linux_dsm_epyc7002/drivers/gpu/drm/drm_mm.c
Chris Wilson cc98e6ce6a drm: Initialise drm_mm.head_node.allocated
commit 202b52b7fb ("drm: Track drm_mm nodes with an interval tree")
introduced a requirement that the special drm_mm.head_node was
initialised and marked as not being allocated. It is a very special node
that has no side but has a hole that represents the drm_mm address
space, and holds the list of nodes. Since it is not a real node, it is
not part of the node rbtree and we detect this as it being unallocated.
This presumed that drm_mm_init() was initialising it to zero. It happens
that i915 kzallocs its objects and so it was accidentally setting it,
but for generic use we cannot make that assumption.

[   22.981519] general protection fault: 0000 [#1] SMP
[   22.981521] Modules linked in: test_drm_mm(+) ctr ccm arc4 rt2800usb rt2x00usb rt2800lib rt2x00lib crc_ccitt mac80211 cmac rfcomm bnep snd_hda_codec_realtek snd_hda_codec_hdmi snd_hda_codec_generic snd_hda_intel dcdbas snd_hda_codec x86_pkg_temp_thermal intel_powerclamp btusb snd_hda_core coretemp crct10dif_pclmul cfg80211 btrtl btbcm btintel bluetooth crc32_pclmul ghash_clmulni_intel aesni_intel snd_pcm i2c_hid aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd snd_timer hid_multitouch snd joydev serio_raw lpc_ich mfd_core i2c_designware_platform i2c_designware_core 8250_dw binfmt_misc soundcore acpi_pad nls_iso8859_1 usbhid hid psmouse ahci libahci [last unloaded: test_drm_mm]
[   22.981544] CPU: 1 PID: 2088 Comm: drm_mm Tainted: G        W       4.9.0-rc7+ #234
[   22.981545] Hardware name: Dell Inc. XPS 13 9343/0310JH, BIOS A07 11/11/2015
[   22.981546] task: ffff88020c971cc0 task.stack: ffffc90001728000
[   22.981547] RIP: 0010:[<ffffffff814050f0>]  [<ffffffff814050f0>] drm_mm_interval_tree_add_node+0xa0/0xd0
[   22.981551] RSP: 0018:ffffc9000172ba98  EFLAGS: 00010202
[   22.981552] RAX: 0f0000c69cf63d80 RBX: ffff88020be00000 RCX: ffff88020be00000
[   22.981553] RDX: 0000000000000fff RSI: ffffc9000172bc48 RDI: ffffffff810ac4df
[   22.981553] RBP: ffffc9000172bb08 R08: ffffc9000172bc70 R09: 0000000000000fff
[   22.981554] R10: ffffffff810ac4d7 R11: 4dc04d8b4cffffe5 R12: 0000000000001000
[   22.981555] R13: ffffc9000172bbd0 R14: ffffc9000172bbe0 R15: 0000000002000000
[   22.981556] FS:  00007f80c9fab740(0000) GS:ffff88021f480000(0000) knlGS:0000000000000000
[   22.981557] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   22.981558] CR2: 00007f80c9fd5000 CR3: 000000020c191000 CR4: 00000000003406e0
[   22.981559] Stack:
[   22.981560]  ffffffff81405d09 ffff88020be00000 ffffc9000172bbe0 000000000172bb08
[   22.981562]  ffffffffffffffff 0000000000000000 0000000000000000 0000000000000000
[   22.981563]  0000000002000000 0000000002000000 ffffffffa02f3000 ffff88020be00000
[   22.981565] Call Trace:
[   22.981568]  [<ffffffff81405d09>] ? drm_mm_insert_node_generic+0x229/0x310
[   22.981570]  [<ffffffffa02f3000>] ? 0xffffffffa02f3000
[   22.981572]  [<ffffffffa02903c1>] __subtest_insert_range.constprop.7+0xd1/0x5b0 [test_drm_mm]
[   22.981575]  [<ffffffff81081222>] ? default_wake_function+0x12/0x20
[   22.981576]  [<ffffffff81096905>] ? __wake_up_common+0x55/0x90
[   22.981578]  [<ffffffff81085f42>] ? sched_clock_cpu+0x72/0xa0
[   22.981581]  [<ffffffff811308ad>] ? irq_work_queue+0xd/0x80
[   22.981582]  [<ffffffff810abcc4>] ? wake_up_klogd+0x34/0x40
[   22.981584]  [<ffffffff810ac19d>] ? console_unlock+0x4cd/0x530
[   22.981585]  [<ffffffff810ac4d7>] ? vprintk_emit+0x2d7/0x490
[   22.981587]  [<ffffffff810ac82f>] ? vprintk_default+0x1f/0x30
[   22.981589]  [<ffffffff81146e1c>] ? printk+0x4d/0x4f
[   22.981590]  [<ffffffffa02f3000>] ? 0xffffffffa02f3000
[   22.981592]  [<ffffffffa02908b5>] subtest_insert_range+0x15/0x80 [test_drm_mm]
[   22.981594]  [<ffffffffa02f3088>] test_drm_mm_init+0x88/0x1000 [test_drm_mm]
[   22.981597]  [<ffffffff8100043d>] do_one_initcall+0x3d/0x150
[   22.981600]  [<ffffffff8119dfbf>] ? kfree+0x13f/0x180
[   22.981602]  [<ffffffff811471f2>] do_init_module+0x60/0x1f1
[   22.981606]  [<ffffffff810db878>] load_module+0x2228/0x2790
