Commit Graph

12068 Commits

Author SHA1 Message Date
Dmitry Chagin
e5d81ef1b5 Extend struct sysvec with new method sv_schedtail, which is used for an
explicit process at fork trampoline path instead of eventhadler(schedtail)
invocation for each child process.

Remove eventhandler(schedtail) code and change linux ABI to use newly added
sysvec method.

While here replace explicit comparing of module sysentvec structure with the
newly created process sysentvec to detect the linux ABI.

Discussed with:	kib

MFC after:	2 Week
2011-03-08 19:01:45 +00:00
Robert Watson
74b5505e5d Continue to introduce Capsicum capability mode:
White list sysarch calls allowed in capability mode; arguably, there
should be some link between the capability mode model and the privilege
model here.  Sysarch is a morass similar to ioctl, in many senses.

Submitted by:	anderson
Discussed with:	benl, kris, pjd
Sponsored by:	Google, Inc.
Obtained from:	Capsicum Project
MFC after:	3 months
2011-03-01 13:35:48 +00:00
John Baldwin
4cef260f42 Fix whitespace nit. 2011-02-22 14:58:14 +00:00
Rebecca Cran
6bccea7c2b Fix typos - remove duplicate "the".
PR:	bin/154928
Submitted by:	Eitan Adler <lists at eitanadler.com>
MFC after: 	3 days
2011-02-21 09:01:34 +00:00
Alan Cox
e6ffa21488 Remove pmap fields that are either unused or not fully implemented.
Discussed with:	kib
2011-02-17 15:36:29 +00:00
Dmitry Chagin
dc4f0a9e11 To avoid excessive code duplication create wrapper for fill regs
from stack frame. Change the trap() code to use newly created function
instead of explicit regs assignment.
2011-02-16 17:50:21 +00:00
Dmitry Chagin
09d6cb0a23 For realtime signals fill the sigval value. 2011-02-15 21:46:36 +00:00
Dmitry Chagin
fde6316272 Sort include files in the alphabetical order. 2011-02-13 19:07:48 +00:00
Dmitry Chagin
222198ab0b Move linux_clone(), linux_fork(), linux_vfork() to a MI path. 2011-02-12 18:17:12 +00:00
Dmitry Chagin
c8d6845e9e In preparation for moving linux_clone() to a MI path
introduce linux_set_upcall_kse().
2011-02-12 16:33:00 +00:00
Dmitry Chagin
2c7660ba3e In preparation for moving linux_clone () to a MI path
move the TLS code in a separate function.

Use function parameter instead of direct using register.
2011-02-12 15:50:21 +00:00
Dmitry Chagin
9bd9b52478 Regen for r218610. 2011-02-12 15:36:25 +00:00
Dmitry Chagin
f91ea2518b The fourth argument of linux_clone is a pointer to the TLS. Change clone syscall definition to match actual linux one. 2011-02-12 15:33:25 +00:00
Alan Cox
02e5228ca0 Setting VV_TEXT here is redundant. It is already set by do_execve().
Reviewed by:	kib
2011-02-09 18:45:33 +00:00
Konstantin Belousov
b17ef03604 Fix linking of the kernel without device npx.
MFC after:	2 weeks
2011-02-05 15:37:10 +00:00
Konstantin Belousov
6f9ec5aab0 Clear the padding when returning context to the usermode, for
MI ucontext_t and x86 MD parts.
Kernel allocates the structures on the stack, and not clearing
reserved fields and paddings causes leakage.

Noted and discussed with:	bde
MFC after:	2 weeks
2011-02-05 15:10:27 +00:00
Matthew D Fleming
08b163fa51 Put the general logic for being a CPU hog into a new function
should_yield().  Use this in various places.  Encapsulate the common
case of check-and-yield into a new function maybe_yield().

Change several checks for a magic number of iterations to use
should_yield() instead.

MFC after:	1 week
2011-02-02 16:35:10 +00:00
Dmitry Chagin
77192fddeb Regen for r218101.
MFC after:	1 Month.
2011-01-30 20:38:26 +00:00
Dmitry Chagin
8d73c2bfd1 Change linux futex syscall definition to match actual linux one.
MFC after:	1 Month.
2011-01-30 20:31:43 +00:00
Dmitry Chagin
9adaae9403 The kern_wait() code already removes the SIGCHLD signal for the waited
process. Removing other SIGCHLD signals is not needed and may cause
problems.

