Memory allocation failure should generally be handled gracefully by
callers. In particular, with transparent hugepage support, attempts
to allocate huge pages can fail under memory pressure, but the callers
fall back to allocating individual pages instead. In that case, there
would be spurious
[TTM] Unable to get page %u
error messages in dmesg.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We need to figure out first how to correctly map them into the CPU page tables.
bug: https://bugs.freedesktop.org/show_bug.cgi?id=103138
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Try to allocate huge pages when it makes sense.
v2: fix comment and use ifdef
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Correctly handle different page sizes in the memory accounting.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Nobody is actually using that, remove it.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Remove unused defines and variables. Also stop computing the
gfp_flags when they aren't used.
No intended functional change.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
set_memory_* functions have moved to set_memory.h. Switch to this
explicitly.
[akpm@linux-foundation.org: track drivers/gpu/drm/i915/i915_gem_gtt.c linux-next changes]
Link: http://lkml.kernel.org/r/1488920133-27229-8-git-send-email-labbott@redhat.com
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch 3d50d4dcb0 exposed the CPU address of DMA-allocated pages as
returned by dma_alloc_coherent because Nouveau on Tegra needed it.
This is not required anymore - as there were no other users for it,
remove it and save some memory for everyone.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It tries to do fancy things with excluding agp support if ttm is
built-in, but agp isn't. Instead just express this depency like drm
does and use CONFIG_AGP everywhere.
Also use the neat Makefile magic to make the entire ttm_agp_backend
file optional.
v2: Use IS_ENABLED(CONFIG_AGP) as suggested by Ville
v3: Review from Emil.
v4: Actually get it right as spotted by 0-day.
Cc: Emil Velikov <emil.l.velikov@gmail.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1459337046-25882-1-git-send-email-daniel.vetter@ffwll.ch
Calls to set_memory_wb() incure heavy TLB flush and IPI cost. To
minimize those wait until pool grow beyond batch size before
draining the pool.
Signed-off-by: Jérôme Glisse <jglisse@redhat.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-and-Tested-by: Michel Dänzer <michel@daenzer.net>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Current code never allowed the page pool to actualy fill in anyway.
This fix it, so that we only start freeing page from the pool when
we go over the pool size.
Changed since v1:
- Move the page batching optimization to its separate patch.
Changed since v2:
- Do not remove code part of the batching optimization with
this patch.
- Better commit message.
Signed-off-by: Jérôme Glisse <jglisse@redhat.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
dma_alloc_coherent() can return memory in the vmalloc range.
virt_to_page() cannot handle such addresses and crashes. This
patch detects such cases and obtains the struct page * using
vmalloc_to_page() instead.
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Andrew Morton wrote:
> On Wed, 12 Nov 2014 13:08:55 +0900 Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp> wrote:
>
> > Andrew Morton wrote:
> > > Poor ttm guys - this is a bit of a trap we set for them.
> >
> > Commit a91576d791 ("drm/ttm: Pass GFP flags in order to avoid deadlock.")
> > changed to use sc->gfp_mask rather than GFP_KERNEL.
> >
> > - pages_to_free = kmalloc(npages_to_free * sizeof(struct page *),
> > - GFP_KERNEL);
> > + pages_to_free = kmalloc(npages_to_free * sizeof(struct page *), gfp);
> >
> > But this bug is caused by sc->gfp_mask containing some flags which are not
> > in GFP_KERNEL, right? Then, I think
> >
> > - pages_to_free = kmalloc(npages_to_free * sizeof(struct page *), gfp);
> > + pages_to_free = kmalloc(npages_to_free * sizeof(struct page *), gfp & GFP_KERNEL);
> >
> > would hide this bug.
> >
> > But I think we should use GFP_ATOMIC (or drop __GFP_WAIT flag)
>
> Well no - ttm_page_pool_free() should stop calling kmalloc altogether.
> Just do
>
> struct page *pages_to_free[16];
>
> and rework the code to free 16 pages at a time. Easy.
Well, ttm code wants to process 512 pages at a time for performance.
Memory footprint increased by 512 * sizeof(struct page *) buffer is
only 4096 bytes. What about using static buffer like below?
