Commit Graph

57 Commits

Author SHA1 Message Date
Jeff Roberson
50c202c592 Implement a facility for dynamic per-cpu variables.
- Modules and kernel code alike may use DPCPU_DEFINE(),
   DPCPU_GET(), DPCPU_SET(), etc. akin to the statically defined
   PCPU_*.  Requires only one extra instruction more than PCPU_* and is
   virtually the same as __thread for builtin and much faster for shared
   objects.  DPCPU variables can be initialized when defined.
 - Modules are supported by relocating the module's per-cpu linker set
   over space reserved in the kernel.  Modules may fail to load if there
   is insufficient space available.
 - Track space available for modules with a one-off extent allocator.
   Free may block for memory to allocate space for an extent.

Reviewed by:    jhb, rwatson, kan, sam, grehan, marius, marcel, stas
2009-06-23 22:42:39 +00:00
Ed Schouten
5942207fb4 Revert my change; reintroduce __gnu89_inline.
It turns out our compiler in stable/7 can't build this code anymore.
Even though my opinion is that those people should just run `make
kernel-toolchain' before building a kernel, I am willing to wait and
commit this after we've branched stable/8.

Requested by:	rwatson
2009-06-08 18:23:43 +00:00
Ed Schouten
032e3d1d19 Remove __gnu89_inline.
Now that we use C99 almost everywhere, just use C99-style in the pmap
code. Since the pmap code is the only consumer of __gnu89_inline, remove
it from cdefs.h as well. Because the flag was only introduced 17 months
ago, I don't expect any problems.

Reviewed by:	alc
2009-06-08 17:27:25 +00:00
Adrian Chadd
c22ca7f04f Fix the MP IPI code to differentiate between bitmapped IPIs and function IPIs.
This attempts to fix the IPI handling code to correctly differentiate
between bitmapped IPIs and function IPIs. The Xen IPIs were on low numbers
which clashed with the bitmapped IPIs.

This commit bumps those IPI numbers up to 240 and above (just like in the i386
code) and fiddles with the ipi_vectors[] logic to call the correct function.

This still isn't "right". Specifically, the IPI code may work fine for TLB
shootdown events but the rendezvous/lazypmap IPIs are thrown by calling ipi_*()
routines which don't set the call_func stuff (function id, addr1, addr2) that
the TLB shootdown events are. So the Xen SMP support is still broken.

PR:		135069
2009-05-31 08:11:39 +00:00
Adrian Chadd
23ca223bfa Remove some unused code in ipi_selected() .
The code path this was copied from (sys/i386/i386/mp_machdep.c:ipi_selected())
handles bitmap'ed IPIs and normal IPIs via separate notification paths. Xen
SMP handles them the same way.
2009-05-31 07:25:24 +00:00
Adrian Chadd
eee45687e9 Even though I'm not quite sure that the call_func stuff will work properly
in all the places/cases IPI messages will be generated, at least be consistent
with how the call_data pointer is assigned and cleared (ie, all done inside
the spinlock.

Ensure that its NULL before continuing, just to try and identify situations
where things are going horribly wrong.
2009-05-30 15:20:25 +00:00
Adrian Chadd
d68bbb81fb Don't schedule a CALL_FUNCTION_VECTOR software IPI if the IPI was signaled
via the bitmap (and thus sent via RESCHEDULE_VECTOR.)
2009-05-30 14:59:08 +00:00
Adrian Chadd
a235aceb1a Correctly report the IPI IRQs being created; make it clear what vectors they are for. 2009-05-30 06:37:03 +00:00
Adrian Chadd
fd27593538 Fix the Xen TOD update when the hypervisor wall clock is nudged.
The "wall clock" in the current code is actually the hypervisor start time.
The time of day is the "start time" plus the hypervisor "uptime".

Large enough bumps in the dom0 clock lead to a hypervisor "bump" which is
implemented as a bump in the start time, not the uptime. The clock.c routines
were reading in the hypervisor start time and then using this as the TOD.
This meant that any hypervisor time bump would cause the FreeBSD DomU to
set its TOD to the hypervisor start time, rather than the actual TOD.

This fix is a bit hacky and some reshuffling should be done later on
to clarify what is going on. I've left the wall clock code alone.
(The code which updates shadow_tv and shadow_tv_version.)
A new routine adds the uptime to the shadow_tv, which is then used to
update the TOD.

I've included some debugging so it is obvious when the clock is nudged.

