]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoPrint the correct v9 opcodes.
Jake Burkholder [Sat, 29 Dec 2001 08:52:50 +0000 (08:52 +0000)]
Print the correct v9 opcodes.

Submitted by: tmm

22 years agoDon't concatenate __func__.
Jake Burkholder [Sat, 29 Dec 2001 08:49:28 +0000 (08:49 +0000)]
Don't concatenate __func__.

22 years agosparc64 needs the same alingment fixes that alpha and ia64 do.
Jake Burkholder [Sat, 29 Dec 2001 08:45:17 +0000 (08:45 +0000)]
sparc64 needs the same alingment fixes that alpha and ia64 do.

Submitted by: tmm

22 years agosparc64 needs the same alignment fixes that ia64 and alpha need.
Jake Burkholder [Sat, 29 Dec 2001 08:44:08 +0000 (08:44 +0000)]
sparc64 needs the same alignment fixes that ia64 and alpha need.

Submitted by: tmm

22 years agoChange traces in hardclock and statclock to use the KTR_CLK trace
Jake Burkholder [Sat, 29 Dec 2001 08:39:57 +0000 (08:39 +0000)]
Change traces in hardclock and statclock to use the KTR_CLK trace
facility, rather than KTR_INTR.

22 years agoUpdate to new constants.
Jake Burkholder [Sat, 29 Dec 2001 08:37:58 +0000 (08:37 +0000)]
Update to new constants.

22 years agoMake cont in ddb work.
Jake Burkholder [Sat, 29 Dec 2001 08:37:35 +0000 (08:37 +0000)]
Make cont in ddb work.

22 years agoPrototype dcache_inval_phys.
Jake Burkholder [Sat, 29 Dec 2001 08:35:49 +0000 (08:35 +0000)]
Prototype dcache_inval_phys.

Submitted by: tmm

22 years agoForced commit to note that the wierd soft bit is actually a wired bit.
Jake Burkholder [Sat, 29 Dec 2001 08:33:53 +0000 (08:33 +0000)]
Forced commit to note that the wierd soft bit is actually a wired bit.

22 years agoAdd comments as to why VM_MAXUSER_ADDRESS is magic (abitrary).
Jake Burkholder [Sat, 29 Dec 2001 08:25:43 +0000 (08:25 +0000)]
Add comments as to why VM_MAXUSER_ADDRESS is magic (abitrary).
Define the KVA_RANGE in terms of ttes, not sttes.
Remove UPT_MIN_ADDRESS.  We no longer use a hard coded address for
the user tsb.

22 years agoMake tte bit constants explicitly unsigned and long.
Jake Burkholder [Sat, 29 Dec 2001 08:24:02 +0000 (08:24 +0000)]
Make tte bit constants explicitly unsigned and long.
Add a wierd soft bit.
Remove struct stte.

22 years agoAdd definitions for dcache color bits, which may move to cache.h.
Jake Burkholder [Sat, 29 Dec 2001 08:19:24 +0000 (08:19 +0000)]
Add definitions for dcache color bits, which may move to cache.h.
Add fields to md_page for tracking virtual page color, and pv entry
lists.
Fix pmap_track_modified to work for non-kernel pmaps.  This is due to
kernel virtual addresses potentially overlapping with userland addresses.

22 years agoImplement pv entries as separate structures from the ttes, like other
Jake Burkholder [Sat, 29 Dec 2001 08:17:16 +0000 (08:17 +0000)]
Implement pv entries as separate structures from the ttes, like other
architectures do.

22 years agoRemove support for multi level tsbs, making this code much simpler and
Jake Burkholder [Sat, 29 Dec 2001 08:15:43 +0000 (08:15 +0000)]
Remove support for multi level tsbs, making this code much simpler and
much less magic, fragile, broken.  Use ttes rather than sttes.
We still use the replacement scheme used by the original code, which
is pretty cool.

Many crucial bug fixes from: tmm

22 years agoImplement pv entries as separate structures from the ttes, as other
Jake Burkholder [Sat, 29 Dec 2001 08:12:22 +0000 (08:12 +0000)]
Implement pv entries as separate structures from the ttes, as other
architectures do.

