Commit Graph

223 Commits

Author SHA1 Message Date
Mateusz Guzik
4ce60b6549 agp: clean up empty lines in .c and .h files 2020-09-01 21:35:39 +00:00
Warner Losh
4f8959b9f4 Remove support for FreeBSD 4.x and earlier from agp driver
Compile tested only, but do we still need this driver?
2020-02-26 19:39:52 +00:00
Mark Johnston
b119329d81 Complete the removal of the "wire_count" field from struct vm_page.
Convert all remaining references to that field to "ref_count" and update
comments accordingly.  No functional change intended.

Reviewed by:	alc, kib
Sponsored by:	Intel, Netflix
Differential Revision:	https://reviews.freebsd.org/D21768
2019-09-25 16:11:35 +00:00
Mark Johnston
fee2a2fa39 Change synchonization rules for vm_page reference counting.
There are several mechanisms by which a vm_page reference is held,
preventing the page from being freed back to the page allocator.  In
particular, holding the page's object lock is sufficient to prevent the
page from being freed; holding the busy lock or a wiring is sufficent as
well.  These references are protected by the page lock, which must
therefore be acquired for many per-page operations.  This results in
false sharing since the page locks are external to the vm_page
structures themselves and each lock protects multiple structures.

Transition to using an atomically updated per-page reference counter.
The object's reference is counted using a flag bit in the counter.  A
second flag bit is used to atomically block new references via
pmap_extract_and_hold() while removing managed mappings of a page.
Thus, the reference count of a page is guaranteed not to increase if the
page is unbusied, unmapped, and the object's write lock is held.  As
a consequence of this, the page lock no longer protects a page's
identity; operations which move pages between objects are now
synchronized solely by the objects' locks.

The vm_page_wire() and vm_page_unwire() KPIs are changed.  The former
requires that either the object lock or the busy lock is held.  The
latter no longer has a return value and may free the page if it releases
the last reference to that page.  vm_page_unwire_noq() behaves the same
as before; the caller is responsible for checking its return value and
freeing or enqueuing the page as appropriate.  vm_page_wire_mapped() is
introduced for use in pmap_extract_and_hold().  It fails if the page is
concurrently being unmapped, typically triggering a fallback to the
fault handler.  vm_page_wire() no longer requires the page lock and
vm_page_unwire() now internally acquires the page lock when releasing
the last wiring of a page (since the page lock still protects a page's
queue state).  In particular, synchronization details are no longer
leaked into the caller.

The change excises the page lock from several frequently executed code
paths.  In particular, vm_object_terminate() no longer bounces between
page locks as it releases an object's pages, and direct I/O and
sendfile(SF_NOCACHE) completions no longer require the page lock.  In
these latter cases we now get linear scalability in the common scenario
where different threads are operating on different files.

__FreeBSD_version is bumped.  The DRM ports have been updated to
accomodate the KPI changes.

Reviewed by:	jeff (earlier version)
Tested by:	gallatin (earlier version), pho
Sponsored by:	Netflix
Differential Revision:	https://reviews.freebsd.org/D20486
2019-09-09 21:32:42 +00:00
Alan Cox
49bfa624ac Eliminate the arena parameter to kmem_free(). Implicitly this corrects an
error in the function hypercall_memfree(), where the wrong arena was being
passed to kmem_free().

Introduce a per-page flag, VPO_KMEM_EXEC, to mark physical pages that are
mapped in kmem with execute permissions.  Use this flag to determine which
arena the kmem virtual addresses are returned to.

Eliminate UMA_SLAB_KRWX.  The introduction of VPO_KMEM_EXEC makes it
redundant.

Update the nearby comment for UMA_SLAB_KERNEL.

