to gem_attach() as the former access softc members not yet initialized
at that time and gem_reset() actually is enough to stop the chip. [1]
o Revise the use of gem_bitwait(); add bus_barrier() calls before calling
gem_bitwait() to ensure the respective bit has been written before we
starting polling on it and poll for the right bits to change, f.e. even
though we only reset RX we have to actually wait for both GEM_RESET_RX
and GEM_RESET_TX to clear. Add some additional gem_bitwait() calls in
places we've been missing them according to the GEM documentation.
Along with this some excessive DELAYs, which probably only were added
because of bugs in gem_bitwait() and its use in the first place, as
well as as have of an gem_bitwait() reimplementation in gem_reset_tx()
were removed.
o Add gem_reset_rxdma() and use it to deal with GEM_MAC_RX_OVERFLOW errors
more gracefully as unlike gem_init_locked() it resets the RX DMA engine
only, causing no link loss and the FIFOs not to be cleared. Also use it
deal with GEM_INTR_RX_TAG_ERR errors, with previously were unhandled.
This was based on information obtained from the Linux GEM and OpenSolaris
ERI drivers.
o Turn on workarounds for silicon bugs in the Apple GMAC variants.
This was based on information obtained from the Darwin GMAC and Linux GEM
drivers.
o Turn on "infinite" (i.e. maximum 31 * 64 bytes in length) DMA bursts.
This greatly improves especially RX performance.
o Optimize the RX path, this consists of:
- kicking the receiver as soon as we've a spare descriptor in gem_rint()
again instead of just once after all the ready ones have been handled;
- kicking the receiver the right way, i.e. as outlined in the GEM
documentation in batches of 4 and by pointing it to the descriptor
after the last valid one;
- calling gem_rint() before gem_tint() in gem_intr() as gem_tint() may
take quite a while;
- doubling the size of the RX ring to 256 descriptors.
Overall the RX performance of a GEM in a 1GHz Sun Fire V210 was improved
from ~100Mbit/s to ~850Mbit/s.
o In gem_add_rxbuf() don't assign the newly allocated mbuf to rxs_mbuf
before calling bus_dmamap_load_mbuf_sg(), if bus_dmamap_load_mbuf_sg()
fails we'll free the newly allocated mbuf, unable to recycle the
previous one but a NULL pointer dereference instead.
o In gem_init_locked() honor the return value of gem_meminit().
o Simplify gem_ringsize() and dont' return garbage in the default case.
Based on OpenBSD.
o Don't turn on MAC control, MIF and PCS interrupts unless GEM_DEBUG is
defined as we don't need/use these interrupts for operation.
o In gem_start_locked() sync the DMA maps of the descriptor rings before
every kick of the transmitter and not just once after enqueuing all
packets as the NIC might instantly start transmitting after we kicked
it the first time.
o Keep state of the link state and use it to enable or disable the MAC
in gem_mii_statchg() accordingly as well as to return early from
gem_start_locked() in case the link is down. [3]
o Initialize the maximum frame size to a sane value.
o In gem_mii_statchg() enable carrier extension if appropriate.
o Increment if_ierrors in case of an GEM_MAC_RX_OVERFLOW error and in
gem_eint(). [3]
o Handle IFF_ALLMULTI correctly; don't set it if we've turned promiscuous
group mode on and don't clear the flag if we've disabled promiscuous
group mode (these were mostly NOPs though). [2]
o Let gem_eint() also report GEM_INTR_PERR errors.
o Move setting sc_variant from gem_pci_probe() to gem_pci_attach() as
device probe methods are not supposed to touch the softc.
o Collapse sc_inited and sc_pci into bits for sc_flags.
o Add CTASSERTs ensuring that GEM_NRXDESC and GEM_NTXDESC are set to
legal values.
o Correctly set up for 802.3x flow control, though #ifdef out the code
that actually enables it as this needs more testing and mainly a proper
framework to support it.