----------
>From d3cb5393c9c8099d6b37e769f78c31af1541fe8c Mon Sep 17 00:00:00 2001
From: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Date: Thu, 13 Nov 2014 22:21:54 +0900
Subject: [PATCH] drm/ttm: Avoid memory allocation from shrinker functions.
Commit a91576d791 ("drm/ttm: Pass GFP flags in order to avoid
deadlock.") caused BUG_ON() due to sc->gfp_mask containing flags
which are not in GFP_KERNEL.
https://bugzilla.kernel.org/show_bug.cgi?id=87891
Changing from sc->gfp_mask to (sc->gfp_mask & GFP_KERNEL) would
avoid the BUG_ON(), but avoiding memory allocation from shrinker
function is better and reliable fix.
Shrinker function is already serialized by global lock, and
clean up function is called after shrinker function is unregistered.
Thus, we can use static buffer when called from shrinker function
and clean up function.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: stable <stable@kernel.org> [2.6.35+]
Signed-off-by: Dave Airlie <airlied@redhat.com>
Pages allocated using the DMA API have a coherent memory mapping. Make
this mapping visible to drivers so they can decide to use it instead of
creating their own redundant one.
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
Acked-by: David Airlie <airlied@linux.ie>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Commit 7dc19d5a "drivers: convert shrinkers to new count/scan API" added
deadlock warnings that ttm_page_pool_free() and ttm_dma_page_pool_free()
are currently doing GFP_KERNEL allocation.
But these functions did not get updated to receive gfp_t argument.
This patch explicitly passes sc->gfp_mask or GFP_KERNEL to these functions,
and removes the deadlock warning.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: stable <stable@kernel.org> [2.6.35+]
Signed-off-by: Dave Airlie <airlied@redhat.com>
I can observe that RHEL7 environment stalls with 100% CPU usage when a
certain type of memory pressure is given. While the shrinker functions
are called by shrink_slab() before the OOM killer is triggered, the stall
lasts for many minutes.
One of reasons of this stall is that
ttm_dma_pool_shrink_count()/ttm_dma_pool_shrink_scan() are called and
are blocked at mutex_lock(&_manager->lock). GFP_KERNEL allocation with
_manager->lock held causes someone (including kswapd) to deadlock when
these functions are called due to memory pressure. This patch changes
"mutex_lock();" to "if (!mutex_trylock()) return ...;" in order to
avoid deadlock.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: stable <stable@kernel.org> [3.3+]
Signed-off-by: Dave Airlie <airlied@redhat.com>
We can use "unsigned int" instead of "atomic_t" by updating start_pool
variable under _manager->lock. This patch will make it possible to avoid
skipping when choosing a pool to shrink in round-robin style, after next
patch changes mutex_lock(_manager->lock) to !mutex_trylock(_manager->lork).
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: stable <stable@kernel.org> [3.3+]
Signed-off-by: Dave Airlie <airlied@redhat.com>
list_empty(&_manager->pools) being false before taking _manager->lock
does not guarantee that _manager->npools != 0 after taking _manager->lock
because _manager->npools is updated under _manager->lock.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: stable <stable@kernel.org> [3.3+]
Signed-off-by: Dave Airlie <airlied@redhat.com>
Used by the vmwgfx driver
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Convert the driver shrinkers to the new API. Most changes are compile
tested only because I either don't have the hardware or it's staging
stuff.
FWIW, the md and android code is pretty good, but the rest of it makes me
want to claw my eyes out. The amount of broken code I just encountered is
mind boggling. I've added comments explaining what is broken, but I fear
that some of the code would be best dealt with by being dragged behind the
bike shed, burying in mud up to it's neck and then run over repeatedly
with a blunt lawn mower.
Special mention goes to the zcache/zcache2 drivers. They can't co-exist
in the build at the same time, they are under different menu options in
menuconfig, they only show up when you've got the right set of mm
subsystem options configured and so even compile testing is an exercise in
pulling teeth. And that doesn't even take into account the horrible,
broken code...