[   22.981608]  [<ffffffff810d8590>] ? __symbol_put+0x40/0x40
[   22.981612]  [<ffffffff811c52b1>] ? kernel_read+0x41/0x60
[   22.981614]  [<ffffffff810dbfb6>] SYSC_finit_module+0x96/0xd0
[   22.981617]  [<ffffffff810dc00e>] SyS_finit_module+0xe/0x10
[   22.981620]  [<ffffffff816e7aa4>] entry_SYSCALL_64_fastpath+0x17/0x98
[   22.981622] Code: c7 41 30 00 00 00 00 48 89 e5 48 89 3a 48 c7 c2 20 4e 40 81 e8 b2 a1 f0 ff 5d c3 48 8d 56 78 45 31 d2 48 89 d6 eb 25 48 8b 51 58 <48> 39 50 38 73 04 48 89 50 38 4c 8b 58 28 4c 39 59 48 48 8d 50
[   22.981651] RIP  [<ffffffff814050f0>] drm_mm_interval_tree_add_node+0xa0/0xd0
[   22.981655]  RSP <ffffc9000172ba98>

Testcase: igt/drm_mm
Fixes: 202b52b7fb ("drm: Track drm_mm nodes with an interval tree")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: dri-devel@lists.freedesktop.org
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.9-rc1+
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20161130205126.31106-1-chris@chris-wilson.co.uk
2016-12-01 10:35:45 +01:00

1025 lines
29 KiB
C

/**************************************************************************
*
* Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sub license, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
* DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
* OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
* USE OR OTHER DEALINGS IN THE SOFTWARE.
*
*
**************************************************************************/
/*
* Generic simple memory manager implementation. Intended to be used as a base
* class implementation for more advanced memory managers.
*
* Note that the algorithm used is quite simple and there might be substantial
* performance gains if a smarter free list is implemented. Currently it is just an
* unordered stack of free regions. This could easily be improved if an RB-tree
* is used instead. At least if we expect heavy fragmentation.
*
* Aligned allocations can also see improvement.
*
* Authors:
* Thomas Hellström <thomas-at-tungstengraphics-dot-com>
*/
#include <drm/drmP.h>
#include <drm/drm_mm.h>
#include <linux/slab.h>
#include <linux/seq_file.h>
#include <linux/export.h>
#include <linux/interval_tree_generic.h>
/**
* DOC: Overview
*
* drm_mm provides a simple range allocator. The drivers are free to use the
* resource allocator from the linux core if it suits them, the upside of drm_mm
* is that it's in the DRM core. Which means that it's easier to extend for
* some of the crazier special purpose needs of gpus.
*
* The main data struct is &drm_mm, allocations are tracked in &drm_mm_node.
* Drivers are free to embed either of them into their own suitable
* datastructures. drm_mm itself will not do any allocations of its own, so if
* drivers choose not to embed nodes they need to still allocate them
* themselves.
*
* The range allocator also supports reservation of preallocated blocks. This is
* useful for taking over initial mode setting configurations from the firmware,
* where an object needs to be created which exactly matches the firmware's
* scanout target. As long as the range is still free it can be inserted anytime
* after the allocator is initialized, which helps with avoiding looped
* depencies in the driver load sequence.
*
* drm_mm maintains a stack of most recently freed holes, which of all
* simplistic datastructures seems to be a fairly decent approach to clustering
* allocations and avoiding too much fragmentation. This means free space
* searches are O(num_holes). Given that all the fancy features drm_mm supports
* something better would be fairly complex and since gfx thrashing is a fairly
* steep cliff not a real concern. Removing a node again is O(1).
*
* drm_mm supports a few features: Alignment and range restrictions can be
* supplied. Further more every &drm_mm_node has a color value (which is just an
* opaqua unsigned long) which in conjunction with a driver callback can be used
* to implement sophisticated placement restrictions. The i915 DRM driver uses
* this to implement guard pages between incompatible caching domains in the
* graphics TT.