o Correct and add some conversions from hard-coded functions names to
__func__ which were borked or forgotten in if_gem.c rev. 1.42.
o Use PCIR_BAR instead of a homegrown macro.
o Replace sc_enaddr[6] with sc_enaddr[ETHER_ADDR_LEN].
o In gem_pci_attach() in case attaching fails release the resources in
the opposite order they were allocated.
o Make gem_reset() static to if_gem.c as it's not needed outside that
module.
o Remove the GEM_GIGABIT flag and the associated code; GEM_GIGABIT was
never set and the associated code was in the wrong place.
o Remove sc_mif_config; it was only used to cache the contents of the
respective register within gem_attach().
o Remove the #ifdef'ed out NetBSD/OpenBSD code for establishing a suspend
hook as it will never be used on FreeBSD.
o Also probe Apple Intrepid 2 GMAC and Apple Shasta GMAC, add support for
Apple K2 GMAC. Based on OpenBSD.
o Add support for Sun GBE/P cards, or in other words actually add support
for cards based on GEM to gem(4). This mainly consists of adding support
for the TBI of these chips. Along with this the PHY selection code was
rewritten to hardcode the PHY number for certain configurations as for
example the PHY of the on-board ERI of Blade 1000 shows up twice causing
no link as the second incarnation is isolated.
These changes were ported from OpenBSD with some additional improvements
and modulo some bugs.
o Add code to if_gem_pci.c allowing to read the MAC-address from the VPD on
systems without Open Firmware.
This is an improved version of my variant of the respective code in
if_hme_pci.c
o Now that gem(4) is MI enable it for all archs.
Pointed out by: yongari [1]
Suggested by: rwatson [2], yongari [3]
Tested on: i386 (GEM), powerpc (GMACs by marcel and yongari),
sparc64 (ERI and GEM)
Reviewed by: yongari
Approved by: re (kensmith)
of the register rather than in the offset describing the register.
- In gem_reset_rx() let gem_bitwait() check for the Rx reset bit
rather than the Tx reset bit to clear.
Obtained from: OpenBSD (same/similar bugs being fixed)
GEMs is unable to discriminate UDP from TCP packets such that
it can generate 0x0000 checksum value for the UDP datagram. So the
UDP checksum offload was disabled by default. You can enable it
by setting link0 flag with ifconfig(8).
o bus_dma(9) clean up. It now correctly set number of required DMA
segments/size and removed incorrect use of BUS_DMA_ALLOCNOW flag
in static allocations done via bus_dmamem_alloc(9).
o Implemented ALTQ(9) support.
o Implemented Tx side bus_dmamap_load_mbuf_sg(9) which can remove
several book keeping chores orginated from call-back mechanism.
Therefore gem_txdma_callback() was removed and its functionality
was reimplemented in gem_load_txmbuf().
o Don't set GEM_TD_START_OF_PACKET flag until all remaining mbuf
chains are set. I think it was a long standing bug and it caused
fluctuating interrupts/CPU usage patterns while netperf test
is in progress. Previously it seems that we race with the device.
Because I don't have a documentation for GEM I'm not sure this is
correct but almost all other documentations I have stated this
implications on setting SOP mark in descriptor ring(e.g. hme(4)).
o Borrowed gem_defrag() from ath(4) which is supposed to be much
faster than m_defrag(9) since it's not need to defrag all
mbuf chains.
o gem_load_txmbuf() was changed to allow passed mbuf chains to free.
Caller of gem_load_txmbuf() correctly handles freed mbuf chains.
o In gem_start_locked(), added checks for availability of Tx
descriptors before trying to load DMA maps which could save CPU
cycles when number of available descriptors are low. Also, simplyfy
IFF_DRV_OACTIVE detection logic.
o Removed hard-coded function names in CTR macros and replaced it
with __func__.
o Moved statistics counter register access to gem_tick() to reduce
number of PCI bus accesses. There is no reason to update statistics
counters in interrupt handler.
o Removed unnecessary call of gem_start_locked() in gem_ioctl().
