- Addresses the issue described in PR/38828.
- Some simplification in threading and sleepq subsystems.
- Eliminates pmap_collect() and, as a side note, allows pmap optimisations.
- Eliminates XS_CTL_DATA_ONSTACK in scsipi code.
- Avoids few scans on LWP list and thus potentially long holds of proc_lock.
- Cuts ~1.5k lines of code. Reduces amd64 kernel size by ~4k.
- Removes __SWAP_BROKEN cases.
Tested on x86, mips, acorn32 (thanks <mpumford>) and partly tested on
acorn26 (thanks to <bjh21>).
Discussed on <tech-kern>, reviewed by <ad>.
controllers as well myself and another on viaide.
Stops errors such as the following when probing SATA drives through
controllers that offer the legacy pciide interface:
viaide1 channel 0: reset failed for drive 0
OK bouyer@
Tested on Intel SS4200-E (ICH7), and Acorp 6A815EPD (ICH2) motherboards,
on amd64 and i386 ports respectively.
It should be noted that the majority of boards with ICH chips do not
expose the GPIO pins for off-board use. For instance, aside from the
three exposed-on-a-header pins on the 6A815EPD, another pin is also
used to control write protect on the FWH. The SS4200 exposes the GPIO
on a header that connects to the 10 LEDs on the front panel, as well
as a tact switch on the back panel.
- remove powerhook_establish(9) because all stuff in cs_power()
will be handled in pmf_class_network
Briefly tested on CS8900 at ofisa on DNARD compatible.
and make suspension by self, by drvctl(8), and by ACPI system sleep
play nice together. Start solidifying some temporary API changes.
1. Extract a new header file, <sys/device_if.h>, from <sys/device.h> and
#include it from <sys/pmf.h> instead of <sys/device.h> to break the
circular dependency between <sys/device.h> and <sys/pmf.h>.
2. Introduce pmf_qual_t, an aggregate of qualifications on a PMF
suspend/resume call. Start to replace instances of PMF_FN_PROTO,
PMF_FN_ARGS, et cetera, with a pmf_qual_t.
3. Introduce the notion of a "suspensor," an entity that holds a
device in suspension. More than one suspensor may hold a device
at once. A device stays suspended as long as at least one
suspensor holds it. A device resumes when the last suspensor
releases it.
Currently, the kernel defines three suspensors,
3a the system-suspensor: for system suspension, initiated
by 'sysctl -w machdep.sleep_state=3', by lid closure, by
power-button press, et cetera,
3b the drvctl-suspensor: for device suspension by /dev/drvctl
ioctl, e.g., drvctl -S sip0.
3c the system self-suspensor: for device drivers that suspend
themselves and their children. Several drivers for network
interfaces put the network device to sleep while it is not
administratively up, that is, after the kernel calls if_stop(,
1). The self-suspensor should not be used directly. See
the description of suspensor delegates, below.
A suspensor can have one or more "delegates". A suspensor can
release devices that its delegates hold suspended. Right now,
only the system self-suspensor has delegates. For each device
that a self-suspending driver attaches, it creates the device's
self-suspensor, a delegate of the system self-suspensor.
Suspensors stop a system-wide suspend/resume cycle from waking
devices that the operator put to sleep with drvctl before the cycle.
They also help self-suspension to work more simply, safely, and in
accord with expectations.
4. Add the notion of device activation level, devact_level_t,
and a routine for checking the current activation level,
device_activation(). Current activation levels are DEVACT_LEVEL_BUS,
DEVACT_LEVEL_DRIVER, and DEVACT_LEVEL_CLASS, which respectively
indicate that the device's bus is active, that the bus and device are
active, and that the bus, device, and the functions of the device's
class (network, audio) are active.
Suspend/resume calls can be qualified with a devact_level_t.
The power-management framework treats a devact_level_t that
qualifies a device suspension as the device's current activation
level; it only runs hooks to reduce the activation level from
the presumed current level to the fully suspended state. The
framework treats a devact_level_t qualifying device resumption
as the target activation level; it only runs hooks to raise the
activation level to the target.