*
* Two behaviors are supported for searching and allocating: bottom-up and top-down.
* The default is bottom-up. Top-down allocation can be used if the memory area
* has different restrictions, or just to reduce fragmentation.
*
* Finally iteration helpers to walk all nodes and all holes are provided as are
* some basic allocator dumpers for debugging.
*/
static struct drm_mm_node *drm_mm_search_free_generic(const struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color,
enum drm_mm_search_flags flags);
static struct drm_mm_node *drm_mm_search_free_in_range_generic(const struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color,
u64 start,
u64 end,
enum drm_mm_search_flags flags);
#ifdef CONFIG_DRM_DEBUG_MM
#include <linux/stackdepot.h>
#define STACKDEPTH 32
#define BUFSZ 4096
static noinline void save_stack(struct drm_mm_node *node)
{
unsigned long entries[STACKDEPTH];
struct stack_trace trace = {
.entries = entries,
.max_entries = STACKDEPTH,
.skip = 1
};
save_stack_trace(&trace);
if (trace.nr_entries != 0 &&
trace.entries[trace.nr_entries-1] == ULONG_MAX)
trace.nr_entries--;
/* May be called under spinlock, so avoid sleeping */
node->stack = depot_save_stack(&trace, GFP_NOWAIT);
}
static void show_leaks(struct drm_mm *mm)
{
struct drm_mm_node *node;
unsigned long entries[STACKDEPTH];
char *buf;
buf = kmalloc(BUFSZ, GFP_KERNEL);
if (!buf)
return;
list_for_each_entry(node, &mm->head_node.node_list, node_list) {
struct stack_trace trace = {
.entries = entries,
.max_entries = STACKDEPTH
};
if (!node->stack) {
DRM_ERROR("node [%08llx + %08llx]: unknown owner\n",
node->start, node->size);
continue;
}
depot_fetch_stack(node->stack, &trace);
snprint_stack_trace(buf, BUFSZ, &trace, 0);
DRM_ERROR("node [%08llx + %08llx]: inserted at\n%s",
node->start, node->size, buf);
}
kfree(buf);
}
#undef STACKDEPTH
#undef BUFSZ
#else
static void save_stack(struct drm_mm_node *node) { }
static void show_leaks(struct drm_mm *mm) { }
#endif
#define START(node) ((node)->start)
#define LAST(node) ((node)->start + (node)->size - 1)
INTERVAL_TREE_DEFINE(struct drm_mm_node, rb,
u64, __subtree_last,
START, LAST, static inline, drm_mm_interval_tree)
struct drm_mm_node *
__drm_mm_interval_first(struct drm_mm *mm, u64 start, u64 last)
{
return drm_mm_interval_tree_iter_first(&mm->interval_tree,
start, last);
}
EXPORT_SYMBOL(__drm_mm_interval_first);
static void drm_mm_interval_tree_add_node(struct drm_mm_node *hole_node,
struct drm_mm_node *node)
{
struct drm_mm *mm = hole_node->mm;
struct rb_node **link, *rb;
struct drm_mm_node *parent;
node->__subtree_last = LAST(node);
if (hole_node->allocated) {
rb = &hole_node->rb;
while (rb) {
parent = rb_entry(rb, struct drm_mm_node, rb);
if (parent->__subtree_last >= node->__subtree_last)
break;
parent->__subtree_last = node->__subtree_last;
rb = rb_parent(rb);
}
rb = &hole_node->rb;
link = &hole_node->rb.rb_right;
} else {
rb = NULL;
link = &mm->interval_tree.rb_node;
}
while (*link) {
rb = *link;
parent = rb_entry(rb, struct drm_mm_node, rb);
if (parent->__subtree_last < node->__subtree_last)
parent->__subtree_last = node->__subtree_last;
if (node->start < parent->start)
link = &parent->rb.rb_left;
else
link = &parent->rb.rb_right;
}
rb_link_node(&node->rb, rb, link);
rb_insert_augmented(&node->rb,
&mm->interval_tree,
&drm_mm_interval_tree_augment);
}
static void drm_mm_insert_helper(struct drm_mm_node *hole_node,
struct drm_mm_node *node,
u64 size, unsigned alignment,
