Commit graph

119 commits

Author SHA1 Message Date
Frans Kaashoek e25b74ca80 Fix layout issues for printed version 2011-09-01 10:25:20 -04:00
Robert Morris 5e08357827 enterpgdir -> entrypgdir 2011-08-30 20:50:19 -04:00
Austin Clements a7061b4f97 Style nits; indentation and tabs 2011-08-29 16:12:01 -04:00
Frans Kaashoek d10d324e79 Remove sys_init syscall
Invoke initlog from forkret on first user process
2011-08-22 20:05:15 -04:00
Frans Kaashoek a4b213cf49 Avoid "boot" in xv6 2011-08-15 20:11:13 -04:00
Frans Kaashoek 9449646853 Use 4Mbyte pages during boot 2011-08-15 17:41:58 -04:00
Frans Kaashoek c60a3551c2 Separate more clearly bootloader from xv6 by renaming multiboot.S to entry.S etc.
Maybe the string boot shouldn't appear in xv6 code?
2011-08-15 12:02:59 -04:00
Frans Kaashoek c95ce31c59 Oops 2011-08-12 12:02:17 -04:00
Frans Kaashoek e577a62f0d Some comments 2011-08-12 07:31:52 -04:00
Frans Kaashoek bd71a45046 Make AP processors boot using bootpgdir
Remove device mapping from bootpgdir
Remove unnecessary vmenable
Set CPUS back to 2 in Makefile
Passes all usertests
2011-08-11 12:25:10 -04:00
Frans Kaashoek 832af025a3 Remove jmpkstack 2011-08-09 21:47:40 -04:00
Frans Kaashoek 66ba8079c7 Use static page table for boot, mapping first 4Mbyte; no more segment trick
Allocate proper kernel page table immediately in main using boot allocator
Remove pginit
Simplify address space layout a tiny bit
More to come (e.g., superpages to simplify static table)
2011-08-09 21:37:35 -04:00
Frans Kaashoek 9aa0337dc1 Map kernel high
Very important to give qemu memory through PHYSTOP :(
2011-07-29 07:31:27 -04:00
Frans Kaashoek 13a96baefc Dirt simple logging
Passes usertests and stressfs
Seems to recover correctly in a number of simple cases
2011-07-27 20:35:46 -04:00
Russ Cox cf4b1ad90b xv6: formatting, cleanup, rev5 (take 2) 2011-02-19 21:17:55 -05:00
Russ Cox 1a81e38b17 make new code like old code
Variable declarations at top of function,
separate from initialization.

Use == 0 instead of ! for checking pointers.

Consistent spacing around {, *, casts.

Declare 0-parameter functions as (void) not ().

Integer valued functions return -1 on failure, 0 on success.
2011-01-11 13:01:13 -05:00
Robert Morris faad047ab2 change some comments, maybe more informative
delete most comments from bootother.S (since copy of bootasm.S)
ksegment() -> seginit()
move more stuff from main() to mainc()
2010-09-13 15:34:44 -04:00
Austin Clements b0751a3e9b Space police 2010-09-01 00:41:25 -04:00
Robert Morris 8d774afb2d no more pminit, or ELF header at 0x10000
kinit() knows about end and PHYSTOP
map all of kernel read/write (rather than r/o instructions)
thanks, austin
2010-08-31 15:39:25 -04:00
Robert Morris 7d7dc9331b kalloc/kfree now only a page at a time
do not keep sorted contiguous free list
2010-08-31 12:54:47 -04:00
Robert Morris ac090078c6 xx 2010-08-30 10:13:49 -04:00
Robert Morris 5ab868fd90 set only PG and WP in vminit; the rest don't seem to be needed and are confusing 2010-08-30 06:38:58 -04:00
Robert Morris 1afc9d3fca add some comments
find out the hard way why user and kernel must have separate segment descriptors
2010-08-05 21:16:55 -04:00
Robert Morris 2cf6b32d4d move jkstack to main.c
replace jstack with asm()s
2010-08-05 14:15:03 -04:00
Frans Kaashoek 4714c20521 Checkpoint page-table version for SMP
Includes code for TLB shootdown (which actually seems unnecessary for xv6)
2010-07-23 07:41:13 -04:00
Frans Kaashoek 40889627ba Initial version of single-cpu xv6 with page tables 2010-07-02 14:51:53 -04:00
Russ Cox 7e0cc8e36e another attempt at cpu-local variables.
this time do it ourselves instead of piggybacking on TLS.
add -fno-pic to Makefile; pic code breaks our fake TLS.
2009-09-02 10:07:59 -07:00
Russ Cox 48755214c9 assorted fixes:
* rename c/cp to cpu/proc
 * rename cpu.context to cpu.scheduler
 * fix some comments
 * formatting for printout
2009-08-30 23:02:08 -07:00
rsc 19333efb9e Some proc cleanup, moving some of copyproc into allocproc.
Also, an experiment: use "thread-local" storage for c and cp
instead of the #define macro for curproc[cpu()].
2009-05-31 00:28:45 +00:00
rsc 2157576107 be consistent: no underscores in function names 2009-03-08 22:07:13 +00:00
rsc b7f653dc49 xv6: boot loader adjustments
do Bochs breakpoint and spin in bootasm.S.
not needed in bootmain too.
fix readseg bug (rounding of va).
zero segments when memsz > filesz.
no need to clear BSS in kernel main.
make bootother.S like bootasm.S
2009-03-08 21:41:30 +00:00
kolya 5c5470a2fa fix obvious printf nits after reading through code 2008-08-21 23:24:02 +00:00
rsc 943fd378a1 Incorporate new understanding of/with Intel SMP spec.
Dropped cmpxchg in favor of xchg, to match lecture notes.