5. Use pmf_qual_t, devact_level_t, and self-suspensors in several
drivers.
6. Temporarily add an unused power-management workqueue that I will
remove or replace, soon.
Delete unused atw_voodoo and constants.
Export Tx/Rx statistics with evcnt(9).
Correct the Short Inter-Frame Space (SIFS) that we write to ADM8211's
registers; I do not recall if that corrected the SIFS that I observed
"on the air." Use the constant IEEE80211_DUR_DS_EIFS to configure
the ADM8211's EIFS, instead of writing the same "magic" number,
0x64, that my reference driver wrote.
Do not clear OACTIVE in atw_init(), because atw_stop() cleared it
previously by calling atw_txdrain().
Use the net80211 short-preamble flag and instead of ATW_SHPREAMBLE.
Add an ADM8211 workaround from the reference driver, atw_workaround1(),
but don't compile it right now.
In at_intr(), don't stop processing the interrupt status after
restarting the receive ring, but process Tx interrupt status. If
a packet's Tx lifetime is exceeded, reinitialize the device to get
packets moving again. If the Tx FIFO underflows, restart the
transmitter, not the receiver!
Avoid losing synchronization with the Rx ring by replicating one
of Charles Hannum's fixes to rtw(4) here: receiving a management
packet may, as a side-effect, reset the Rx ring, so refer to the
softc's Rx ring pointer, sc_rxptr, every time through the loop in
atw_rxintr(), instead of refering to a pointer on the stack, i.
Re-synchronize DMA after reading the OWN bit on an Rx/Tx descriptor.
XXX This needs more work.
Reset sc_tx_timer as Tx descriptors are reclaimed from the device.
Shorten staircases in atw_watchdog().
Remove from softc an unused member, sc_intr_ack.
hme_chipreset() to make hme_stop() match struct ifnet API
- set ifp->if_timer in hme_start() if any TX packets are queued
- also clear ifp->if_timer and ifp->if_flags in hme_stop()
- replace shutdownhook_establish(9) with pmf_device_reigster1(9)
Briefly checked hme at pci.
as well as pre-pmf(9)'fied kernel did via shutdownhook_establish(9).
Tested on O2.
ahc_suspend() and ahc_resume() are still disabled due to lack of hardware.
Anyone who can test Cardbus APA-1480?
since pmf_device_deregister(9) is called from ahc_detach() so that
cardbus backend also gets proper pmf(9) calls.
PCI backend is tested on on O2, but cardbus is untested.
The sysmon subsystem is marked MPSAFE and so runs without the kernel lock.
So get the kernel lock in mfi_sensor_refresh() before calling
mfi_ioctl_vol().
This fixes command list corruption seen on heavy I/O load on the mfi
driver(4).
XXX More testing is needed. I've tested this on a Dell Dimension 3000,
XXX but that system does not attach every possible device that I try to
XXX detach with this code:
ichlpcib0 at pci0 dev 31 function 0
ichlpcib0: vendor 0x8086 product 0x24d0 (rev. 0x02)
timecounter: Timecounter "ichlpcib0" frequency 3579545 Hz quality 1000
ichlpcib0: 24-bit timer
ichlpcib0: TCO (watchdog) timer configured.
isa0 at ichlpcib0
not finished the device_t/softc split for the SBus attachment
because I don't have an SBus gem(4) to test with.
Convert from legacy shutdownhooks to a PMF shutdown hook. Add PMF
suspend/resume handlers.
Factor a detachment hook out of gem_attach(). During device
attachment, track which resources are reserved in sc_attach_state,
and release only those resources during detachment.
Tested on gem0 and gem1 at pci1 on a Sun Fire V120.
Detachment may fail after freeing some but not all resources, so
take care not to re-release any resource during detachment.
Tested on a Dell PowerEdge 1950.
Reset ch_status and ch_error at the beginning of any operation.
The chip only writes the RDH FIS to the SRAM when the error bit is
set in a RDH FIS (per the datasheet). Thus, satafis_sdb_parse() is
replaced with satafis_rhd_parse(). A valid FIS type field seems to
not be written to the SRAM, so don't bother checking it for the
correct magic.