unsigned long color,
enum drm_mm_allocator_flags flags)
{
struct drm_mm *mm = hole_node->mm;
u64 hole_start = drm_mm_hole_node_start(hole_node);
u64 hole_end = drm_mm_hole_node_end(hole_node);
u64 adj_start = hole_start;
u64 adj_end = hole_end;
BUG_ON(node->allocated);
if (mm->color_adjust)
mm->color_adjust(hole_node, color, &adj_start, &adj_end);
if (flags & DRM_MM_CREATE_TOP)
adj_start = adj_end - size;
if (alignment) {
u64 tmp = adj_start;
unsigned rem;
rem = do_div(tmp, alignment);
if (rem) {
if (flags & DRM_MM_CREATE_TOP)
adj_start -= rem;
else
adj_start += alignment - rem;
}
}
BUG_ON(adj_start < hole_start);
BUG_ON(adj_end > hole_end);
if (adj_start == hole_start) {
hole_node->hole_follows = 0;
list_del(&hole_node->hole_stack);
}
node->start = adj_start;
node->size = size;
node->mm = mm;
node->color = color;
node->allocated = 1;
list_add(&node->node_list, &hole_node->node_list);
drm_mm_interval_tree_add_node(hole_node, node);
BUG_ON(node->start + node->size > adj_end);
node->hole_follows = 0;
if (__drm_mm_hole_node_start(node) < hole_end) {
list_add(&node->hole_stack, &mm->hole_stack);
node->hole_follows = 1;
}
save_stack(node);
}
/**
* drm_mm_reserve_node - insert an pre-initialized node
* @mm: drm_mm allocator to insert @node into
* @node: drm_mm_node to insert
*
* This functions inserts an already set-up drm_mm_node into the allocator,
* meaning that start, size and color must be set by the caller. This is useful
* to initialize the allocator with preallocated objects which must be set-up
* before the range allocator can be set-up, e.g. when taking over a firmware
* framebuffer.
*
* Returns:
* 0 on success, -ENOSPC if there's no hole where @node is.
*/
int drm_mm_reserve_node(struct drm_mm *mm, struct drm_mm_node *node)
{
u64 end = node->start + node->size;
struct drm_mm_node *hole;
u64 hole_start, hole_end;
u64 adj_start, adj_end;
if (WARN_ON(node->size == 0))
return -EINVAL;
end = node->start + node->size;
/* Find the relevant hole to add our node to */
hole = drm_mm_interval_tree_iter_first(&mm->interval_tree,
node->start, ~(u64)0);
if (hole) {
if (hole->start < end)
return -ENOSPC;
} else {
hole = list_entry(&mm->head_node.node_list,
typeof(*hole), node_list);
}
hole = list_last_entry(&hole->node_list, typeof(*hole), node_list);
if (!hole->hole_follows)
return -ENOSPC;
adj_start = hole_start = __drm_mm_hole_node_start(hole);
adj_end = hole_end = __drm_mm_hole_node_end(hole);
if (mm->color_adjust)
mm->color_adjust(hole, node->color, &adj_start, &adj_end);
if (adj_start > node->start || adj_end < end)
return -ENOSPC;
node->mm = mm;
node->allocated = 1;
list_add(&node->node_list, &hole->node_list);
drm_mm_interval_tree_add_node(hole, node);
if (node->start == hole_start) {
hole->hole_follows = 0;
list_del(&hole->hole_stack);
}
node->hole_follows = 0;
if (end != hole_end) {
list_add(&node->hole_stack, &mm->hole_stack);
node->hole_follows = 1;
}
save_stack(node);
return 0;
}
EXPORT_SYMBOL(drm_mm_reserve_node);
/**
* drm_mm_insert_node_generic - search for space and insert @node
* @mm: drm_mm to allocate from
* @node: preallocate node to insert
* @size: size of the allocation
* @alignment: alignment of the allocation
* @color: opaque tag value to use for this node
* @sflags: flags to fine-tune the allocation search
* @aflags: flags to fine-tune the allocation behavior
*
* The preallocated node must be cleared to 0.
*
* Returns:
* 0 on success, -ENOSPC if there's no suitable hole.
