src - OpenBSD base system

Age	Commit message (Collapse)	Author
2007-05-19	hook up interrupts, but a little different to what i usually do. this maps	David Gwynne
	the interrupt before the bus independant code is attached, and establishes the interrupt after that.
2007-05-18	spelling in comment fix	David Gwynne

2007-05-16	add definitions for the bits in the "register" bus_space region.	David Gwynne

2007-05-16	map the pci BARs that we're interested in. this gives us access to the	David Gwynne
	chips registers and io interface, but whats actually in them is still a mystery to me.
2007-05-15	start work on a driver for emulex fibre channel controllers.	David Gwynne
	it is split up into a bus independant chunk (dev/ic/sli*) and the attach glue (dev/pci/sli_pci.c) cos there are sbus varaints of this chip which i hope to get my hands on one day. this does nothing except attach to the LP8000 controllers.
2007-05-14	delete unused code; ok krw	Theo de Raadt

2007-05-09	Remove an obsolete comment which guesses about a register write an never	Marcus Glocker
	was used anyway. OK claudio@
2007-05-09	diable -> disable	Miod Vallat

2007-05-09	remove the internal wrapper functions for ieee80211_mhz2ieee and	Reyk Floeter
	ieee80211_ieee2mhz which also removes some dead code. figured out by deraadt@
2007-05-08	backout while the amd64 puc situation is rethought	Theo de Raadt

2007-05-08	non-static inline functions waste space	Theo de Raadt

2007-05-08	move com_activate() to where it belongs	Theo de Raadt

2007-05-08	put more foo_detach() routines where they are used	Theo de Raadt

2007-05-08	rl_detach() is only used for cardbus case	Theo de Raadt

2007-05-08	wdcrestart() is never used	Theo de Raadt

2007-05-08	atapi_print() is never used	Theo de Raadt

2007-05-08	ami chips have no mechanism to take commands off them once we've submitted	David Gwynne
	them. this means that we cant reliably complete an io before the chip says we can because it can possibly complete later and overwrite memory it no longer owns, or write garbage to disk. so this diff forces the timeout on a scsi io to be as long as the chip should ever take, which is sixty seconds. after much discussion with marco we also decided to add a bit more so the time spent in the software runq wouldnt affect the ios lifetime too much. instead of completing ios out of the timeout, this simply warns the operator to check the state of the volumes if things are starting to slow down. ok marco@
2007-05-08	turn off SIOP_STATS; ok krw	Theo de Raadt

2007-05-08	dc_detach() is only used by cardbus code, so move it there; ok jsg	Theo de Raadt

2007-05-07	floating unused variable (except in debug case)	Theo de Raadt

2007-05-05	malo_hexdump() is only used by code under MALO_DEBUG, so ifdef it	Jonathan Gray
	ok mglocker@ claudio@
2007-05-05	move xl_detach() -- which is only used by cardbus -- to the cardbus code.	Theo de Raadt
	ok jsg
2007-05-05	Properly ifdef debug bits to save a bit of space.	Jonathan Gray
	ok reyk@
2007-05-03	Add monitor mode. Add fast channel switching. Do some simplifications	Marcus Glocker
	while there. Comments, tested, and OK claudio@
2007-05-03	Enable fast channel switching in monitor mode, which makes apps like	Marcus Glocker
	kismet happy. Previously system got very slow because at every channel switch an ENETRESET was issued. Reported and diff by Steffen Schuetz <st.sch@gmx.net>. Closes PR 5452. OK claudio@
2007-05-02	Eliminate unused inline functions and replace ahc_lock() which	Kenneth R Westerback
	resolves to s = splbio() and ahc_unlock() which resolves to splx(s) with splbio/splx calls. ok marco@ dlg@
2007-04-28	missing splbio/splx pair (which splasserted me per-device probe) in	Theo de Raadt
	polling mode code; ok dlg marco krw
2007-04-28	remove abstraction around splbio/splx; they just confuse things; ok dlg ↵	Theo de Raadt
	marco krw
2007-04-26	Improve debug statements. No binary change without AMI_DEBUG.	Alexander Bluhm
	ok marco
2007-04-23	Kill inlines to shave some bytes.	Artur Grabowski
	krw@
2007-04-22	disable debug output here too. it's working fine (even on strange archs)	David Gwynne

