Welcome everyone again!
This is yet another quick update on Contiguous Memory Allocator patches.
This version includes another set of code cleanups requested by Mel
Gorman and a few minor bug fixes. I really hope that this version will
be accepted for merging and future development will be handled by
incremental patches.
ARM integration code has not been changed since v16. It provides
implementation of the ideas that has been discussed during Linaro Sprint
meeting in Cambourne, August 2011. Here are the details:
This version provides a solution for complete integration of CMA to
DMA mapping subsystem on ARM architecture. The issue caused by double
dma pages mapping and possible aliasing in coherent memory mapping has
been finally resolved, both for GFP_ATOMIC case (allocations comes from
coherent memory pool) and non-GFP_ATOMIC case (allocations comes from
CMA managed areas).
For coherent, nommu, ARMv4 and ARMv5 systems the current DMA-mapping
implementation has been kept.
For ARMv6+ systems, CMA has been enabled and a special pool of coherent
memory for atomic allocations has been created. The size of this pool
defaults to DEFAULT_CONSISTEN_DMA_SIZE/8, but can be changed with
coherent_pool kernel parameter (if really required).
All atomic allocations are served from this pool. I've did a little
simplification here, because there is no separate pool for writecombine
memory - such requests are also served from coherent pool. I don't
think that such simplification is a problem here - I found no driver
that use dma_alloc_writecombine with GFP_ATOMIC flags.
All non-atomic allocation are served from CMA area. Kernel mappings are
updated to reflect required memory attributes changes. This is possible
because during early boot, all CMA area are remapped with 4KiB pages in
kernel low-memory.
This version have been tested on Samsung S5PC110 based Goni machine and
Exynos4 UniversalC210 board with various V4L2 multimedia drivers.
Coherent atomic allocations has been tested by manually enabling the dma
bounce for the s3c-sdhci device.
All patches are prepared on top of Linux Kernel v3.3-rc2.
A few words for these who see CMA for the first time:
The Contiguous Memory Allocator (CMA) makes it possible for device
drivers to allocate big contiguous chunks of memory after the system
has booted.
The main difference from the similar frameworks is the fact that CMA
allows to transparently reuse memory region reserved for the big
chunk allocation as a system memory, so no memory is wasted when no
big chunk is allocated. Once the alloc request is issued, the
framework will migrate system pages to create a required big chunk of
physically contiguous memory.
For more information you can refer to nice LWN articles:
http://lwn.net/Articles/447405/ and http://lwn.net/Articles/450286/
as well as links to previous versions of the CMA framework.
The CMA framework has been initially developed by Michal Nazarewicz
at Samsung Poland R&D Center. Since version 9, I've taken over the
development, because Michal has left the company. Since version v17
Michal is working again on CMA patches and the current version is
the result of our joint open-source effort.
Best regards
Marek Szyprowski
Samsung Poland R&D Center
Links to previous versions of the patchset:
v19: <http://www.spinics.net/lists/linux-mm/msg29145.html>
v18: <http://www.spinics.net/lists/linux-mm/msg28125.html>
v17: <http://www.spinics.net/lists/arm-kernel/msg148499.html>
v16: <http://www.spinics.net/lists/linux-mm/msg25066.html>
v15: <http://www.spinics.net/lists/linux-mm/msg23365.html>
v14: <http://www.spinics.net/lists/linux-media/msg36536.html>
v13: (internal, intentionally not released)
v12: <http://www.spinics.net/lists/linux-media/msg35674.html>
v11: <http://www.spinics.net/lists/linux-mm/msg21868.html>
v10: <http://www.spinics.net/lists/linux-mm/msg20761.html>
v9: <http://article.gmane.org/gmane.linux.kernel.mm/60787>
v8: <http://article.gmane.org/gmane.linux.kernel.mm/56855>
v7: <http://article.gmane.org/gmane.linux.kernel.mm/55626>
v6: <http://article.gmane.org/gmane.linux.kernel.mm/55626>
v5: (intentionally left out as CMA v5 was identical to CMA v4)
v4: <http://article.gmane.org/gmane.linux.kernel.mm/52010>
v3: <http://article.gmane.org/gmane.linux.kernel.mm/51573>
v2: <http://article.gmane.org/gmane.linux.kernel.mm/50986>
v1: <http://article.gmane.org/gmane.linux.kernel.mm/50669>
Changelog:
v20:
1. Addressed even more comments from Mel Gorman and added his Acked-by tag
on most of the core memory management patches.
2. Squashed a few minor fixes here and there (corrected alignment calculation
for region limit, added adjusting low watermark level on reclaim, fixed
return value of __alloc_contig_migrate_range function)
3. Removed problematic "mm: page_alloc: update migrate type of pages on pcp
when isolating" patch and sligtly altered MIGRATE_CMA type handling what
solved the problem
4. Rebased onto v3.3-rc2
v19:
1. Addressed another set of comments and suggestions from Mel Gorman, mainly
related to breaking patches into smaller, single-feature related chunks
and rewriting already existing functions in memory compaction code.
2. Reworked completely page reclaim code, removed it from split_free_page()
and introduce direct call from alloc_contig_range().
3. Merged a fix from Mans Rullgard for correct cma area limit alignment.
4. Replaced broken "mm: page_alloc: set_migratetype_isolate: drain PCP prior
to isolating" patch with "mm: page_alloc: update migrate type of pages on
pcp when isolating" which is another attempt to solve this issue without
touching free_pcppages_bulk().
5. Rebased onto v3.3-rc1
v18:
1. Addressed comments and suggestions from Mel Gorman related to changes
in memory compaction code, most important points:
- removed "mm: page_alloc: handle MIGRATE_ISOLATE in free_pcppages_bulk()"
and moved all the logic to set_migratetype_isolate - see
"mm: page_alloc: set_migratetype_isolate: drain PCP prior to isolating"
patch
- code in "mm: compaction: introduce isolate_{free,migrate}pages_range()"
patch have been simplified and improved
- removed "mm: mmzone: introduce zone_pfn_same_memmap()" patch
2. Fixed crash on initialization if HIGHMEM is available on ARM platforms
3. Fixed problems with allocation of contiguous memory if all free pages
are occupied by page cache and reclaim is required.
4. Added a workaround for temporary migration failures (now CMA tries
to allocate different memory block in such case), what heavily increased
reliability of the CMA.
5. Minor cleanup here and there.
6. Rebased onto v3.2-rc7 kernel tree.
v17:
1. Replaced whole CMA core memory migration code to the new one kindly
provided by Michal Nazarewicz. The new code is based on memory
compaction framework not the memory hotplug, like it was before. This
change has been suggested by Mel Godman.
2. Addressed most of the comments from Andrew Morton and Mel Gorman in
the rest of the CMA code.
3. Fixed broken initialization on ARM systems with DMA zone enabled.
4. Rebased onto v3.2-rc2 kernel.
v16:
1. merged a fixup from Michal Nazarewicz to address comments from Dave
Hansen about checking if pfns belong to the same memory zone
2. merged a fix from Michal Nazarewicz for incorrect handling of pages
which belong to page block that is in MIGRATE_ISOLATE state, in very
rare cases the migrate type of page block might have been changed
from MIGRATE_CMA to MIGRATE_MOVABLE because of this bug
3. moved some common code to include/asm-generic
4. added support for x86 DMA-mapping framework for pci-dma hardware,
CMA can be now even more widely tested on KVM/QEMU and a lot of common
x86 boxes
5. rebased onto next-20111005 kernel tree, which includes changes in ARM
DMA-mapping subsystem (CONSISTENT_DMA_SIZE removal)
6. removed patch for CMA s5p-fimc device private regions (served only as
example) and provided the one that matches real life case - s5p-mfc
device
v15:
1. fixed calculation of the total memory after activating CMA area (was
broken from v12)
2. more code cleanup in drivers/base/dma-contiguous.c
3. added address limit for default CMA area
4. rewrote ARM DMA integration:
- removed "ARM: DMA: steal memory for DMA coherent mappings" patch
- kept current DMA mapping implementation for coherent, nommu and
ARMv4/ARMv5 systems
- enabled CMA for all ARMv6+ systems
- added separate, small pool for coherent atomic allocations, defaults
to CONSISTENT_DMA_SIZE/8, but can be changed with kernel parameter
coherent_pool=[size]
v14:
1. Merged with "ARM: DMA: steal memory for DMA coherent mappings"
patch, added support for GFP_ATOMIC allocations.