*/
int drm_mm_insert_node_generic(struct drm_mm *mm, struct drm_mm_node *node,
u64 size, unsigned alignment,
unsigned long color,
enum drm_mm_search_flags sflags,
enum drm_mm_allocator_flags aflags)
{
struct drm_mm_node *hole_node;
if (WARN_ON(size == 0))
return -EINVAL;
hole_node = drm_mm_search_free_generic(mm, size, alignment,
color, sflags);
if (!hole_node)
return -ENOSPC;
drm_mm_insert_helper(hole_node, node, size, alignment, color, aflags);
return 0;
}
EXPORT_SYMBOL(drm_mm_insert_node_generic);
static void drm_mm_insert_helper_range(struct drm_mm_node *hole_node,
struct drm_mm_node *node,
u64 size, unsigned alignment,
unsigned long color,
u64 start, u64 end,
enum drm_mm_allocator_flags flags)
{
struct drm_mm *mm = hole_node->mm;
u64 hole_start = drm_mm_hole_node_start(hole_node);
u64 hole_end = drm_mm_hole_node_end(hole_node);
u64 adj_start = hole_start;
u64 adj_end = hole_end;
BUG_ON(!hole_node->hole_follows || node->allocated);
if (adj_start < start)
adj_start = start;
if (adj_end > end)
adj_end = end;
if (mm->color_adjust)
mm->color_adjust(hole_node, color, &adj_start, &adj_end);
if (flags & DRM_MM_CREATE_TOP)
adj_start = adj_end - size;
if (alignment) {
u64 tmp = adj_start;
unsigned rem;
rem = do_div(tmp, alignment);
if (rem) {
if (flags & DRM_MM_CREATE_TOP)
adj_start -= rem;
else
adj_start += alignment - rem;
}
}
if (adj_start == hole_start) {
hole_node->hole_follows = 0;
list_del(&hole_node->hole_stack);
}
node->start = adj_start;
node->size = size;
node->mm = mm;
node->color = color;
node->allocated = 1;
list_add(&node->node_list, &hole_node->node_list);
drm_mm_interval_tree_add_node(hole_node, node);
BUG_ON(node->start < start);
BUG_ON(node->start < adj_start);
BUG_ON(node->start + node->size > adj_end);
BUG_ON(node->start + node->size > end);
node->hole_follows = 0;
if (__drm_mm_hole_node_start(node) < hole_end) {
list_add(&node->hole_stack, &mm->hole_stack);
node->hole_follows = 1;
}
save_stack(node);
}
/**
* drm_mm_insert_node_in_range_generic - ranged search for space and insert @node
* @mm: drm_mm to allocate from
* @node: preallocate node to insert
* @size: size of the allocation
* @alignment: alignment of the allocation
* @color: opaque tag value to use for this node
* @start: start of the allowed range for this node
* @end: end of the allowed range for this node
* @sflags: flags to fine-tune the allocation search
* @aflags: flags to fine-tune the allocation behavior
*
* The preallocated node must be cleared to 0.
*
* Returns:
* 0 on success, -ENOSPC if there's no suitable hole.
*/
int drm_mm_insert_node_in_range_generic(struct drm_mm *mm, struct drm_mm_node *node,
u64 size, unsigned alignment,
unsigned long color,
u64 start, u64 end,
enum drm_mm_search_flags sflags,
enum drm_mm_allocator_flags aflags)
{
struct drm_mm_node *hole_node;
if (WARN_ON(size == 0))
return -EINVAL;
hole_node = drm_mm_search_free_in_range_generic(mm,
size, alignment, color,
start, end, sflags);
if (!hole_node)
return -ENOSPC;
drm_mm_insert_helper_range(hole_node, node,
size, alignment, color,
start, end, aflags);
return 0;
}
EXPORT_SYMBOL(drm_mm_insert_node_in_range_generic);
/**
* drm_mm_remove_node - Remove a memory node from the allocator.
* @node: drm_mm_node to remove
*
* This just removes a node from its drm_mm allocator. The node does not need to
* be cleared again before it can be re-inserted into this or any other drm_mm
* allocator. It is a bug to call this function on a un-allocated node.