Many crucial bug fixes from: tmm

22 years agoGreat pmap rewrite to use a much simpler one level tsb of ttes, instead
Jake Burkholder [Sat, 29 Dec 2001 08:11:02 +0000 (08:11 +0000)]
Great pmap rewrite to use a much simpler one level tsb of ttes, instead
of sttes.  Also removes many differences between this and the other pmaps.
Reserve the kva space used by the openfirmware translations.
Use physical addresses directly in pmap_zero_page and pmap_copy_page, now
that we have the cache line shooting support.
Add code to track the virtual cachability of mapped pages.  The dmmu
requires that multiple mappings of the same phsyical address have the
save virtual address bits up to a colour boundary.  Violating this
requires all mappings to be mapped uncacheable.  We do not yet handle
the case of a badly aliased mapping becoming cachable again.

Many crucial bug fixes from: tmm

22 years agoUse fprs to track floating register usage. Clear it once we've saved
Jake Burkholder [Sat, 29 Dec 2001 08:03:15 +0000 (08:03 +0000)]
Use fprs to track floating register usage.  Clear it once we've saved
the registers so we don't uselessly save them over and over again for
each context switch until another floating point instruction is executed.
Use a non-specific tlb slot for the tsb, which needs to have a locked
entry.
Remove overly verbose traces.

22 years agoAdd .register directives for gcc3.
Jake Burkholder [Sat, 29 Dec 2001 07:59:04 +0000 (07:59 +0000)]
Add .register directives for gcc3.
Add macros to atomically increment an integer variable in the data
section and to atomically set a bit in a tte.  Note that the latter
does not return the new value.
Rewrite RESUME_SPILLFILL_MAGIC to use more sensical calculations, and
to preserve all alternate globals religiously.  Must now be called on
alternate globals.
Defer switching to the kernel stack until inside the syscall, trap,
interrupt wrappers.  Splitting the windows is all that's really urgent.
Adapt to new trap types.
Add %xcc where appropriate in order to not use v8 opcodes inadvertantly
(which work fine).
Modify the low level tlb fault handlers to operate on a tsb made up of
ttes, not sttes.  This effectively makes the tsb twice as large.
After atomically updating tte bits in memory, also set the bit in the
register that holds the data which will be loaded into the tlb.  The
macro returns the old value.
Use the preloaded mmu global which holds the address of the current
user tsb.
Add back a low level protection fault handler instead of just punting
into the vm system.  This effectively saves a soft fault per COW fault.
Add a trace to intr_enqueue.
Pass arguments to the trap, interrupt, syscall wrappers in the out
registers instead of some on the stack, some in registers.
Use the preloaded alternate global pcb register.

22 years agoAdd needed include of fsr.h.
Jake Burkholder [Sat, 29 Dec 2001 07:47:33 +0000 (07:47 +0000)]
Add needed include of fsr.h.
Use fprs to track floating point usage.
Remove misguided comment.
Clear fprs in a child process's new trapframe.

22 years ago1. Adapt to new trap types.
Jake Burkholder [Sat, 29 Dec 2001 07:45:49 +0000 (07:45 +0000)]
1. Adapt to new trap types.
2. Make trap_pfault more like it is on other architectures.
3. Fix a bug in syscall() which caused system calls with more than
   six arguments that are called through the wild card syscall to
   have their arguments scrambled.  This affected mmap due to the
   (bogus) wrapper in libc.

Submitted by: tmm (3)

22 years agoAdd .register directives for gcc3.
Jake Burkholder [Sat, 29 Dec 2001 07:34:42 +0000 (07:34 +0000)]
Add .register directives for gcc3.
Add some traces that can be useful but are also very loud.
Use defines for offsets into jmpbuf instead of magic numbers.
Fix a style bug.
Fixup comments.

22 years agoAssert at compile time that structures which need to be a power of
Jake Burkholder [Sat, 29 Dec 2001 07:32:56 +0000 (07:32 +0000)]
Assert at compile time that structures which need to be a power of
2 really are.
Move a trace to before flushw in case it goes off the deep end.