Reviewed by:	kib, markj
Discussed with:	jeff
Approved by:	re (marius)
Differential Revision:	https://reviews.freebsd.org/D16845
2018-08-25 19:38:08 +00:00
Alan Cox
44d0efb215 Eliminate kmem_alloc_contig()'s unused arena parameter.
Reviewed by:	hselasky, kib, markj
Discussed with:	jeff
Differential Revision:	https://reviews.freebsd.org/D16799
2018-08-20 15:57:27 +00:00
Alan Cox
94d0f0877d Oops. r338030 didn't eliminate the unused arena argument from all of
kmem_alloc_attr()'s callers.  Correct that mistake.
2018-08-18 22:35:19 +00:00
Pedro F. Giffuni
718cf2ccb9 sys/dev: further adoption of SPDX licensing ID tags.
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.

The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
2017-11-27 14:52:40 +00:00
Mark Johnston
8ee96d1452 Add a helper function to agp(4) which installs a single GTT entry.
Submitted by:	dumbbell
MFC after:	1 week
2017-07-09 22:53:24 +00:00
Pedro F. Giffuni
4d24901ac9 sys/dev: Replace zero with NULL for pointers.
Makes things easier to read, plus architectures may set NULL to something
different than zero.

Found with:	devel/coccinelle
MFC after:	3 weeks
2017-02-20 03:43:12 +00:00
Gleb Smirnoff
bfc8c24c73 Move bogus_page declaration to vm_page.h and initialization to vm_page.c.
Reviewed by:	kib
2017-01-04 22:27:19 +00:00
Pedro F. Giffuni
d9c9c81c08 sys: use our roundup2/rounddown2() macros when param.h is available.
rounddown2 tends to produce longer lines than the original code
and when the code has a high indentation level it was not really
advantageous to do the replacement.

This tries to strike a balance between readability using the macros
and flexibility of having the expressions, so not everything is
converted.
2016-04-21 19:57:40 +00:00
Pedro F. Giffuni
4ec642f1d6 dev/agp: use our nitems() macro when it is avaliable through param.h.
Consistently capitalize the macros used in the driver.

No functional change.
2016-04-19 23:31:35 +00:00
Jean-Sébastien Pédron
85a60a7756 agp: Do not attach to Intel GEN6+
The i915 video driver doesn't depend on agp(4) anymore for Sandybridge
and later GPUs. Therefore, there is no need to attach agp(4) to those
devices.

While here, fix `agp_i965_res_spec` to include the aperture base for
GEN4 and GEN5.

Reviewed by:	kib
Approved by:	kib
Differential Revision:	https://reviews.freebsd.org/D5586
2016-03-12 09:44:23 +00:00
Jean-Sébastien Pédron
740be6d755 drm/i915: Update to match Linux 3.8.13
This update brings initial support for Haswell GPUs.

Tested by:	Many users of FreeBSD, PC-BSD and HardenedBSD
Relnotes:	yes
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D5554
2016-03-08 20:33:02 +00:00
Kevin Lo
128236c022 Remove break after return. 2015-01-23 15:14:30 +00:00
Konstantin Belousov
a537a017e8 Initial attachment of the agp(4) to Haswell IGP. There is no handling
of cacheablility control bits in GTT PTEs yet.

Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2014-11-11 12:56:30 +00:00
Konstantin Belousov
d6fe56e5d7 Based on some BIOS configuration (GGC register in host bridge, bit 1),
IGP may declare subclass as either VGA-compatible, or non-VGA.  The
difference is that in the later case, IGP does not claim VGA cycles.
Other than that, the device functions normally, and agp_i810 should
attach to it.

Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2014-11-11 12:52:45 +00:00
Konstantin Belousov
a41f4046f5 On 965 and higher, map GTT as write-combining.
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2014-11-11 12:37:13 +00:00
Konstantin Belousov
53dc09a8e6 Revert agp_intel.c part of the r274040 for now. There is a confusion
on my part about north bridge/GPU pci ids and use of aperture.

Leave the agp_intel.c out of static compilation on amd64, it makes the
things consistent with agp.ko.

Pointed out by:	tijl
Sponsored by:	The FreeBSD Foundation
MFC after:	13 days
2014-11-04 09:56:58 +00:00
Konstantin Belousov
a0cc5254d9 agp_intel.c provides support for AGP on Intel chipsets from 440/BX to
875.  This intersects with the agp_i810.c, which supports all Intels
from i810 to Core i5/7.  Both agp_intel.c and agp_i810.c are compiled
into kernel when device agp is specified in config, and agp_i810
attach seems to be selected by chance due to linking order.