*/
void drm_mm_remove_node(struct drm_mm_node *node)
{
struct drm_mm *mm = node->mm;
struct drm_mm_node *prev_node;
if (WARN_ON(!node->allocated))
return;
BUG_ON(node->scanned_block || node->scanned_prev_free
|| node->scanned_next_free);
prev_node =
list_entry(node->node_list.prev, struct drm_mm_node, node_list);
if (node->hole_follows) {
BUG_ON(__drm_mm_hole_node_start(node) ==
__drm_mm_hole_node_end(node));
list_del(&node->hole_stack);
} else
BUG_ON(__drm_mm_hole_node_start(node) !=
__drm_mm_hole_node_end(node));
if (!prev_node->hole_follows) {
prev_node->hole_follows = 1;
list_add(&prev_node->hole_stack, &mm->hole_stack);
} else
list_move(&prev_node->hole_stack, &mm->hole_stack);
drm_mm_interval_tree_remove(node, &mm->interval_tree);
list_del(&node->node_list);
node->allocated = 0;
}
EXPORT_SYMBOL(drm_mm_remove_node);
static int check_free_hole(u64 start, u64 end, u64 size, unsigned alignment)
{
if (end - start < size)
return 0;
if (alignment) {
u64 tmp = start;
unsigned rem;
rem = do_div(tmp, alignment);
if (rem)
start += alignment - rem;
}
return end >= start + size;
}
static struct drm_mm_node *drm_mm_search_free_generic(const struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color,
enum drm_mm_search_flags flags)
{
struct drm_mm_node *entry;
struct drm_mm_node *best;
u64 adj_start;
u64 adj_end;
u64 best_size;
BUG_ON(mm->scanned_blocks);
best = NULL;
best_size = ~0UL;
__drm_mm_for_each_hole(entry, mm, adj_start, adj_end,
flags & DRM_MM_SEARCH_BELOW) {
u64 hole_size = adj_end - adj_start;
if (mm->color_adjust) {
mm->color_adjust(entry, color, &adj_start, &adj_end);
if (adj_end <= adj_start)
continue;
}
if (!check_free_hole(adj_start, adj_end, size, alignment))
continue;
if (!(flags & DRM_MM_SEARCH_BEST))
return entry;
if (hole_size < best_size) {
best = entry;
best_size = hole_size;
}
}
return best;
}
static struct drm_mm_node *drm_mm_search_free_in_range_generic(const struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color,
u64 start,
u64 end,
enum drm_mm_search_flags flags)
{
struct drm_mm_node *entry;
struct drm_mm_node *best;
u64 adj_start;
u64 adj_end;
u64 best_size;
BUG_ON(mm->scanned_blocks);
best = NULL;
best_size = ~0UL;
__drm_mm_for_each_hole(entry, mm, adj_start, adj_end,
flags & DRM_MM_SEARCH_BELOW) {
u64 hole_size = adj_end - adj_start;
if (adj_start < start)
adj_start = start;
if (adj_end > end)
adj_end = end;
if (mm->color_adjust) {
mm->color_adjust(entry, color, &adj_start, &adj_end);
if (adj_end <= adj_start)
continue;
}
if (!check_free_hole(adj_start, adj_end, size, alignment))
continue;
if (!(flags & DRM_MM_SEARCH_BEST))
return entry;
if (hole_size < best_size) {
best = entry;
best_size = hole_size;
}
}
return best;
}
/**
* drm_mm_replace_node - move an allocation from @old to @new
* @old: drm_mm_node to remove from the allocator
* @new: drm_mm_node which should inherit @old's allocation
*
* This is useful for when drivers embed the drm_mm_node structure and hence
* can't move allocations by reassigning pointers. It's a combination of remove
* and insert with the guarantee that the allocation start will match.
*/
void drm_mm_replace_node(struct drm_mm_node *old, struct drm_mm_node *new)
{
list_replace(&old->node_list, &new->node_list);
list_replace(&old->hole_stack, &new->hole_stack);
rb_replace_node(&old->rb, &new->rb, &old->mm->interval_tree);
new->hole_follows = old->hole_follows;
new->mm = old->mm;
new->start = old->start;
new->size = old->size;
new->color = old->color;
new->__subtree_last = old->__subtree_last;
old->allocated = 0;
new->allocated = 1;
}
EXPORT_SYMBOL(drm_mm_replace_node);
/**
* DOC: lru scan roaster
*
* Very often GPUs need to have continuous allocations for a given object. When
* evicting objects to make space for a new one it is therefore not most
* efficient when we simply start to select all objects from the tail of an LRU
* until there's a suitable hole: Especially for big objects or nodes that
* otherwise have special allocation constraints there's a good chance we evict
* lots of (smaller) objects unecessarily.
*
* The DRM range allocator supports this use-case through the scanning
* interfaces. First a scan operation needs to be initialized with
* drm_mm_init_scan() or drm_mm_init_scan_with_range(). The the driver adds
* objects to the roaster (probably by walking an LRU list, but this can be
* freely implemented) until a suitable hole is found or there's no further
* evitable object.
*
* The the driver must walk through all objects again in exactly the reverse
* order to restore the allocator state. Note that while the allocator is used
* in the scan mode no other operation is allowed.
*
* Finally the driver evicts all objects selected in the scan. Adding and
* removing an object is O(1), and since freeing a node is also O(1) the overall
* complexity is O(scanned_objects). So like the free stack which needs to be
* walked before a scan operation even begins this is linear in the number of
* objects. It doesn't seem to hurt badly.
*/
/**
* drm_mm_init_scan - initialize lru scanning
* @mm: drm_mm to scan
* @size: size of the allocation
* @alignment: alignment of the allocation
* @color: opaque tag value to use for the allocation
*
* This simply sets up the scanning routines with the parameters for the desired
* hole. Note that there's no need to specify allocation flags, since they only
* change the place a node is allocated from within a suitable hole.
*
* Warning:
* As long as the scan list is non-empty, no other operations than
* adding/removing nodes to/from the scan list are allowed.