Pointed out by:	jilles

MFC after:	1 Month.
2011-01-30 18:17:38 +00:00
Dmitry Chagin
adc7ece00a Implement a variation of the linux_common_wait() which should
be used by linuxolator itself.

Move linux_wait4() to MD path as it requires native struct
rusage translation to struct l_rusage on linux32/amd64.

MFC after:	1 Month.
2011-01-28 18:47:07 +00:00
Dmitry Chagin
a5c1afadeb Add macro to test the sv_flags of any process. Change some places to test
the flags instead of explicit comparing with address of known sysentvec
structures.

MFC after:	1 month
2011-01-26 20:03:58 +00:00
Matthew D Fleming
f89f7ada8d Set td_kstack_pages for thread0. This was already being done for most
architectures, but i386 and amd64 were missing it.

Submitted by:	Mohd Fahadullah <mfahadullah AT isilon DOT com>
2011-01-26 17:06:13 +00:00
Sergey Kandaurov
4053b05b91 Make MSGBUF_SIZE kernel option a loader tunable kern.msgbufsize.
Submitted by:	perryh pluto.rain.com (previous version)
Reviewed by:	jhb
Approved by:	kib (mentor)
Tested by:	universe
2011-01-21 10:26:26 +00:00
Jung-uk Kim
fd240d6d9f Fix yet another fallout from r208833. VM86 BIOS call may cause page fault
when FPU is in use.

Reported by:	Marc UBM Bocklet (ubm dot freebsd at googlemail dot com)
Tested by:	b. f. (bf1783 at googlemail dot com)
MFC after:	3 days
2011-01-19 17:09:07 +00:00
Konstantin Belousov
55aabb7fd1 For architectures not using direct map , and requiring real KVA page for
sf buf allocation, use wakeup() instead of wakeup_one() to notify sf
buffer waiters about free buffer.

sf_buf_alloc() calls msleep(PCATCH) when SFB_CATCH flag was given,
and for simultaneous wakeup and signal delivery, msleep() returns
EINTR/ERESTART despite the thread was selected for wakeup_one(). As
result, we loose a wakeup, and some other waiter will not be woken up.

Reported and tested by:	az
Reviewed by:	alc, jhb
MFC after:	1 week
2011-01-18 21:57:02 +00:00
John Baldwin
6bd823f334 - Remove some always-true checks (checking for unsigned < 0).
- Only check largs->num against max_ldt_segment on amd64 for I386_SET_LDT
  when descriptors are provided.  Specifically, allow the 'start == 0'
  and 'num == 0' special case used to free all LDT entries that previously
  failed with EINVAL.

Submitted by:	clang via rdivacky (some of 1)
Reviewed by:	kib
2011-01-18 16:43:01 +00:00
Jung-uk Kim
2fea643112 Add reader/writer lock around mem_range_attr_get() and mem_range_attr_set().
Compile sys/dev/mem/memutil.c for all supported platforms and remove now
unnecessary dev_mem_md_init().  Consistently define mem_range_softc from
mem.c for all platforms.  Add missing #include guards for machine/memdev.h
and sys/memrange.h.  Clean up some nearby style(9) nits.

MFC after:	1 month
2011-01-17 22:58:28 +00:00
Jung-uk Kim
df74996c3d Avoid preemption while manipulating CRs and MTRRs.
Tested by:	ariff
2011-01-17 17:30:35 +00:00
John Baldwin
072e9838e2 If an interrupt on an I/O APIC is moved to a different CPU after it has
started to execute, it seems that the corresponding ISR bit in the "old"
local APIC can be cleared.  This causes the local APIC interrupt routine
to fail to find an interrupt to service.  Rather than panic'ing in this
case, simply return from the interrupt without sending an EOI to the
local APIC.  If there are any other pending interrupts in other ISR
registers, the local APIC will assert a new interrupt.

Tested by:	steve
2011-01-13 17:00:22 +00:00
Konstantin Belousov
50a57dfbec Move repeated MAXSLP definition from machine/vmparam.h to sys/vmmeter.h.
Update the outdated comments describing MAXSLP and the process
selection algorithm for swap out.

