Arrange to update cached LWP credentials in userret() rather than during
syscall/trap entry, eliminating a test+branch on every syscall/trap.

This wasn't possible in the 3.99.x timeframe when l->l_cred came about
because there wasn't a reliable/timely way to force an ONPROC LWP running on
a remote CPU into the kernel (which is just about the only new thing in
this scheme).

Allow syscall_establish() to install new syscalls when the existing
entry-point is either sys_nomodule or sys_nosys. Update the
makesyscalls.sh script to create a const array of bits to allow
syscall_disestablish() to properly restore the original entry-point.
Update all the initializers of struct emul to initialize the pointer
to the bit array struct emul.

XXX Regen of all files created by makesyscalls.sh will come soon,
XXX followed by a kernel version bump (since struct emul is being
XXX modified).

This commit should address PR kern/45781 and also removes the need
for the work-around for that PR in file

sys/arch/usermode/modules/syscallemu/syscallemu.c

branches: 1.53.2; 1.53.4;
Remove vm86.

Pass 3.

adjust to new trace_{enter,exit} and dtrace

branches: 1.51.12; 1.51.30;
syscall (setcontext) can alter eax, so don't attempt to get code again from it
in order to avoid register spills.

Use lwp_getpcb() on x86 MD code, clean from struct user usage.

branches: 1.49.8;
Provide a small inline wrapper for sysent::sy_call() and use it to store
the active syscall description in struct lwp. To be used at some future
point to prevent unloading of modules that provide syscalls, while the
syscalls are in use.

Undo revivesa damage to userret().

Merge wrstuden-revivesa into HEAD.

branches: 1.46.2; 1.46.6;
Remove clause 3 and 4 from TNF licenses

branches: 1.45.2;
- Retire SYCALL_MPSAFE. With the exceptions of darwin and irix emulations,
all system calls are now MPSAFE.
- Remove unneeded acquire/release of kernel_lock.

branches: 1.44.2;
Make context switch + syscall counters optionally per-CPU and accumulate
in schedclock() at "about 16 hz".

Set p->p_trace_enabled in fork and whenever the controlling falgs change
instead of doing it in syscall_intern().
Note that syscall_intern() must still be called when the flags change
since many ports use a different copy of the syscall entry code when
tracing is enabled.

branches: 1.42.2; 1.42.6;
Remove the 'args' parameter to 'trace_exit()' it is no longer used.
Instead of passing the (un)real system call code and syscall table pointer,
just pass the number of arguments - which is what ktrace really wants.
Ride forthcoming 4.99.53

Don't pass 'curlwp' into trace_enter() and trace_exit().

Kill syscall fancy.
Optimise matching native code.

branches: 1.39.24; 1.39.30; 1.39.32; 1.39.36;
Merge newlock2 to head.

branches: 1.38.4;
- Hold a reference to the process credentials in each struct lwp.
- Update the reference on syscall and user trap if p_cred has changed.
- Collect accounting flags in the LWP, and collate on LWP exit.

branches: 1.37.2;
Syscall debug tracing is handled by trace_enter() / trace_exit(). Change
trace_is_enabled() to return TRUE if SYSCALL_DEBUG is defined, and g/c
all of the SYSCALL_DEBUG handling from individual system call dispatch
routines.

Clean up fallout proc_is_traced_p() change:
- proc_is_traced_p() -> trace_is_enabled(), to match trace_enter() and
trace_exit().
- trace_is_enabled() becomes a real function.
- Remove unnecessary include files from various files that used to care
about KTRACE and SYSTRACE, but do no more.

Add a proc_is_traced_p() macro and use it, instead of copying the same code
in many places. Idea from thorpej.

make these compile w/o SYSTRACE.

branches: 1.33.2;
implement PT_SYSCALL

branches: 1.32.4; 1.32.6;
merge ktrace-lwp.

branches: 1.31.2;
Delete now unused "bad" label

PR/29607: Christian Biere systrace doesn't handle interrupted syscalls properly
Instead of jumping to the default "bad" case, jump to the error handling
switch, so that we can deal with ERESTART/EJUSTRETURN properly.

plug kernel lock leaks

Don't pass the (unused) return value args to the
trace_enter()/systrace_enter() functions.

Remove some assigned-to but otherwise unused variables.

Nuke __P().

Pass pointers to frames from assembly, do not use the 'frame on stack
as argument passed by value' trick, as gcc 3.3.x makes (valid) assumptions
about the stack that will not be true. Costs 2 instructions per trap/syscall
on i386, 4 per interrupt for MP. One instruction per trap/syscall on amd64,
2 per interrupt for MP. I expect gcc 3.3.1 to make up for this by better
optimization (it'd better..)

While here, make amd64 compile again by using subr_mbr_disk.c

branches: 1.24.2;
Merge the nathanw_sa branch.

Pass the system call table to trace_enter() and ktrsys() so that it is
possible to use alternate system call tables. This is usefull for
displaying correctly the arguments in Mach binaries traces.

If NULL is given, then the regular systam call table for the process is used.

Add a realcode argument to trace_enter and ktrsyscall. realcode is the
original system call number, which can be negative for a Mach trap.
We cannot just replace code by realcode, because ktrsyscall uses it as
an index in the system call table, thus crashing the kernel when the
value is negative.

Merge Bill Sommerfeld's i386 MP branch. This code has some known
caveats, but works quite well in a lot of MP cases, and all
UP cases that I have tested. Parts of this will hopefully be
reworked in the not-too-distant future.

MD systrace gluons.

branches: 1.19.2;
remove debugging printf.

linux passes the 6th argument in %ebp; handle that so that we can get mmap2
working. Also tell us which syscall has too many args. Thanks Frank!

add RCSID

branches: 1.16.2;
use _KERNEL_OPT

branches: 1.15.2; 1.15.4;
Clean up #includes.

Remove unnecessary #includes.

Whoops. Previous change was not correct.

Minor code rearrangement.

#ifdef KTRACE some stuff.

Handle the VM86-mode syscall problem by changing the syscall vector when
jumping in or out of VM86 mode.

Separate the syscall handlers into two versions -- one which does ktrace, and
one which doesn't. Dynamically switch.
There's more to do on this, but I have to go to work.

There is no Berkeley code in any of these files.

Introduce 2 new flags in types.h:
* __HAVE_SYSCALL_INTERN. If this is defined, e_syscall is replaced by
e_syscall_intern, which is called at key places in the kernel. This can be
used to set a MD syscall handler pointer. This obsoletes and replaces the
*_HAS_SEPARATED_SYSCALL flags.
* __HAVE_MINIMAL_EMUL. If this is defined, certain (deprecated) elements in
struct emul are omitted.

Introduce PROC_PC(), which is used to get a process's user PC. If this is
defined, call addupc_intr() directly from statclock() in the system time case,
using the same P_OWEUPC path if the copyin/copyout fails.
Use this in i386 to remove profiling code from the normal userret() path.

since this is now standalone and not called via syscall(), we need to
update statistics and set p->p_md.md_regs accordingly here
do the errno translation uncoditionally

make locore.s call emulation-specific syscall code directly
this should improve speed of emulation syscall path (avoids one function call,
and emulations syscall can use the trapframe trick)
Idea: Charles Hannum

put userret() into header file machine/userret.h and pull it where needed

GC garbage in userret() and *_syscall().

branches: 1.1.2;
as discussed on tech-kern, put linux and ibcs2 syscall code to separate
files and change syscall() to call p->p_emul->e_syscall() if it's not NULL