2. Added checks for NULL device pointer
v13: (internal, intentionally not released)
v12:
1. Fixed 2 nasty bugs in dma-contiguous allocator:
- alignment argument was not passed correctly
- range for dma_release_from_contiguous was not checked correctly
2. Added support for architecture specfic dma_contiguous_early_fixup()
function
3. CMA and DMA-mapping integration for ARM architechture has been
rewritten to take care of the memory aliasing issue that might
happen for newer ARM CPUs (mapping of the same pages with different
cache attributes is forbidden). TODO: add support for GFP_ATOMIC
allocations basing on the "ARM: DMA: steal memory for DMA coherent
mappings" patch and implement support for contiguous memory areas
that are placed in HIGHMEM zone
v11:
1. Removed genalloc usage and replaced it with direct calls to
bitmap_* functions, dropped patches that are not needed
anymore (genalloc extensions)
2. Moved all contiguous area management code from mm/cma.c
to drivers/base/dma-contiguous.c
3. Renamed cm_alloc/free to dma_alloc/release_from_contiguous
4. Introduced global, system wide (default) contiguous area
configured with kernel config and kernel cmdline parameters
5. Simplified initialization to just one function:
dma_declare_contiguous()
6. Added example of device private memory contiguous area
v10:
1. Rebased onto 3.0-rc2 and resolved all conflicts
2. Simplified CMA to be just a pure memory allocator, for use
with platfrom/bus specific subsystems, like dma-mapping.
Removed all device specific functions are calls.
3. Integrated with ARM DMA-mapping subsystem.
4. Code cleanup here and there.
5. Removed private context support.
v9: 1. Rebased onto 2.6.39-rc1 and resolved all conflicts
2. Fixed a bunch of nasty bugs that happened when the allocation
failed (mainly kernel oops due to NULL ptr dereference).
3. Introduced testing code: cma-regions compatibility layer and
videobuf2-cma memory allocator module.
v8: 1. The alloc_contig_range() function has now been separated from
CMA and put in page_allocator.c. This function tries to
migrate all LRU pages in specified range and then allocate the
range using alloc_contig_freed_pages().
2. Support for MIGRATE_CMA has been separated from the CMA code.
I have not tested if CMA works with ZONE_MOVABLE but I see no
reasons why it shouldn't.
3. I have added a @private argument when creating CMA contexts so
that one can reserve memory and not share it with the rest of
the system. This way, CMA acts only as allocation algorithm.
v7: 1. A lot of functionality that handled driver->allocator_context
mapping has been removed from the patchset. This is not to say
that this code is not needed, it's just not worth posting
everything in one patchset.
Currently, CMA is "just" an allocator. It uses it's own
migratetype (MIGRATE_CMA) for defining ranges of pageblokcs
which behave just like ZONE_MOVABLE but dispite the latter can
be put in arbitrary places.
2. The migration code that was introduced in the previous version
actually started working.
v6: 1. Most importantly, v6 introduces support for memory migration.
The implementation is not yet complete though.
Migration support means that when CMA is not using memory
reserved for it, page allocator can allocate pages from it.
When CMA wants to use the memory, the pages have to be moved
and/or evicted as to make room for CMA.
To make it possible it must be guaranteed that only movable and
reclaimable pages are allocated in CMA controlled regions.
This is done by introducing a MIGRATE_CMA migrate type that
guarantees exactly that.
Some of the migration code is "borrowed" from Kamezawa
Hiroyuki's alloc_contig_pages() implementation. The main
difference is that thanks to MIGRATE_CMA migrate type CMA
assumes that memory controlled by CMA are is always movable or
reclaimable so that it makes allocation decisions regardless of
the whether some pages are actually allocated and migrates them
if needed.
The most interesting patches from the patchset that implement
the functionality are:
09/13: mm: alloc_contig_free_pages() added
10/13: mm: MIGRATE_CMA migration type added
11/13: mm: MIGRATE_CMA isolation functions added
12/13: mm: cma: Migration support added [wip]
Currently, kernel panics in some situations which I am trying
to investigate.
2. cma_pin() and cma_unpin() functions has been added (after
a conversation with Johan Mossberg). The idea is that whenever
hardware does not use the memory (no transaction is on) the
chunk can be moved around. This would allow defragmentation to
be implemented if desired. No defragmentation algorithm is
provided at this time.
3. Sysfs support has been replaced with debugfs. I always felt
unsure about the sysfs interface and when Greg KH pointed it
out I finally got to rewrite it to debugfs.
v5: (intentionally left out as CMA v5 was identical to CMA v4)
v4: 1. The "asterisk" flag has been removed in favour of requiring
that platform will provide a "*=<regions>" rule in the map
attribute.
2. The terminology has been changed slightly renaming "kind" to
"type" of memory. In the previous revisions, the documentation
indicated that device drivers define memory kinds and now,
v3: 1. The command line parameters have been removed (and moved to
a separate patch, the fourth one). As a consequence, the
cma_set_defaults() function has been changed -- it no longer
accepts a string with list of regions but an array of regions.
2. The "asterisk" attribute has been removed. Now, each region
has an "asterisk" flag which lets one specify whether this
region should by considered "asterisk" region.
3. SysFS support has been moved to a separate patch (the third one
in the series) and now also includes list of regions.
v2: 1. The "cma_map" command line have been removed. In exchange,
a SysFS entry has been created under kernel/mm/contiguous.
The intended way of specifying the attributes is
a cma_set_defaults() function called by platform initialisation
code. "regions" attribute (the string specified by "cma"
command line parameter) can be overwritten with command line
parameter; the other attributes can be changed during run-time
using the SysFS entries.
2. The behaviour of the "map" attribute has been modified
slightly. Currently, if no rule matches given device it is
assigned regions specified by the "asterisk" attribute. It is
by default built from the region names given in "regions"
attribute.
3. Devices can register private regions as well as regions that
can be shared but are not reserved using standard CMA
mechanisms. A private region has no name and can be accessed
only by devices that have the pointer to it.
4. The way allocators are registered has changed. Currently,
a cma_allocator_register() function is used for that purpose.
Moreover, allocators are attached to regions the first time
memory is registered from the region or when allocator is
registered which means that allocators can be dynamic modules
that are loaded after the kernel booted (of course, it won't be
possible to allocate a chunk of memory from a region if
allocator is not loaded).
5. Index of new functions:
+static inline dma_addr_t __must_check
+cma_alloc_from(const char *regions, size_t size,
+ dma_addr_t alignment)
+static inline int
+cma_info_about(struct cma_info *info, const const char *regions)
+int __must_check cma_region_register(struct cma_region *reg);
+dma_addr_t __must_check
+cma_alloc_from_region(struct cma_region *reg,
+ size_t size, dma_addr_t alignment);
+static inline dma_addr_t __must_check
+cma_alloc_from(const char *regions,
+ size_t size, dma_addr_t alignment);
+int cma_allocator_register(struct cma_allocator *alloc);
Patches in this patchset:
Marek Szyprowski (6):
mm: extract reclaim code from __alloc_pages_direct_reclaim()
mm: trigger page reclaim in alloc_contig_range() to stabilize
watermarks
drivers: add Contiguous Memory Allocator
X86: integrate CMA with DMA-mapping subsystem
ARM: integrate CMA with DMA-mapping subsystem
ARM: Samsung: use CMA for 2 memory banks for s5p-mfc device
Michal Nazarewicz (9):
mm: page_alloc: remove trailing whitespace
mm: compaction: introduce isolate_migratepages_range().