[glommer@openvz.org: fixes for i915, android lowmem, zcache, bcache]
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Glauber Costa <glommer@openvz.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Kent Overstreet <koverstreet@google.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Cc: Arve Hjønnevåg <arve@android.com>
Cc: Carlos Maiolino <cmaiolino@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: David Rientjes <rientjes@google.com>
Cc: Gleb Natapov <gleb@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: J. Bruce Fields <bfields@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Kent Overstreet <koverstreet@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pull drm merge (part 1) from Dave Airlie:
"So first of all my tree and uapi stuff has a conflict mess, its my
fault as the nouveau stuff didn't hit -next as were trying to rebase
regressions out of it before we merged.
Highlights:
- SH mobile modesetting driver and associated helpers
- some DRM core documentation
- i915 modesetting rework, haswell hdmi, haswell and vlv fixes, write
combined pte writing, ilk rc6 support,
- nouveau: major driver rework into a hw core driver, makes features
like SLI a lot saner to implement,
- psb: add eDP/DP support for Cedarview
- radeon: 2 layer page tables, async VM pte updates, better PLL
selection for > 2 screens, better ACPI interactions
The rest is general grab bag of fixes.
So why part 1? well I have the exynos pull req which came in a bit
late but was waiting for me to do something they shouldn't have and it
looks fairly safe, and David Howells has some more header cleanups
he'd like me to pull, that seem like a good idea, but I'd like to get
this merge out of the way so -next dosen't get blocked."
Tons of conflicts mostly due to silly include line changes, but mostly
mindless. A few other small semantic conflicts too, noted from Dave's
pre-merged branch.
* 'drm-next' of git://people.freedesktop.org/~airlied/linux: (447 commits)
drm/nv98/crypt: fix fuc build with latest envyas
drm/nouveau/devinit: fixup various issues with subdev ctor/init ordering
drm/nv41/vm: fix and enable use of "real" pciegart
drm/nv44/vm: fix and enable use of "real" pciegart
drm/nv04/dmaobj: fixup vm target handling in preparation for nv4x pcie
drm/nouveau: store supported dma mask in vmmgr
drm/nvc0/ibus: initial implementation of subdev
drm/nouveau/therm: add support for fan-control modes
drm/nouveau/hwmon: rename pwm0* to pmw1* to follow hwmon's rules
drm/nouveau/therm: calculate the pwm divisor on nv50+
drm/nouveau/fan: rewrite the fan tachometer driver to get more precision, faster
drm/nouveau/therm: move thermal-related functions to the therm subdev
drm/nouveau/bios: parse the pwm divisor from the perf table
drm/nouveau/therm: use the EXTDEV table to detect i2c monitoring devices
drm/nouveau/therm: rework thermal table parsing
drm/nouveau/gpio: expose the PWM/TOGGLE parameter found in the gpio vbios table
drm/nouveau: fix pm initialization order
drm/nouveau/bios: check that fixed tvdac gpio data is valid before using it
drm/nouveau: log channel debug/error messages from client object rather than drm client
drm/nouveau: have drm debugging macros build on top of core macros
...
Convert #include "..." to #include <path/...> in drivers/gpu/.
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Dave Jones <davej@redhat.com>
Remove useless kfree() and clean up code related to the removal.
The semantic patch that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@r exists@
position p1,p2;
expression x;
@@
if (x@p1 == NULL) { ... kfree@p2(x); ... return ...; }
@unchanged exists@
position r.p1,r.p2;
expression e <= r.x,x,e1;
iterator I;
statement S;
@@
if (x@p1 == NULL) { ... when != I(x,...) S
when != e = e1
when != e += e1
when != e -= e1
when != ++e
when != --e
when != e++
when != e--
when != &e
kfree@p2(x); ... return ...; }
@ok depends on unchanged exists@
position any r.p1;
position r.p2;
expression x;
@@
... when != true x@p1 == NULL
kfree@p2(x);
@depends on !ok && unchanged@
position r.p2;
expression x;
@@
*kfree@p2(x);
// </smpl>
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Use the more current logging style.
Add pr_fmt and remove the TTM_PFX uses.