Comments wording and reviewed by:	alc
2011-01-09 12:50:44 +00:00
Tijl Coosemans
d22e78d6b9 Copy powerpc/include/_inttypes.h to x86 and replace i386/amd64/pc98
headers with stubs.

Approved by:	kib (mentor)
2011-01-08 18:09:48 +00:00
Tijl Coosemans
a56e818f29 On mixed 32/64 bit architectures (mips, powerpc) use __LP64__ rather than
architecture macros (__mips_n64, __powerpc64__) when 64 bit types (and
corresponding macros) are different from 32 bit. [1]

Correct the type of INT64_MIN, INT64_MAX and UINT64_MAX.

Define (U)INTMAX_C as an alias for (U)INT64_C matching the type definition
for (u)intmax_t. Do this on all architectures for consistency.

Suggested by:	bde [1]
Approved by:	kib (mentor)
2011-01-08 12:43:05 +00:00
Tijl Coosemans
d942996baf On 32 bit architectures define (u)int64_t as (unsigned) long long instead
of (unsigned) int __attribute__((__mode__(__DI__))). This aligns better
with macros such as (U)INT64_C, (U)INT64_MAX, etc. which assume (u)int64_t
has type (unsigned) long long.

The mode attribute was used because long long wasn't standardised until
C99. Nowadays compilers should support long long and use of the mode
attribute is discouraged according to GCC Internals documentation.

The type definition has to be marked with __extension__ to support
compilation with "-std=c89 -pedantic".

Discussed with:	bde
Approved by:	kib (mentor)
2011-01-08 11:47:55 +00:00
Tijl Coosemans
9858863cd4 Fix types of some values in machine/_limits.h.
On some architectures UCHAR_MAX and USHRT_MAX had type unsigned int.
However, lacking integer suffixes for types smaller than int, their type
should correspond to that of an object of type unsigned char (or short)
when used in an expression with objects of type int. In that case unsigned
char (short) are promoted to int (i.e. signed) so the type of UCHAR_MAX and
USHRT_MAX should also be int.

Where MIN/MAX constants implicitly have the correct type the suffix has
been removed.

While here, correct some comments.

Reviewed by:	bde
Approved by:	kib (mentor)
2011-01-08 11:13:34 +00:00
Tijl Coosemans
911127a0d6 Remove unused support for 64 bit long on 32 bit architectures.
It was used mainly to discover and fix some 64-bit portability problems
before 64-bit arches were widely available.

Discussed with:	bde
Approved by:	kib (mentor)
2011-01-07 22:57:31 +00:00
Konstantin Belousov
39198f15ee Add AT_STACKPROT elf aux vector. Will be used to inform rtld about the
initial stack protection set by the kernel image activator.
2011-01-07 14:22:34 +00:00
John Baldwin
c305730dc0 Remove bogus usage of INTR_FAST. "Fast" interrupts are now indicated by
registering a filter handler rather than a threaded handler.  Also remove
a bogus use of INTR_MPSAFE for a filter.
2011-01-06 21:08:06 +00:00
Colin Percival
b5e61aab00 Spell CRITICAL_ASSERT correctly.
Submitted by:	jhb
MFC with:	r216944
2011-01-04 16:29:07 +00:00
Colin Percival
aa829cef6a Add hamfisted locking to the Xen/PV pmap code: Only allow one thread to
be in {pmap_pinit, pmap_copy, pmap_release} at a time.

This reduces the rate of panics when running 'make index' from ~0.6/hour
to ~0.02/hour (p < 10^-30).

At a later date this locking will be removed, and for this reason, it is
wrapped in #ifdef HAMFISTED_LOCKING; this temporary hack is being put in
place with the intention of shipping somewhat-stable Xen bits in FreeBSD
8.2-RELEASE.

PR:		kern/153672
MFC after:	3 days
2011-01-04 15:55:15 +00:00
Robert Watson
2913e88c91 Make "options XENHVM" compile for i386, not just amd64 -- a largely
mechanical change.  This opens the door for using PV device drivers
under Xen HVM on i386, as well as more general harmonisation of i386
and amd64 Xen support in FreeBSD.

Reviewed by:    cperciva
MFC after:      3 weeks
2011-01-04 14:49:54 +00:00
Colin Percival
d01b2dad73 Adjust the critical section protecting _xen_flush_queue to cover the
entire range where the page mapping request queue needs to be atomically
examined and modified.

