Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
77,896 Commits 606 Branches 0 Tags 3.1 GiB
b54ea24924
Commit Graph

2777 Commits

Author SHA1 Message Date
pk
b54ea24924 Merge in4_cksum(). 2001-03-21 00:38:47 +00:00
chs
e533061fce in mem_access_fault*(), when invoking the pcb_onfault mechanism, pass the 
return value of uvm_fault() to the onfault routine via %o0.  users of
pcb_onfault now return this value to their callers rather than always
returning EFAULT. this allows i/o errors in VOP_GETPAGES() to be returned
back to read(), etc.
 2001-03-18 17:11:22 +00:00
mrg
3bb2b205f9 show the cpu a process is running on in proc 2001-03-18 14:24:10 +00:00
mrg
45d39a2030 redo previous slightly differently. 2001-03-18 14:10:34 +00:00
mrg
e82b5bb630 "mach slock" for simple_locks 2001-03-18 14:06:10 +00:00
mrg
0fcd32a415 "mach cpu" is MP only. 2001-03-17 01:54:17 +00:00
mrg
e44f962032 steal a few mach commands from sparc64. add a #if 0'ed "mach cpu" 2001-03-16 10:28:53 +00:00
mrg
48fc32e16e #ifdef DDB, add new function cpu_debug_dump(). define CPU_INFO_ITERATOR and 
CPU_INFO_FOREACH.
 2001-03-16 10:26:11 +00:00
chs
ac3bc537bd eliminate the KERN_* error codes in favor of the traditional E* codes. 
the mapping is:

KERN_SUCCESS			0
KERN_INVALID_ADDRESS		EFAULT
KERN_PROTECTION_FAILURE		EACCES
KERN_NO_SPACE			ENOMEM
KERN_INVALID_ARGUMENT		EINVAL
KERN_FAILURE			various, mostly turn into KASSERTs
KERN_RESOURCE_SHORTAGE		ENOMEM
KERN_NOT_RECEIVER		<unused>
KERN_NO_ACCESS			<unused>
KERN_PAGES_LOCKED		<unused>
 2001-03-15 06:10:32 +00:00
mrg
ef26546e15 lock the kernel around calling interrupt stubs. 2001-03-15 03:20:43 +00:00
mrg
122b3a0e67 sprinkle KERNEL_LOCK() and KERNEL_PROC_LOCK() where they're missing. add 
SYSCALL_DEBUG support to trap.c.
 2001-03-15 03:01:39 +00:00
tsutsui
43bd337868 Update for recent changes to the MI i82586 driver. 
(added 'why' argument to the channel attention hook function)
 2001-03-13 16:31:14 +00:00
pk
1b84e6fb33 pasto: sc_maddr => sc_mem 2001-03-09 10:19:54 +00:00
thorpej
5dedbd1c97 Duh, don't forget the proc argument. 2001-03-08 01:46:18 +00:00
thorpej
3423154d78 Use bus_dmamap_load(), not bus_dmamap_load_raw(). 2001-03-08 01:38:57 +00:00
thorpej
2c4c690f14 Add the BUS_DMA_STREAMING flag. 2001-03-07 22:42:16 +00:00
pk
a917eabaec * A couple of #if defined(SUN4M)' should be #if defined(MULTIPROCESSOR)' 
* Print notice if not all CPUs were attached
* Switch to SMP version of cache flush routines in just one place: in
  cpu_attach() after all known CPUs have been configured.
 2001-03-06 13:39:22 +00:00
pk
b91b8796d9 Implement MSGBUFSIZE option. 2001-03-05 16:46:09 +00:00
pk
8eeb7a8a79 mp_{pause,resume}_cpu(): avoid crash when cpu(s) not yet configured. 2001-03-05 16:45:22 +00:00
pk
f3c74253b0 Fix `flags' argument to pmap_enter(). 2001-03-05 16:43:34 +00:00
pk
90e5095df2 Fix syntax error in previous change. 2001-03-05 07:16:19 +00:00
pk
4e14c96788 In debug printf: missing quote on format string. 2001-03-05 07:04:01 +00:00
pk
a1b18f9435 Fix syntax error in previous. 2001-03-04 21:28:11 +00:00
pk
bac5be6461 Simplify `bad alias' handling in pv_link()/pv_unlink(): 
- there's no need to have different PV_NC flags for sun4[c] & sun4m

	- deal with kvm_uncache() being called on a managed page in a
	  separate helper function that deals with the caching state
	  within the pv list for that page, avoiding interference with
	  the `bad alias' removals in pv_unlink().