Additionally, append 2009 to copyright year set, and remove
obsolete CVS Id lines from when this was in my repository.
Clean up siisata pci attachment.
noteable changes:
- Use aprint_*_dev where appropriate.
- Condense board match table and use C99 initializers.
- Use aprint_verbose for extra info.
Also, add 2009 to my copyright, as well as remove my local CVS tags.
- Move clearing of interrupts to before atastart() is called in the
xfer interrupt handler. Should fix kern/41579.
- Using cv_timedwait(9) is not possible in code that can be called from
interrupt context, fall back to DELAY(9).
- Correctly poll Port Slot Status register for soft reset PRBs.
- Only use the Recive Transfer Count register on reads, when it is valid.
- Activate PRBs in a way that takes the whole physical address into account,
even when the PRB is beyond 4GiB.
- consistently use DELAY(9)
- Use DELAY() constants in completion polling loops that are consistent with
the loop count limit. (i.e. timeout in 10 rather than 100 seconds)
- Support detachment. From KIYOHARA Takashi.
- Add PCI detachment functionality (albeit not very interesting when
the bus can not yet be rescanned).
- Rework interrupt handlers to reduce near-duplicate code.
Borrowed from ahcisata(4).
- Attempt to make polled I/O work. Untested.
- Fix formatting of some messages.
- For always-polled commands, disable interrupt
at slot level rather than port level.
- Instead of busy-waiting indefinitely for completion of some commands
move on after 31 seconds. Use cv_timedwait(9) instead of DELAY(9).
- Use abstracted SATA FIS code.
- Enable use of disks that don't respond with the standard signature.
* On checking TCPv4/UDPv4 RX checksum on DESCV2 chips, also check
RE_RDESC_VLANCTL_IPV4 bit because those DESCV2 chips may also recognize
IPv6 packets and set RE_PROTOID_TCPIP or RE_PROTOID_UDPIP bits for
TCPv6/UDPv6 packets. This may fix PR kern/40605.
* According to Realtek's Linux driver, DESCV2 chips don't set RE_PROTOID_IP
for non-TCP/UDP IP packets (set only RE_RDESC_VLANCTL_IPV[46]) so
remove PROTOID check for IPv4 RX cheksum on DESCV2 chips.
and remove bogus casts around ahc_detach().
XXX: There is a pmf_device_deregister(9) call in ahc_detach()
XXX: while there is no pmf_device_register(9) in ahc_attach().
XXX: It looks more pmf(9) stuff is required for cardbus detach.
- some KNF
- u_intNN_t -> uintNN_t
- wrap long lines and fix indent
- remove return statements at the end of void functions
- use __arraycount()
XXX: many inconsistent aprint_error_dev(9) vs printf(9)
No crash on:
esiop0 at pci0 dev 11 function 0: Symbios Logic 53c875 (ultra-wide scsi)
siop0 at pci0 dev 9 function 0: Symbios Logic 53c810a (fast scsi)
siop0 at gsc0 hpa 0xf0830000 path 2/0/7 irq 3 ipl 6: NCR53C720 rev 2
- calculate each descriptor sizes and offsets in iee_attach() and store them
into softc, rather than re-calculating them everywhere via macros
- prepare macros to sync DMA shmem per each descriptor
- sync only necessary descriptors in iee_intr() and iee_start()
- make sure SCB_FOO macros take softc as an arg properly
- use bus_dmamap_load_mbuf(9) for RX mbufs
- put 2 byte alignments to RX mbufs
XXX: still slower than ie(4) in i82586 compat mode on HP9000 735/125.
DMA descriptor members in shared memory.
The previous ones could return wrong values if cache line size is
smaller than structures, and we can assume cache line size is
power of two to avoid division ops.
from MD attachments to MI iee_attach().
Tested on HP9000 735/125. Compile test only on ews4800mips.
XXX: more strict bus_dmamap_sync(9) calls are required in various places.
we have an MII transeiver attached.