mm: compaction: introduce map_pages()
mm: compaction: introduce isolate_freepages_range()
mm: compaction: export some of the functions
mm: page_alloc: introduce alloc_contig_range()
mm: page_alloc: change fallbacks array handling
mm: mmzone: MIGRATE_CMA migration type added
mm: page_isolation: MIGRATE_CMA isolation functions added
Documentation/kernel-parameters.txt | 9 +
arch/Kconfig | 3 +
arch/arm/Kconfig | 2 +
arch/arm/include/asm/dma-contiguous.h | 16 ++
arch/arm/include/asm/mach/map.h | 1 +
arch/arm/kernel/setup.c | 9 +-
arch/arm/mm/dma-mapping.c | 368 ++++++++++++++++++++++++------
arch/arm/mm/init.c | 22 ++-
arch/arm/mm/mm.h | 3 +
arch/arm/mm/mmu.c | 31 ++-
arch/arm/plat-s5p/dev-mfc.c | 51 +----
arch/x86/Kconfig | 1 +
arch/x86/include/asm/dma-contiguous.h | 13 +
arch/x86/include/asm/dma-mapping.h | 4 +
arch/x86/kernel/pci-dma.c | 18 ++-
arch/x86/kernel/pci-nommu.c | 8 +-
arch/x86/kernel/setup.c | 2 +
drivers/base/Kconfig | 89 +++++++
drivers/base/Makefile | 1 +
drivers/base/dma-contiguous.c | 405 ++++++++++++++++++++++++++++++++
include/asm-generic/dma-contiguous.h | 27 ++
include/linux/device.h | 4 +
include/linux/dma-contiguous.h | 110 +++++++++
include/linux/gfp.h | 12 +
include/linux/mmzone.h | 38 +++-
include/linux/page-isolation.h | 18 +-
mm/Kconfig | 2 +-
mm/Makefile | 3 +-
mm/compaction.c | 418 +++++++++++++++++++++------------
mm/internal.h | 33 +++
mm/memory-failure.c | 2 +-
mm/memory_hotplug.c | 6 +-
mm/page_alloc.c | 377 ++++++++++++++++++++++++++---
mm/page_isolation.c | 15 +-
mm/vmstat.c | 3 +
35 files changed, 1757 insertions(+), 367 deletions(-)
create mode 100644 arch/arm/include/asm/dma-contiguous.h
create mode 100644 arch/x86/include/asm/dma-contiguous.h
create mode 100644 drivers/base/dma-contiguous.c
create mode 100644 include/asm-generic/dma-contiguous.h
create mode 100644 include/linux/dma-contiguous.h
--
1.7.1.569.g6f426
alloc_contig_range() performs memory allocation so it also should keep
track on keeping the correct level of memory watermarks. This commit adds
a call to *_slowpath style reclaim to grab enough pages to make sure that
the final collection of contiguous pages from freelists will not starve
the system.
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
CC: Michal Nazarewicz <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/page_alloc.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 47 insertions(+), 0 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 983ccba..371a79f 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5632,6 +5632,46 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
return ret > 0 ? 0 : ret;
}
+/*
+ * Trigger memory pressure bump to reclaim some pages in order to be able to
+ * allocate 'count' pages in single page units. Does similar work as
+ *__alloc_pages_slowpath() function.
+ */
+static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
+{
+ enum zone_type high_zoneidx = gfp_zone(gfp_mask);
+ struct zonelist *zonelist = node_zonelist(0, gfp_mask);
+ int did_some_progress = 0;
+ int order = 1;
+ unsigned long watermark;
+
+ /*
+ * Increase level of watermarks to force kswapd do his job
+ * to stabilize at new watermark level.
+ */
+ min_free_kbytes += count * PAGE_SIZE / 1024;
+ setup_per_zone_wmarks();
+
+ /* Obey watermarks as if the page was being allocated */
+ watermark = low_wmark_pages(zone) + count;
+ while (!zone_watermark_ok(zone, 0, watermark, 0, 0)) {
+ wake_all_kswapd(order, zonelist, high_zoneidx, zone_idx(zone));
+
+ did_some_progress = __perform_reclaim(gfp_mask, order, zonelist,
+ NULL);
+ if (!did_some_progress) {
+ /* Exhausted what can be done so it's blamo time */
+ out_of_memory(zonelist, gfp_mask, order, NULL);
+ }
+ }
+
+ /* Restore original watermark levels. */
+ min_free_kbytes -= count * PAGE_SIZE / 1024;
+ setup_per_zone_wmarks();
+
+ return count;
+}
+
/**
* alloc_contig_range() -- tries to allocate given range of pages
* @start: start PFN to allocate
@@ -5730,6 +5770,13 @@ int alloc_contig_range(unsigned long start, unsigned long end,
goto done;
}
+ /*
+ * Reclaim enough pages to make sure that contiguous allocation
+ * will not starve the system.
+ */
+ __reclaim_pages(zone, GFP_HIGHUSER_MOVABLE, end-start);
+
+ /* Grab isolated pages from freelists. */
outer_end = isolate_freepages_range(outer_start, end);
if (!outer_end) {
ret = -EBUSY;
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
The MIGRATE_CMA migration type has two main characteristics:
(i) only movable pages can be allocated from MIGRATE_CMA
pageblocks and (ii) page allocator will never change migration
type of MIGRATE_CMA pageblocks.
This guarantees (to some degree) that page in a MIGRATE_CMA page
block can always be migrated somewhere else (unless there's no
memory left in the system).
It is designed to be used for allocating big chunks (eg. 10MiB)
of physically contiguous memory. Once driver requests
contiguous memory, pages from MIGRATE_CMA pageblocks may be
migrated away to create a contiguous block.
To minimise number of migrations, MIGRATE_CMA migration type
is the last type tried when page allocator falls back to other
migration types then requested.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
include/linux/gfp.h | 3 ++
include/linux/mmzone.h | 38 +++++++++++++++++++----
mm/Kconfig | 2 +-
mm/compaction.c | 11 +++++--
mm/page_alloc.c | 78 ++++++++++++++++++++++++++++++++++++++----------
mm/vmstat.c | 3 ++
6 files changed, 108 insertions(+), 27 deletions(-)
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 052a5b6..78d32a7 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -397,6 +397,9 @@ static inline bool pm_suspended_storage(void)
extern int alloc_contig_range(unsigned long start, unsigned long end);
extern void free_contig_range(unsigned long pfn, unsigned nr_pages);
+/* CMA stuff */
+extern void init_cma_reserved_pageblock(struct page *page);
+
#endif
#endif /* __LINUX_GFP_H */
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 650ba2f..82f4fa5 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -35,13 +35,37 @@
*/
#define PAGE_ALLOC_COSTLY_ORDER 3
-#define MIGRATE_UNMOVABLE 0
-#define MIGRATE_RECLAIMABLE 1
-#define MIGRATE_MOVABLE 2
-#define MIGRATE_PCPTYPES 3 /* the number of types on the pcp lists */
-#define MIGRATE_RESERVE 3
-#define MIGRATE_ISOLATE 4 /* can't allocate from here */
-#define MIGRATE_TYPES 5
+enum {
+ MIGRATE_UNMOVABLE,
+ MIGRATE_RECLAIMABLE,
+ MIGRATE_MOVABLE,
+ MIGRATE_PCPTYPES, /* the number of types on the pcp lists */
+ MIGRATE_RESERVE = MIGRATE_PCPTYPES,
+#ifdef CONFIG_CMA
+ /*
+ * MIGRATE_CMA migration type is designed to mimic the way
+ * ZONE_MOVABLE works. Only movable pages can be allocated
+ * from MIGRATE_CMA pageblocks and page allocator never
+ * implicitly change migration type of MIGRATE_CMA pageblock.
+ *
+ * The way to use it is to change migratetype of a range of
+ * pageblocks to MIGRATE_CMA which can be done by
+ * __free_pageblock_cma() function. What is important though
+ * is that a range of pageblocks must be aligned to
+ * MAX_ORDER_NR_PAGES should biggest page be bigger then
+ * a single pageblock.