22 years agoRemove a debug printf.
Jake Burkholder [Sat, 29 Dec 2001 07:28:20 +0000 (07:28 +0000)]
Remove a debug printf.
Setup new dedicated global register (alternate, and mmu).
Make setregs readable again.
Adapt to moving of many things into trapframe.

22 years agointr_handlers is an array of function pointers, not small structures.
Jake Burkholder [Sat, 29 Dec 2001 07:24:57 +0000 (07:24 +0000)]
intr_handlers is an array of function pointers, not small structures.
Assert at compile time that structures which need to be a power of 2
in size really are.

22 years agoAdd .register directive for gcc3.
Jake Burkholder [Sat, 29 Dec 2001 07:22:27 +0000 (07:22 +0000)]
Add .register directive for gcc3.
Enforce in hardware the non-use of floating point in the kernel.

22 years ago1. Make this more name space friendly (* -> emul_*).
Jake Burkholder [Sat, 29 Dec 2001 07:20:20 +0000 (07:20 +0000)]
1. Make this more name space friendly (* -> emul_*).
2. Don't whine about unaligned fixups by default.
3. Adapt to removal of mmuframe.

Submitted by: tmm (1, 2)

22 years agoBe paranoid about the sizeof passed to db_get_value, not all trapframe
Jake Burkholder [Sat, 29 Dec 2001 07:16:45 +0000 (07:16 +0000)]
Be paranoid about the sizeof passed to db_get_value, not all trapframe
fields are u_long.
Print useful information about traps in backtraces.
GC some dead code.

22 years agoImplement dcache_inval_phys, which shoots the cache lines that correspond
Jake Burkholder [Sat, 29 Dec 2001 07:14:07 +0000 (07:14 +0000)]
Implement dcache_inval_phys, which shoots the cache lines that correspond
to a specific physical address.  This is used for page copy and zero
routines which use physical addresses directly.

Submitted by: tmm

22 years agoMake AIO a loadable module.
Alfred Perlstein [Sat, 29 Dec 2001 07:13:47 +0000 (07:13 +0000)]
Make AIO a loadable module.

Remove the explicit call to aio_proc_rundown() from exit1(), instead AIO
will use at_exit(9).

Add functions at_exec(9), rm_at_exec(9) which function nearly the
same as at_exec(9) and rm_at_exec(9), these functions are called
on behalf of modules at the time of execve(2) after the image
activator has run.

Use a modified version of tegge's suggestion via at_exec(9) to close
an exploitable race in AIO.

Fix SYSCALL_MODULE_HELPER such that it's archetecuterally neutral,
the problem was that one had to pass it a paramater indicating the
number of arguments which were actually the number of "int".  Fix
it by using an inline version of the AS macro against the syscall
arguments.  (AS should be available globally but we'll get to that
later.)

Add a primative system for dynamically adding kqueue ops, it's really
not as sophisticated as it should be, but I'll discuss with jlemon when
he's around.

22 years agoRename definitions to better match the hardware wstate fields.
Jake Burkholder [Sat, 29 Dec 2001 07:12:30 +0000 (07:12 +0000)]
Rename definitions to better match the hardware wstate fields.
Don't include WSTATE_TRANSITION in WSTATE_NORMAL_MASK.

22 years agoAdd definitions for TSTATE_MM_* and TSTATE_{I,X}CC_*.
Jake Burkholder [Sat, 29 Dec 2001 07:11:00 +0000 (07:11 +0000)]
Add definitions for TSTATE_MM_* and TSTATE_{I,X}CC_*.
Implement TSTATE_SECURE in terms of PSTATE_SECURE.

22 years agoRename and renumber trap types to comply with the user trap types as
Jake Burkholder [Sat, 29 Dec 2001 07:09:50 +0000 (07:09 +0000)]
Rename and renumber trap types to comply with the user trap types as
specified by the sparc abi.  We use numerically higher values for all
internal kernel types.
Remove soft trap types which need to be exposed to userland.  They will
move to utrap.h.