Strip support for 810 and later from agp_intel.c.  Since 440-class
chipsets do not support any long-mode capable CPUs, remove agp_intel.c
from amd64 kernel file list.  Note that agp_intel.c is not compiled
into agp.ko on amd64 already.

Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
2014-11-03 11:49:35 +00:00
Tijl Coosemans
0de6ac2e3c - agp_generic_unbind_memory: flush AGP TLB before unwiring pages
- agp_bind_pages: assert that pages have been wired down

MFC after:	1 month
2014-11-02 11:47:40 +00:00
Tijl Coosemans
a8ac51bbb6 In agp_amd_bind_page don't flush the AGP TLB. It's done by the calling
function.
2014-11-02 11:28:15 +00:00
Tijl Coosemans
4e612cddb9 In agp(4) avoid the need to flush all cpu caches with wbinvd between
updating the GTT and flushing the AGP TLB by storing the GTT in
write-combining memory.

On x86 flushing the AGP TLB is done by an I/O operation or a store to a
MMIO register in uncacheable memory.  Both cases imply that WC buffers are
flushed so no memory barriers are needed.

On powerpc there is no WC memory type.  It maps to uncacheable memory and
two stores to uncacheable memory, such as to the GTT and then to an MMIO
register, are strongly ordered, so no memory barriers are needed either.

MFC after:	1 month
2014-11-02 11:26:37 +00:00
Tijl Coosemans
cf99ea5d13 Avoid possible overflow in agp_generic_alloc_memory.
MFC after:	1 week
2014-10-30 14:36:27 +00:00
Tijl Coosemans
903fb143cf Add two new functions to the AGP driver KPI to bind/unbind arbitrary sets
of pages into the GTT.

Reviewed by:	kib
MFC after:	1 month
2014-10-30 11:27:03 +00:00
Attilio Rao
3ae10f7477 - Modify vm_page_unwire() and vm_page_enqueue() to directly accept
the queue where to enqueue pages that are going to be unwired.
- Add stronger checks to the enqueue/dequeue for the pagequeues when
  adding and removing pages to them.

Of course, for unmanaged pages the queue parameter of vm_page_unwire() will
be ignored, just as the active parameter today.
This makes adding new pagequeues quicker.

This change effectively modifies the KPI.  __FreeBSD_version will be,
however, bumped just when the full cache of free pages will be
evicted.

Sponsored by:	EMC / Isilon storage division
Reviewed by:	alc
Tested by:	pho
2014-06-16 18:15:27 +00:00
Eitan Adler
7a22215c53 Fix undefined behavior: (1 << 31) is not defined as 1 is an int and this
shifts into the sign bit.  Instead use (1U << 31) which gets the
expected result.

This fix is not ideal as it assumes a 32 bit int, but does fix the issue
for most cases.

A similar change was made in OpenBSD.

Discussed with:	-arch, rdivacky
Reviewed by:	cperciva
2013-11-30 22:17:27 +00:00
Konstantin Belousov
5944de8ecd Remove the deprecated VM_ALLOC_RETRY flag for the vm_page_grab(9).
The flag was mandatory since r209792, where vm_page_grab(9) was
changed to only support the alloc retry semantic.

Suggested and reviewed by:	alc
Sponsored by:	The FreeBSD Foundation
2013-08-22 07:39:53 +00:00
Attilio Rao
c7aebda8a1 The soft and hard busy mechanism rely on the vm object lock to work.
Unify the 2 concept into a real, minimal, sxlock where the shared
acquisition represent the soft busy and the exclusive acquisition
represent the hard busy.
The old VPO_WANTED mechanism becames the hard-path for this new lock
and it becomes per-page rather than per-object.
The vm_object lock becames an interlock for this functionality:
it can be held in both read or write mode.
However, if the vm_object lock is held in read mode while acquiring
or releasing the busy state, the thread owner cannot make any
assumption on the busy state unless it is also busying it.