*/
void drm_mm_init_scan(struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color)
{
mm->scan_color = color;
mm->scan_alignment = alignment;
mm->scan_size = size;
mm->scanned_blocks = 0;
mm->scan_hit_start = 0;
mm->scan_hit_end = 0;
mm->scan_check_range = 0;
mm->prev_scanned_node = NULL;
}
EXPORT_SYMBOL(drm_mm_init_scan);
/**
* drm_mm_init_scan - initialize range-restricted lru scanning
* @mm: drm_mm to scan
* @size: size of the allocation
* @alignment: alignment of the allocation
* @color: opaque tag value to use for the allocation
* @start: start of the allowed range for the allocation
* @end: end of the allowed range for the allocation
*
* This simply sets up the scanning routines with the parameters for the desired
* hole. Note that there's no need to specify allocation flags, since they only
* change the place a node is allocated from within a suitable hole.
*
* Warning:
* As long as the scan list is non-empty, no other operations than
* adding/removing nodes to/from the scan list are allowed.
*/
void drm_mm_init_scan_with_range(struct drm_mm *mm,
u64 size,
unsigned alignment,
unsigned long color,
u64 start,
u64 end)
{
mm->scan_color = color;
mm->scan_alignment = alignment;
mm->scan_size = size;
mm->scanned_blocks = 0;
mm->scan_hit_start = 0;
mm->scan_hit_end = 0;
mm->scan_start = start;
mm->scan_end = end;
mm->scan_check_range = 1;
mm->prev_scanned_node = NULL;
}
EXPORT_SYMBOL(drm_mm_init_scan_with_range);
/**
* drm_mm_scan_add_block - add a node to the scan list
* @node: drm_mm_node to add
*
* Add a node to the scan list that might be freed to make space for the desired
* hole.
*
* Returns:
* True if a hole has been found, false otherwise.
*/
bool drm_mm_scan_add_block(struct drm_mm_node *node)
{
struct drm_mm *mm = node->mm;
struct drm_mm_node *prev_node;
u64 hole_start, hole_end;
u64 adj_start, adj_end;
mm->scanned_blocks++;
BUG_ON(node->scanned_block);
node->scanned_block = 1;
prev_node = list_entry(node->node_list.prev, struct drm_mm_node,
node_list);
node->scanned_preceeds_hole = prev_node->hole_follows;
prev_node->hole_follows = 1;
list_del(&node->node_list);
node->node_list.prev = &prev_node->node_list;
node->node_list.next = &mm->prev_scanned_node->node_list;
mm->prev_scanned_node = node;
adj_start = hole_start = drm_mm_hole_node_start(prev_node);
adj_end = hole_end = drm_mm_hole_node_end(prev_node);
if (mm->scan_check_range) {
if (adj_start < mm->scan_start)
adj_start = mm->scan_start;
if (adj_end > mm->scan_end)
adj_end = mm->scan_end;
}
if (mm->color_adjust)
mm->color_adjust(prev_node, mm->scan_color,
&adj_start, &adj_end);
if (check_free_hole(adj_start, adj_end,
mm->scan_size, mm->scan_alignment)) {
mm->scan_hit_start = hole_start;
mm->scan_hit_end = hole_end;
return true;
}
return false;
}
EXPORT_SYMBOL(drm_mm_scan_add_block);
/**
* drm_mm_scan_remove_block - remove a node from the scan list
* @node: drm_mm_node to remove
*
* Nodes _must_ be removed in the exact same order from the scan list as they
* have been added, otherwise the internal state of the memory manager will be
* corrupted.
*
* When the scan list is empty, the selected memory nodes can be freed. An
* immediately following drm_mm_search_free with !DRM_MM_SEARCH_BEST will then
* return the just freed block (because its at the top of the free_stack list).
*
* Returns:
* True if this block should be evicted, false otherwise. Will always
* return false when no hole has been found.
*/
bool drm_mm_scan_remove_block(struct drm_mm_node *node)
{
struct drm_mm *mm = node->mm;
struct drm_mm_node *prev_node;
mm->scanned_blocks--;
BUG_ON(!node->scanned_block);
node->scanned_block = 0;
prev_node = list_entry(node->node_list.prev, struct drm_mm_node,
node_list);
prev_node->hole_follows = node->scanned_preceeds_hole;
list_add(&node->node_list, &prev_node->node_list);
return (drm_mm_hole_node_end(node) > mm->scan_hit_start &&
node->start < mm->scan_hit_end);
}
EXPORT_SYMBOL(drm_mm_scan_remove_block);
/**
* drm_mm_clean - checks whether an allocator is clean
* @mm: drm_mm allocator to check
*
* Returns:
* True if the allocator is completely free, false if there's still a node
* allocated in it.