+ */
+ MIGRATE_CMA,
+#endif
+ MIGRATE_ISOLATE, /* can't allocate from here */
+ MIGRATE_TYPES
+};
+
+#ifdef CONFIG_CMA
+# define is_migrate_cma(migratetype) unlikely((migratetype) == MIGRATE_CMA)
+#else
+# define is_migrate_cma(migratetype) false
+#endif
#define for_each_migratetype_order(order, type) \
for (order = 0; order < MAX_ORDER; order++) \
diff --git a/mm/Kconfig b/mm/Kconfig
index e338407..3922002 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -198,7 +198,7 @@ config COMPACTION
config MIGRATION
bool "Page migration"
def_bool y
- depends on NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION
+ depends on NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA
help
Allows the migration of the physical location of pages of processes
while the virtual addresses are not changed. This is useful in
diff --git a/mm/compaction.c b/mm/compaction.c
index d5174c4..a6e7c64 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -45,6 +45,11 @@ static void map_pages(struct list_head *list)
}
}
+static inline bool migrate_async_suitable(int migratetype)
+{
+ return is_migrate_cma(migratetype) || migratetype == MIGRATE_MOVABLE;
+}
+
/*
* Isolate free pages onto a private freelist. Caller must hold zone->lock.
* If @strict is true, will abort returning 0 on any invalid PFNs or non-free
@@ -277,7 +282,7 @@ isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
*/
pageblock_nr = low_pfn >> pageblock_order;
if (!cc->sync && last_pageblock_nr != pageblock_nr &&
- get_pageblock_migratetype(page) != MIGRATE_MOVABLE) {
+ migrate_async_suitable(get_pageblock_migratetype(page))) {
low_pfn += pageblock_nr_pages;
low_pfn = ALIGN(low_pfn, pageblock_nr_pages) - 1;
last_pageblock_nr = pageblock_nr;
@@ -345,8 +350,8 @@ static bool suitable_migration_target(struct page *page)
if (PageBuddy(page) && page_order(page) >= pageblock_order)
return true;
- /* If the block is MIGRATE_MOVABLE, allow migration */
- if (migratetype == MIGRATE_MOVABLE)
+ /* If the block is MIGRATE_MOVABLE or MIGRATE_CMA, allow migration */
+ if (migrate_async_suitable(migratetype))
return true;
/* Otherwise skip the block */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 238fcec..993c375 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -750,6 +750,26 @@ void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
__free_pages(page, order);
}
+#ifdef CONFIG_CMA
+/*
+ * Free whole pageblock and set it's migration type to MIGRATE_CMA.
+ */
+void __init init_cma_reserved_pageblock(struct page *page)
+{
+ unsigned i = pageblock_nr_pages;
+ struct page *p = page;
+
+ do {
+ __ClearPageReserved(p);
+ set_page_count(p, 0);
+ } while (++p, --i);
+
+ set_page_refcounted(page);
+ set_pageblock_migratetype(page, MIGRATE_CMA);
+ __free_pages(page, pageblock_order);
+ totalram_pages += pageblock_nr_pages;
+}
+#endif
/*
* The order of subdivision here is critical for the IO subsystem.
@@ -875,10 +895,15 @@ struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
* This array describes the order lists are fallen back to when
* the free lists for the desirable migrate type are depleted
*/
-static int fallbacks[MIGRATE_TYPES][3] = {
- [MIGRATE_UNMOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
- [MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
- [MIGRATE_MOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
+static int fallbacks[MIGRATE_TYPES][4] = {
+ [MIGRATE_UNMOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
+ [MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
+#ifdef CONFIG_CMA
+ [MIGRATE_MOVABLE] = { MIGRATE_CMA, MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
+ [MIGRATE_CMA] = { MIGRATE_RESERVE }, /* Never used */
+#else
+ [MIGRATE_MOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
+#endif
[MIGRATE_RESERVE] = { MIGRATE_RESERVE }, /* Never used */
[MIGRATE_ISOLATE] = { MIGRATE_RESERVE }, /* Never used */
};
@@ -995,11 +1020,18 @@ __rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
* pages to the preferred allocation list. If falling
* back for a reclaimable kernel allocation, be more
* aggressive about taking ownership of free pages
+ *
+ * On the other hand, never change migration
+ * type of MIGRATE_CMA pageblocks nor move CMA
+ * pages on different free lists. We don't
+ * want unmovable pages to be allocated from
+ * MIGRATE_CMA areas.
*/
- if (unlikely(current_order >= (pageblock_order >> 1)) ||
- start_migratetype == MIGRATE_RECLAIMABLE ||
- page_group_by_mobility_disabled) {
- unsigned long pages;
+ if (!is_migrate_cma(migratetype) &&
+ (unlikely(current_order >= pageblock_order / 2) ||
+ start_migratetype == MIGRATE_RECLAIMABLE ||
+ page_group_by_mobility_disabled)) {
+ int pages;
pages = move_freepages_block(zone, page,
start_migratetype);
@@ -1017,11 +1049,14 @@ __rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
rmv_page_order(page);
/* Take ownership for orders >= pageblock_order */
- if (current_order >= pageblock_order)
+ if (current_order >= pageblock_order &&
+ !is_migrate_cma(migratetype))
change_pageblock_range(page, current_order,
start_migratetype);
- expand(zone, page, order, current_order, area, migratetype);
+ expand(zone, page, order, current_order, area,
+ is_migrate_cma(migratetype)
+ ? migratetype : start_migratetype);
trace_mm_page_alloc_extfrag(page, order, current_order,
start_migratetype, migratetype);
@@ -1072,7 +1107,7 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
unsigned long count, struct list_head *list,
int migratetype, int cold)
{
- int i;
+ int mt = migratetype, i;
spin_lock(&zone->lock);
for (i = 0; i < count; ++i) {
@@ -1093,7 +1128,12 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
list_add(&page->lru, list);
else
list_add_tail(&page->lru, list);
- set_page_private(page, migratetype);
+#ifdef CONFIG_CMA
+ mt = get_pageblock_migratetype(page);
+ if (!is_migrate_cma(mt) && mt != MIGRATE_ISOLATE)
+ mt = migratetype;
+#endif
+ set_page_private(page, mt);
list = &page->lru;
}
__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
@@ -1337,8 +1377,12 @@ int split_free_page(struct page *page)
if (order >= pageblock_order - 1) {
struct page *endpage = page + (1 << order) - 1;
- for (; page < endpage; page += pageblock_nr_pages)
- set_pageblock_migratetype(page, MIGRATE_MOVABLE);
+ for (; page < endpage; page += pageblock_nr_pages) {
+ int mt = get_pageblock_migratetype(page);
+ if (mt != MIGRATE_ISOLATE && !is_migrate_cma(mt))
+ set_pageblock_migratetype(page,
+ MIGRATE_MOVABLE);
+ }
}
return 1 << order;
@@ -5369,14 +5413,16 @@ static int
__count_immobile_pages(struct zone *zone, struct page *page, int count)
{
unsigned long pfn, iter, found;
+ int mt;
+
/*
* For avoiding noise data, lru_add_drain_all() should be called
* If ZONE_MOVABLE, the zone never contains immobile pages
*/
if (zone_idx(zone) == ZONE_MOVABLE)
return true;
-
- if (get_pageblock_migratetype(page) == MIGRATE_MOVABLE)
+ mt = get_pageblock_migratetype(page);
+ if (mt == MIGRATE_MOVABLE || is_migrate_cma(mt))
return true;
pfn = page_to_pfn(page);
diff --git a/mm/vmstat.c b/mm/vmstat.c
index f600557..c47ac13 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -613,6 +613,9 @@ static char * const migratetype_names[MIGRATE_TYPES] = {
"Reclaimable",
"Movable",
"Reserve",
+#ifdef CONFIG_CMA
+ "CMA",
+#endif
"Isolate",
};
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit adds the alloc_contig_range() function which tries
to allocate given range of pages. It tries to migrate all
already allocated pages that fall in the range thus freeing them.