Also:
- Add a new flag to directly shared busy pages while vm_page_alloc
  and vm_page_grab are being executed.  This will be very helpful
  once these functions happen under a read object lock.
- Move the swapping sleep into its own per-object flag

The KPI is heavilly changed this is why the version is bumped.
It is very likely that some VM ports users will need to change
their own code.

Sponsored by:	EMC / Isilon storage division
Discussed with:	alc
Reviewed by:	jeff, kib
Tested by:	gavin, bapt (older version)
Tested by:	pho, scottl
2013-08-09 11:11:11 +00:00
Attilio Rao
89f6b8632c Switch the vm_object mutex to be a rwlock. This will enable in the
future further optimizations where the vm_object lock will be held
in read mode most of the time the page cache resident pool of pages
are accessed for reading purposes.

The change is mostly mechanical but few notes are reported:
* The KPI changes as follow:
  - VM_OBJECT_LOCK() -> VM_OBJECT_WLOCK()
  - VM_OBJECT_TRYLOCK() -> VM_OBJECT_TRYWLOCK()
  - VM_OBJECT_UNLOCK() -> VM_OBJECT_WUNLOCK()
  - VM_OBJECT_LOCK_ASSERT(MA_OWNED) -> VM_OBJECT_ASSERT_WLOCKED()
    (in order to avoid visibility of implementation details)
  - The read-mode operations are added:
    VM_OBJECT_RLOCK(), VM_OBJECT_TRYRLOCK(), VM_OBJECT_RUNLOCK(),
    VM_OBJECT_ASSERT_RLOCKED(), VM_OBJECT_ASSERT_LOCKED()
* The vm/vm_pager.h namespace pollution avoidance (forcing requiring
  sys/mutex.h in consumers directly to cater its inlining functions
  using VM_OBJECT_LOCK()) imposes that all the vm/vm_pager.h
  consumers now must include also sys/rwlock.h.
* zfs requires a quite convoluted fix to include FreeBSD rwlocks into
  the compat layer because the name clash between FreeBSD and solaris
  versions must be avoided.
  At this purpose zfs redefines the vm_object locking functions
  directly, isolating the FreeBSD components in specific compat stubs.

The KPI results heavilly broken by this commit.  Thirdy part ports must
be updated accordingly (I can think off-hand of VirtualBox, for example).

Sponsored by:	EMC / Isilon storage division
Reviewed by:	jeff
Reviewed by:	pjd (ZFS specific review)
Discussed with:	alc
Tested by:	pho
2013-03-09 02:32:23 +00:00
Konstantin Belousov
5f451cd59b Fix reversed condition in the logic to wait for the chipset buffers
flush wait on the Gen2 chipsets.  Confirmed by the inspection of the
Linux agp code.

Submitted by:	Taku YAMAMOTO <taku@tackymt.homeip.net>
MFC after:	2 weeks
2013-01-27 09:31:11 +00:00
Antoine Brodin
dfa4d7fdb9 Remove unneeded semicolons.
Reviewed by:	md5 of the object files
2013-01-01 18:16:49 +00:00
Baptiste Daroussin
af62061042 Add pci id for the xeon hd4000 (IvyBridge server GT2)
Submitted by:	François Tigeot <ftigeot@wolfpond.org>
Obtained from:	dragonfly
MFC after:	3 days
2012-12-11 09:38:12 +00:00
Eitan Adler
9e93aa2336 Remove unneeded header from agp: opt_bus.h
Tested with "make universe"

Approved by:	cperciva
MFC after:	1 week
2012-11-15 18:49:17 +00:00
Eitan Adler
a8de37b024 This isn't functionally identical. In some cases a hint to disable
unit 0 would in fact disable all units.

This reverts r241856

Approved by: cperciva (implicit)
2012-10-22 13:06:09 +00:00
Eitan Adler
76b7512247 Now that device disabling is generic, remove extraneous code from the
device drivers that used to provide this feature.