22 years ago1. Certain tlb operations need to be atomic, so disable interrupts for
Jake Burkholder [Sat, 29 Dec 2001 07:07:35 +0000 (07:07 +0000)]
1. Certain tlb operations need to be atomic, so disable interrupts for
   their duration.  This is still only effective as long as they are
   only used in the static kernel.  Code in modules may cause instruction
   faults which makes these break in different ways anyway.
2. Add a load bearing membar #Sync.
3. Add an inline for demapping an entire context.

Submitted by: tmm (1, 2)

22 years agojmpbuf is no longer a ucontext_t since it does not need to be passed
Jake Burkholder [Sat, 29 Dec 2001 07:03:09 +0000 (07:03 +0000)]
jmpbuf is no longer a ucontext_t since it does not need to be passed
to sigreturn.
Add definitions for array offsets.

22 years agoAdd fprs to struct fpreg.
Jake Burkholder [Sat, 29 Dec 2001 07:02:12 +0000 (07:02 +0000)]
Add fprs to struct fpreg.

22 years agoDefine PSTATE_MM_MASK in terms of PSTATE_MM_SIZE.
Jake Burkholder [Sat, 29 Dec 2001 07:01:34 +0000 (07:01 +0000)]
Define PSTATE_MM_MASK in terms of PSTATE_MM_SIZE.
Implement PSTATE_SECURE.

22 years agoRemove pcb_y. It has moved to trapframe.
Jake Burkholder [Sat, 29 Dec 2001 07:00:30 +0000 (07:00 +0000)]
Remove pcb_y.  It has moved to trapframe.

22 years agoDon't concatentate __func__.
Jake Burkholder [Sat, 29 Dec 2001 07:00:04 +0000 (07:00 +0000)]
Don't concatentate __func__.
Make page size constants explicitly long and unsigned.
Add a macro for compile time assertions.

22 years agoMake it clear that IH_SHIFT is expected to be that of a pointer.
Jake Burkholder [Sat, 29 Dec 2001 06:57:55 +0000 (06:57 +0000)]
Make it clear that IH_SHIFT is expected to be that of a pointer.
Make intr_handlers an array of function pointers instead of
small structures.

22 years agoAdd definitions for magic numbers used in asm.
Jake Burkholder [Sat, 29 Dec 2001 06:56:11 +0000 (06:56 +0000)]
Add definitions for magic numbers used in asm.
Bloat trapframe with many extra fields so we don't need extra structures.
Use small data types where possible.
Remove second copy of TF_DONE.
Remove mmuframe.

22 years agoChange fpblock to be an array of ints instead longs.
Jake Burkholder [Sat, 29 Dec 2001 06:53:19 +0000 (06:53 +0000)]
Change fpblock to be an array of ints instead longs.
Change fp_init_thread to take a thread instead of a pcb so we
can get at the trapframe.

22 years agoAdd "memory" to the clobber list for membars.
Jake Burkholder [Sat, 29 Dec 2001 06:51:50 +0000 (06:51 +0000)]
Add "memory" to the clobber list for membars.

Submitted by: tmm

22 years agoImplement CLKF_USERMODE so that user time is accounted properly.
Jake Burkholder [Sat, 29 Dec 2001 06:51:14 +0000 (06:51 +0000)]
Implement CLKF_USERMODE so that user time is accounted properly.

Submitted by: tmm

22 years agoAdd definitions for the number of bits in the icc and xcc fields
Jake Burkholder [Sat, 29 Dec 2001 06:50:25 +0000 (06:50 +0000)]
Add definitions for the number of bits in the icc and xcc fields
of the ccr, as well as the shifts and masks for each.

Submitted by: tmm

22 years agoUse ASI_P instead of ASI_N if _KERNEL isn't defined so that these
Jake Burkholder [Sat, 29 Dec 2001 06:48:43 +0000 (06:48 +0000)]
Use ASI_P instead of ASI_N if _KERNEL isn't defined so that these
can be used in userland.

Submitted by: tmm

22 years agoAdd macros for dedicated register variables, for use in assmebler.
Jake Burkholder [Sat, 29 Dec 2001 06:47:50 +0000 (06:47 +0000)]
Add macros for dedicated register variables, for use in assmebler.
Add a PUTS macro.