Coalesce formats and align arguments.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
. It was useful during development, but now on a production system
we can get this (if the user forgot to upload the firmware):
[drm] radeon: irq initialized.
[drm] GART: num cpu pages 131072, num gpu pages 131072
[drm] radeon: ib pool ready.
[drm] Loading SUMO Microcode
r600_cp: Failed to load firmware "radeon/SUMO_pfp.bin"
atl1c 0000:03:00.0: version 1.0.1.0-NAPI.213057] [drm:evergreen_startup] *ERROR* Failed to load firmware!
radeon 0000:00:01.0: disabling GPU acceleration
88] radeon 0000:00:01.0: ffff8801bb782400 unpin not necessary
------------[ cut here ]------------
WARNING: at /home/konrad/linux-linus/drivers/gpu/drm/ttm/ttm_page_alloc_dma.c:956 ttm_dma_unpopulate+0x79/0x300 [ttm]()
Hardware name: System Product Name
Modules linked in: e1000e atl1c radeon(+) ahci libahci libata scsi_mod fbcon tileblit font ttm bitblit softcursor drm_kms_helper wmi xen_blkfront xen_netfront fb_sys_fops sysimgblt sysfillrect syscopyarea xenfs xen_privcmd
Pid: 1600, comm: modprobe Not tainted 3.2.0-06100-ge343a89 #1
Call Trace:
[<ffffffff8108973a>] warn_slowpath_common+0x7a/0xb0
[<ffffffff81089785>] warn_slowpath_null+0x15/0x20
[<ffffffffa0060309>] ttm_dma_unpopulate+0x79/0x300 [ttm]
[<ffffffffa01341c0>] radeon_ttm_tt_unpopulate+0x120/0x130 [radeon]
[<ffffffffa0056e0c>] ttm_tt_destroy+0x2c/0x70 [ttm]
[<ffffffffa0057a4e>] ttm_bo_cleanup_memtype_use+0x3e/0x80 [ttm]
[<ffffffffa00595a1>] ttm_bo_release+0x251/0x280 [ttm]
[<ffffffffa0059610>] ttm_bo_unref+0x40/0x60 [ttm]
[<ffffffffa0134d02>] radeon_bo_unref+0x42/0x80 [radeon]
[<ffffffffa0186dfb>] radeon_sa_bo_manager_fini+0x6b/0x80 [radeon]
[<ffffffffa0146b8f>] radeon_ib_pool_fini+0x6f/0x90 [radeon]
[<ffffffffa014be49>] r100_ib_fini+0x19/0x20 [radeon]
[<ffffffffa017b47e>] evergreen_init+0x1ee/0x2d0 [radeon]
The big WARN() has nothing to do with the culprit - which is that
the firmware was not loaded. So lets remove the WARN() from the TTM DMA code.
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The "if (!p && !p->dev)" condition isn't right because || was intended
instead of &&. But actually, "p" is the list cursor and so it's always
non-NULL and we can just remove that bit. We can remove the another
similar check as well.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The code to figure out how many pages to shrink the pool
ends up capping the 'count' at _manager->options.max_size - which is OK.
Except that the 'count' is also used when accounting for how many pages
are recycled - which we end up with the invalid values. This fixes
it by using a different value for the amount of pages to shrink.
On top of that we would free the cached page pool - which is nonsense
as they are deleted from the pool - so there are no free pages in that
pool..
Also we also missed the opportunity to batch the amount of pages
to free (similar to how ttm_page_alloc.c does it). This reintroduces
the code that was lost during rebasing.
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Move dma data to a superset ttm_dma_tt structure which herit
from ttm_tt. This allow driver that don't use dma functionalities
to not have to waste memory for it.
V2 Rebase on top of no memory account changes (where/when is my
delorean when i need it ?)