Count all collision and error counters.
Handle counter overflow and RXTERR.
Tested on U60 HME, PCI HME (501-5019) and SBus Sunswift (501-2739)
big endian mode:
- add IEE_REV_A flag to indicate if chip support the 32 bit BE mode or not
- add IEE_SWAPA32() macro and use it on necessary 32 bit DMA pointers
- rename IEE_SWAP() macro for the SCP address pointer and statistics
counters which require word swap even on Rev B/C chips to IEE_SWAP32()
for clarification
- add comments about these BE mode quirks
Tested on HP9000 735/125 by me and also tested on 715/50 by skrll@
with netbsd-5 branch, and fixes MI part of PR port-hp700/35531.
- pull MACSTAT and CMDSTOP quirks for 8168/8111 chips
- always set CPLUSCMD_PCI_MRW on reset
- set VLANSTRIP and RXCSUM_ENB bits on CPLUS register per if_capenable
Tested on 8111C and 8111D by several users, and
no bad side effect on my old 8169S.
corresponding attimer(4) resource is not perpetually busy: now it
is possible to detach and to reattach pcppi(4). Make attimer(4)'s
device-detachment hook return EBUSY if pcppi(4) is still attached, so
that pcppi(4) cannot end up with a dangling pointer to attimer(4).
sort of support hardware WEP. Only change the WEP key index in a Tx
descriptor from 0 if the type of our transmit key is _WEP, not _TKIP:
i.e., only if we're really doing WEP crypto in the hardware.
Ignore a watchdog timeout on any Tx ring if we can collect some packets
from that ring. Restart both the receiver and the transmitter when a
watchdog timeout occurs instead of restarting only the transmitter.
Instead, always check ether_type and use appropriate offsets to adjust
the hardware RX sum value.
XXX: vlan(4) doesn't seem to use csum_data and csum_flags in mbufs anyway.
per device info taken from FreeBSD driver. Tested by snj@ on 8111C.
Should closes PR kern/40955.
Note on old 8169 chips IP hw csum must be enabled to use TCP/UDP hw csums,
but I'm not sure if these newer chips still have the same restriction.
z8530sc.c:
Check pending interrupts in a loop until all requests are handled.
The old comments said it would cause horrible latency to sun3x floppy etc,
but serial ports should have higher priority than disks anyway.
z8530tty.c:
Don't enable and disable TX interrupts on each transmit start and completion
because it could cause possible race conditions.
Instead, set ZSWR0_RESET_TXINT on each TIE interrupt to clear the request
as other kbd drivers attached at zs(4).
Tested on cobalt, macppc, news68k, sparc, and sun3.
> Fix a bug in calculation of checksum deduction:
> - To get 16 bit one's complement value from uint32_t variable,
> higher 16 bits should be ignored.
> - RFC 1624 describes methods to recalculate checksum field in headers,
> i.e. one's complement of one's complement sum that could be 0x0000,
> but we don't have to use the strategy to deduct one's complement sum
> itself which won't be zero but should be 0xffff.
- To get 16 bit one's complement value from uint32_t variable,
higher 16 bits should be ignored.
- RFC 1624 describes methods to recalculate checksum field in headers,
i.e. one's complement of one's complement sum that could be 0x0000,
but we don't have to use the strategy to deduct one's complement sum
itself which won't be zero but should be 0xffff.
Found on debugging mec(4) on sgimips O2.
There are still about 1600 left, but they have ',' or /* ... */
in the actual variable definitions - which my awk script doesn't handle.
There are also many that need () -> (void).
(The script does handle misordered arguments.)
in IP headers, so we have to deduct not only IP option headers but all
IP headers. But in TCP/UDP layer we can assume the IP header is valid
and a sum of the IP header part should be 0xffff, so we don't have to
bother to deduct it from the computed checksum.
which don't have EXT_RFA and IPCB support. From hme(4) driver and
FreeBSD's fxp(4). Tested on i82559.
XXX: Probably we should have a common function to parse RX packet headers
XXX: to handle a raw checksum value and share it among hme(4) and gem(4) etc.