yamt
175fe41cad
stack frames can be overwrapped for eg. softintr.
2002-06-26 08:18:01 +00:00
matt
1a5c0cf685
When not using the OLD pmap, bump kernel KVA space to 512MB (OLD pmap stays
...
at 256MB).
2002-06-26 01:16:22 +00:00
matt
7a12a9737c
Revamp how SR(s) are loaded on the user/kernel boundary. We now load all
...
16 SR registers when transitioning between kernel and user. Also, don't
reload the kernel SR(s) on every trap but only on traps from user space.
Instead of loading magic SRs for the kernel, load the kernel SRs from the
kernel_pmap_. This makes trap_subr.S completely ignorant of SR uses and
so they can change with having to change trap_subr.S. Also note that
since the user and kernel get complete SR sets, user VA space can now be
increased to 4GB if desired.
2002-06-26 01:14:45 +00:00
matt
7c77963b01
Set SR_PRKEY for user pmaps. For the kernel SR(s) set SR_SUKEY|SR_PRKEY.
...
Note that we never use a PTE PP of 0 or 1 (supervisor protection) so the
"key" is basically unused. However, use SR_PRKEY for user space is
conceptionally the right thing to do. Currently the kernel_pmap SR(s) are
ignored but that is going to be fixed shortly.
2002-06-26 01:10:20 +00:00
thorpej
a572f8f4a5
Report stray interrupts.
2002-06-26 01:06:44 +00:00
drochner
a548b203d4
use PCI_ID_CODE instead of a local macro (cosmetics)
2002-06-25 21:18:32 +00:00
thorpej
7704072be3
Correct a comment.
2002-06-25 19:41:08 +00:00
thorpej
fea38885e8
* Interrupt status is in cp13.4, not cp13.1 (D'oh!)
...
* Fix an inverted test.
2002-06-25 19:40:46 +00:00
thorpej
f2bff71e47
Interrupt steering register is cp13.8, not cp13.2 (D'oh!).
2002-06-25 19:39:51 +00:00
eeh
debc8b26aa
Fix ktrace call for 32-bit kernels.
2002-06-25 17:37:03 +00:00
lukem
214a01678d
- replace all "scsibus* at XXXX" with a single "scsibus* at scsi?"
...
- replace all "atapibus* at XXXX" with a single "atapibus* at atapi?"
- replace all "audio* at XXXX" with a single "audio* at audiobus?"
- replace all "midi* at XXXX" (except "midi* at pcppi?") with a single
"midi* at midibus?"
2002-06-25 03:12:04 +00:00
thorpej
e09f5065d0
During signal delivery, arrange to have the signal handler invoked
...
directly, using the trampoline only for the return path.
2002-06-25 01:24:49 +00:00
itojun
4aac356c0e
fwnode* has idhi/idlo locator
2002-06-25 00:32:24 +00:00
itojun
92f9b53eea
sync firewire scsi config to the latest reality
2002-06-25 00:29:20 +00:00
nathanw
4ce40595e2
Pass through the PCI configuration ioctls, like vga_pci.c does.
2002-06-24 21:08:37 +00:00
eeh
691c4d1179
Fix compilation problems in DEBUG code.
2002-06-24 18:48:13 +00:00
lukem
e3b23eaafd
Switch from pccons to wscons, so that USB keyboards have a chance of
...
working. From Grant Beattie <grant@netbsd.org>, with changes by me.
2002-06-24 16:14:13 +00:00
mrg
383ef0f9dc
make booting from FS_RAID work. tested on vs4000/90.
2002-06-24 14:53:16 +00:00
itojun
9f911945d5
handle IDT-less case just as linux does
2002-06-24 10:10:17 +00:00
itojun
fe3778040b
tabify
2002-06-24 10:10:16 +00:00
itojun
241389f5d8
make it at least compile/link - copy change to i386/locore.s (1.255 -> 1.256)
...
not sure if it is correct
2002-06-24 10:05:54 +00:00
pk
cc3f5791ea
Comments update: indicate which counters are being incremented.
2002-06-24 09:26:47 +00:00
pk
0e8ea4f752
`fdcio_intrcnt.ev_count' is a 64-bit variable.
2002-06-24 09:24:37 +00:00
jdolecek
afb66555e4
make vectors.o explicitly depend on assym.h, so that 'make depend' step
...
is optional
2002-06-24 07:15:24 +00:00
simonb
8e38e28f7f
White space nits: space after a comma.
2002-06-24 05:52:28 +00:00
simonb
d4068eac65
Add 64MB and 256MB tlb page masks.