Use xchg to release lock, for future protection and to
keep gcc from acting clever.
2007-10-01 20:43:15 +00:00
rsc 9fd9f80431 Re: why cpuid() in locking code?
rtm wrote:
> Why does acquire() call cpuid()? Why does release() call cpuid()?

The cpuid in acquire is redundant with the cmpxchg, as you said.
I have removed the cpuid from acquire.

The cpuid in release is actually doing something important,
but not on the hardware.  It keeps gcc from reordering the
lock->locked assignment above the other two during optimization.
(Not that current gcc -O2 would choose to do that, but it is allowed to.)
I have replaced the cpuid in release with a "gcc barrier" that
keeps gcc from moving things around but has no hardware effect.

On a related note, I don't think the cpuid in mpmain is necessary,
for the same reason that the cpuid wasn't needed in release.

As to the question of whether

  acquire();
  x = protected;
  release();

might read protected after release(), I still haven't convinced
myself whether it can.  I'll put the cpuid back into release if
we determine that it can.

Russ
2007-09-30 14:30:04 +00:00
rsc ab08960f64 Final word on the locking fiasco?
Change pushcli / popcli so that they can never turn on
interrupts unexpectedly.  That is, if interrupts are on,
then pushcli(); popcli(); turns them off and back on, but
if they are off to begin with, then pushcli(); popcli(); is
a no-op.

I think our fundamental mistake was having a primitive
(release and then popcli nee spllo) that could turn
interrupts on at unexpected moments instead of being
explicit about when we want to start allowing interrupts.

With the new semantics, all the manual fiddling of ncli
to force interrupts off in certain sections goes away.
In return, we must explicitly mark the places where
we want to enable interrupts unconditionally, by calling sti().
There is only one: inside the scheduler loop.
2007-09-27 21:25:37 +00:00
rsc c95bde8163 yank out stack overflow checking ugliness 2007-09-27 20:38:53 +00:00
rsc 4f74de0edc okay, that was long enough - revert 2007-09-27 20:32:45 +00:00
rsc ce2e751555 test: store curproc at top of stack
I don't actually think this is worthwhile, but I figured
I would check it in before reverting it, so that it can
be in the revision history.

Pros:
  * curproc doesn't need to turn on/off interrupts
  * scheduler doesn't have to edit curproc anymore

Cons:
  * it's ugly
  * all the stack computation is more complicated.
  * it doesn't actually simplify anything but curproc,
    and even curproc is harder to follow.
2007-09-27 20:29:50 +00:00
rsc 3807c1f20b rename splhi/spllo to pushcli/popcli 2007-09-27 20:09:40 +00:00
rsc 4721271961 use larger, allocated cpu stacks 2007-09-27 19:32:43 +00:00
rsc c8919e6537 kernel SMP interruptibility fixes.
Last year, right before I sent xv6 to the printer, I changed the
SETGATE calls so that interrupts would be disabled on entry to
interrupt handlers, and I added the nlock++ / nlock-- in trap()
so that interrupts would stay disabled while the hw handlers
(but not the syscall handler) did their work.  I did this because
the kernel was otherwise causing Bochs to triple-fault in SMP
mode, and time was short.

Robert observed yesterday that something was keeping the SMP
preemption user test from working.  It turned out that when I
simplified the lapic code I swapped the order of two register
writes that I didn't realize were order dependent.  I fixed that
and then since I had everything paged in kept going and tried
to figure out why you can't leave interrupts on during interrupt
handlers.  There are a few issues.

First, there must be some way to keep interrupts from "stacking
up" and overflowing the stack.  Keeping interrupts off the whole
time solves this problem -- even if the clock tick handler runs
long enough that the next clock tick is waiting when it finishes,
keeping interrupts off means that the handler runs all the way
through the "iret" before the next handler begins.  This is not
really a problem unless you are putting too many prints in trap
-- if the OS is doing its job right, the handlers should run
quickly and not stack up.