Once all pages in the range are freed they are removed from the
buddy system thus allocated for the caller to use.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
include/linux/gfp.h | 8 ++
mm/page_alloc.c | 185 +++++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 193 insertions(+), 0 deletions(-)
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 581e74b..052a5b6 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -391,4 +391,12 @@ static inline bool pm_suspended_storage(void)
}
#endif /* CONFIG_PM_SLEEP */
+#ifdef CONFIG_CMA
+
+/* The below functions must be run on a range from a single zone. */
+extern int alloc_contig_range(unsigned long start, unsigned long end);
+extern void free_contig_range(unsigned long pfn, unsigned nr_pages);
+
+#endif
+
#endif /* __LINUX_GFP_H */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 7fe7697..9006e69 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -57,6 +57,7 @@
#include <linux/ftrace_event.h>
#include <linux/memcontrol.h>
#include <linux/prefetch.h>
+#include <linux/migrate.h>
#include <linux/page-debug-flags.h>
#include <asm/tlbflush.h>
@@ -5505,6 +5506,190 @@ out:
spin_unlock_irqrestore(&zone->lock, flags);
}
+#ifdef CONFIG_CMA
+
+static unsigned long pfn_align_to_maxpage_down(unsigned long pfn)
+{
+ return pfn & ~(MAX_ORDER_NR_PAGES - 1);
+}
+
+static unsigned long pfn_align_to_maxpage_up(unsigned long pfn)
+{
+ return ALIGN(pfn, MAX_ORDER_NR_PAGES);
+}
+
+static struct page *
+__alloc_contig_migrate_alloc(struct page *page, unsigned long private,
+ int **resultp)
+{
+ return alloc_page(GFP_HIGHUSER_MOVABLE);
+}
+
+/* [start, end) must belong to a single zone. */
+static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
+{
+ /* This function is based on compact_zone() from compaction.c. */
+
+ unsigned long pfn = start;
+ unsigned int tries = 0;
+ int ret = 0;
+
+ struct compact_control cc = {
+ .nr_migratepages = 0,
+ .order = -1,
+ .zone = page_zone(pfn_to_page(start)),
+ .sync = true,
+ };
+ INIT_LIST_HEAD(&cc.migratepages);
+
+ migrate_prep_local();
+
+ while (pfn < end || !list_empty(&cc.migratepages)) {
+ if (fatal_signal_pending(current)) {
+ ret = -EINTR;
+ break;
+ }
+
+ if (list_empty(&cc.migratepages)) {
+ cc.nr_migratepages = 0;
+ pfn = isolate_migratepages_range(cc.zone, &cc,
+ pfn, end);
+ if (!pfn) {
+ ret = -EINTR;
+ break;
+ }
+ tries = 0;
+ } else if (++tries == 5) {
+ ret = ret < 0 ? ret : -EBUSY;
+ break;
+ }
+
+ ret = migrate_pages(&cc.migratepages,
+ __alloc_contig_migrate_alloc,
+ 0, false, true);
+ }
+
+ putback_lru_pages(&cc.migratepages);
+ return ret > 0 ? 0 : ret;
+}
+
+/**
+ * alloc_contig_range() -- tries to allocate given range of pages
+ * @start: start PFN to allocate
+ * @end: one-past-the-last PFN to allocate
+ *
+ * The PFN range does not have to be pageblock or MAX_ORDER_NR_PAGES
+ * aligned, however it's the caller's responsibility to guarantee that
+ * we are the only thread that changes migrate type of pageblocks the
+ * pages fall in.
+ *
+ * The PFN range must belong to a single zone.
+ *
+ * Returns zero on success or negative error code. On success all
+ * pages which PFN is in [start, end) are allocated for the caller and
+ * need to be freed with free_contig_range().
+ */
+int alloc_contig_range(unsigned long start, unsigned long end)
+{
+ struct zone *zone = page_zone(pfn_to_page(start));
+ unsigned long outer_start, outer_end;
+ int ret = 0, order;
+
+ /*
+ * What we do here is we mark all pageblocks in range as
+ * MIGRATE_ISOLATE. Because of the way page allocator work, we
+ * align the range to MAX_ORDER pages so that page allocator
+ * won't try to merge buddies from different pageblocks and
+ * change MIGRATE_ISOLATE to some other migration type.
+ *
+ * Once the pageblocks are marked as MIGRATE_ISOLATE, we
+ * migrate the pages from an unaligned range (ie. pages that
+ * we are interested in). This will put all the pages in
+ * range back to page allocator as MIGRATE_ISOLATE.
+ *
+ * When this is done, we take the pages in range from page
+ * allocator removing them from the buddy system. This way
+ * page allocator will never consider using them.
+ *
+ * This lets us mark the pageblocks back as
+ * MIGRATE_CMA/MIGRATE_MOVABLE so that free pages in the
+ * MAX_ORDER aligned range but not in the unaligned, original
+ * range are put back to page allocator so that buddy can use
+ * them.
+ */
+
+ ret = start_isolate_page_range(pfn_align_to_maxpage_down(start),
+ pfn_align_to_maxpage_up(end));
+ if (ret)
+ goto done;
+
+ ret = __alloc_contig_migrate_range(start, end);
+ if (ret)
+ goto done;
+
+ /*
+ * Pages from [start, end) are within a MAX_ORDER_NR_PAGES
+ * aligned blocks that are marked as MIGRATE_ISOLATE. What's
+ * more, all pages in [start, end) are free in page allocator.
+ * What we are going to do is to allocate all pages from
+ * [start, end) (that is remove them from page allocater).
+ *
+ * The only problem is that pages at the beginning and at the
+ * end of interesting range may be not aligned with pages that
+ * page allocator holds, ie. they can be part of higher order
+ * pages. Because of this, we reserve the bigger range and
+ * once this is done free the pages we are not interested in.
+ *
+ * We don't have to hold zone->lock here because the pages are
+ * isolated thus they won't get removed from buddy.
+ */
+
+ lru_add_drain_all();
+ drain_all_pages();
+
+ order = 0;
+ outer_start = start;
+ while (!PageBuddy(pfn_to_page(outer_start))) {
+ if (++order >= MAX_ORDER) {
+ ret = -EBUSY;
+ goto done;
+ }
+ outer_start &= ~0UL << order;
+ }
+
+ /* Make sure the range is really isolated. */
+ if (test_pages_isolated(outer_start, end)) {
+ pr_warn("alloc_contig_range test_pages_isolated(%lx, %lx) failed\n",
+ outer_start, end);
+ ret = -EBUSY;
+ goto done;
+ }
+
+ outer_end = isolate_freepages_range(outer_start, end);
+ if (!outer_end) {
+ ret = -EBUSY;
+ goto done;
+ }
+
+ /* Free head and tail (if any) */
+ if (start != outer_start)
+ free_contig_range(outer_start, start - outer_start);
+ if (end != outer_end)
+ free_contig_range(end, outer_end - end);
+
+done:
+ undo_isolate_page_range(pfn_align_to_maxpage_down(start),
+ pfn_align_to_maxpage_up(end));
+ return ret;
+}
+
+void free_contig_range(unsigned long pfn, unsigned nr_pages)
+{
+ for (; nr_pages--; ++pfn)
+ __free_page(pfn_to_page(pfn));
+}
+#endif
+
#ifdef CONFIG_MEMORY_HOTREMOVE
/*
* All pages in the range must be isolated before calling this.
--
1.7.1.569.g6f426
The Contiguous Memory Allocator is a set of helper functions for DMA
mapping framework that improves allocations of contiguous memory chunks.
CMA grabs memory on system boot, marks it with CMA_MIGRATE_TYPE and
gives back to the system. Kernel is allowed to allocate movable pages
within CMA's managed memory so that it can be used for example for page
cache when DMA mapping do not use it. On dma_alloc_from_contiguous()
request such pages are migrated out of CMA area to free required
contiguous block and fulfill the request. This allows to allocate large
contiguous chunks of memory at any time assuming that there is enough
free memory available in the system.
This code is heavily based on earlier works by Michal Nazarewicz.
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
CC: Michal Nazarewicz <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
Documentation/kernel-parameters.txt | 5 +
arch/Kconfig | 3 +
drivers/base/Kconfig | 89 ++++++++
drivers/base/Makefile | 1 +
drivers/base/dma-contiguous.c | 405 ++++++++++++++++++++++++++++++++++
include/asm-generic/dma-contiguous.h | 27 +++
include/linux/device.h | 4 +
include/linux/dma-contiguous.h | 110 +++++++++
8 files changed, 644 insertions(+), 0 deletions(-)
create mode 100644 drivers/base/dma-contiguous.c
create mode 100644 include/asm-generic/dma-contiguous.h
create mode 100644 include/linux/dma-contiguous.h
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 033d4e6..84982e2 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -508,6 +508,11 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
Also note the kernel might malfunction if you disable
some critical bits.
