the interface up front and
building up a list of queues.

the send queue is now mutexed, which raises the expectation that
packets can be enqueued or purged on one cpu while another cpu is
dequeueing them in a driver for transmission. a lot of drivers use
IFQ_POLL to peek at an mbuf and attempt to fit it on the ring before
committing to it with a later IFQ_DEQUEUE operation. if the mbuf
gets freed in between the POLL and DEQUEUE operations, fireworks
will ensue.

to avoid this, the ifq api introduces ifq_deq_begin, ifq_deq_rollback,
and ifq_deq_commit. ifq_deq_begin allows a driver to take the ifq
mutex and get a reference to the mbuf they wish to try and tx. if
there's space, they can ifq_deq_commit it to remove the mbuf and
release the mutex. if there's no space, ifq_deq_rollback simply
releases the mutex. this api was developed to make updating the
drivers using IFQ_POLL easy, instead of having to do significant
semantic changes to avoid POLL that we cannot test on all the
hardware.

the common code has been tested pretty hard, and all the driver
modifications are straightforward except for de(4). if that breaks
it can be dealt with later.

ok mpi@ jmatthew@

arp_ifinit() is no longer needed.

Increment if_ipackets in if_input().

Note that pseudo-drivers not using if_input() are not affected by this
conversion.

ok mikeb@, kettenis@, claudio@, dlg@

simply do if_input() between splnet() and splx(), like other drivers
(such as if_udav.c).

ok by mpi@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Convert to if_input().

ok dlg@

unifdef INET

No need for <netinet/in_systm.h> here.

Kill a bunch of DVACT_DEACTIVATE handlers whose only purpose were to
set the 'dying' flag of a device. Such handlers are useless now that
usbd_detach() already set this flag.

Even if the purpose of this flag is questionnable on OpenBSD because
DVACT_DEACTIVATE acts as a pre-detach event, this is just a first small
step towards the simplification of the autoconf(9) device states.

This cleaning is now possible thanks to the work of pirofti@ to convert
all the USB drivers to properly use usbd_is_dying().

Discussed many times with deraadt@

Clean-up usbd_abort_pipe() usage.


This function never fails. So change it's return type to void and
adjust the copy-pasted callers that were checking the return.

"If it compiles, ok" mpi@

Most network drivers include netinet/in_var.h, but apparently they
don't have to. Just remove these include lines.
Compiled on amd64 i386 sparc64; OK henning@ mikeb@

Get rid of various 'typedef struct' definitions and use plain structure
definitions instead. We don't change usb.h for now to stay compatible
with userland.

Tested by mpi@ on macppc and myself on i386.

ok mpi@

proc.h is way too much header for usb to handle.

Rewrite the receive filter handling code and cleanup the ioctl bits.

ok sthen@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

a bit more ansi; ok tedu

garbage collect "usb events". without /dev/usb there is no way to access
them from userland, and nothing in the kernel uses them.

ok krw@, miod@

* replace per-driver dying and/or other state variables with use of
usbd_deactivete() and usbd_is_dying()
* use usbd_deactivate() in activate()/DEACTIVATE
* convert a few more direct checks of the associated bus' dying flag
with usbd_is_dying()

don't free network related resources if they were not allocated

check that a timeout(9) has been initialized before deleting it

ok phessler

create another kthread to run xfer abort tasks. xfer abort tasks cannot
be run from the generic task kthread, because xfers that need to be
aborted block newly queued tasks from running (i.e. the xfer to be aborted
blocks the abort of that task). as there are now three types of usb
tasks, add an argument to usb_init_task() and another member to struct
usb_task to specify the task type.

fixes boot hangs that are showing up because we now use usb tasks to
attach/detach usb devices.

Delete comments about these drivers needing powerhook support; they don't.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Eliminate the redundant bits of code for MTU and multicast handling
from the individual drivers now that ether_ioctl() handles this.