2007-04-20	Increase the delay while uploading a firmware buffer in malo_load_firmware().	Claudio Jeker
	We're doing DMA transfers without interrupts or some other indication and 100 microseconds is not enough on my X40 and so the firmware gets corrupted. Increasing it to 500 microseconds this should give us enough safety margin. OK mglocker@
2007-04-19	Cleanup phy selection code, simplifying the code considerably. Attach only	Mark Kettenis
	one phy on Sun ERI; gets rid of the duplicate phy on the blade1k. tested by many, ok drahn@
2007-04-19	make it compile again if LMDEBUG is defined;	Robert Nagy
	ok kettenis@
2007-04-17	The debug printf in malo_newstate() is too verbose so require debug level 2	Claudio Jeker
	to print it.
2007-04-16	During the drives detection only check the status register to be not	Alexander Yurchenko
	0xff, this will allow to detect some certain microdrives plugged into soekris. Patch from Jochen <jochen@penguin-breeder.org> via PR 5340.
2007-04-13	scsi_generic is now 16 bytes. Adjust padding so struct is still	Kenneth R Westerback
	256 bytes.
2007-04-13	scsi_generic is now 16 bytes. Adjust offsets into structure being	Kenneth R Westerback
	passed to controller so they point where expected. Fixes martin@'s alpha and sparc machines.
2007-04-13	The integrated Atheros NICs found in IBM/Lenovo ThinkPads use the same	Reyk Floeter
	device ID for old AR5212-based 32bit and new AR5424-based 64bit (PCI Express Mini Card) interfaces. Use an extra check to look if the card is 64bit and attach it as a single chip device. This prevents a panic when attaching the device on some laptops like the T60. Tested on a T42 (old AR5212 Mini PCI interface), Tested on a T60 (new AR5424 Mini Card interface) Nevertheless, the AR5424 does not work yet. Thanks to Stefan Konrath
2007-04-12	Oops, a code rearrangement meant that I didn't initialise the failed ccb	Christopher Pascoe
	pointer for problematic non-NCQ commands. spotted by and ok dlg@
2007-04-11	Remove debug leftovers to make console output quite again in hostap mode.	Marcus Glocker
	ok claudio@
2007-04-11	Cleanup makros even more. Remove unused ones and reduce the amount of	Claudio Jeker
	insane nesting. This driver is slowly getting to a sane level of makro usage. OK mglocker@
2007-04-11	As probe response mgmt frames are already handled by the firmware, ignore	Marcus Glocker
	the ones which come from the 802.11 stack. Seen in the DragonFly driver. ok claudio@
2007-04-10	``it's'' -> ``its'' when the grammar gods require this change.	Miod Vallat

2007-04-10	wrap reading of the error fis up into a func of its own, and use	David Gwynne
	bus_space_read_raw_region_4 to do it. more archs are happy with it.
2007-04-08	Change the API to sili_port_intr so it can timeout a command and reuse the	Christopher Pascoe
	existing reset/requeue infrastructure, and implement command timeouts. sili_port_intr now always processes all pending CCBs, so switch to use the auto-clearing interrupt status reads.
2007-04-08	Add support for NCQ error recovery and enable NCQ command submission.	Christopher Pascoe

2007-04-08	Track the order that CCBs are submitted in, so that after an error we are	Christopher Pascoe
	sure to reissue commands in the correct order. Defer completion of commands when we an error has occurred so that new commands don't go active before queeud ones.
2007-04-08	sili uses one queue for NCQ and legacy commands. No need for the AHCI-style	Christopher Pascoe
	indirection when determining which commands are complete.
2007-04-07	Always assume write commands had zero residual; the chip only counts	Christopher Pascoe
	received bytes.