Greg Lehey [Tue, 24 Aug 1999 02:20:30 +0000 (02:20 +0000)]
give_sd_to_drive: If we don't have space, deallocate the subdisk.
Previously we could end up with a valid subdisk
entry pointing to offset -1 on the drive.
Bruce Evans [Tue, 24 Aug 1999 00:48:19 +0000 (00:48 +0000)]
Cast pointers to [u]intptr_t instead of casting them to [u_]long. Don't
depend on gcc's feature of casting lvalues, especially for direct
assignment where it doesn't even simplify the syntax. Cosmetic.
Bruce Evans [Tue, 24 Aug 1999 00:38:52 +0000 (00:38 +0000)]
Cast pointers to uintptr_t instead of casting them to u_long. They
are still converted to u_long by assignment of the uintptr_t, and
address calculations are still done using u_long. This is OK for
currently supported machines, but addresses should be represented
by vm_offset_t or uintptr_t in case pointers are longer than longs.
"Fixed" size of linker_path[]. MAXPATHLEN + 1 was 1 too large for
search paths with only one file path in them, but much too small
for search paths with several long file paths in them.
Now that we can bind cdevsw to the individual dev_t, divorce the PERFMON
stuff from mem.c. If PERFMON is there, it will "steal" a minor from
mem.c, but mem.c doesn't need to know about this.
Bruce Evans [Mon, 23 Aug 1999 21:07:13 +0000 (21:07 +0000)]
Initialise fsids with (user) device numbers again. Bitrot when dev_t's
were changed to pointers was obscured by casting dev_t's to longs.
fsids haven't even been comprised of longs since the Lite2 merge.
Peter Wemm [Mon, 23 Aug 1999 19:23:33 +0000 (19:23 +0000)]
The nexus_attach() code works a lot better if it's actually connected to
the device methods... Also, don't fail to add eisa/isa because a previous
device failed to attach.
Alan Cox [Mon, 23 Aug 1999 18:08:34 +0000 (18:08 +0000)]
struct vm_map:
The lock structure cannot be the first element of the vm_map
because this can result in livelock between two or more system
processes trying to kmem_alloc_wait.
Nick Hibma [Mon, 23 Aug 1999 14:49:22 +0000 (14:49 +0000)]
Recognise the GPS receiver from Trimble
in NMEA mode (without date).
Trimble OEM Ace-II receiver. Low cost PCB with single
voltage input, external active antenna and two serial
ports with either NMEA and ITAPs output. Programmable
to be tuned for 'time' accuracy in fixed station config.
Author of tool notified.
Submitted by: Dirk van Gulik <dirk.vangulik@jrc.it>
Peter Wemm [Mon, 23 Aug 1999 12:59:03 +0000 (12:59 +0000)]
Fix a few problems on the Alpha, and fix -Wall while here.
Before:
isab0@pci0:8:0: class=0x060100 card=0x00000000 chip=0xffffffffc6931080 ...
After:
isab0@pci0:8:0: class=0x060100 card=0x00000000 chip=0xc6931080 ...
Doug White [Sun, 22 Aug 1999 21:45:24 +0000 (21:45 +0000)]
Major PicoBSD update by luigi with some debugging help from myself.
. Now builds on -STABLE (-CURRENT is broken due to bugs)
. etc directory contents centralized instead of in each
type directory (can exclude & override as desired)
. Removed extraneous language files (lang files for rc really necessary?)
. dialog-based build tool with support for custom floppy builds
. MFS image loads as a mfs_root module instead of compiled into kernel
THIS IS BROKEN ON CURRENT. I'll MFC to -STABLE immediately following.
luigi tells me I have an OK from jkh on the MFC.
Peter Wemm [Sun, 22 Aug 1999 19:59:29 +0000 (19:59 +0000)]
Zap a heap of unused cruft now. We don't need the ISA/EISA/PCI hooks
here any more as they are self identifying. Only PNP remains but that
will be replaced any day now.
Also reword a comment that had been XXX'ed to death to make it clear[er]
why we don't enable interrupts before probing.
PCIBIOS interrupt routing controls may make this possible to fix one day.
Peter Wemm [Sun, 22 Aug 1999 19:56:55 +0000 (19:56 +0000)]
Take advantage of the apm/npx code and let them identify themselves rather
than having explicit hooks here.
Treat the eisa/isa attach a little differently so that we defer the
decision about to attach eisa/isa to the motherboard directly only if
the PCI probe (if it exists) fails to turn up a PCI->EISA/ISA bridge.
This restores the original device geometry where ISA and/or EISA attach
to their bridge rather than bypassing and going to the root.
Peter Wemm [Sun, 22 Aug 1999 19:54:11 +0000 (19:54 +0000)]
Make the identify routine add itself with priority 100 to make sure it
goes after the npx/apm devices and any other motherboard devices that
may get added down the track.
John Polstra [Sun, 22 Aug 1999 16:50:30 +0000 (16:50 +0000)]
Go back to using microtime() to get the timestamps for {f,l,}utimes(path,
NULL) for now. Bruce says I jumped the gun with my change in
revision 1.131, or maybe it should use nanotime(), or maybe it
shouldn't be decided in the VFS layer at all. I'm leaving it with
the old behavior until the Trans-Pacific Internet Vulcan Mind Meld
yields fuller understanding.