Shrinks the i386 kernels by..
RAMDISK - 2176 bytes
RAMDISKB - 1504 bytes
RAMDISKC - 736 bytes

Tested by naddy@/okan@/sthen@/brad@/todd@/jmc@ and lots of users.
Build tested on almost all archs by todd@/brad@

ok naddy@

Set the IFCAP_VLAN_MTU capabilities flag so these interfaces are allowed
to transmit full sized VLAN tagged frames.

Use ether_ioctl().

Convert timeout_add() calls using multiples of hz to timeout_add_sec()

Really just the low-hanging fruit of (hopefully) forthcoming timeout
conversions.

ok art@, krw@

Sanitize the use of timeouts:

Instead of calling timeout_set(..., NULL, NULL) in attach routines and
later
timeout_del(...)
timeout_set(..., func, arg)
timeout_add(..., time)
set the function and argument in the initial timeout_set() call and only
use timeout_add(..., time) later.

ok dlg, fgsch, krw, winiger

treat usb vendor/product names as a locator, and have usbd_print handle it,
so that it shows up before the :. as a result, all the usb devices do not
need to have name printing code anymore. all this now works and prints
nicely because usbd_probe_and_attach() is serialized. ok kettenis

Remove the definition and usage of the USB_DECLARE_DRIVER_CLASS and
USB_DECLARE_DRIVER macros.

No binary change.

ok dlg.

Remove the definition and usage of the IF_INPUT macro which was defined
as ether_input_mbuf which is itself a macro for ether_input.

No binary change.

ok dlg.

Remove the definition and use of the device_ptr_t which was a struct device *.

No binary change.

ok mk.

Remove the definition and usage of the USBGETSOFTC macro, which was really
only a cast to (void *).

ok jsg.

Remove the definition and use of the USBDEVNAME macro.

(This might look easy, but it was a big diff. Thanks to dlg and especially jsg
for looking over it; we found at least four mistakes in the initial diff.)

ok jsg.

Remove the definition and use of the USBDEV macro. It only created confusion
and the address of it's argument: USBDEV(sc->sc_dev) yields &sc->sc_dev.

No binary changes.

ok jsg.

Remove definitions and usage of usb_callout and related macros. These macros
were used as a layer of confusion^Wabstraction around the timeout(9) API.

No binary change.

ok jsg.

Remove the definition and use of if_deactivate(). It was defined empty and
thus produced no code at all.

No binary change.

ok jsg.

Remove the "Static" declaration of many functions. It was defined to be empty
and it was not consistently used. It was confusing as it suggested these
functions were static, which they were not.

discussed with dlg and jsg, ok jsg.

Last part of FreeBSD/NetBSD sepcific code removal.

ok jsg@

Farewell USB_{ATTACH,MATCH,DETACH}* you will not be missed.
ok deraadt@ krw@ mbalmer@

Remove logprintf macro

Remove Ether_ifattach macro

More lockmgr -> rwlock low hanging fruit.

ok dlg@

Remove _KERNEL_OPT leftovers.

add ZyXEL USB id, remove useless comments and reorder USB devices.

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

allow bpf(4) to ignore packets based on their direction (inbound or
outbound), using a new BIOCSDIRFILT ioctl;
guidance, feedback and ok canacar@

Remove last NRND NetBSDisms from tree.

ok deraadt@ brad@

typos from alexey dobriyan;

Use usbd_clear_endpoint_stall_async() when clearing endpoint stalls in
an interrupt context.

From NetBSD

ok dlg@

{TrendNet,Trendware} -> TRENDnet, ok jsg@

Don't keep the devinfo string on the stack, instead use malloc/free.
This should cure some rare stack overflows.