+ cma=nn[MG] [ARM,KNL]
+ Sets the size of kernel global memory area for contiguous
+ memory allocations. For more information, see
+ include/linux/dma-contiguous.h
+
cmo_free_hint= [PPC] Format: { yes | no }
Specify whether pages are marked as being inactive
when they are freed. This is used in CMO environments
diff --git a/arch/Kconfig b/arch/Kconfig
index 4f55c73..8ec200c 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -128,6 +128,9 @@ config HAVE_ARCH_TRACEHOOK
config HAVE_DMA_ATTRS
bool
+config HAVE_DMA_CONTIGUOUS
+ bool
+
config USE_GENERIC_SMP_HELPERS
bool
diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig
index 7be9f79..f56cb20 100644
--- a/drivers/base/Kconfig
+++ b/drivers/base/Kconfig
@@ -189,4 +189,93 @@ config DMA_SHARED_BUFFER
APIs extension; the file's descriptor can then be passed on to other
driver.
+config CMA
+ bool "Contiguous Memory Allocator (EXPERIMENTAL)"
+ depends on HAVE_DMA_CONTIGUOUS && HAVE_MEMBLOCK && EXPERIMENTAL
+ select MIGRATION
+ help
+ This enables the Contiguous Memory Allocator which allows drivers
+ to allocate big physically-contiguous blocks of memory for use with
+ hardware components that do not support I/O map nor scatter-gather.
+
+ For more information see <include/linux/dma-contiguous.h>.
+ If unsure, say "n".
+
+if CMA
+
+config CMA_DEBUG
+ bool "CMA debug messages (DEVELOPMENT)"
+ depends on DEBUG_KERNEL
+ help
+ Turns on debug messages in CMA. This produces KERN_DEBUG
+ messages for every CMA call as well as various messages while
+ processing calls such as dma_alloc_from_contiguous().
+ This option does not affect warning and error messages.
+
+comment "Default contiguous memory area size:"
+
+config CMA_SIZE_MBYTES
+ int "Size in Mega Bytes"
+ depends on !CMA_SIZE_SEL_PERCENTAGE
+ default 16
+ help
+ Defines the size (in MiB) of the default memory area for Contiguous
+ Memory Allocator.
+
+config CMA_SIZE_PERCENTAGE
+ int "Percentage of total memory"
+ depends on !CMA_SIZE_SEL_MBYTES
+ default 10
+ help
+ Defines the size of the default memory area for Contiguous Memory
+ Allocator as a percentage of the total memory in the system.
+
+choice
+ prompt "Selected region size"
+ default CMA_SIZE_SEL_ABSOLUTE
+
+config CMA_SIZE_SEL_MBYTES
+ bool "Use mega bytes value only"
+
+config CMA_SIZE_SEL_PERCENTAGE
+ bool "Use percentage value only"
+
+config CMA_SIZE_SEL_MIN
+ bool "Use lower value (minimum)"
+
+config CMA_SIZE_SEL_MAX
+ bool "Use higher value (maximum)"
+
+endchoice
+
+config CMA_ALIGNMENT
+ int "Maximum PAGE_SIZE order of alignment for contiguous buffers"
+ range 4 9
+ default 8
+ help
+ DMA mapping framework by default aligns all buffers to the smallest
+ PAGE_SIZE order which is greater than or equal to the requested buffer
+ size. This works well for buffers up to a few hundreds kilobytes, but
+ for larger buffers it just a memory waste. With this parameter you can
+ specify the maximum PAGE_SIZE order for contiguous buffers. Larger
+ buffers will be aligned only to this specified order. The order is
+ expressed as a power of two multiplied by the PAGE_SIZE.
+
+ For example, if your system defaults to 4KiB pages, the order value
+ of 8 means that the buffers will be aligned up to 1MiB only.
+
+ If unsure, leave the default value "8".
+
+config CMA_AREAS
+ int "Maximum count of the CMA device-private areas"
+ default 7
+ help
+ CMA allows to create CMA areas for particular devices. This parameter
+ sets the maximum number of such device private CMA areas in the
+ system.
+
+ If unsure, leave the default value "7".
+
+endif
+
endmenu
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 610f999..ffd9dae 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -6,6 +6,7 @@ obj-y := core.o bus.o dd.o syscore.o \
attribute_container.o transport_class.o \
topology.o
obj-$(CONFIG_DEVTMPFS) += devtmpfs.o
+obj-$(CONFIG_CMA) += dma-contiguous.o
obj-y += power/
obj-$(CONFIG_HAS_DMA) += dma-mapping.o
obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o
diff --git a/drivers/base/dma-contiguous.c b/drivers/base/dma-contiguous.c
new file mode 100644
index 0000000..c33f771
--- /dev/null
+++ b/drivers/base/dma-contiguous.c
@@ -0,0 +1,405 @@
+/*
+ * Contiguous Memory Allocator for DMA mapping framework
+ * Copyright (c) 2010-2011 by Samsung Electronics.
+ * Written by:
+ * Marek Szyprowski <[email protected]>
+ * Michal Nazarewicz <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of the
+ * License or (at your optional) any later version of the license.
+ */
+
+#define pr_fmt(fmt) "cma: " fmt
+
+#ifdef CONFIG_CMA_DEBUG
+#ifndef DEBUG
+# define DEBUG
+#endif
+#endif
+
+#include <asm/page.h>
+#include <asm/dma-contiguous.h>
+
+#include <linux/memblock.h>
+#include <linux/err.h>
+#include <linux/mm.h>
+#include <linux/mutex.h>
+#include <linux/page-isolation.h>
+#include <linux/slab.h>
+#include <linux/swap.h>
+#include <linux/mm_types.h>
+#include <linux/dma-contiguous.h>
+
+#ifndef SZ_1M
+#define SZ_1M (1 << 20)
+#endif
+
+struct cma {
+ unsigned long base_pfn;
+ unsigned long count;
+ unsigned long *bitmap;
+};
+
+struct cma *dma_contiguous_default_area;
+
+#ifdef CONFIG_CMA_SIZE_MBYTES
+#define CMA_SIZE_MBYTES CONFIG_CMA_SIZE_MBYTES
+#else
+#define CMA_SIZE_MBYTES 0
+#endif
+
+#ifdef CONFIG_CMA_SIZE_PERCENTAGE
+#define CMA_SIZE_PERCENTAGE CONFIG_CMA_SIZE_PERCENTAGE
+#else
+#define CMA_SIZE_PERCENTAGE 0
+#endif
+
+/*
+ * Default global CMA area size can be defined in kernel's .config.
+ * This is usefull mainly for distro maintainers to create a kernel
+ * that works correctly for most supported systems.
+ * The size can be set in bytes or as a percentage of the total memory
+ * in the system.
+ *
+ * Users, who want to set the size of global CMA area for their system
+ * should use cma= kernel parameter.
+ */
+static unsigned long size_bytes = CMA_SIZE_MBYTES * SZ_1M;
+static unsigned long size_percent = CMA_SIZE_PERCENTAGE;
+static long size_cmdline = -1;
+
+static int __init early_cma(char *p)
+{
+ pr_debug("%s(%s)\n", __func__, p);
+ size_cmdline = memparse(p, &p);
+ return 0;
+}
+early_param("cma", early_cma);
+
+static unsigned long __init cma_early_get_total_pages(void)
+{
+ struct memblock_region *reg;
+ unsigned long total_pages = 0;
+
+ /*
+ * We cannot use memblock_phys_mem_size() here, because
+ * memblock_analyze() has not been called yet.
+ */
+ for_each_memblock(memory, reg)
+ total_pages += memblock_region_memory_end_pfn(reg) -
+ memblock_region_memory_base_pfn(reg);
+ return total_pages;
+}
+
+/**
+ * dma_contiguous_reserve() - reserve area for contiguous memory handling
+ * @limit: End address of the reserved memory (optional, 0 for any).
+ *
+ * This funtion reserves memory from early allocator. It should be
+ * called by arch specific code once the early allocator (memblock or bootmem)
+ * has been activated and all other subsystems have already allocated/reserved
+ * memory.