Reviewed by: grehan (initial version), marius (with improvements and suggestions)
Tested by: grehan (ppc), marius(sparc64)
correct network drivers with respect to busmaster DMA, go over it
with at duster to make other aspects of it a role model:
Eliminate the pci specific softc, it serves no rational purpose.
Use convenience resource allocation/deallocation functions to save
code and errorhandling.
Switch from bus_space_{read|write}_%u() to bus_{read|write}_%u()
functions and forget about tags and handles, the resource will know
about those, should they be needed. This also eliminates a number
of inconsistently named local variables.
gem_watchdog() in order to avoid races accessing if_timer.
While at it relax the watchdog a bit by reloading it in gem_tint()
if there are still packets enqueued.
- Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that.
- Fix inconsistencies in prototypes.
required by arches like sparc64 (not yet implemented) and sun4v where there
are seperate IOMMU's for each PCI bus... For all other arches, it will
end up returning NULL, which makes it a no-op...
Convert a few drivers (the ones we've been working w/ on sun4v) to the
new convection... Eventually all drivers will need to replace the parent
tag of NULL, w/ bus_get_dma_tag(dev), though dev is usually different for
each driver, and will require hand inspection...
Reviewed by: scottl (earlier version)
rather than in ifindex_table[]; all (except one) accesses are
through ifp anyway. IF_LLADDR() works faster, and all (except
one) ifaddr_byindex() users were converted to use ifp->if_addr.
- Stop storing a (pointer to) Ethernet address in "struct arpcom",
and drop the IFP2ENADDR() macro; all users have been converted
to use IF_LLADDR() instead.
the switch statement in order to make this driver more like other
Ethernet NIC drivers.
- In gem_attach() call gem_stop() in addition to gem_reset() to make
sure the chip actually is stopped and not just reset.
- In gem_stop() also stop the gem_rint_timeout() callout in case the
driver is compiled with GEM_RINT_TIMEOUT defined.
Merge some locking improvements from hme(4):
- Use callout_init_mtx() to close races between gem_stop() and gem_tick()
as weel as gem_stop() and gem_rint() in case the driver is compiled
with GEM_RINT_TIMEOUT defined.
- Use the driver lock instead of Giant in a bus dma callback.
- Lock the driver lock around mii operations.
- Cleanup locking in gem_ioctl().
- Remove redundant assertions that the driver lock is not held in
gem_attach() and gem_detach() since mtx_lock() will assert that
already since the driver lock is not recursive.
- Add callout_drain()'s to gem_detach() after calling gem_stop() to make
sure that if softclock is running on another CPU and is blocked on our
driver lock, we will wait until it has acquired the lock, seen that it
was cancelled, dropped the lock, and awakened us so that we can safely
destroy the mutex.
- On resume all registers have to be initialized again like after
power-on so reset sc_inited in gem_suspend() in order get all of
the registers set next time gem_init_regs() is called.
- On at least some ERI and GEM revisions GEM_MAC_RX_OVERFLOW happen
often due to a silicon bug and re-initializing is all we can do
about these errors so make handling them non-verbose.
- Remove a superfluous memset(3) call in gem_meminit(), all elements
are initialized to 0 anyway.
MFC after: 1 week
IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to
ifnet.if_drv_flags. Device drivers are now responsible for
synchronizing access to these flags, as they are in if_drv_flags. This
helps prevent races between the network stack and device driver in
maintaining the interface flags field.
Many __FreeBSD__ and __FreeBSD_version checks maintained and continued;
some less so.
Reviewed by: pjd, bz
MFC after: 7 days
over iteration of their multicast address lists when synchronizing the
hardware address filter with the network stack-maintained list.
Problem reported by: Ed Maste (emaste at phaedrus dot sandvine dot ca>
MFC after: 1 week
descriptors that are still marked owned in #ifdef GEM_RINT_TIMEOUT
instead of #if 0 for convenience.