*/
bool drm_mm_clean(struct drm_mm * mm)
{
struct list_head *head = &mm->head_node.node_list;
return (head->next->next == head);
}
EXPORT_SYMBOL(drm_mm_clean);
/**
* drm_mm_init - initialize a drm-mm allocator
* @mm: the drm_mm structure to initialize
* @start: start of the range managed by @mm
* @size: end of the range managed by @mm
*
* Note that @mm must be cleared to 0 before calling this function.
*/
void drm_mm_init(struct drm_mm * mm, u64 start, u64 size)
{
INIT_LIST_HEAD(&mm->hole_stack);
mm->scanned_blocks = 0;
/* Clever trick to avoid a special case in the free hole tracking. */
INIT_LIST_HEAD(&mm->head_node.node_list);
mm->head_node.allocated = 0;
mm->head_node.hole_follows = 1;
mm->head_node.scanned_block = 0;
mm->head_node.scanned_prev_free = 0;
mm->head_node.scanned_next_free = 0;
mm->head_node.mm = mm;
mm->head_node.start = start + size;
mm->head_node.size = start - mm->head_node.start;
list_add_tail(&mm->head_node.hole_stack, &mm->hole_stack);
mm->interval_tree = RB_ROOT;
mm->color_adjust = NULL;
}
EXPORT_SYMBOL(drm_mm_init);
/**
* drm_mm_takedown - clean up a drm_mm allocator
* @mm: drm_mm allocator to clean up
*
* Note that it is a bug to call this function on an allocator which is not
* clean.
*/
void drm_mm_takedown(struct drm_mm *mm)
{
if (WARN(!list_empty(&mm->head_node.node_list),
"Memory manager not clean during takedown.\n"))
show_leaks(mm);
}
EXPORT_SYMBOL(drm_mm_takedown);
static u64 drm_mm_debug_hole(struct drm_mm_node *entry,
const char *prefix)
{
u64 hole_start, hole_end, hole_size;
if (entry->hole_follows) {
hole_start = drm_mm_hole_node_start(entry);
hole_end = drm_mm_hole_node_end(entry);
hole_size = hole_end - hole_start;
pr_debug("%s %#llx-%#llx: %llu: free\n", prefix, hole_start,
hole_end, hole_size);
return hole_size;
}
return 0;
}
/**
* drm_mm_debug_table - dump allocator state to dmesg
* @mm: drm_mm allocator to dump
* @prefix: prefix to use for dumping to dmesg
*/
void drm_mm_debug_table(struct drm_mm *mm, const char *prefix)
{
struct drm_mm_node *entry;
u64 total_used = 0, total_free = 0, total = 0;
total_free += drm_mm_debug_hole(&mm->head_node, prefix);
drm_mm_for_each_node(entry, mm) {
pr_debug("%s %#llx-%#llx: %llu: used\n", prefix, entry->start,
entry->start + entry->size, entry->size);
total_used += entry->size;
total_free += drm_mm_debug_hole(entry, prefix);
}
total = total_free + total_used;
pr_debug("%s total: %llu, used %llu free %llu\n", prefix, total,
total_used, total_free);
}
EXPORT_SYMBOL(drm_mm_debug_table);
#if defined(CONFIG_DEBUG_FS)
static u64 drm_mm_dump_hole(struct seq_file *m, struct drm_mm_node *entry)
{
u64 hole_start, hole_end, hole_size;
if (entry->hole_follows) {
hole_start = drm_mm_hole_node_start(entry);
hole_end = drm_mm_hole_node_end(entry);
hole_size = hole_end - hole_start;
seq_printf(m, "%#018llx-%#018llx: %llu: free\n", hole_start,
hole_end, hole_size);
return hole_size;
}
return 0;
}
/**
* drm_mm_dump_table - dump allocator state to a seq_file
* @m: seq_file to dump to
* @mm: drm_mm allocator to dump
*/
int drm_mm_dump_table(struct seq_file *m, struct drm_mm *mm)
{
struct drm_mm_node *entry;
u64 total_used = 0, total_free = 0, total = 0;
total_free += drm_mm_dump_hole(m, &mm->head_node);
drm_mm_for_each_node(entry, mm) {
seq_printf(m, "%#018llx-%#018llx: %llu: used\n", entry->start,
entry->start + entry->size, entry->size);
total_used += entry->size;
total_free += drm_mm_dump_hole(m, entry);
}
total = total_free + total_used;
seq_printf(m, "total: %llu, used %llu free %llu\n", total,
total_used, total_free);
return 0;
}
EXPORT_SYMBOL(drm_mm_dump_table);
#endif