+ */
+void __init dma_contiguous_reserve(phys_addr_t limit)
+{
+ unsigned long selected_size = 0;
+ unsigned long total_pages;
+
+ pr_debug("%s(limit %08lx)\n", __func__, (unsigned long)limit);
+
+ total_pages = cma_early_get_total_pages();
+ size_percent *= (total_pages << PAGE_SHIFT) / 100;
+
+ pr_debug("%s: total available: %ld MiB, size absolute: %ld MiB, size percentage: %ld MiB\n",
+ __func__, (total_pages << PAGE_SHIFT) / SZ_1M,
+ size_bytes / SZ_1M, size_percent / SZ_1M);
+
+#ifdef CONFIG_CMA_SIZE_SEL_MBYTES
+ selected_size = size_bytes;
+#elif defined(CONFIG_CMA_SIZE_SEL_PERCENTAGE)
+ selected_size = size_percent;
+#elif defined(CONFIG_CMA_SIZE_SEL_MIN)
+ selected_size = min(size_bytes, size_percent);
+#elif defined(CONFIG_CMA_SIZE_SEL_MAX)
+ selected_size = max(size_bytes, size_percent);
+#endif
+
+ if (size_cmdline != -1)
+ selected_size = size_cmdline;
+
+ if (!selected_size)
+ return;
+
+ pr_debug("%s: reserving %ld MiB for global area\n", __func__,
+ selected_size / SZ_1M);
+
+ dma_declare_contiguous(NULL, selected_size, 0, limit);
+};
+
+static DEFINE_MUTEX(cma_mutex);
+
+static int cma_activate_area(unsigned long base_pfn, unsigned long count)
+{
+ unsigned long pfn = base_pfn;
+ unsigned i = count >> pageblock_order;
+ struct zone *zone;
+
+ WARN_ON_ONCE(!pfn_valid(pfn));
+ zone = page_zone(pfn_to_page(pfn));
+
+ do {
+ unsigned j;
+ base_pfn = pfn;
+ for (j = pageblock_nr_pages; j; --j, pfn++) {
+ WARN_ON_ONCE(!pfn_valid(pfn));
+ if (page_zone(pfn_to_page(pfn)) != zone)
+ return -EINVAL;
+ }
+ init_cma_reserved_pageblock(pfn_to_page(base_pfn));
+ } while (--i);
+ return 0;
+}
+
+static struct cma *cma_create_area(unsigned long base_pfn,
+ unsigned long count)
+{
+ int bitmap_size = BITS_TO_LONGS(count) * sizeof(long);
+ struct cma *cma;
+ int ret = -ENOMEM;
+
+ pr_debug("%s(base %08lx, count %lx)\n", __func__, base_pfn, count);
+
+ cma = kmalloc(sizeof *cma, GFP_KERNEL);
+ if (!cma)
+ return ERR_PTR(-ENOMEM);
+
+ cma->base_pfn = base_pfn;
+ cma->count = count;
+ cma->bitmap = kzalloc(bitmap_size, GFP_KERNEL);
+
+ if (!cma->bitmap)
+ goto no_mem;
+
+ ret = cma_activate_area(base_pfn, count);
+ if (ret)
+ goto error;
+
+ pr_debug("%s: returned %p\n", __func__, (void *)cma);
+ return cma;
+
+error:
+ kfree(cma->bitmap);
+no_mem:
+ kfree(cma);
+ return ERR_PTR(ret);
+}
+
+static struct cma_reserved {
+ phys_addr_t start;
+ unsigned long size;
+ struct device *dev;
+} cma_reserved[MAX_CMA_AREAS] __initdata;
+static unsigned cma_reserved_count __initdata;
+
+static int __init cma_init_reserved_areas(void)
+{
+ struct cma_reserved *r = cma_reserved;
+ unsigned i = cma_reserved_count;
+
+ pr_debug("%s()\n", __func__);
+
+ for (; i; --i, ++r) {
+ struct cma *cma;
+ cma = cma_create_area(PFN_DOWN(r->start),
+ r->size >> PAGE_SHIFT);
+ if (!IS_ERR(cma))
+ dev_set_cma_area(r->dev, cma);
+ }
+ return 0;
+}
+core_initcall(cma_init_reserved_areas);
+
+/**
+ * dma_declare_contiguous() - reserve area for contiguous memory handling
+ * for particular device
+ * @dev: Pointer to device structure.
+ * @size: Size of the reserved memory.
+ * @start: Start address of the reserved memory (optional, 0 for any).
+ * @limit: End address of the reserved memory (optional, 0 for any).
+ *
+ * This funtion reserves memory for specified device. It should be
+ * called by board specific code when early allocator (memblock or bootmem)
+ * is still activate.
+ */
+int __init dma_declare_contiguous(struct device *dev, unsigned long size,
+ phys_addr_t base, phys_addr_t limit)
+{
+ struct cma_reserved *r = &cma_reserved[cma_reserved_count];
+ unsigned long alignment;
+
+ pr_debug("%s(size %lx, base %08lx, limit %08lx)\n", __func__,
+ (unsigned long)size, (unsigned long)base,
+ (unsigned long)limit);
+
+ /* Sanity checks */
+ if (cma_reserved_count == ARRAY_SIZE(cma_reserved)) {
+ pr_err("Not enough slots for CMA reserved regions!\n");
+ return -ENOSPC;
+ }
+
+ if (!size)
+ return -EINVAL;
+
+ /* Sanitise input arguments */
+ alignment = PAGE_SIZE << max(MAX_ORDER, pageblock_order);
+ base = ALIGN(base, alignment);
+ size = ALIGN(size, alignment);
+ limit &= ~(alignment - 1);
+
+ /* Reserve memory */
+ if (base) {
+ if (memblock_is_region_reserved(base, size) ||
+ memblock_reserve(base, size) < 0) {
+ base = -EBUSY;
+ goto err;
+ }
+ } else {
+ /*
+ * Use __memblock_alloc_base() since
+ * memblock_alloc_base() panic()s.
+ */
+ phys_addr_t addr = __memblock_alloc_base(size, alignment, limit);
+ if (!addr) {
+ base = -ENOMEM;
+ goto err;
+ } else if (addr + size > ~(unsigned long)0) {
+ memblock_free(addr, size);
+ base = -EINVAL;
+ goto err;
+ } else {
+ base = addr;
+ }
+ }
+
+ /*
+ * Each reserved area must be initialised later, when more kernel
+ * subsystems (like slab allocator) are available.
+ */
+ r->start = base;
+ r->size = size;
+ r->dev = dev;
+ cma_reserved_count++;
+ pr_info("CMA: reserved %ld MiB at %08lx\n", size / SZ_1M,
+ (unsigned long)base);
+
+ /*
+ * Architecture specific contiguous memory fixup.
+ */
+ dma_contiguous_early_fixup(base, size);
+ return 0;
+err:
+ pr_err("CMA: failed to reserve %ld MiB\n", size / SZ_1M);
+ return base;
+}
+
+/**
+ * dma_alloc_from_contiguous() - allocate pages from contiguous area
+ * @dev: Pointer to device for which the allocation is performed.
+ * @count: Requested number of pages.
+ * @align: Requested alignment of pages (in PAGE_SIZE order).
+ *
+ * This funtion allocates memory buffer for specified device. It uses
+ * device specific contiguous memory area if available or the default
+ * global one. Requires architecture specific get_dev_cma_area() helper
+ * function.
+ */
+struct page *dma_alloc_from_contiguous(struct device *dev, int count,
+ unsigned int align)
+{
+ unsigned long mask, pfn, pageno, start = 0;
+ struct cma *cma = dev_get_cma_area(dev);
+ int ret;
+
+ if (!cma || !cma->count)
+ return NULL;
+
+ if (align > CONFIG_CMA_ALIGNMENT)
+ align = CONFIG_CMA_ALIGNMENT;
+
+ pr_debug("%s(cma %p, count %d, align %d)\n", __func__, (void *)cma,
+ count, align);
+
+ if (!count)
+ return NULL;
+
+ mask = (1 << align) - 1;
+
+ mutex_lock(&cma_mutex);
+
+ for (;;) {
+ pageno = bitmap_find_next_zero_area(cma->bitmap, cma->count,
+ start, count, mask);
+ if (pageno >= cma->count) {
+ ret = -ENOMEM;
+ goto error;
+ }
+
+ pfn = cma->base_pfn + pageno;
+ ret = alloc_contig_range(pfn, pfn + count, MIGRATE_CMA);
+ if (ret == 0) {
+ bitmap_set(cma->bitmap, pageno, count);
+ break;
+ } else if (ret != -EBUSY) {
+ goto error;
+ }
+ pr_debug("%s(): memory range at %p is busy, retrying\n",
+ __func__, pfn_to_page(pfn));
+ /* try again with a bit different memory target */
+ start = pageno + mask + 1;
+ }
+
+ mutex_unlock(&cma_mutex);
+
+ pr_debug("%s(): returned %p\n", __func__, pfn_to_page(pfn));
+ return pfn_to_page(pfn);
+error:
+ mutex_unlock(&cma_mutex);
+ return NULL;
+}
+
+/**
+ * dma_release_from_contiguous() - release allocated pages
+ * @dev: Pointer to device for which the pages were allocated.