From augustss NetBSD

ok dlg@ pascoe@

Update for GREENHOUSE2 -> REALTEK USB vid change.

clear IFF_RUNNING & IFF_OACTIVE in foo_stop() before de-allocating resources.

don't call mii_pollstat() right after mii_tick() in foo_tick_task()

remove netns crud.
some drivers actually had hooks for SIOCSIFADDR, most just useless includes
"looks good" deraadt miod brad

ethernet on OQO model 01 device cable. dlg ok

- make sure int is in running state before touching the multicast filters
- call foo_setmulti only instead of init'ing the chip
- don't overwrite potential error return with success when calling
ether_addmulti/ether_delmulti

ok dlg@

Three mode devices found in the wild, I think

strncpy->strlcpy
ok henning@

Add Longshine LCS-8138TX and Micronet SP128AR found in linux driver
ok dlg@

from netbsd: if_url.c 1.8, if_urlreg.h 1.2, urlphy.c 1.9 and urlphyreg.h 1.2
via Jonathan Gray
log message:

Remove advertising clauses from code written by Shingo WATANABE.
He has kindly consented it.

don't need to set ifp->if_mtu or ifp->if_output in each driver,
{ether,atm,fddi}_ifattach already does this.

ok mcbride@ markus@ henning@

put "do { } while (0)" wrappers on all the debug maroc functions

init ether address correctly, since our Ether_ifattach macro does not do
that job; 3612, rspmn@arcor.de

Set devclass to DV_IFNET for all USB network drivers.
ok deraadt@ nate@

string cleanup; nate ok

branches: 1.10.4;
pad minimum frames with 0's; based on netbsd

sync with NetBSD

Plug memory leak; nyar-junk@int0rnet.net

get rid of trailing whitespace

shorten address printout

Kill __FUNCTION__.

branches: 1.4.2;
Fix ioctl handling
Add default MTU
fix typo

errors noted by Jason Ackley <jason@ackley.net>

Add a driver for Realtek RTL8150L based ethernet devices
From NetBSD

Meant to commit only ulpt and committed everything. Most things weren't ready

Sync ulpt driver with NetBSD

Change users of IFQ_SET_MAXLEN() and IFQ_IS_EMPTY() to use the "new" API.

ok dlg@ tobhe@

use the UE_GET_XFERTYPE macro where applicable

ok mpi@

remove duplicate device ids

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

Use LIST_FOREACH construct instead of homebrew loop.

ok bluhm@, mpi@

move counting if_opackets next to counting if_obytes in if_enqueue.

this means packets are consistently counted in one place, unlike the
many and various ways that drivers thought they should do it.

ok mpi@ deraadt@

Avoid calling usbd_set_config_no() in *_attach() and let the stack do
it instead.

If anything bad happen due to a malformed descriptor it makes no sense
to try to attach a driver, and bail before probing.

This is similar to the change to avoid calling usbd_set_config_index().

G/C IFQ_SET_READY().

replace IFF_OACTIVE manipulation with mpsafe operations.

there are two things shared between the network stack and drivers
in the send path: the send queue and the IFF_OACTIVE flag. the send
queue is now protected by a mutex. this diff makes the oactive
functionality mpsafe too.

IFF_OACTIVE is part of if_flags. there are two problems with that.
firstly, if_flags is a short and we dont have any MI atomic operations
to manipulate a short. secondly, while we could make the IFF_OACTIVE
operates mpsafe, all changes to other flags would have to be made
safe at the same time, otherwise a read-modify-write cycle on their
updates could clobber the oactive change.

instead, this moves the oactive mark into struct ifqueue and provides
an API for changing it. there's ifq_set_oactive, ifq_clr_oactive,
and ifq_is_oactive. these are modelled on ifsq_set_oactive,
ifsq_clr_oactive, and ifsq_is_oactive in dragonflybsd.

this diff includes changes to all the drivers manipulating IFF_OACTIVE
to now use the ifsq_{set,clr_is}_oactive API too.

ok kettenis@ mpi@ jmatthew@ deraadt@

You only need <net/if_dl.h> if you're using LLADDR() or a sockaddr_dl.

shuffle struct ifqueue so in flight mbufs are protected by a mutex.

the code is refactored so the IFQ macros call newly implemented ifq
functions. the ifq code is split so each discipline (priq and hfsc
in our case) is an opaque set of operations that the common ifq
code can call. the common code does the locking, accounting (ifq_len
manipulation), and freeing of the mbuf if the disciplines enqueue
function rejects it. theyre kind of like bufqs in the block layer
with their fifo and nscan disciplines.

