src - OpenBSD base system

Age	Commit message (Collapse)	Author
2011-01-10	Only use MMCFG extended PCIe config space. Gets us back the on-CPU PCI ↵	Mark Kettenis
	devices on AMD Family 0Fh processors.
2011-01-09	Check in resettodr() if inittodr() has been called, otherwise we end up	Jasper Lievisse Adriaanse
	resetting the clock when we don't need to. Found out with booting hppa64 kernels, and the problem also exists on hppa when booting with '-a' and hitting 'exit' when asked for the root filesystem. help & ok jsing@ also ok kettenis@ (who suggested naming the variable like amd64/i386 to prevent creating yet another variant of this code)
2011-01-09	Only use memory mapped extended PCIe config space access for the busses	Mark Kettenis
	advertised in the MCFG table, and fall back on the traditional method for other busses. Fixes issue reported by henning@.
2011-01-08	forward decl struct proc	Theo de Raadt

2011-01-08	rename imask[] to cpu_imask[] to because imask is too loud as far as	Theo de Raadt
	kernel namespace ok miod
2011-01-08	The Ubiquiti WiFiStation EXT works on the Sun Blade 1500.	Damien Bergamini
	Enable athn at uhub on all USB 2.0 capable arches.
2011-01-07	do not special case ep0 at pci/eisa. ep* will be fine. We should not	Theo de Raadt
	use 'foo0 at smartbus?' unless we need to ok miod
2011-01-07	increase i386 maxdsiz to 2GB. comments from ariane deraadt kettenis.	Ted Unangst
	ok deraadt
2011-01-07	Detect wether a disk image is an ISO image and set the media type to	Mark Kettenis
	VD_MEDIA_TYPE_CD to vDisk 1.1 clients. Makes installing from a Solaris 11 Express bootable install CD work.
2011-01-07	Fix malloc(9) call with swapped type and flags arguments.	Mark Kettenis

2011-01-06	Add support for the vDisk 1.1 protocol. Improve Solaris vtoc/disklabel	Mark Kettenis
	support and implement VD_OP_SET_VTOC. This makes Solaris actualy run on top of an OpenBSD control domain. Installing from CD needs some further tweaking but a network install will probably work. OpenBSD clients are still happy.
2011-01-06	Solaris insists on getting the right sequence numbers in DRING_DATA messages	Mark Kettenis
	and it seems we were off by one.
2011-01-06	enable athn at uhub	Damien Bergamini

2011-01-05	Pass pmap_bootstrap() the memory range used by the PROM/BUG/firmware/whatever,	Miod Vallat
	instead of assuming it is ``everything below the kernel image''.
2011-01-05	Minor optimizations:	Miod Vallat
	- move pte validity checks from pmap_remove_pte() to its callers. - when iterating over a segment, compute pte address once and increment it in the loop, instead of recomputing it on every iteration.
2011-01-05	Make copypage() and zeropage() per-cpu function pointers, and use a	Miod Vallat
	different version on 88110, which does load allocate of to-be-completely-overwritten cache lines.
2011-01-05	Now that pmap_copy_page() no longer needs to flush a couple contiguous tlb	Miod Vallat
	entries, drop the count parameter to cmmu_tlb_inv(), and introduce cmmu_tlb_inv_all() to drop all user tlb entries (to be used during context switches).
2011-01-05	- remove unused headers wrapped in #if defined(DDB)	Jasper Lievisse Adriaanse
	ok jsing@ kettenis@
2011-01-04	Support machdep.lidsuspend support in apm(4) land, at least on thinkpads	Theo de Raadt
	which have aps(4). You can now turn off the BIOS option for lid suspends, and control it with sysctl.
2011-01-04	Add support for Memory Mapped Configuration space access. This gives us	Mark Kettenis
	access to PCIe extended configuration space access on modern i386 and amd64 machines.
2011-01-04	Use pmap_uncache_page() to alter cacheability of pages in bus_dmamem_map()	Miod Vallat
	instead of playing with pte bits directly; this will cause cacheability to be restored eventually; makes the zaurus textmode console memory cached again after exiting from X. ok drahn@
2011-01-04	Introduce pmap_uncache_page(), created from the guts of pmap_kenter_cache()	Miod Vallat
	to make a page uncached and maintain this both at the pte and pv list level, and make pmap_kenter_cache() use it. ok drahn@
2011-01-04	It is actually possible to run CoolThreads servers with a graphics console,	Mark Kettenis
	so don't assume vcons(4) always is the console.
2011-01-04	XVR-300 is boring. For starters it has a very dull looking OpenPROM logo.	Mark Kettenis
	And radeonfb(4) just works on it.
2011-01-04	- de-register. no binary change.	Jasper Lievisse Adriaanse
	ok jsing@
2011-01-03	remove unused file	Alexandr Shadchin
	ok krw, miod
2011-01-03	Rework backtrace code, and try to cope with variadic functions better	Miod Vallat
	(those play nasty games with the stack pointer).
2011-01-02	More changes to pacify the Solaris virtual disk client. Implement	Mark Kettenis
	VD_OP_GET_VTOC and VD_OP_GET_DISKGEOM. Factor out the code to ACK descriptors and fix a bug copying in/out a descriptor that crosses a page boundary.
2011-01-02	Make need_resched() call cpu_unidle() such that we wake up a process running on	Mark Kettenis
	another CPU. ok miod@
2011-01-02	Figure out how much L2 cache is available on 88410-blessed designs, and	Miod Vallat
	report it in dmesg.
2011-01-02	Enable udl(4) on hppa, together with rasops and rawkbd for Xorg.	Matthieu Herrb
	Xorg is not yet built in xenocara, this will come later. ok miod@
2011-01-02	Kill pmap_table_t typedef.	Miod Vallat