Oddly, while this doesn't seem to affect the overall rate of panics
(running 'make index' on EC2 t1.micro instances, there are 0.6 +/- 0.1
panics per hour, both before and after this change), it eliminates
vm_fault from panic backtraces, leaving only backtraces going through
vmspace_fork.
2011-01-04 00:16:38 +00:00
Colin Percival
aaaf607148 Make i386_set_ldt work on i386/XEN, step 5/5.
When cleaning up a thread, reset its LDT to the default LDT.

Note: Casting the LDT pointer to an int and storing it in pc_currentldt is
wildly bogus, but is harmless since pc_currentldt is a write-only variable.

MFC after:	3 days
2010-12-31 17:42:25 +00:00
Colin Percival
698cc19d6b Make i386_set_ldt work on i386/XEN, step 4/5.
Use xen_update_descriptor to update the LDT rather than bcopy.  Under Xen,
pages used for holding LDTs must be read-only, so we can't make the change
ourselves.

Ths obvious alternative of "remap the page read-write, make the change, then
map it read-only again" doesn't work since Xen won't allow an LDT page to be
remapped as R/W.  An arguably better solution is used by NetBSD: They don't
modify LDTs in-place at all, but instead copy the entire LDT, modify the new
version, then atomically swap.

MFC after:	3 days
2010-12-31 17:41:14 +00:00
Colin Percival
de187b8df2 Make i386_set_ldt work on i386/XEN, step 3/5.
Synchronize reality with comment: The user_ldt_alloc function is supposed to
return with dt_lock held.  Due to broken locking in i386/xen/pmap.c, we drop
dt_lock during the call to pmap_map_readonly and then pick it up again; this
can be removed once the Xen pmap locking is fixed.

MFC after:	3 days
2010-12-31 17:40:30 +00:00
Colin Percival
90b7d33458 Make i386_set_ldt work on i386/XEN, step 2/5.
Don't map physical to machine page numbers in pte_load_store, since it uses
PT_SET_VA (which takes a physical page number and converts it to a machine
page number).

MFC after:	3 days
2010-12-31 17:39:58 +00:00
Colin Percival
d262f2dcfc Make i386_set_ldt work on i386/XEN, step 1/5.
Lock the vm page queue mutex around calls to pte_store.  As with many other
uses of the vm page queue mutex in i386/xen/pmap.c, this is bogus and needs
to be replaced at some future date by a spin lock dedicated to protecting
the queue of pending xen page mapping hypervisor calls.  (But for now, bogus
locking is better than a panic.)

MFC after:	3 days
2010-12-31 17:39:31 +00:00
Pyun YongHyeon
2608aefc0b Add driver for DM&P Vortex86 RDC R6040 Fast Ethernet.
The controller is commonly found on DM&P Vortex86 x86 SoC.  The
driver supports all hardware features except flow control.  The
flow control was intentionally disabled due to silicon bug.

DM&P Electronics, Inc. provided all necessary information including
sample board to write driver and answered many questions I had.
Many thanks for their support of FreeBSD.

H/W donated by:	DM&P Electronics, Inc.
2010-12-31 00:21:41 +00:00
Warner Losh
714cf6c0df Revert r216777, per jhb@ 2010-12-28 22:45:29 +00:00
Warner Losh
1977f3f168 Comment out npx and isa from NOTES file. We don't need them here
since DEFAULTS already pulls them in.
2010-12-28 21:22:08 +00:00
Warner Losh
78b92d19e0 Remove mem, io, isa and npx since they are duplicative of the entries
in DEFAULTS.  Saves 8 lines of warnings when we build XBOX.
2010-12-28 21:20:58 +00:00
Colin Percival
4a416f8375 Remove a "not strictly correct" (and panic-inducing) workaround for a bug
which doesn't seem to exist.

PR:		kern/141328
MFC after:	3 days
2010-12-28 14:36:32 +00:00
Colin Percival
76c9650713 Build the modules which can be built. The excluded modules fall into two
categories: Those which can't build with PAE because they attempt to cast
a pointer to a bus_addr_t (mostly scsi drivers); and those which can't be
built with XEN because they conflict with something in xen-os.h (e.g., in
cxgb there is a conflicting definition of test_and_clear_bit).

