| Age | Commit message (Collapse) | Author |
|---|
|
ok various
|
|
this makes it consistent with the rest of the code.
|
|
OK stsp@
|
|
Found by LLVM/Clang Static Analyzer.
ok dlg@
|
|
Pointed out by jsg@ and Nathanael Rensen, <nathanael at
list ! polymorpheus ! com>, thanks!
|
|
runs the mailbox interface knows about a MAC address that appears to be
derived from a unique serial number along with the raspberry pi
foundation oui.
It modifies the device tree when booting to store the MAC address
in /axi/usb/hub/ethernet/mac-address so fetch and use this value
for the integrated smsc(4) Ethernet.
A different smsc adapter plugged into one of the USB ports
probes later with a different unit number and skips this path.
|
|
currently tests the return value.
ok stsp@
|
|
on some or all original mfii boards (2208) the aen path fires repeatedly
without reporting anything, causes enough load to start actual io.
found by naddy@
|
|
Alexey Suslikov <alexey.suslikov at gmail.com>.
Thanks.
|
|
Thanks.
|
|
code. From Michael W. Bombardieri" <mb at ii.net>. Thanks.
|
|
Thanks.
|
|
|
|
|
|
When destroying the ring all transfers should be already gone so there
should be no need to repeat ourselves after xbf_stop has done its work.
Get rid of the yield() that was probably masking some issues that have
been since fixed.
|
|
|
|
|
|
|
|
intr_barrier(9) is useful to make sure that after an interrupt is
masked, the interrupt handler for the device has finished executing
before proceeding with further device configuration.
However, since Xen interrupt handlers run in the thread context, we
need to make sure that they have finished as well. By scheduling a
xen_barrier_task modelled after (or rather copied ;) ifq_barrier_task
we can ensure that the interrupt handler is no longer running.
|
|
An xnf & xbf attach/detach loop has revealed that sometimes when we're
about to free a grant table entry that is still in use which is a grave
mistake code wise. Turned out we could allocate an entry twice because
of an incorrect test that took flags value into account when making the
decision regarding availability of a given entry. At the same time,
upon releasing the entry we explicitly CAS in 0 into the flags making
this check bogus.
While here be explicit about starting flags by initializing them to 0
and always panic when the "double free" condition is encountered.
rzalamena@ has lent me his eyes and has double-checked the condition.
|
|
basically use htolem64 to set the address and htolem32 for the
length, not the other way round. lucky this is mostly run on x86.
found by Jon Kloske
|
|
by pre-allocating two cryptodesc objects and storing them in an array
instead of a linked list. If more than two cryptodesc objects are
required use mallocarray to fetch them. Adapt the drivers to the new
API.
This change results in one pool-get per ESP packet instead of three.
It also simplifies softraid crypto where more cryptodesc objects are
allocated than used.
From, with and ok markus@, ok bluhm@
"looks sane" mpi@
|
|
useful to propagate the error. When an error occurs in an asynchronous
network path, incrementing a counter is the right thing. There are
four places where an error is not accounted, just add a comment for
now.
OK mpi@ visa@
|
|
|
|
This prevents a deadlock with the X server and some wireless drivers.
The real fix is to take unix domain socket code out of the NET_LOCK().
Issue reported by pirofti@ and ajacoutot@
ok tb@, stsp@, pirofti@
|
|
this lets me pass the specific argument to an aen handler in mfii.
it also unbreaks the tree.
found by jmatthew@
|
|
this replaces individual calls to pool_init, pool_set_constraints, and
pool_sethardlimit with calls to m_pool_init. m_pool_init inits the
mbuf pools with the mbuf pool allocator, and because of that doesnt
set per pool limits.
ok bluhm@ as part of a larger diff
|
|
|
|
more specificially we probe the disk if it goes from UNCONFIGURED_GOOD
to a SYSTEM disk, and detach it if goes from being a SYSTEM disk
to anything else.
this semantic comes from the lsi^Wavago code in the illumos mr_sas
driver. seems to work fine.
i think this covers all the ways a passthru disk can transition on
these boards.
|
|
|
|
this only handles MFI_EVT_PD_INSERTED_EXT and MFI_EVT_PD_REMOVED_EXT so
far. if this code is to be reused in mfi, it should probably change to
use MFI_EVT_PD_INSERTED and MFI_EVT_PD_REMOVED instead.
unlike mpii and mpi, it looks like the firmware aborts outstanding
commands against a disk when it's physically removed, so we dont
have to explicitly abort them. this is probably a carry over from
mfi generation boards which dont have an explicit abort command
they can use.
|
|
this submits MR_DCMD_CTRL_EVENT_WAIT commands via the async dcmd
path to read all events from boot onward, and eventually ends up
waiting after the boot messages are consumed.
right now none of the events are handled, but this can be added now
this framework is in place.
the board does generate human readable log messages for every event.
we can send them somewhere (dmesg or syslog for example), but for
now theyre masked by #if 0.
|
|
async dcmds are submitted via an mpii request (like the scsi commands
are) which uses the ccb_request buffer, meaning that the dcmd itself
has to go somewhere else. this reuses the sense buffer on the ccb
for the dcmd, and provides wrappers for accessing that space and
submitting a dcmd via the passthru command.
|
|
|
|
Now that we can attach and detach devices, we need to make sure we
can do so while interrupts are running. Thankfully, in the meantime
the refcnt_init(9) API came around to help us out.
|
|
|
|
|
|
|
|
there's a struct timeout in scsi_xfer for this purpose, which is
used to schedule a timeout of the command in the future. the timeout
adds the xs to a list in mfii_softc of outstanding commands that
are to be aborted. this list is processed in a task so we can sleep
for an mfii_ccb. the new ccb is used to issue an abort against the
specific command that timed out.
to avoid having a timeout complete at the same time as a command
on the chip, a refcnt is added to ccbs. the chip and the timeout
get a ref each. the mfii completion path will attempt to timeout_del,
and if that's succesful it will subtract the timeouts ref as well
as its own. if it fails, the abort path owns the ccb and becomes
responsible for calling scsi_done on behalf of the mfii completion
path.
|
|
|
|
ok visa@
|
|
ok stsp@
|
|
The issue appears at least with 7265 and 8260 chips as confirmed by
stsp@, tb@ and myself.
Fixed by unlocking NET_LOCK() at the beginning of iwmioctl() and
relocking it at the end.
Guidance and ok mpi@, ok stsp@.
|
|
of a shutdown. Correct a mistake in rev 1.11 to make this trigger a
shutdown again.
ok reyk@
|
|
Found by LLVM/Clang Static Analyzer.
ok mpi@
|
|
makes bioctl work on mfi(4) again.
while here memset a few mboxes before using them.
bioctl breakage reported by jason george
|
|
From Jan Klemkow
|
|
Works much better than firmware-based rate scaling in my testing.
ok mpi@
|
|
not tell rate scaling algos about failed retries. All attempts use the same
rate anyway, so checking whether the frame succeeded or failed is good enough.
ok mpi@
|
|
correspond to the flag stored in our stack as ac->ac_acm.
ok mpi@
|