2002-06-24 05:46:47 +00:00
thorpej
f427b80c27
Add ENTRY_NOPROFILE().
2002-06-24 01:13:34 +00:00
thorpej
f89b062ae7
During native signal delivery (and iBCS2, since it uses the native
...
sigcode), arrange to have the signal handler invoked directly, using
the trampoline only for the return path, saving a call insn. Play
some other stack trickery in the trampoline to turn 2 pushl's into
one movl.
2002-06-23 22:18:49 +00:00
thorpej
4c0113def4
Swap the order of the sigcontext and hardware state in the sigframe.
2002-06-23 21:31:12 +00:00
manu
d0c5097f05
Typo
2002-06-23 20:36:36 +00:00
thorpej
9cb2f482f0
When delivering a signal, arrange to have the handler invoked directly,
...
using the trampoline only for the return. This saves two instructions
in the trampoline, one of them being a branch.
2002-06-23 19:16:43 +00:00
thorpej
7f7f0f86bd
When delivering a signal, arrange for the handler to be invoked
...
directly. The trampoline is now used only for the return, and
makes it another 2 instructions shorter.
2002-06-23 18:49:33 +00:00
thorpej
508e2dcf41
When delivering a signal, don't push the signum, code, and context pointer,
...
or handler onto the stack. Instead, just stuff them into the correct
argument registers (handler is a "4th arg").
2002-06-23 18:35:05 +00:00
thorpej
aaf6e7902d
Add ENTRY_NOPROFILE().
2002-06-23 17:26:58 +00:00
perry
90062e3f5d
Remove (completely unused) definition of CLK_TCK. Nothing in the
...
kernel uses it, it is protected by _KERNEL, and it conflicts with the
definition in userland.
2002-06-23 17:07:58 +00:00
perry
fc9464af40
remove definition of CLOCKS_PER_SEC protected by #ifdef KERNEL. It is
...
used nowhere, and KERNEL is defined nowhere (vs. _KERNEL).
2002-06-23 16:54:00 +00:00
mrg
9dcb300af2
finish previous; it would be nice if people were to use the lovely
...
cross compiling mechanisms we have when making sweeping changes...
2002-06-23 03:00:19 +00:00
thorpej
43e7ad972b
Garbage-collect sigframe references.
2002-06-23 00:16:59 +00:00
thorpej
31404c3f2e
When delivering a signal, there is no need to push the signal number,
...
code, context pointer, or handler onto the stack, so don't do so.
2002-06-23 00:16:20 +00:00
fvdl
73f40b9030
Add bge@pci attachment.
2002-06-22 23:13:37 +00:00
chris
5f8eb07bb7
Back out last change, following jason's advice, it's lsof that's actually
...
broken not the headers.
2002-06-22 20:39:18 +00:00
toshii
01b6ded4f1
Comment out MEMORY_DISK_HOOKS as a workarund for the following problem.
...
hpcarm has a bug which prevents booting if kernel text+data size exceeds
3.75MB.
2002-06-22 13:04:00 +00:00
isaki
cfde97e2fe
Fix compilation error when scsibus* doesn't exist.
2002-06-22 10:59:54 +00:00
chris
6052d38107
Allow userland to see machine/vmparam.h. This allows ntop to compile, as it uses the struct uvm from uvm.h, which need the VM_NFREELIST definition.
...
Of course this does mean that ntop could potentially be platform specific, ie different versions for cats, acorn32 etc as struct uvm maybe a different size, but running on the same cpu architecture.
2002-06-22 10:15:30 +00:00
ross
789df24a75
a "no" token was added recently, so we must now quote any "no" symbols
...
no => "no"
2002-06-22 00:01:10 +00:00
gmcgarry
02ca40e27f
Make this compile with NETIF_DEBUG.
2002-06-21 23:18:45 +00:00
ross
f46b777e48
ca* at cac? ==> ld* at cac?
2002-06-21 22:57:55 +00:00
thorpej
3d7cdd4ec5
Add support for auxillary busses on the Intel 82452 PCI-Host
...
bridge. PR 17353, from MOCHIDA Shuji.
2002-06-21 16:03:33 +00:00
lukem
e78903dcee
add "SUNW,hme" as an an alias for "hme".
...
my ss5 that netboots off an sbus swift card now autodetects the root device.
2002-06-21 14:38:36 +00:00
eeh
73fd336386
Use paddr_t for avail_start and avail_end so we can handle machines with
...
RAM above the 2GB mark.
2002-06-21 02:57:29 +00:00