the new api also supports atomic switching of disciplines at runtime.
the hfsc setup in pf_ioctl.c has been tweaked to build a complete
hfsc_if structure which it attaches to the send queue in a single
operation, rather than attaching to the interface up front and
building up a list of queues.

the send queue is now mutexed, which raises the expectation that
packets can be enqueued or purged on one cpu while another cpu is
dequeueing them in a driver for transmission. a lot of drivers use
IFQ_POLL to peek at an mbuf and attempt to fit it on the ring before
committing to it with a later IFQ_DEQUEUE operation. if the mbuf
gets freed in between the POLL and DEQUEUE operations, fireworks
will ensue.

to avoid this, the ifq api introduces ifq_deq_begin, ifq_deq_rollback,
and ifq_deq_commit. ifq_deq_begin allows a driver to take the ifq
mutex and get a reference to the mbuf they wish to try and tx. if
there's space, they can ifq_deq_commit it to remove the mbuf and
release the mutex. if there's no space, ifq_deq_rollback simply
releases the mutex. this api was developed to make updating the
drivers using IFQ_POLL easy, instead of having to do significant
semantic changes to avoid POLL that we cannot test on all the
hardware.

the common code has been tested pretty hard, and all the driver
modifications are straightforward except for de(4). if that breaks
it can be dealt with later.

ok mpi@ jmatthew@

arp_ifinit() is no longer needed.

Increment if_ipackets in if_input().

Note that pseudo-drivers not using if_input() are not affected by this
conversion.

ok mikeb@, kettenis@, claudio@, dlg@

simply do if_input() between splnet() and splx(), like other drivers
(such as if_udav.c).

ok by mpi@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Convert to if_input().

ok dlg@

unifdef INET

No need for <netinet/in_systm.h> here.

Kill a bunch of DVACT_DEACTIVATE handlers whose only purpose were to
set the 'dying' flag of a device. Such handlers are useless now that
usbd_detach() already set this flag.

Even if the purpose of this flag is questionnable on OpenBSD because
DVACT_DEACTIVATE acts as a pre-detach event, this is just a first small
step towards the simplification of the autoconf(9) device states.

This cleaning is now possible thanks to the work of pirofti@ to convert
all the USB drivers to properly use usbd_is_dying().

Discussed many times with deraadt@

Clean-up usbd_abort_pipe() usage.


This function never fails. So change it's return type to void and
adjust the copy-pasted callers that were checking the return.

"If it compiles, ok" mpi@

Most network drivers include netinet/in_var.h, but apparently they
don't have to. Just remove these include lines.
Compiled on amd64 i386 sparc64; OK henning@ mikeb@

Get rid of various 'typedef struct' definitions and use plain structure
definitions instead. We don't change usb.h for now to stay compatible
with userland.

Tested by mpi@ on macppc and myself on i386.

ok mpi@

proc.h is way too much header for usb to handle.

Rewrite the receive filter handling code and cleanup the ioctl bits.

ok sthen@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

a bit more ansi; ok tedu

garbage collect "usb events". without /dev/usb there is no way to access
them from userland, and nothing in the kernel uses them.

ok krw@, miod@

* replace per-driver dying and/or other state variables with use of
usbd_deactivete() and usbd_is_dying()
* use usbd_deactivate() in activate()/DEACTIVATE
* convert a few more direct checks of the associated bus' dying flag
with usbd_is_dying()

don't free network related resources if they were not allocated

check that a timeout(9) has been initialized before deleting it

ok phessler

create another kthread to run xfer abort tasks. xfer abort tasks cannot
be run from the generic task kthread, because xfers that need to be
aborted block newly queued tasks from running (i.e. the xfer to be aborted
blocks the abort of that task). as there are now three types of usb
tasks, add an argument to usb_init_task() and another member to struct
usb_task to specify the task type.

fixes boot hangs that are showing up because we now use usb tasks to
attach/detach usb devices.

Delete comments about these drivers needing powerhook support; they don't.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Eliminate the redundant bits of code for MTU and multicast handling
from the individual drivers now that ether_ioctl() handles this.