22 years agoAdapt to new kernel tsb structure.
Jake Burkholder [Sat, 29 Dec 2001 06:43:36 +0000 (06:43 +0000)]
Adapt to new kernel tsb structure.

22 years agoMake sure curbrk is well aligned.
Jake Burkholder [Sat, 29 Dec 2001 06:40:58 +0000 (06:40 +0000)]
Make sure curbrk is well aligned.

22 years agoUse jmpbuf offsets, not ucontext_t.
Jake Burkholder [Sat, 29 Dec 2001 06:40:35 +0000 (06:40 +0000)]
Use jmpbuf offsets, not ucontext_t.

22 years agoAdd .register directives for gcc3.
Jake Burkholder [Sat, 29 Dec 2001 06:39:53 +0000 (06:39 +0000)]
Add .register directives for gcc3.
Adapt to jmpbuf no longer being a ucontext_t.
Restore the context "by hand" in longjmp and call sigprocmask, instead
of just using sigreturn.

22 years ago__infinity is not const.
Jake Burkholder [Sat, 29 Dec 2001 06:38:31 +0000 (06:38 +0000)]
__infinity is not const.

22 years agoAdd jmpbuf offsets. Remove ucontext offsets.
Jake Burkholder [Sat, 29 Dec 2001 06:38:06 +0000 (06:38 +0000)]
Add jmpbuf offsets.  Remove ucontext offsets.

22 years agoAdd .register directives for gcc3.
Jake Burkholder [Sat, 29 Dec 2001 06:37:33 +0000 (06:37 +0000)]
Add .register directives for gcc3.
Adapt to jmpbuf no longer being a ucontext_t.
Call abort if longjmperror returns.

22 years agoInclude utrap.h for soft trap types.
Jake Burkholder [Sat, 29 Dec 2001 06:36:07 +0000 (06:36 +0000)]
Include utrap.h for soft trap types.

22 years agoWe use HIDENAME(minbrk) on sparc64 too.
Jake Burkholder [Sat, 29 Dec 2001 06:35:19 +0000 (06:35 +0000)]
We use HIDENAME(minbrk) on sparc64 too.

22 years agoAdd device ofw_console so this kernel is usable.
Jake Burkholder [Sat, 29 Dec 2001 05:58:07 +0000 (05:58 +0000)]
Add device ofw_console so this kernel is usable.

This commit brought to you by a sun ultra 10 running freebsd.

22 years ago"10/100Base FastEthernet PC CARD" (LNA-100) from Billionton Systems Inc.
Warner Losh [Sat, 29 Dec 2001 02:15:46 +0000 (02:15 +0000)]
"10/100Base FastEthernet PC CARD" (LNA-100) from Billionton Systems Inc.

Submitted by: "Torfinn Ingolfsen" <torfinn.ingolfsen@oslo.online.no>

22 years agoRemove part of a diff committed at the end of the file.
Ben Smithurst [Sat, 29 Dec 2001 00:51:23 +0000 (00:51 +0000)]
Remove part of a diff committed at the end of the file.

22 years agoForce raw printing of non-printable characters via the -w option.
Josef Karthauser [Sat, 29 Dec 2001 00:22:29 +0000 (00:22 +0000)]
Force raw printing of non-printable characters via the -w option.

PR: bin/28007

22 years agoRestore these files to shiny KNF.
Josef Karthauser [Sat, 29 Dec 2001 00:18:16 +0000 (00:18 +0000)]
Restore these files to shiny KNF.

22 years agoFix missing comma in previous commit.
Dag-Erling Smørgrav [Fri, 28 Dec 2001 23:51:48 +0000 (23:51 +0000)]
Fix missing comma in previous commit.

Pointy hat to: jhb

22 years agoConvert sppp_params() to use a malloced structure in order to reduce
Joerg Wunsch [Fri, 28 Dec 2001 23:36:35 +0000 (23:36 +0000)]
Convert sppp_params() to use a malloced structure in order to reduce
kernel stack usage.

This effectively merges rev 1.3 of i4b's i4b_ispppsubr.c.

