the configuration. The match/attach code was assuming whole-device
attach args all the time.
-Use the first (ie index 0) configuration for ubsa -- it makes the
code work in at least one case (PR kern/39211 by Frank Wille), and
there is no indication that an alternative configuration would
be needed. (I've admittedly never seen a usb device with more than
one configuration.)
This fixes mistakes when uhmodem support was added, and the changes were
not tested with a Huawei device yet. So please test if you can.
per-interface attachment into individual functions, to ease
maintainance and allow easier plugin of new attachment functions
-keep a counter of USB interfaces in use on a device, and try to
keep track of interfaces claimed by drivers behind the framework's
back
as proposed on tech-kern@.
While all DMA memory used in the USB framework is mapped BUS_DMAMAP_COHERENT
(including data memory, which is memcpy'd from/to USB-private buffers),
the CPU can reorder loads or stores from/to main memory, causing the
controller to have an incoherent view of the DMA descriptors lists for a
short time. bus_dmamap_sync() should contain memory barriers that prevents
the CPU from reordering load/store. Note that BUS_DMAMAP_COHERENT is
still required for the DMA descriptor lists - these can't work properly
with software cache coherency.
This fixes "host controller process error/host controller halted" errors
I'm occasionally seeing with a umodem device on uhci on x86.
Thanks to Michael Lorenz for testing it on his O2, and Izumi Tsutsui on
his Cobalt Qube 2700.
XXX:
"modload uaudio.kmod" doesn't attach the uaudio driver to
plugged USB devices for now because uhub_rescan() is not
implemented yet. We need to re-plug USB devices after modload.
-avoid unnecessary tree walks and port status requests:
-acknowledge hub status change notifications (but do nothing)
-clear cached port status change bits earlier (but not on ehci, due
to an ehci driver bug)
-do the ehci check on attach rather than a string comparision at runtime
to a crash reported by Christoph Egger in a followup to PR kern/38528.
For consistency, keep track of the device_t pointer to child devices
rather than the softc. We really shouldn't mess with child's softc data.
-unifdef
-since the roothub attach doesn't use locators, don't call
config_stdsubmatch() -- it is a no-op in that case
-ifsubmatch has configuration and interface always set to useful values,
remove unnecessary checks
-remove now unused locator definitions from shared header
than a zero-terminated list; this makes the code simpler and also
hopefully fixes the recent "childdet" botch, see PR kern/38528
-handle the root hub specially a bit earlier, this allows to kick out
the "submatch" functions completely which needed to second-guess
from the port number (where "0" meant root hub")
(we could handle the root hub specially even earlier, but as done
now big parts of the hub emulation code are exercised regularely,
this would bitrot otherwise)