Shrinks the i386 kernels by..
RAMDISK - 2176 bytes
RAMDISKB - 1504 bytes
RAMDISKC - 736 bytes

Tested by naddy@/okan@/sthen@/brad@/todd@/jmc@ and lots of users.
Build tested on almost all archs by todd@/brad@

ok naddy@

Set the IFCAP_VLAN_MTU capabilities flag so these interfaces are allowed
to transmit full sized VLAN tagged frames.

Use ether_ioctl().

Convert timeout_add() calls using multiples of hz to timeout_add_sec()

Really just the low-hanging fruit of (hopefully) forthcoming timeout
conversions.

ok art@, krw@

Sanitize the use of timeouts:

Instead of calling timeout_set(..., NULL, NULL) in attach routines and
later
timeout_del(...)
timeout_set(..., func, arg)
timeout_add(..., time)
set the function and argument in the initial timeout_set() call and only
use timeout_add(..., time) later.

ok dlg, fgsch, krw, winiger

treat usb vendor/product names as a locator, and have usbd_print handle it,
so that it shows up before the :. as a result, all the usb devices do not
need to have name printing code anymore. all this now works and prints
nicely because usbd_probe_and_attach() is serialized. ok kettenis

Remove the definition and usage of the USB_DECLARE_DRIVER_CLASS and
USB_DECLARE_DRIVER macros.

No binary change.

ok dlg.

Remove the definition and usage of the IF_INPUT macro which was defined
as ether_input_mbuf which is itself a macro for ether_input.

No binary change.

ok dlg.

Remove the definition and use of the device_ptr_t which was a struct device *.

No binary change.

ok mk.

Remove the definition and usage of the USBGETSOFTC macro, which was really
only a cast to (void *).

ok jsg.

Remove the definition and use of the USBDEVNAME macro.

(This might look easy, but it was a big diff. Thanks to dlg and especially jsg
for looking over it; we found at least four mistakes in the initial diff.)

ok jsg.

Remove the definition and use of the USBDEV macro. It only created confusion
and the address of it's argument: USBDEV(sc->sc_dev) yields &sc->sc_dev.

No binary changes.

ok jsg.

Remove definitions and usage of usb_callout and related macros. These macros
were used as a layer of confusion^Wabstraction around the timeout(9) API.

No binary change.

ok jsg.

Remove the definition and use of if_deactivate(). It was defined empty and
thus produced no code at all.

No binary change.

ok jsg.

Remove the "Static" declaration of many functions. It was defined to be empty
and it was not consistently used. It was confusing as it suggested these
functions were static, which they were not.

discussed with dlg and jsg, ok jsg.

Last part of FreeBSD/NetBSD sepcific code removal.

ok jsg@

Farewell USB_{ATTACH,MATCH,DETACH}* you will not be missed.
ok deraadt@ krw@ mbalmer@

Remove logprintf macro

Remove Ether_ifattach macro

More lockmgr -> rwlock low hanging fruit.

ok dlg@

Remove _KERNEL_OPT leftovers.

add ZyXEL USB id, remove useless comments and reorder USB devices.

In a drivers activate() entry point, if on DVACT_DEACTIVATE it does
something, then it MUST return zero on DVACT_ACTIVATE, not EOPNOTSUPP;
this very popular bug has been cut and pasted a lot of times...
ok deraadt@ mickey@

allow bpf(4) to ignore packets based on their direction (inbound or
outbound), using a new BIOCSDIRFILT ioctl;
guidance, feedback and ok canacar@

Remove last NRND NetBSDisms from tree.

ok deraadt@ brad@

typos from alexey dobriyan;

Use usbd_clear_endpoint_stall_async() when clearing endpoint stalls in
an interrupt context.

From NetBSD

ok dlg@

{TrendNet,Trendware} -> TRENDnet, ok jsg@

Don't keep the devinfo string on the stack, instead use malloc/free.
This should cure some rare stack overflows.