MFC after: 1 month

22 years agoFix my breakage to the low-level hardware sync drivers brought by the
Joerg Wunsch [Fri, 28 Dec 2001 23:20:12 +0000 (23:20 +0000)]
Fix my breakage to the low-level hardware sync drivers brought by the
inclusion of VJ compression into sppp.

Now, instead of the need to include this and that and everything plus
the kitchensink in each of those drivers, struct sppp uses struct
slcompress as an opaque structure only referenced by a pointer.  The
actual structure is then malloced at initialization time.

While i was at it, also fixed a bug where received VJ packets would only
be recognized if INET6 was defined.

22 years agoFix documentation to match reality
Julian Elischer [Fri, 28 Dec 2001 22:24:26 +0000 (22:24 +0000)]
Fix documentation to match reality

22 years agoAvoid having the text and data sections merged into one program
Robert Drehmel [Fri, 28 Dec 2001 22:12:38 +0000 (22:12 +0000)]
Avoid having the text and data sections merged into one program
header table entry by removing the -N flag from the LDFLAGS.

22 years agoDocument some behaviour I'm depending on.
Julian Elischer [Fri, 28 Dec 2001 22:00:27 +0000 (22:00 +0000)]
Document some behaviour I'm depending on.
(TAILQ_FOREACH leaves the variable NULL if there were no more to do)

22 years agoConvert some spaces into tabs that I missed first time around.
Josef Karthauser [Fri, 28 Dec 2001 21:55:23 +0000 (21:55 +0000)]
Convert some spaces into tabs that I missed first time around.

22 years agoRevert most of rev 1.42, to restore KNF style.
Josef Karthauser [Fri, 28 Dec 2001 21:47:33 +0000 (21:47 +0000)]
Revert most of rev 1.42, to restore KNF style.

Requested by: bde

22 years agoFix ipfw fwd so that it acts as the docs say
Julian Elischer [Fri, 28 Dec 2001 21:21:57 +0000 (21:21 +0000)]
Fix ipfw fwd so that it acts as the docs say
when forwarding an incoming packet to another machine.

Obtained from: Vicor Production tree
MFC after: 3 weeks

22 years agoFixed a missing "const".
Bruce Evans [Fri, 28 Dec 2001 20:59:44 +0000 (20:59 +0000)]
Fixed a missing "const".

22 years agoAdd a new flag, -h which when combined with the -l option causes
Josef Karthauser [Fri, 28 Dec 2001 20:50:12 +0000 (20:50 +0000)]
Add a new flag, -h which when combined with the -l option causes
file sizes to be displayed with unit suffixes; Byte, Kilobyte,
Megabyte, Gigabyte, Terabyte and Petabyte in order to reduce the
number of digits to three or less.

Submitted by: nik

22 years agoSpelling/typo fix: `bood' -> `boot'
Chris Costello [Fri, 28 Dec 2001 20:12:24 +0000 (20:12 +0000)]
Spelling/typo fix: `bood' -> `boot'

22 years agoMake it clearer what changed to a user using the old format of LSCOLORS.
Josef Karthauser [Fri, 28 Dec 2001 19:27:30 +0000 (19:27 +0000)]
Make it clearer what changed to a user using the old format of LSCOLORS.

22 years agoMop up some warnings.
Josef Karthauser [Fri, 28 Dec 2001 19:26:06 +0000 (19:26 +0000)]
Mop up some warnings.

22 years agoMake the style more consistent throughout the file.
Josef Karthauser [Fri, 28 Dec 2001 18:43:32 +0000 (18:43 +0000)]
Make the style more consistent throughout the file.

22 years agoFixed an apparent typo ("-" before ":") and an English error (comma
Bruce Evans [Fri, 28 Dec 2001 18:32:13 +0000 (18:32 +0000)]
Fixed an apparent typo ("-" before ":") and an English error (comma
splice) in the "already exists" message.

Fixed some minor style bugs (KNFization to "return (foo)" had rotted
in 2 out of 177 cases).

22 years agoUse the default colour instead if a bad colour specification is given.
Josef Karthauser [Fri, 28 Dec 2001 18:20:23 +0000 (18:20 +0000)]
Use the default colour instead if a bad colour specification is given.