Reviewed by:	des
Approved by:	cperciva
MFC after:	1 week
2012-10-22 03:41:14 +00:00
Konstantin Belousov
1c771f9222 After the PHYS_TO_VM_PAGE() function was de-inlined, the main reason
to pull vm_param.h was removed.  Other big dependency of vm_page.h on
vm_param.h are PA_LOCK* definitions, which are only needed for
in-kernel code, because modules use KBI-safe functions to lock the
pages.

Stop including vm_param.h into vm_page.h. Include vm_param.h
explicitely for the kernel code which needs it.

Suggested and reviewed by:	alc
MFC after:    2 weeks
2012-08-05 14:11:42 +00:00
Marcel Moolenaar
cbf1d2e3f6 Revert revision 238172 of agp_i810.c. Correctness is considered more
important than avoiding confusion.

Feedback from: kib, jhb
2012-07-09 16:23:59 +00:00
Marcel Moolenaar
e11379e9f8 agp.c:
Don't use Maxmem when the amount of memory is meant. Use realmem instead.
Maxmem is not only a MD variable, it represents the highest physical memory
address in use. On systems where memory is sparsely layed-out the highest
memory address and the amount of memory are not interchangeable. Scaling the
AGP aperture based on the actual amount of memory (= realmem) rather than
the available memory (= physmem) makes sure there's consistent behaviour
across architectures.

agp_i810.c:
While arguably the use of Maxmem can be considered correct, replace its use
with realmem anyway. agp_i810.c is specific to amd64, i386 & pc98, which
have a dense physical memory layout. Avoiding Maxmem here is done with an
eye on copy-n-paste behaviour in general and to avoid confusion caused by
using realmem in agp.c and Maxmem in agp_i810.c.

In both cases, remove the inclusion of md_var.h
2012-07-06 15:57:03 +00:00
Konstantin Belousov
e1f4f1f860 Correct device id for GPU on some server SandyBridge model.
Submitted and tested by:	Thomas Zander <thomas.e.zander googlemail com>
MFC after:	3 days
2012-06-23 15:36:32 +00:00
Konstantin Belousov
28d86329af A rewrite of the i810 bits of the agp(4) driver. New driver supports
operations required by GEMified i915.ko. It also attaches to SandyBridge
and IvyBridge CPU northbridges now.

Sponsored by:	The FreeBSD Foundation
MFC after:	1 month
2012-05-22 10:59:26 +00:00
John Baldwin
3b0a4aef96 Do a sweep of the tree replacing calls to pci_find_extcap() with calls to
pci_find_cap() instead.
2011-03-23 13:10:15 +00:00
Nathan Whitehorn
50fd2a5b9c Add a driver for the Apple Uninorth AGP host bridge found in all PowerPC
Macintoshes with an AGP bus.
2010-10-31 18:27:05 +00:00
Konstantin Belousov
83d5d2963e Do not mention VM_ALLOC_RETRY in comment, and normalize the terminology
(blocking -> sleeping).

Reviewed by:	alc
MFC after:	3 days
2010-07-08 08:39:02 +00:00
Brian Somers
5d6d222816 Add a missing linefeed
PR:		147337
Submitted by:	cyberleo at cyberleo dot net
MFC after:	1 week
2010-06-19 08:42:29 +00:00
Alan Cox
e3ef0d2fcf Push down the acquisition of the page queues lock into vm_page_unwire().
Update the comment describing which lock should be held on entry to
vm_page_wire().

Reviewed by:	kib
2010-05-05 03:45:46 +00:00
Alan Cox
5fdd0a335f Acquire the page lock around vm_page_unwire(). For consistency, extend the
scope of the object lock in agp_i810.c.  (In this specific case, the scope
of the object lock shouldn't matter, but I don't want to create a bad
example that might be copied to a case where it did matter.)

Reviewed by:	kib
2010-05-03 16:55:50 +00:00
Robert Noland
c7bbe8bdaf Add support of Intel Pineview chips, aka IGD.
MFC after:	3 days
2010-03-12 21:34:23 +00:00
Robert Noland
9ccf1d6a0b Add pci ids for Intel Ironlake chipsets.
These behave just like g45 for agp.

Tested by:	Torfinn Ingolfsen

MFC after:	3 days
2010-03-12 19:43:39 +00:00