Bill Paul [Sun, 22 Aug 1999 16:34:52 +0000 (16:34 +0000)]
Mention that the 3Com 3cSOHO100-TX adapter (cheap version of the 3c905B) is
supported, as well as the 3c980B server adapter (newer rev of the 3c980,
same PCI ID) and the 3c905C-TX.
Nick Hibma [Sun, 22 Aug 1999 16:21:29 +0000 (16:21 +0000)]
1) Add HEADER section. Similar to CODE.
2) deindent the two sections (CODE and HEADER), based on the indent found
on the first line of the section; to make the output more readable.
Bill Paul [Sun, 22 Aug 1999 15:34:47 +0000 (15:34 +0000)]
Modify yp_mkdb so that it only checks for bogus '+' and '-' characters
in source input if the -f flag is used, and modify Makefile.yp to only
use -f for the passwd, master.passwd and group maps. These should be
the only ones for which the + and - characters have special meaning
that make it important for us to avoid letting them into any of the map
databases. In some cases (namely the automounter maps) we have to allow
at least the - character through in order to create the map properly.
Mitsuru IWASAKI [Sun, 22 Aug 1999 14:48:00 +0000 (14:48 +0000)]
Fix `key release event prevent suspend' problem. We don't need
`sleep 1; zzz' trick now.
- APM BIOS Call for suspend/standby now should be issued with delay.
- Delay for suspend/standby can be adjusted by using sysctl(8) interface
(eg. sysctl -w machdep.apm_suspend_delay=3).
Kazutaka YOKOTA [Sun, 22 Aug 1999 09:52:33 +0000 (09:52 +0000)]
- Remove cdevsw entry points in individual keyboard drivers;
instead, use generic entry points for all drivers.
- Eliminate bogus makedev().
- Eliminate softc in the lower drivers, as it is no longer necessary.
Kazutaka YOKOTA [Sun, 22 Aug 1999 06:12:58 +0000 (06:12 +0000)]
Revise description on moused options:
- Clearly distinguish PS/2, bus, and serial protocols.
- Explicitly state that serial mouse protocols don't work with
the PS/2 and bus mice.
Alan Cox [Sun, 22 Aug 1999 05:37:18 +0000 (05:37 +0000)]
Make "s_unlock" an inline function. (Inlining this function takes
less space than calling it. A callable version still exists for
use by some assembly code.)
Kazutaka YOKOTA [Sun, 22 Aug 1999 03:31:13 +0000 (03:31 +0000)]
Assorted bug fixes.
keyboard.c
- Call tcsetattr() in VGLKeyboardEnd() to restore tty, only when
tty attributes have been previously saved.
PR: misc/9524
Submitted by: Katusyuki 'kei' Maeda (kei@nanet.co.jp)
- Set up the tty raw mode correctly.
main.c
- Restore VESA_800x600 raster text mode correctly in VGLEnd().
Submitted by: des
text.c
- Allocate the correct size of a font buffer in VGLSetFontFile().
I forgot the submitter ;-(
simple.c, bitmap.c
- Fix address calculation for the VGA mode X in VGLGetXY() and
VGLBitmapCopy().
- Fix typo (dsty -> dstx) in __VGLBitmapCopy().
John Polstra [Sun, 22 Aug 1999 01:46:57 +0000 (01:46 +0000)]
Use the new vfs_timestamp() function to create the timestamps used
by utimes(path, NULL). This gives them the same precision as the
timestamps produced by write operations. Do likewise for lutimes()
and futimes().
John Polstra [Sun, 22 Aug 1999 00:15:16 +0000 (00:15 +0000)]
Support full-precision file timestamps. Until now, only the seconds
have been maintained, and that is still the default. A new sysctl
variable "vfs.timestamp_precision" can be used to enable higher
levels of precision:
0 = seconds only; nanoseconds zeroed (default).
1 = seconds and nanoseconds, accurate within 1/HZ.
2 = seconds and nanoseconds, truncated to microseconds.
>=3 = seconds and nanoseconds, maximum precision.
Level 1 uses getnanotime(), which is fast but can be wrong by up
to 1/HZ. Level 2 uses microtime(). It might be desirable for
consistency with utimes() and friends, which take timeval structures
rather than timespecs. Level 3 uses nanotime() for the higest
precision.
I benchmarked levels 0, 1, and 3 by copying a 550 MB tree with
"cpio -pdu". There was almost negligible difference in the system
times -- much less than 1%, and less than the variation among
multiple runs at the same level. Bruce Evans dreamed up a torture
test involving 1-byte reads with intervening fstat() calls, but
the cpio test seems more realistic to me.
This feature is currently implemented only for the UFS (FFS and
MFS) filesystems. But I think it should be easy to support it in
the others as well.
An earlier version of this was reviewed by Bruce. He's not to
blame for any breakage I've introduced since then.
Mike Smith [Sat, 21 Aug 1999 22:10:49 +0000 (22:10 +0000)]
Clean up after removing sys/eventhandler.h from sys/systm.h at the last
minute. This should cover all of the missed cases (and should let LINT
build again).