22 years agoRevamp the colour support to allow for bold characters. Colours
Josef Karthauser [Fri, 28 Dec 2001 18:14:50 +0000 (18:14 +0000)]
Revamp the colour support to allow for bold characters.  Colours
are now defined using the characters a-h and A-H for the bold
variants.  The old way using 0-7 for the colours still works, but
prints a message asking the user to switch.

PR: bin/27374

22 years agoFixed locking bugs in rev.1.346:
Bruce Evans [Fri, 28 Dec 2001 18:08:10 +0000 (18:08 +0000)]
Fixed locking bugs in rev.1.346:

(1) Don't attempt aquire the non-recursive lock sio_lock recursively.
    Doing so caused unbounded recursion in some setups.  E.g., if DDB,
    BREAK_TO_DEBUGGER and WITNESS are configured; if the debugger is
    entered using a break, then WITNESS will actually detect the invalid
    recursion and will add to it attempting to print a message about it.

(2) Don't use sio_lock before it has been initialized.  The old check
    (sio_inited != 0) didn't work when sio_inited was boolean because
    sio_inited was set too early, and became just wrong when sio_inited
    was changed to a tri-state variable in rev.1.348.

Reported and fixed in another way by: fenner (1)

22 years agoFix typo (unclosed comment).
Jake Burkholder [Fri, 28 Dec 2001 17:55:15 +0000 (17:55 +0000)]
Fix typo (unclosed comment).

22 years agoModify the release-building process so the srelease distribution
Bruce A. Mah [Fri, 28 Dec 2001 17:21:28 +0000 (17:21 +0000)]
Modify the release-building process so the srelease distribution
doesn't get spammed by RELNOTESng files.

Previously, we built the RELNOTESng files "in place" and then
populated the media (e.g. floppy, cdrom, ftp) areas from the build
directories.  However, the distributions got rolled after the
RELNOTESng build, and thus picked up the rendered RELNOTESng files.

Now, we install the RELNOTESng documents to R/stage/release.doc, then
clean out the original directories in src/release/doc.  We use the
renditions in the staging area to populate the media areas.

PR: 33141
MFC after: 3 days
(plus re approval)

22 years agoMove UFS and FFS docs to section 7.
Chris Costello [Fri, 28 Dec 2001 17:10:13 +0000 (17:10 +0000)]
Move UFS and FFS docs to section 7.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

22 years agoImplement timestamps so i4b/driver/i4b_isppp.c can derive the idle
Joerg Wunsch [Fri, 28 Dec 2001 15:59:54 +0000 (15:59 +0000)]
Implement timestamps so i4b/driver/i4b_isppp.c can derive the idle
time from the PPP packets sent.  This effectively merges rev 1.2 of
the old i4b_ispppsubr.c, with the exception that i eventually ended up
in debugging and fixing it so the idle time is now really
detected. ;-)  (The version in i4b simply doesn't work right since it
still accounts for incoming LCP echo packets which it is supposed to
ignore for idle time considerations...)

Obtained from: i4b
MFC after: 1 month

22 years agognu/libexec has gone so remove it from the most targets.
Paul Richards [Fri, 28 Dec 2001 11:53:18 +0000 (11:53 +0000)]
gnu/libexec has gone so remove it from the most targets.

22 years agoAdd xref to jail(2).
Mike Pritchard [Fri, 28 Dec 2001 09:26:45 +0000 (09:26 +0000)]
Add xref to jail(2).

PR: docs/33177
Submitted by: Tom Rhodes <darklogik@pittgoth.com>

22 years agoAdd xref to jail(8)
Mike Pritchard [Fri, 28 Dec 2001 09:24:19 +0000 (09:24 +0000)]
Add xref to jail(8)

PR: docs/33178
Submitted by: Tom Rhodes <darklogik@pittgoth.com>

22 years agoadd note that truss works again as of dec 09.
Warner Losh [Fri, 28 Dec 2001 05:35:32 +0000 (05:35 +0000)]
add note that truss works again as of dec 09.