From augustss NetBSD

ok dlg@ pascoe@

Update for GREENHOUSE2 -> REALTEK USB vid change.

clear IFF_RUNNING & IFF_OACTIVE in foo_stop() before de-allocating resources.

don't call mii_pollstat() right after mii_tick() in foo_tick_task()

remove netns crud.
some drivers actually had hooks for SIOCSIFADDR, most just useless includes
"looks good" deraadt miod brad

ethernet on OQO model 01 device cable. dlg ok

- make sure int is in running state before touching the multicast filters
- call foo_setmulti only instead of init'ing the chip
- don't overwrite potential error return with success when calling
ether_addmulti/ether_delmulti

ok dlg@

Three mode devices found in the wild, I think

strncpy->strlcpy
ok henning@

Add Longshine LCS-8138TX and Micronet SP128AR found in linux driver
ok dlg@

from netbsd: if_url.c 1.8, if_urlreg.h 1.2, urlphy.c 1.9 and urlphyreg.h 1.2
via Jonathan Gray
log message:

Remove advertising clauses from code written by Shingo WATANABE.
He has kindly consented it.

don't need to set ifp->if_mtu or ifp->if_output in each driver,
{ether,atm,fddi}_ifattach already does this.

ok mcbride@ markus@ henning@

put "do { } while (0)" wrappers on all the debug maroc functions

init ether address correctly, since our Ether_ifattach macro does not do
that job; 3612, rspmn@arcor.de

Set devclass to DV_IFNET for all USB network drivers.
ok deraadt@ nate@

string cleanup; nate ok

branches: 1.10.4;
pad minimum frames with 0's; based on netbsd

sync with NetBSD

Plug memory leak; nyar-junk@int0rnet.net

get rid of trailing whitespace

shorten address printout

Kill __FUNCTION__.

branches: 1.4.2;
Fix ioctl handling
Add default MTU
fix typo

errors noted by Jason Ackley <jason@ackley.net>

Add a driver for Realtek RTL8150L based ethernet devices
From NetBSD

Meant to commit only ulpt and committed everything. Most things weren't ready

Sync ulpt driver with NetBSD

use the UE_GET_XFERTYPE macro where applicable

ok mpi@

remove duplicate device ids

Make some USB ioctls return ENXIO, instead of EIO, if usbd_is_dying().
Brings us one step closer towards making this condition's error code
consistent across all USB drivers.
Patch by Moritz Buhl
ok mpi@ bluhm@

Use LIST_FOREACH construct instead of homebrew loop.

ok bluhm@, mpi@

move counting if_opackets next to counting if_obytes in if_enqueue.

this means packets are consistently counted in one place, unlike the
many and various ways that drivers thought they should do it.

ok mpi@ deraadt@

Avoid calling usbd_set_config_no() in *_attach() and let the stack do
it instead.

If anything bad happen due to a malformed descriptor it makes no sense
to try to attach a driver, and bail before probing.

This is similar to the change to avoid calling usbd_set_config_index().

G/C IFQ_SET_READY().

replace IFF_OACTIVE manipulation with mpsafe operations.

there are two things shared between the network stack and drivers
in the send path: the send queue and the IFF_OACTIVE flag. the send
queue is now protected by a mutex. this diff makes the oactive
functionality mpsafe too.

IFF_OACTIVE is part of if_flags. there are two problems with that.
firstly, if_flags is a short and we dont have any MI atomic operations
to manipulate a short. secondly, while we could make the IFF_OACTIVE
operates mpsafe, all changes to other flags would have to be made
safe at the same time, otherwise a read-modify-write cycle on their
updates could clobber the oactive change.

instead, this moves the oactive mark into struct ifqueue and provides
an API for changing it. there's ifq_set_oactive, ifq_clr_oactive,
and ifq_is_oactive. these are modelled on ifsq_set_oactive,
ifsq_clr_oactive, and ifsq_is_oactive in dragonflybsd.

this diff includes changes to all the drivers manipulating IFF_OACTIVE
to now use the ifsq_{set,clr_is}_oactive API too.