MFC after:	1 week
2010-12-27 23:59:27 +00:00
Colin Percival
8ea0b3bb2f Lock the vm page queue mutex in pmap_pte_release around the call
to PMAP_SET_VA; this fixes a mutex-not-held panic when a process
which called mlock(2) exits, and parallels a change made in
pmap_pte 10 months ago (svn r204160).

Note: The locking in this code is utterly broken.  We should not
be using the VM page queue mutex to protect the queue of pending
Xen page mapping hypervisor calls.  Even if it made sense to do
so, this commit and r204160 introduce LORs between the vm page
queue mutex and PMAP2mutex.

(However, a possible deadlock is better than a guaranteed panic,
and this change will hopefully make life easier for whoever fixes
the Xen pmap locking in the future.)

PR:		kern/140313
MFC after:	3 days
2010-12-26 13:05:43 +00:00
Tijl Coosemans
81bd5041a2 Merge amd64 and i386 bus.h and move the resulting header to x86. Replace
the original amd64 and i386 headers with stubs.

Rename (AMD64|I386)_BUS_SPACE_* to X86_BUS_SPACE_* everywhere.

Reviewed by:	imp (previous version), jhb
Approved by:	kib (mentor)
2010-12-20 16:39:43 +00:00
Alan Cox
9d555e459c Redo some parts of r216333, specifically, the locking changes to
pmap_extract_and_hold(), and undo the rest.  In particular, I forgot
that PG_PS and PG_PTE_PAT are the same bit.
2010-12-19 07:31:56 +00:00
Konstantin Belousov
7222d2fbee Inform a compiler which asm statements in the x86 implementation of
atomics change eflags.

Reviewed by:	jhb
MFC after:	2 weeks
2010-12-18 16:41:11 +00:00
Konstantin Belousov
a9b31c256e In pmap_extract(), unlock pmap lock earlier. The calculation does not need
the lock when operating on local variables.

Reviewed by:	alc
2010-12-18 11:31:32 +00:00
Jung-uk Kim
e1c9d39ebe Stop lying about supporting cpu_est_clockrate() when TSC is invariant. This
function always returned the nominal frequency instead of current frequency
because we use RDTSC instruction to calculate difference in CPU ticks, which
is supposedly constant for the case.  Now we support cpu_get_nominal_mhz()
for the case, instead.  Note it should be just enough for most usage cases
because cpu_est_clockrate() is often times abused to find maximum frequency
of the processor.
2010-12-14 20:07:51 +00:00
Konstantin Belousov
60c7c84e85 In fpudna()/npxdna(), mark FPU context initialized and optionally
mark user FPU context initialized, if current context is user context.
It was reversed in r215865, by inadequate change of this code fragment
to a call to fpuuserinited()/npxuserinited().

The issue is only relevant for in-kernel users of FPU.

Reported by:	Jan Henrik Sylvester <me janh de>, Mike Tancsa <mike sentex net>
Tested by:	Mike Tancsa
MFC after:	3 days
2010-12-12 16:16:39 +00:00
Colin Percival
20d1a304b3 Reduce the Xen timecounter from 1GHz to 2^-9 GHz, thereby increasing the
timecounter period from 2^32 ns (~4.3s) to 2^41 ns (~36m39s).  Some time
sharing systems can skip clock interrupts for a few seconds when under
load (e.g., if we've recently used more than our fair share of CPU and
someone else wants a burst of CPU) and we were losing time in quanta of
2^32 ns due to timecounter wrapping.

Increasing the timecounter period up to 2^41 ns is definitely overkill,
but we still have microsecond timecounter precision, and anyone using
paravirtualized hardware when they need submicrosecond timing is crazy.
2010-12-11 22:33:33 +00:00
Colin Percival
0f30ed5bc6 Make the machdep.independent_wallclock sysctl do what it says on the box. 2010-12-11 20:12:42 +00:00
Alan Cox
d1cf854b5d When r207410 eliminated the acquisition and release of the page queues
lock from pmap_extract_and_hold(), it didn't take into account that
pmap_pte_quick() sometimes requires the page queues lock to be held.
This change reimplements pmap_extract_and_hold() such that it no
longer uses pmap_pte_quick(), and thus never requires the page queues
lock.