2011-01-02	Fix dump to work with PMAP_DIRECT, as done on mvme88k.	Miod Vallat

2011-01-02	Only store bus/dev/func numbers when the corresponding pointer argument	Jasper Lievisse Adriaanse
	isn't NULL. Original to hppa's elroy(4) by, and ok kettenis@
2011-01-02	- move resettodr and inittodr to clock.c, like the other clock related functions	Jasper Lievisse Adriaanse

2011-01-02	We need db_machdep.h even when TRAPDEBUG is enabled.	Joel Sing

2011-01-02	Include files.bluetooth so that hppa64 compiles.	Joel Sing

2011-01-02	Move away from using fixed-sized descriptors and fill in some missing bits	Mark Kettenis
	in the messages that ACK descriptors. Makes OpenSolaris virtual disk clients much happier and able to boot from a bootable automatic installer cdrom.
2011-01-02	Use the session ID passed to use by the virtual disk client, and pass back	Mark Kettenis
	a bit more information to the client during the attribute negotiation phase. Gets us past the attribute negotiation phase against an OpenSolaris virtual disk client.
2011-01-02	Declare ourselves to be a virtual disk server during the version negotiation	Mark Kettenis
	phase. Gets us past the version negotiation against an OpenSolaris virtual disk client.
2011-01-01	Now that we __HAVE_PMAP_DIRECT, it gets easy to simply save partial cache	Miod Vallat
	lines and restore them after invalidating rounded-to-cacheline-boundary ranges.
2011-01-01	Better errata #17 handling logic. Only concerns MULTIPROCESSOR kernels.	Miod Vallat

2011-01-01	Corretctly return 0xffffffff for the returned LBA if the disk is too large	Mark Kettenis
	for the READ CAPACITY command.
2011-01-01	Implement READ CAPACITY (16).	Mark Kettenis

2011-01-01	Move setting cpu_hzticks to the clock initialization function, like	Jasper Lievisse Adriaanse
	other ports (e.g. hppa64) do. ok jsing@ kettenis@
2011-01-01	- ansify	Jasper Lievisse Adriaanse
	ok jsing@ kettenis@
2011-01-01	Fix horribly broken softintr_establish(). The old code was not managing the	Mark Kettenis
	linked list of handlers properly. Fixes issues with usb(4) reported by matthieu@. tested by matthieu@, ok jsing@
2011-01-01	Only store bus/dev/func numbers when the corresponding pointer argument	Mark Kettenis
	isn't NULL.
2011-01-01	Only store bus/dev/func numbers when the corresponding pointer argument	Mark Kettenis
	isn't NULL.
2010-12-31	Massive overhauling of the m88k pmap, though I can't pretend it's a new pmap	Miod Vallat
	since a large part of the structures and logic remains. Since m88k has separate supervisor/user spaces, we can map physical memory 1:1 in supervisor space, and have the kernel virtual address space start from the end of physical memory. This allows us to switch to __HAVE_PMAP_DIRECT. And to get rid of the double mapped sdt, since now their virtual and physical addresses will always match. The upper bound of the kernel virtual memory space is now platform dependent, until the code which relies upon some hardware devices being mapped 1:1 in supervisor mode is updated to no longer require this (this is mainly a PITA on luna88k, where onboard devices start at 0x40000000, leaving only 1GB of KVA at the moment - still much better than the previous 512MB). Tested on mvme88k only (187, 188, 197LE, 197DP). Other platforms ought to work, aviion will be checked shortly and fixed if necessary. No known OpenBSD/luna88k system in working condition at the moment.