ok kettenis@ mpi@ jmatthew@ deraadt@

You only need <net/if_dl.h> if you're using LLADDR() or a sockaddr_dl.

shuffle struct ifqueue so in flight mbufs are protected by a mutex.

the code is refactored so the IFQ macros call newly implemented ifq
functions. the ifq code is split so each discipline (priq and hfsc
in our case) is an opaque set of operations that the common ifq
code can call. the common code does the locking, accounting (ifq_len
manipulation), and freeing of the mbuf if the disciplines enqueue
function rejects it. theyre kind of like bufqs in the block layer
with their fifo and nscan disciplines.

the new api also supports atomic switching of disciplines at runtime.
the hfsc setup in pf_ioctl.c has been tweaked to build a complete
hfsc_if structure which it attaches to the send queue in a single
operation, rather than attaching to the interface up front and
building up a list of queues.

the send queue is now mutexed, which raises the expectation that
packets can be enqueued or purged on one cpu while another cpu is
dequeueing them in a driver for transmission. a lot of drivers use
IFQ_POLL to peek at an mbuf and attempt to fit it on the ring before
committing to it with a later IFQ_DEQUEUE operation. if the mbuf
gets freed in between the POLL and DEQUEUE operations, fireworks
will ensue.

to avoid this, the ifq api introduces ifq_deq_begin, ifq_deq_rollback,
and ifq_deq_commit. ifq_deq_begin allows a driver to take the ifq
mutex and get a reference to the mbuf they wish to try and tx. if
there's space, they can ifq_deq_commit it to remove the mbuf and
release the mutex. if there's no space, ifq_deq_rollback simply
releases the mutex. this api was developed to make updating the
drivers using IFQ_POLL easy, instead of having to do significant
semantic changes to avoid POLL that we cannot test on all the
hardware.

the common code has been tested pretty hard, and all the driver
modifications are straightforward except for de(4). if that breaks
it can be dealt with later.

ok mpi@ jmatthew@

arp_ifinit() is no longer needed.

Increment if_ipackets in if_input().

Note that pseudo-drivers not using if_input() are not affected by this
conversion.

ok mikeb@, kettenis@, claudio@, dlg@

simply do if_input() between splnet() and splx(), like other drivers
(such as if_udav.c).

ok by mpi@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Convert to if_input().

ok dlg@

unifdef INET

No need for <netinet/in_systm.h> here.

Kill a bunch of DVACT_DEACTIVATE handlers whose only purpose were to
set the 'dying' flag of a device. Such handlers are useless now that
usbd_detach() already set this flag.

Even if the purpose of this flag is questionnable on OpenBSD because
DVACT_DEACTIVATE acts as a pre-detach event, this is just a first small
step towards the simplification of the autoconf(9) device states.

This cleaning is now possible thanks to the work of pirofti@ to convert
all the USB drivers to properly use usbd_is_dying().

Discussed many times with deraadt@

Clean-up usbd_abort_pipe() usage.


This function never fails. So change it's return type to void and
adjust the copy-pasted callers that were checking the return.

"If it compiles, ok" mpi@

Most network drivers include netinet/in_var.h, but apparently they
don't have to. Just remove these include lines.
Compiled on amd64 i386 sparc64; OK henning@ mikeb@

Get rid of various 'typedef struct' definitions and use plain structure
definitions instead. We don't change usb.h for now to stay compatible
with userland.

Tested by mpi@ on macppc and myself on i386.

ok mpi@

proc.h is way too much header for usb to handle.