For consistency, adopt the same idiom as used by the new
implementation of pmap_extract_and_hold() in pmap_extract() and
pmap_mincore().  It also happens to make these functions shorter.

Fix a style error in pmap_pte().

Reviewed by:	kib@
2010-12-09 20:16:00 +00:00
Colin Percival
91ff9dc058 Replace i386/i386/busdma_machdep.c and amd64/amd64/busdma_machdep.c
(which are identical) with a single x86/x86/busdma_machdep.c.
2010-12-09 06:41:50 +00:00
Jung-uk Kim
71e0b05797 Do not subtract 0.5% from estimated frequency if DELAY(9) is driven by TSC.
Remove a confusing comment about converting to MHz as we never did.
2010-12-08 23:40:41 +00:00
Colin Percival
af60888734 On amd64, we have (since r1.72, in December 2005) MAX_BPAGES=8192,
while on i386 we have MAX_BPAGES=512.  Implement this difference via
'#ifdef __i386__'.

With this commit, the i386 and amd64 busdma_machdep.c files become
identical; they will soon be replaced by a single file under sys/x86.
2010-12-08 20:20:10 +00:00
Jung-uk Kim
dd7d207dcb Merge sys/amd64/amd64/tsc.c and sys/i386/i386/tsc.c and move to sys/x86/x86.
Discussed with:	avg
2010-12-08 00:09:24 +00:00
Jung-uk Kim
61d14101dd Use int for 'tsc_present' instead of u_int. It is just a boolean. 2010-12-07 23:19:49 +00:00
Jung-uk Kim
7214d5d75b Remove stale comments about P-state invariant TSC and fix style(9) nits. 2010-12-07 22:43:25 +00:00
Jung-uk Kim
1bcc28295b Do not register a event handler for CPU freqency changes when it is found
P-state invariant.  This is continuation of r216274.
2010-12-07 22:34:51 +00:00
Jung-uk Kim
4a9c4056dc Now the P-state invariant TSC is probed early enough, do not register event
handlers for CPU freqency changes when it is found P-state invariant.
Adjust a comment about non-existent tsc_freq_max() while I am here.
2010-12-07 22:23:26 +00:00
Jung-uk Kim
78a661bbaa Probe P-state invariant TSC from rightful place. 2010-12-07 22:12:02 +00:00
Colin Percival
716d203d6b MFamd64 r204214: Enforce stronger alignment semantics (require that the
end of segments be aligned, not just the start of segments) in order to
allow Xen's blkfront driver to operate correctly.

PR:		kern/152818
MFC after:	3 days
2010-12-05 03:20:55 +00:00
Colin Percival
a39dc31fca Remove gratuitous i386/amd64 inconsistency in favour of the less verbose
version of declaring a variable initialized to zero.
2010-12-04 23:36:40 +00:00
Colin Percival
5c5590862f Remove unnecessary #includes which seem to have been accidentally added
as part of CVS r1.76 (in January 2006).
2010-12-04 23:24:35 +00:00
Jung-uk Kim
2f7ab7e85d Revert r216161. It is not necessary because we zero-fill BSS anyway.
Requested by:	jhb
2010-12-03 22:27:51 +00:00
Jung-uk Kim
b14fe63392 Explicitly initialize TSC frequency. To calibrate TSC frequency, we use
DELAY(9) and it may use TSC in turn if TSC frequency is non-zero.

MFC after:	3 days
2010-12-03 21:54:10 +00:00
Jung-uk Kim
e391a266ed Do not change CPU ticker frequency if TSC is P-state invariant. Note this
change was meant to be committed with r184102 (and its subsequent MFCs) but
it fell off somehow.

Pointyhat to:	jkim
MFC after:	3 days
2010-12-03 21:06:30 +00:00
Rebecca Cran
c90f7d9b44 Revert r216134. This checkin broke platforms where bus_space are macros:
they need to be a single statement, and do { } while (0) doesn't work in this
situation so revert until a solution can be devised.
2010-12-03 07:09:23 +00:00
Rebecca Cran
15b4888a24 Disallow passing in a count of zero bytes to the bus_space(9) functions.
Passing a count of zero on i386 and amd64 for [I386|AMD64]_BUS_SPACE_MEM
causes a crash/hang since the 'loop' instruction decrements the counter
before checking if it's zero.