- Remove stale code and comment about relying on the preset XIF config.
- In case of a watchdog timeout call the init function instead of just
the start function so the chip is properly reset.
Merge from hme(4):
- Convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers.
- Protect from a duplicate mbuf free panic in case the DMA engine hangs.
Reviewed by: yongari
Tested on: powerpc(grehan), sparc64
MFC after: 1 week
struct ifnet or the layer 2 common structure it was embedded in have
been replaced with a struct ifnet pointer to be filled by a call to the
new function, if_alloc(). The layer 2 common structure is also allocated
via if_alloc() based on the interface type. It is hung off the new
struct ifnet member, if_l2com.
This change removes the size of these structures from the kernel ABI and
will allow us to better manage them as interfaces come and go.
Other changes of note:
- Struct arpcom is no longer referenced in normal interface code.
Instead the Ethernet address is accessed via the IFP2ENADDR() macro.
To enforce this ac_enaddr has been renamed to _ac_enaddr.
- The second argument to ether_ifattach is now always the mac address
from driver private storage rather than sometimes being ac_enaddr.
Reviewed by: sobomax, sam
if_xname, if_dname, and if_dunit. if_xname is the name of the interface
and if_dname/unit are the driver name and instance.
This change paves the way for interface renaming and enhanced pseudo
device creation and configuration symantics.
Approved By: re (in principle)
Reviewed By: njl, imp
Tested On: i386, amd64, sparc64
Obtained From: NetBSD (if_xname)
Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg.
Lockfunc allows a driver to provide a function for managing its locking
semantics while using busdma. At the moment, this is used for the
asynchronous busdma_swi and callback mechanism. Two lockfunc implementations
are provided: busdma_lock_mutex() performs standard mutex operations on the
mutex that is specified from lockfuncarg. dftl_lock() is a panic
implementation and is defaulted to when NULL, NULL are passed to
bus_dma_tag_create(). The only time that NULL, NULL should ever be used is
when the driver ensures that bus_dmamap_load() will not be deferred.
Drivers that do not provide their own locking can pass
busdma_lock_mutex,&Giant args in order to preserve the former behaviour.
sparc64 and powerpc do not provide real busdma_swi functions, so this is
largely a noop on those platforms. The busdma_swi on is64 is not properly
locked yet, so warnings will be emitted on this platform when busdma
callback deferrals happen.
If anyone gets panics or warnings from dflt_lock() being called, please
let me know right away.
Reviewed by: tmm, gibbs
- Fix compilation without GEM_DEBUG.
- Do not #define GEM_DEBUG by default; it adds overhead (due to bzero()ing
RX space) and is not needed any more, since the driver is quite stable
now.
- Fix watchdog timeouts when failing to load TX packets.
- Do not forcibly limit the number of descriptors used for a packet to
GEM_NTXSEGS, by passing this number to bus_dma_tag_create(). There is
no requirement for a limit any lower than the total number of
available descriptors, and the present limit caused network problems
due to mbuf chains requiring more descriptors.
GEM_NTXSEGS is still used to estimate the interrupt window size, for
which we just need an estimate.
Approved by: re (rwatson)
network layer (ether).
- Don't abuse module names to facilitate ifconfig module loading;
such abuse isn't really needed. (And if we do need type information
associated with a module then we should make it explicit and not
use hacks.)
- Remove NetBSD-style or-ed together BUS_DMASYNC operations, in some
cases relaxing the (intended) syncing operation a bit.
- Stop pretending that that we can sync part of a dmamap: replace the
GEM_CDTXSYNC and GEM_CDRXSYNC macros with GEM_CDSYNC to sync the
complete control map, and combine syncs wherever possible to avoid
the overhead.
- Sync all maps before unloading them.
- Remove a few syncs which should be unnecessary.
- remove DPRINTF(), there is a CTR*() for any of them, and KTR is
far more useful to debug this driver.
- some cleanups; remove some unused code and definitions.