Submitted by: des
Ignored by: imp for over two weeks

22 years agoAdd note about daily security checks moving to periodic(8)
Warner Losh [Fri, 28 Dec 2001 05:33:34 +0000 (05:33 +0000)]
Add note about daily security checks moving to periodic(8)

22 years agoAdd note to 20010924 entry that tells people about the COMPAT4X option
Warner Losh [Fri, 28 Dec 2001 05:09:19 +0000 (05:09 +0000)]
Add note to 20010924 entry that tells people about the COMPAT4X option
in /etc/make.conf, since sometimes these libraries are updated.

Submitted by: ru

22 years agoMerge the changes in the English version (1.239 --> 1.242).
Jun Kuriyama [Fri, 28 Dec 2001 05:07:23 +0000 (05:07 +0000)]
Merge the changes in the English version (1.239 --> 1.242).

22 years agoFix typo in example.
Bruce A. Mah [Fri, 28 Dec 2001 03:59:19 +0000 (03:59 +0000)]
Fix typo in example.

PR: 33260
Submitted by: Dan Peterson <danp@danp.net>

22 years agoNovac DVD/CD Station(Portable DVD/CD-ROM player
Warner Losh [Fri, 28 Dec 2001 03:42:41 +0000 (03:42 +0000)]
Novac DVD/CD Station(Portable DVD/CD-ROM player

Submitted by: moment@pluto.dti.ne.jp

22 years agoXircom RealPort RE-10
Warner Losh [Fri, 28 Dec 2001 03:39:13 +0000 (03:39 +0000)]
Xircom RealPort RE-10

Submitted by: Willem van Engen <wvengen@stack.nl>

22 years agoAdd Proxim Harmony card (from Bob)
Warner Losh [Fri, 28 Dec 2001 03:34:05 +0000 (03:34 +0000)]
Add Proxim Harmony card (from Bob)
Also fix a minor disordering of a wi card while I'm here.

Submitted by: bob bobing <this_is_my_act@yahoo.com>

22 years agoFix some leftover stray characters from expr(1)-to-$(()) sweeps.
Crist J. Clark [Thu, 27 Dec 2001 22:41:35 +0000 (22:41 +0000)]
Fix some leftover stray characters from expr(1)-to-$(()) sweeps.

PR: bin/33158
Submitted by: Adrian Filipi-Martin <adrian@ubergeeks.com>
MFC after: 1 day

22 years agoClean up half a dozen of header files that are no longer needed now that
Joerg Wunsch [Thu, 27 Dec 2001 22:40:01 +0000 (22:40 +0000)]
Clean up half a dozen of header files that are no longer needed now that
we use struct sppp_parms instead of struct sppp for userland programs.

MFC after: 1 month

22 years agoBreak out the relevant fields from struct sppp into a struct
Joerg Wunsch [Thu, 27 Dec 2001 22:38:50 +0000 (22:38 +0000)]
Break out the relevant fields from struct sppp into a struct
sppp_parms that are needed for the SPPPIO[GS]DEFS ioctl commands.
This allows it to keep struct sppp inside #ifdef _KERNEL (where it
belongs), and prevents userland programs that wish to include
<net/if_sppp.h> from including the earth, the hell, and the universe
before the are able to resolve all the kernel-internal stuff that's in
struct sppp.

Discussed with: hm
MFC after: 1 month

22 years agoAt least I'm consistent. Back out the last accidental rev to this file on
Bruce A. Mah [Thu, 27 Dec 2001 22:32:33 +0000 (22:32 +0000)]
At least I'm consistent.  Back out the last accidental rev to this file on
HEAD too.

Noticed by: ben
Another pointy hat to: bmah

22 years agoNew release note: network filesystems can now be listed in
Bruce A. Mah [Thu, 27 Dec 2001 21:55:02 +0000 (21:55 +0000)]
New release note:  network filesystems can now be listed in
/etc/fstab.

Submitted by: sheldonh

22 years agoNew release notes: pseudofs-ize procfs(5), libstand can write to UFS
Bruce A. Mah [Thu, 27 Dec 2001 21:37:38 +0000 (21:37 +0000)]
New release notes: pseudofs-ize procfs(5), libstand can write to UFS
filesystems, sysinstall(8) auto-partition updates.