PR:	kern/80980
Discussed with:	jhb
2010-12-02 22:19:30 +00:00
Colin Percival
d42446149f Fix bug introduced by r194784: Under XEN, the page(s) allocated to dpcpu
for CPU #0 weren't being properly reserved.  Under VM pressure this would
cause problems when the dpcpu structures were overwritten by arbitrary
data; the most common symptom was a panic when netisr attempted to lock a
mutex.

For some reason the XEN code keeps track of the start of available memory
in the variables 'first', 'physfree', and 'init_first'; as far as I can
tell, we always have first == physfree == init_first * PAGE_SIZE.  The
earlier commit adjusted 'first' (which, on !XEN, is the only variable
which tracks this value) but not the other two variables.

Exercise for reader: Eliminate two of these three variables.
2010-11-29 06:50:30 +00:00
Konstantin Belousov
c6fb218c3c Calling fill_fpregs() for curthread is legitimate, and ELF coredump
does this.

Reported and tested by:	pho
MFC after:	5 days
2010-11-28 17:56:34 +00:00
Konstantin Belousov
5c6eb03790 Remove npxgetregs(), npxsetregs(), fpugetregs() and fpusetregs()
functions, they are unused. Remove 'user' from npxgetuserregs()
etc. names.

For {npx,fpu}{get,set}regs(), always use pcb->pcb_user_save for FPU
context storage. This eliminates the need for ugly copying with
overwrite of the newly added and reserved fields in ucontext on i386
to satisfy alignment requirements for fpusave() and fpurstor().

pc98 version was copied from i386.

Suggested and reviewed by:	bde
Tested by:    pho (i386 and amd64)
MFC after:    1 week
2010-11-26 14:50:42 +00:00
Tijl Coosemans
ce4ec51dbe Merge amd64/i386 _align.h by aligning on the size of register_t (copied
from powerpc).

Reviewed by:	imp, jhb
Approved by:	kib (mentor)
2010-11-26 10:59:20 +00:00
Ulrich Spörlein
02604cd4f4 Remove kernel support for BB profiling, now that kernbb(8) is gone, too.
PR:		bin/83558
Reviewed by:	jkim
2010-11-26 08:11:43 +00:00
Colin Percival
5c0ab2fa8b Revert r215819 and fix the bug properly. In pmap_qremove, paging table
updates were being queued by pmap_kremove, but the queue wasn't being
flushed; as a result, the updates didn't happen until *after* the call
to pmap_invalidate_range, and old entries could stick around in the TLB.
Adding a PT_UPDATES_FLUSH() call immediately before pmap_invalidate_range
ensures that after the invalidation the TLB will be repopulated with the
correct new entries.

Thanks to:	kib, avg, alc
2010-11-25 22:06:07 +00:00
Dimitry Andric
079d7e43ca Use unambiguous inline assembly to load a float variable. GNU as
silently converts 'fld' to 'flds', without taking the actual variable
type into account (!), but clang's integrated assembler rightfully
complains about it.

Discussed with:	cperciva
2010-11-25 18:14:18 +00:00
John Baldwin
9d76324839 Add device IDs for two more ServerWorks Host-PCI bridges so that we can
read their starting PCI bus number for older systems that do not support
ACPI (or have a broken _BBN method).

PR:		kern/148108
MFC after:	1 week
2010-11-25 15:42:33 +00:00
Colin Percival
98702b3990 Work around paging bug. Somehow we seem to be ending up with entries in
the TLB which don't correspond to ptes with PG_V set; prior to this commit
I'm sometimes getting the wrong data when pages are loaded into the buffer
cache (they're being loaded, but the missing TLB invalidation is causing
the wrong data to be visible).
2010-11-25 15:41:34 +00:00
Colin Percival
1a3b2b87de Rename HYPERVISOR_multicall (which performs the multicall hypercall) to
_HYPERVISOR_multicall, and create a new HYPERVISOR_multicall function which
invokes _HYPERVISOR_multicall and checks that the individual hypercalls all
succeeded.
2010-11-25 15:05:21 +00:00
Colin Percival
0bd7a92067 Remove vestigal debugging code which, in fork-heavy workloads, can cause
a 30x slowdown.
2010-11-25 04:45:31 +00:00
Jung-uk Kim
d2d0fda841 Remove a stale tunable introduced in r215703. 2010-11-23 17:28:23 +00:00
Andriy Gapon
9b984feb3d specialreg.h: add definitions for some useful bits found in CPUID.6 EAX and ECX
CPUID.6 is defined as Thermal and Power Management Leaf by both Intel
and AMD.