V3 Make sure page list is initialized empty
V4 typo/syntax fixes
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
In TTM world the pages for the graphic drivers are kept in three different
pools: write combined, uncached, and cached (write-back). When the pages
are used by the graphic driver the graphic adapter via its built in MMU
(or AGP) programs these pages in. The programming requires the virtual address
(from the graphic adapter perspective) and the physical address (either System RAM
or the memory on the card) which is obtained using the pci_map_* calls (which does the
virtual to physical - or bus address translation). During the graphic application's
"life" those pages can be shuffled around, swapped out to disk, moved from the
VRAM to System RAM or vice-versa. This all works with the existing TTM pool code
- except when we want to use the software IOTLB (SWIOTLB) code to "map" the physical
addresses to the graphic adapter MMU. We end up programming the bounce buffer's
physical address instead of the TTM pool memory's and get a non-worky driver.
There are two solutions:
1) using the DMA API to allocate pages that are screened by the DMA API, or
2) using the pci_sync_* calls to copy the pages from the bounce-buffer and back.
This patch fixes the issue by allocating pages using the DMA API. The second
is a viable option - but it has performance drawbacks and potential correctness
issues - think of the write cache page being bounced (SWIOTLB->TTM), the
WC is set on the TTM page and the copy from SWIOTLB not making it to the TTM
page until the page has been recycled in the pool (and used by another application).
The bounce buffer does not get activated often - only in cases where we have
a 32-bit capable card and we want to use a page that is allocated above the
4GB limit. The bounce buffer offers the solution of copying the contents
of that 4GB page to an location below 4GB and then back when the operation has been
completed (or vice-versa). This is done by using the 'pci_sync_*' calls.
Note: If you look carefully enough in the existing TTM page pool code you will
notice the GFP_DMA32 flag is used - which should guarantee that the provided page
is under 4GB. It certainly is the case, except this gets ignored in two cases:
- If user specifies 'swiotlb=force' which bounces _every_ page.
- If user is using a Xen's PV Linux guest (which uses the SWIOTLB and the
underlaying PFN's aren't necessarily under 4GB).
To not have this extra copying done the other option is to allocate the pages
using the DMA API so that there is not need to map the page and perform the
expensive 'pci_sync_*' calls.
This DMA API capable TTM pool requires for this the 'struct device' to
properly call the DMA API. It also has to track the virtual and bus address of
the page being handed out in case it ends up being swapped out or de-allocated -
to make sure it is de-allocated using the proper's 'struct device'.
Implementation wise the code keeps two lists: one that is attached to the
'struct device' (via the dev->dma_pools list) and a global one to be used when
the 'struct device' is unavailable (think shrinker code). The global list can
iterate over all of the 'struct device' and its associated dma_pool. The list
in dev->dma_pools can only iterate the device's dma_pool.
/[struct device_pool]\
/---------------------------------------------------| dev |
/ +-------| dma_pool |
/-----+------\ / \--------------------/
|struct device| /-->[struct dma_pool for WC]</ /[struct device_pool]\
| dma_pools +----+ /-| dev |
| ... | \--->[struct dma_pool for uncached]<-/--| dma_pool |
\-----+------/ / \--------------------/
\----------------------------------------------/
[Two pools associated with the device (WC and UC), and the parallel list
containing the 'struct dev' and 'struct dma_pool' entries]
The maximum amount of dma pools a device can have is six: write-combined,
uncached, and cached; then there are the DMA32 variants which are:
write-combined dma32, uncached dma32, and cached dma32.
Currently this code only gets activated when any variant of the SWIOTLB IOMMU
code is running (Intel without VT-d, AMD without GART, IBM Calgary and Xen PV
with PCI devices).
Tested-by: Michel Dänzer <michel@daenzer.net>
[v1: Using swiotlb_nr_tbl instead of swiotlb_enabled]
[v2: Major overhaul - added 'inuse_list' to seperate used from inuse and reorder
the order of lists to get better performance.]
[v3: Added comments/and some logic based on review, Added Jerome tag]
[v4: rebase on top of ttm_tt & ttm_backend merge]
[v5: rebase on top of ttm memory accounting overhaul]
[v6: New rebase on top of more memory accouting changes]
[v7: well rebase on top of no memory accounting changes]
[v8: make sure pages list is initialized empty]
[v9: calll ttm_mem_global_free_page in unpopulate for accurate accountg]
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Acked-by: Thomas Hellstrom <thellstrom@vmware.com>
Michel Dänzer