PR:	135008
2009-05-29 13:43:21 +00:00
Adrian Chadd
7d18ff9a2b Migrate the Xen hypervisor clock reading routines into something
sharable.
2009-05-29 13:36:06 +00:00
Adrian Chadd
58e3a119e5 Say hello to a very basic, read-only, Xen Hypervisor RTC.
The hypervisor doesn't provide a single "TOD" - it instead provides a
"start time" and a "running time". These are added together to form
the current TOD. The TOD is in UTC.

This RTC is only (initially) designed to be read at startup. There's
some further poking that needs to happen to pick up hypervisor time
changes (ie, by the Dom0 time being adjusted by something). This
time adjustment currently can cause "weird stuff" in the DomU clock;
I'll begin investigating and repairing that in subsequent commits.

PR:		135008
2009-05-28 04:17:05 +00:00
Attilio Rao
120b18d86f FreeBSD right now support 32 CPUs on all the architectures at least.
With the arrival of 128+ cores it is necessary to handle more than that.
One of the first thing to change is the support for cpumask_t that needs
to handle more than 32 bits masking (which happens now).  Some places,
however, still assume that cpumask_t is a 32 bits mask.
Fix that situation by using always correctly cpumask_t when needed.

While here, remove the part under STOP_NMI for the Xen support as it
is broken in any case.

Additively make ipi_nmi_pending as static.

Reviewed by:	jhb, kmacy
Tested by:	Giovanni Trematerra <giovanni dot trematerra at gmail dot com>
2009-05-14 17:43:00 +00:00
Alexander Motin
1703f2b424 Rename statclock_disable variable to atrtcclock_disable that it actually is,
and hide it inside of atrtc driver. Add new tunable hint.atrtc.0.clock
controlling it. Setting it to 0 disables using RTC clock as stat-/
profclock sources.

Teach i386 and amd64 SMP platforms to emulate stat-/profclocks using i8254
hardclock, when LAPIC and RTC clocks are disabled.

This allows to reduce global interrupt rate of idle system down to about
100 interrupts per core, permitting C3 and deeper C-states provide maximum
CPU power efficiency.
2009-05-03 17:47:21 +00:00
Kip Macy
51a70df2db fix XEN compilation 2009-05-02 22:22:00 +00:00
Doug Rabson
3e33218d77 Fix the Xen build for i386 PV mode. 2009-04-01 17:06:28 +00:00
John Baldwin
957f3ee2f7 Remove unused arg from npxinit(). Forgot to commit this file in the
last i386 FPU change.
2009-03-05 18:43:54 +00:00
Kip Macy
69f93ab8b5 - fix formatting
- fix types in ticks_to_system_time
2009-02-15 06:36:02 +00:00
Alan Cox
2f0b74fb06 Remove unnecessary page queues locking around vm_page_wakeup().
Approved by:	kmacy
2009-02-14 22:07:22 +00:00
Kip Macy
ffc4f90610 Don't try to directly update page tables 2009-02-08 21:54:51 +00:00
Kip Macy
2650d60534 pass in smp_processor_id to identify the cpu in use 2009-02-05 04:00:55 +00:00
Kip Macy
bd1c698cab adjust the way that idle happens so as to avoid missing timer interrupts 2009-02-05 02:01:18 +00:00
Kip Macy
476f05d531 make sure that interrupts are disabled when handling page faults et al 2009-02-03 03:43:00 +00:00
Bjoern A. Zeeb
e72f0a071c Bring over the code from sys/i386/i386/mp_machdep.c, r187880
to make XEN config compile again:
- Allocate apic vectors on a per-cpu basis.
2009-01-31 21:40:27 +00:00
Kip Macy
3a6d1fcf9c merge 186535, 186537, and 186538 from releng_7_xen
Log:
 - merge in latest xenbus from dfr's xenhvm
 - fix race condition in xs_read_reply by converting tsleep to mtx_sleep