Rewrite the receive filter handling code and cleanup the ioctl bits.

ok sthen@

Remove config_activate() and DVACT_ACTIVATE. PCMCIA's the only thing
that's ever used it, and it's long since been changed to use
DVACT_{QUIESCE,SUSPEND,RESUME} instead.

ok deraadt@, dlg@; miod@ also agreed with this idea when I brought it
up a few weeks ago

a bit more ansi; ok tedu

garbage collect "usb events". without /dev/usb there is no way to access
them from userland, and nothing in the kernel uses them.

ok krw@, miod@

* replace per-driver dying and/or other state variables with use of
usbd_deactivete() and usbd_is_dying()
* use usbd_deactivate() in activate()/DEACTIVATE
* convert a few more direct checks of the associated bus' dying flag
with usbd_is_dying()

don't free network related resources if they were not allocated

check that a timeout(9) has been initialized before deleting it

ok phessler

create another kthread to run xfer abort tasks. xfer abort tasks cannot
be run from the generic task kthread, because xfers that need to be
aborted block newly queued tasks from running (i.e. the xfer to be aborted
blocks the abort of that task). as there are now three types of usb
tasks, add an argument to usb_init_task() and another member to struct
usb_task to specify the task type.

fixes boot hangs that are showing up because we now use usb tasks to
attach/detach usb devices.

Delete comments about these drivers needing powerhook support; they don't.

Get rid of devact enum, substitute it with an int and coresponding defines.

This is needed for the addition of further suspend/resume actions.

Okay deraadt@, marco@.

Eliminate the redundant bits of code for MTU and multicast handling
from the individual drivers now that ether_ioctl() handles this.

Shrinks the i386 kernels by..
RAMDISK - 2176 bytes
RAMDISKB - 1504 bytes
RAMDISKC - 736 bytes

Tested by naddy@/okan@/sthen@/brad@/todd@/jmc@ and lots of users.
Build tested on almost all archs by todd@/brad@

ok naddy@

Set the IFCAP_VLAN_MTU capabilities flag so these interfaces are allowed
to transmit full sized VLAN tagged frames.

Use ether_ioctl().

Convert timeout_add() calls using multiples of hz to timeout_add_sec()

Really just the low-hanging fruit of (hopefully) forthcoming timeout
conversions.

ok art@, krw@

Sanitize the use of timeouts:

Instead of calling timeout_set(..., NULL, NULL) in attach routines and
later
timeout_del(...)
timeout_set(..., func, arg)
timeout_add(..., time)
set the function and argument in the initial timeout_set() call and only
use timeout_add(..., time) later.

ok dlg, fgsch, krw, winiger

treat usb vendor/product names as a locator, and have usbd_print handle it,
so that it shows up before the :. as a result, all the usb devices do not
need to have name printing code anymore. all this now works and prints
nicely because usbd_probe_and_attach() is serialized. ok kettenis

Remove the definition and usage of the USB_DECLARE_DRIVER_CLASS and
USB_DECLARE_DRIVER macros.

No binary change.

ok dlg.

Remove the definition and usage of the IF_INPUT macro which was defined
as ether_input_mbuf which is itself a macro for ether_input.

No binary change.

ok dlg.

Remove the definition and use of the device_ptr_t which was a struct device *.

No binary change.

ok mk.

Remove the definition and usage of the USBGETSOFTC macro, which was really
only a cast to (void *).

ok jsg.

Remove the definition and use of the USBDEVNAME macro.

(This might look easy, but it was a big diff. Thanks to dlg and especially jsg
for looking over it; we found at least four mistakes in the initial diff.)

ok jsg.

Remove the definition and use of the USBDEV macro. It only created confusion
and the address of it's argument: USBDEV(sc->sc_dev) yields &sc->sc_dev.

No binary changes.

ok jsg.

Remove definitions and usage of usb_callout and related macros. These macros
were used as a layer of confusion^Wabstraction around the timeout(9) API.

No binary change.

ok jsg.

Remove the definition and use of if_deactivate(). It was defined empty and
thus produced no code at all.

No binary change.

ok jsg.

Remove the "Static" declaration of many functions. It was defined to be empty
and it was not consistently used. It was confusing as it suggested these
functions were static, which they were not.

discussed with dlg and jsg, ok jsg.

Last part of FreeBSD/NetBSD sepcific code removal.

ok jsg@

Farewell USB_{ATTACH,MATCH,DETACH}* you will not be missed.
ok deraadt@ krw@ mbalmer@

Remove logprintf macro

Remove Ether_ifattach macro

More lockmgr -> rwlock low hanging fruit.

ok dlg@