Note that currently never called for managed pages, but for now I'm
keeping the support in.
 2001-03-04 21:12:24 +00:00
mrg
b301668cb5 build sparc64 bootblk, ofwboot, & ofwboot.net on sparc 2001-03-04 15:34:45 +00:00
mrg
e85a9ee949 this is a sparc64 program. so -DSUN4U, machine -> sparc64. cleanup. 2001-03-04 14:50:05 +00:00
mrg
0a1cdb5f63 allow sparc "make release" to build a sparc64 capable 32 bit kernel. 2001-03-04 10:12:56 +00:00
pk
51668836b3 Remove unused declaration of `cpcb', which is no longer a global variable. 2001-03-03 19:42:38 +00:00
pk
d74da58795 alloc_cpuinfo: zero all the memory we allocate; not just the `cpuinfo' part. 
cpu_attach: initialize `curpcb->pcb_wim' for each additional CPU.
 2001-03-03 19:40:28 +00:00
pk
bb5f16e6e0 Sprinkle KERNEL_PROC_LOCK()s at the kernel entry points. 2001-03-03 19:11:02 +00:00
pk
bba5d36487 Initialize the IPI message locks. 2001-03-02 17:00:13 +00:00
pk
0548d12369 * proc_trampoline: call `proc_trampoline_mp()' #if MULTIPROCESSOR; 
currently necessary to acquire the `running in kernel mode' lock;
  pointed out by Bill Sommerfeld/Simon Gerraty.

* remove reference to `cpu_set_kpc()' from comments, as we don't have it.
 2001-03-02 10:27:00 +00:00
pk
e999fb6b77 bootstrap area: remove dead code + some trivial reformatting 2001-03-02 09:48:13 +00:00
pk
d98e8d5bca * Correct the `flags' argument in call to pmap_enter(). 
* Enable the message buffer at a temporary virtual address (KERNBASE)
  in bootstrap(), if that can be done.
 2001-03-01 16:14:25 +00:00
pk
813b8d79fd * Correct the `flags' argument in several calls to pmap_enter(). 
* message buffer initialisation:
	- change the buffer size to 8192 on all platforms
	- don't assume the buffer is always located at physical address 0
	- drop the SUN4 hack that left first half of the page unused;
	  I believe this is no longer necessary, since the msgbuf
	  data structure is duely sanity-checked before use in initmsgbuf().
 2001-03-01 16:09:25 +00:00
pk
6ae8b9ec6d Drop MSGBUFSIZE; we don't use it and it can't currently be configured 
to a different value anyway.
 2001-03-01 15:59:04 +00:00
pk
80ba31bd21 Drop the managed() macro; it is no longer used. 
State the intended use of the vm_{first,last,num}_phys variables that are
still exported from pmap.c.
 2001-03-01 15:56:47 +00:00
pk
aad0b0bcd2 Change pv list' management to use the hooks provided in the vm_physmem' 
structure. While this comes with the cost of having to search the
`vm_physmem' array every time need to find a PV entry corresponing to
some physical address, we gain the flexibility needed to support
arbitrary non-contiguous ranges of physical memory addresses.

Also, eliminate the need to sort the memory address ranges as presented
by the machine's PROM, and the requirement that physical memory starts
at address 0 (when possible).
 2001-03-01 15:52:18 +00:00
pk
b87a62e09a Make sure that e.g. myetheraddr() has a valid idprom pointer as soon as 
possible on sun4, since it may be used before the clock attaches.
 2001-02-28 14:45:23 +00:00
mjacob
312d560493 add SES device (unused #4) 2001-02-28 00:06:28 +00:00
pk
18026ec99d * Make PAGE_SIZE et.al. constants, if possible. 
* Add pvlist pointer to `struct pmap_physseg'
* Because of the above, add guard against multiple inclusions.
 2001-02-27 22:40:08 +00:00
cgd
8a986b2e96 convert to use getprogname() 2001-02-19 22:48:57 +00:00
jmc
106522ed47 Add -f option. If used it will perform the low level block operations on the 
file rather than requiring a raw device. Useful when dealing with vnd's
images and building snapshots. Also allows snapshots to be built in
securelevel > 0.
 2001-02-19 06:23:55 +00:00
pk
d850f6f9d2 Remove check for encoded iospace/nocache bits within physical addresses 
in places where they can't possibly occur as the PA is for a VM managed
page.
 2001-02-16 23:00:11 +00:00
pk
063b190de9 Hold off supplying the text/data gap to the VM manager; our pv table 
handling isn't adequate yet.
 2001-02-13 15:54:33 +00:00
pk
e547d3816d pmap_page_upload(): check if `etext_gap' is there, just in case. 2001-02-13 13:48:15 +00:00
pk
460bc0a0fc Recover the ~64KB gap in between the text/rodata & data segments that's being 
generated by the linker in sparc/ELF images. The physical pages in this gap
are unmapped from the kernel and given to the VM manager.
 2001-02-12 22:02:58 +00:00
pk
d713d112c9 Provide symbol `kernel_data_start' as the first address in the .data segment. 2001-02-12 21:56:54 +00:00
pk
203c10194b Supply a kernel link script that puts `etext' at the end of all 
read-only sections. Modelled after a similar change in the i386 port.
 2001-02-11 16:16:58 +00:00
pk
25a91d5c1e _bus_dma_valloc_skewed(): return a vaddr_t, not error codes; make missing 
boundary check return an error, so the caller can decide how to respond.
 2001-01-31 15:17:42 +00:00