Log:
 unmask evtchn in bind_{virq, ipi}_to_irq

Log:
 - remove code for handling case of not being able to sleep
 - eliminate tsleep - make sleeps atomic
2008-12-29 06:31:03 +00:00
Kip Macy
23dc562170 Integrate 185578 from dfr
Use newbus to managed devices
2008-12-04 07:59:05 +00:00
Kip Macy
cbc936b6d8 fix initialization for case of normal kernbase
remove unused shutdown code
2008-12-04 07:28:13 +00:00
Kip Macy
e67fc8f258 only call hardclock on cpu0
pointed out by: Scott Long
2008-10-25 20:42:10 +00:00
Kip Macy
951f87e4ae handle case where eflags represents actual flags value when
restoring interrupts
2008-10-25 04:46:02 +00:00
Kip Macy
1f5aa99363 Fix general issues with IPI support 2008-10-24 07:58:38 +00:00
Kip Macy
b1efbd6b47 Fix IPI support 2008-10-23 07:20:43 +00:00
Kip Macy
a51ac24407 Hook in ipi handlers 2008-10-21 08:03:12 +00:00
Kip Macy
764b7ef593 Implement infrastructure for gluing i386 ipi functions in to xen's infrastructure 2008-10-21 06:39:40 +00:00
Kip Macy
756fb0605f Add routine for initializing AP clock 2008-10-21 06:38:40 +00:00
Kip Macy
39f8ee3024 Import interrupt management defines from latest xenlinux 2008-10-20 05:42:38 +00:00
Kip Macy
9bf38e47a3 - move gdt, ldt allocation to before KPT allocation
- fix bugs where we would:
    - try to map the hypervisors address space
    - accidentally kick out an existing kernel mapping for some domain creation memory allocation sizes
    - accidentally skip a 2MB kernel mapping for some domain creation memory allocation sizes
- don't rely on trapping in to xen to read rcr2, reference through vcpu
- whitespace cleanups
2008-10-19 01:27:40 +00:00
Kip Macy
ba32964d08 GC unused values 2008-10-19 01:23:30 +00:00
Marius Strobl
6f04e7b9aa Remove ipi_all() and ipi_self() as the former hasn't been used at
all to date and the latter also is only used in ia64 and powerpc
code which no longer serves a real purpose after bring-up and just
can be removed as well. Note that architectures like sun4u also
provide no means of implementing IPI'ing a CPU itself natively
in the first place.

Suggested by:	jhb
Reviewed by:	arch, grehan, jhb
2008-09-28 18:34:14 +00:00
Kip Macy
852c25eda2 move ipi_pcpu to evtchn.c 2008-09-26 05:54:24 +00:00
Kip Macy
920ba15bf9 Update xen/interface includes to the latest in mercurial
MFC after:	1 month
2008-09-26 05:29:39 +00:00
Kip Macy
4e01238d67 add initial ipi definitions
MFC after:	1 month
2008-09-25 07:11:04 +00:00
Kip Macy
9d741e6d59 Make nkpt dependent on the size of the initial memory allocation
MFC after:	1 month
2008-09-25 07:03:09 +00:00
Kip Macy
425eaba985 Change order of pcpu initialization so the pc_prvspace is set
MFC after:	1 month
2008-09-18 02:59:19 +00:00
Kip Macy
6cf8efdc48 fix initial page directory setup for APs to work when KERNBASE < 0xc0000000
MFC after:	1 month
2008-09-18 01:09:15 +00:00
Kip Macy
57bd99b3c5 Some people have very strange notions of how large KVA_PAGES should be. The
core of this change generalizes the initial page directory setup so that
the kernel can be given arbitrarily large or small.

- small formatting fixes
- update copyright

MFC after:	1 month
2008-09-17 19:11:37 +00:00
Kip Macy
6859a304c6 Get initial bootstrap of APs working under xen.
Note that the APs still blow up in sched_throw().

MFC after:	1 month
2008-09-10 07:11:08 +00:00
Kip Macy
1fc7c4a654 enable the xen_guest string so that the freebsd xen kernel will
at least pass muster with the loader on 3.0.3

Note that this doesn't actually make it work as Xen 3.0.3
appears to disallow recursive mappings on the page directory
2008-09-03 00:06:10 +00:00
Kip Macy
42e68d4d74 Accomodate the fact that the number of l1 pages varies with the size of the initially
allocated memory - this lets us boot with 3GB

MFC after:	1 month
2008-09-02 02:55:45 +00:00
Kip Macy
32a5d14b26 Fix problem with large memory configuration by ensuring that all NKPT
page table pages have been zeroed before entering them in the page
directory

MFC after:	1 month
2008-09-02 01:32:52 +00:00
Kip Macy
4e683d7252 Fix boot time pmap_growkernel panic for case where vm is allocated >= 768M
MFC after:	1 month
2008-08-21 02:57:02 +00:00
Kip Macy
18bad85737 - clean up interrupt handling for xen a tiny bit
- parse the command line in to kenv
- defer shutdown watcher until later in boot

MFC after:	1 month
2008-08-20 09:16:46 +00:00