Reviewed by:	jhb
MFC after:	7 days
2010-11-23 13:55:30 +00:00
Jung-uk Kim
7dd052c1d9 - Disable caches and flush caches/TLBs when we update PAT as we do for MTRR.
Flushing TLBs is required to ensure cache coherency according to the AMD64
architecture manual.  Flushing caches is only required when changing from a
cacheable memory type (WB, WP, or WT) to an uncacheable type (WC, UC, or
UC-).  Since this function is only used once per processor during startup,
there is no need to take any shortcuts.
- Leave PAT indices 0-3 at the default of WB, WT, UC-, and UC.  Program 5 as
WP (from default WT) and 6 as WC (from default UC-).  Leave 4 and 7 at the
default of WB and UC.  This is to avoid transition from a cacheable memory
type to an uncacheable type to minimize possible cache incoherency.  Since
we perform flushing caches and TLBs now, this change may not be necessary
any more but we do not want to take any chances.
- Remove Apple hardware specific quirks.  With the above changes, it seems
this hack is no longer needed.
- Improve pmap_cache_bits() with an array to map PAT memory type to index.
This array is initialized early from pmap_init_pat(), so that we do not need
to handle special cases in the function any more.  Now this function is
identical on both amd64 and i386.

Reviewed by:	jhb
Tested by:	RM (reuf_m at hotmail dot com)
		Ryszard Czekaj (rychoo at freeshell dot net)
		army.of.root (army dot of dot root at googlemail dot com)
MFC after:	3 days
2010-11-22 19:52:44 +00:00
Colin Percival
c3f128981e In xen_get_timecount, return the full ns-precision time rather than
rounding to 1/HZ precision.

I have no idea why the rounding was introduced in the first place, but
it makes FreeBSD unhappy.
2010-11-22 09:04:29 +00:00
Colin Percival
61381fcf2d Unifdef XEN. This file is only compiled with the XEN kernel option set,
and the !XEN bits get in the way of understanding the code.
2010-11-20 21:36:12 +00:00
Colin Percival
8cdabbaf32 Add VTOM(va) macro as xpmap_ptom(VTOP(va)) to convert to machine addresses.
Clean up the code by converting xpmap_ptom(VTOP(...)) to VTOM(...) and
converting xpmap_ptom(VM_PAGE_TO_PHYS(...)) to VM_PAGE_TO_MACH(...).  In
a few places we take advantage of the fact that xpmap_ptom can commute with
setting PG_* flags.

This commit should have no net effect save to improve the readability of
this code.
2010-11-20 20:04:29 +00:00
Colin Percival
ad520892d7 Make pmap_release consistent with pmap_pinit with respect to unpinning
pages.  The pinning of NPGPTD pages is #if 0ed out in pmap_pinit (I'm
not quite sure why...) and this commit adds a corresponding #if 0 in
pmap_release to avoid unpinning those pages.

Some versions of Xen seem to silently ignore requests to unpin pages
which were never pinned in the first place, but some return an error
(causing FreeBSD to panic) prior to this commit.
2010-11-19 15:12:19 +00:00
Andriy Gapon
b43d292565 specialreg.h: add definitions for MPERF/APERF pair of MSRs
These MSRs can be used to determine actual (average) performance as
compared to a maximum defined performance.
Availability of these MSRs is indicated by bit0 in CPUID.6.ECX on both
Intel and AMD processors.

MFC after:	5 days
2010-11-19 15:07:36 +00:00
Andriy Gapon
7af7c7624a specialreg.h: add AMD-specific "Hardware Configuration Register" MSR
It seems that this MSR has been available in a range of AMD processors
families for quite a while now.

Note1: not all AMD MSRs that are found in amd64 specialreg.h are also in
the i386 version.
Note2: perhaps some additional name component is needed to distinguish
AMD-specific MSRs.

MFC after:	5 days
2010-11-19 15:00:20 +00:00