Second, if xv6 had page faults, then it would be important to
keep interrupts disabled between the start of the interrupt and
the time that cr2 was read, to avoid a scenario like:

   p1 page faults [cr2 set to faulting address]
   p1 starts executing trapasm.S
   clock interrupt, p1 preempted, p2 starts executing
   p2 page faults [cr2 set to another faulting address]
   p2 starts, finishes fault handler
   p1 rescheduled, reads cr2, sees wrong fault address

Alternately p1 could be rescheduled on the other cpu, in which
case it would still see the wrong cr2.  That said, I think cr2
is the only interrupt state that isn't pushed onto the interrupt
stack atomically at fault time, and xv6 doesn't care.  (This isn't
entirely hypothetical -- I debugged this problem on Plan 9.)

Third, and this is the big one, it is not safe to call cpu()
unless interrupts are disabled.  If interrupts are enabled then
there is no guarantee that, between the time cpu() looks up the
cpu id and the time that it the result gets used, the process
has not been rescheduled to the other cpu.  For example, the
very commonly-used expression curproc[cpu()] (aka the macro cp)
can end up referring to the wrong proc: the code stores the
result of cpu() in %eax, gets rescheduled to the other cpu at
just the wrong instant, and then reads curproc[%eax].

We use curproc[cpu()] to get the current process a LOT.  In that
particular case, if we arranged for the current curproc entry
to be addressed by %fs:0 and just use a different %fs on each
CPU, then we could safely get at curproc even with interrupts
disabled, since the read of %fs would be atomic with the read
of %fs:0.  Alternately, we could have a curproc() function that
disables interrupts while computing curproc[cpu()].  I've done
that last one.

Even in the current kernel, with interrupts off on entry to trap,
interrupts are enabled inside release if there are no locks held.
Also, the scheduler's idle loop must be interruptible at times
so that the clock and disk interrupts (which might make processes
runnable) can be handled.

In addition to the rampant use of curproc[cpu()], this little
snippet from acquire is wrong on smp:

  if(cpus[cpu()].nlock == 0)
    cli();
  cpus[cpu()].nlock++;

because if interrupts are off then we might call cpu(), get
rescheduled to a different cpu, look at cpus[oldcpu].nlock, and
wrongly decide not to disable interrupts on the new cpu.  The
fix is to always call cli().  But this is wrong too:

  if(holding(lock))
    panic("acquire");
  cli();
  cpus[cpu()].nlock++;

because holding looks at cpu().  The fix is:

  cli();
  if(holding(lock))
    panic("acquire");
  cpus[cpu()].nlock++;

I've done that, and I changed cpu() to complain the first time
it gets called with interrupts disabled.  (It gets called too
much to complain every time.)

I added new functions splhi and spllo that are like acquire and
release but without the locking:

  void
  splhi(void)
  {
    cli();
    cpus[cpu()].nsplhi++;
  }

  void
  spllo(void)
  {
    if(--cpus[cpu()].nsplhi == 0)
      sti();
  }

and I've used those to protect other sections of code that refer
to cpu() when interrupts would otherwise be disabled (basically
just curproc and setupsegs).  I also use them in acquire/release
and got rid of nlock.

I'm not thrilled with the names, but I think the concept -- a
counted cli/sti -- is sound.  Having them also replaces the
nlock++/nlock-- in trap.c and main.c, which is nice.


Final note: it's still not safe to enable interrupts in
the middle of trap() between lapic_eoi and returning
to user space.  I don't understand why, but we get a
fault on pop %es because 0x10 is a bad segment
descriptor (!) and then the fault faults trying to go into
a new interrupt because 0x8 is a bad segment descriptor too!
Triple fault.  I haven't debugged this yet.
2007-09-27 12:58:42 +00:00
rsc c1b100e930 nits 2007-08-28 18:23:48 +00:00
rsc 9e82bfb04c rename 8253pit.c to timer.c 2007-08-28 04:40:58 +00:00
rsc 43baa1f224 nit 2007-08-28 04:14:32 +00:00
rsc 3341e30f6e nit 2007-08-28 04:13:24 +00:00
rsc 19b42cc078 Rename main0 to main. 2007-08-27 23:32:16 +00:00
rsc 558ab49f13 delete unnecessary #include lines 2007-08-27 23:26:33 +00:00
rsc 99b11b6c64 Simplify MP hardware code.
Mainly delete unused constants and code.

Move mp_startthem to main.c as bootothers.
2007-08-27 22:53:31 +00:00
rsc b63bb0fd00 Clean up lapic code.
One initialization function now, not three.
Use #defines instead of enums (consistent with other code, but sigh).

Still boots in Bochs in SMP mode.
2007-08-27 16:57:13 +00:00
rsc 124f32ae38 tweak 2007-08-24 19:36:52 +00:00