+ * @pages: Allocated pages.
+ * @count: Number of allocated pages.
+ *
+ * This funtion releases memory allocated by dma_alloc_from_contiguous().
+ * It return 0 when provided pages doen't belongs to contiguous area and
+ * 1 on success.
+ */
+int dma_release_from_contiguous(struct device *dev, struct page *pages,
+ int count)
+{
+ struct cma *cma = dev_get_cma_area(dev);
+ unsigned long pfn;
+
+ if (!cma || !pages)
+ return 0;
+
+ pr_debug("%s(page %p)\n", __func__, (void *)pages);
+
+ pfn = page_to_pfn(pages);
+
+ if (pfn < cma->base_pfn || pfn >= cma->base_pfn + cma->count)
+ return 0;
+
+ mutex_lock(&cma_mutex);
+
+ bitmap_clear(cma->bitmap, pfn - cma->base_pfn, count);
+ free_contig_range(pfn, count);
+
+ mutex_unlock(&cma_mutex);
+ return 1;
+}
diff --git a/include/asm-generic/dma-contiguous.h b/include/asm-generic/dma-contiguous.h
new file mode 100644
index 0000000..bf2bccc
--- /dev/null
+++ b/include/asm-generic/dma-contiguous.h
@@ -0,0 +1,27 @@
+#ifndef ASM_DMA_CONTIGUOUS_H
+#define ASM_DMA_CONTIGUOUS_H
+
+#ifdef __KERNEL__
+
+#include <linux/device.h>
+#include <linux/dma-contiguous.h>
+
+#ifdef CONFIG_CMA
+
+static inline struct cma *dev_get_cma_area(struct device *dev)
+{
+ if (dev && dev->cma_area)
+ return dev->cma_area;
+ return dma_contiguous_default_area;
+}
+
+static inline void dev_set_cma_area(struct device *dev, struct cma *cma)
+{
+ if (dev)
+ dev->cma_area = cma;
+ dma_contiguous_default_area = cma;
+}
+
+#endif
+#endif
+#endif
diff --git a/include/linux/device.h b/include/linux/device.h
index b63fb39..9559c18 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -667,6 +667,10 @@ struct device {
struct dma_coherent_mem *dma_mem; /* internal for coherent mem
override */
+#ifdef CONFIG_CMA
+ struct cma *cma_area; /* contiguous memory area for dma
+ allocations */
+#endif
/* arch specific additions */
struct dev_archdata archdata;
diff --git a/include/linux/dma-contiguous.h b/include/linux/dma-contiguous.h
new file mode 100644
index 0000000..ffb4b40
--- /dev/null
+++ b/include/linux/dma-contiguous.h
@@ -0,0 +1,110 @@
+#ifndef __LINUX_CMA_H
+#define __LINUX_CMA_H
+
+/*
+ * Contiguous Memory Allocator for DMA mapping framework
+ * Copyright (c) 2010-2011 by Samsung Electronics.
+ * Written by:
+ * Marek Szyprowski <[email protected]>
+ * Michal Nazarewicz <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of the
+ * License or (at your optional) any later version of the license.
+ */
+
+/*
+ * Contiguous Memory Allocator
+ *
+ * The Contiguous Memory Allocator (CMA) makes it possible to
+ * allocate big contiguous chunks of memory after the system has
+ * booted.
+ *
+ * Why is it needed?
+ *
+ * Various devices on embedded systems have no scatter-getter and/or
+ * IO map support and require contiguous blocks of memory to
+ * operate. They include devices such as cameras, hardware video
+ * coders, etc.
+ *
+ * Such devices often require big memory buffers (a full HD frame
+ * is, for instance, more then 2 mega pixels large, i.e. more than 6
+ * MB of memory), which makes mechanisms such as kmalloc() or
+ * alloc_page() ineffective.
+ *
+ * At the same time, a solution where a big memory region is
+ * reserved for a device is suboptimal since often more memory is
+ * reserved then strictly required and, moreover, the memory is
+ * inaccessible to page system even if device drivers don't use it.
+ *
+ * CMA tries to solve this issue by operating on memory regions
+ * where only movable pages can be allocated from. This way, kernel
+ * can use the memory for pagecache and when device driver requests
+ * it, allocated pages can be migrated.
+ *
+ * Driver usage
+ *
+ * CMA should not be used by the device drivers directly. It is
+ * only a helper framework for dma-mapping subsystem.
+ *
+ * For more information, see kernel-docs in drivers/base/dma-contiguous.c
+ */
+
+#ifdef __KERNEL__
+
+struct cma;
+struct page;
+struct device;
+
+#ifdef CONFIG_CMA
+
+/*
+ * There is always at least global CMA area and a few optional device
+ * private areas configured in kernel .config.
+ */
+#define MAX_CMA_AREAS (1 + CONFIG_CMA_AREAS)
+
+extern struct cma *dma_contiguous_default_area;
+
+void dma_contiguous_reserve(phys_addr_t addr_limit);
+int dma_declare_contiguous(struct device *dev, unsigned long size,
+ phys_addr_t base, phys_addr_t limit);
+
+struct page *dma_alloc_from_contiguous(struct device *dev, int count,
+ unsigned int order);
+int dma_release_from_contiguous(struct device *dev, struct page *pages,
+ int count);
+
+#else
+
+#define MAX_CMA_AREAS (0)
+
+static inline void dma_contiguous_reserve(phys_addr_t limit) { }
+
+static inline
+int dma_declare_contiguous(struct device *dev, unsigned long size,
+ phys_addr_t base, phys_addr_t limit)
+{
+ return -ENOSYS;
+}
+
+static inline
+struct page *dma_alloc_from_contiguous(struct device *dev, int count,
+ unsigned int order)
+{
+ return NULL;
+}
+
+static inline
+int dma_release_from_contiguous(struct device *dev, struct page *pages,
+ int count)
+{
+ return 0;
+}
+
+#endif
+
+#endif
+
+#endif
--
1.7.1.569.g6f426
Replace custom memory bank initialization using memblock_reserve and
dma_declare_coherent with a single call to CMA's dma_declare_contiguous.
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
---
arch/arm/plat-s5p/dev-mfc.c | 51 ++++++-------------------------------------
1 files changed, 7 insertions(+), 44 deletions(-)
diff --git a/arch/arm/plat-s5p/dev-mfc.c b/arch/arm/plat-s5p/dev-mfc.c
index a30d36b..fcb8400 100644
--- a/arch/arm/plat-s5p/dev-mfc.c
+++ b/arch/arm/plat-s5p/dev-mfc.c
@@ -14,6 +14,7 @@
#include <linux/interrupt.h>
#include <linux/platform_device.h>
#include <linux/dma-mapping.h>
+#include <linux/dma-contiguous.h>
#include <linux/memblock.h>
#include <linux/ioport.h>
@@ -22,52 +23,14 @@
#include <plat/irqs.h>
#include <plat/mfc.h>
-struct s5p_mfc_reserved_mem {
- phys_addr_t base;
- unsigned long size;
- struct device *dev;
-};
-
-static struct s5p_mfc_reserved_mem s5p_mfc_mem[2] __initdata;
-
void __init s5p_mfc_reserve_mem(phys_addr_t rbase, unsigned int rsize,
phys_addr_t lbase, unsigned int lsize)
{
- int i;
-
- s5p_mfc_mem[0].dev = &s5p_device_mfc_r.dev;
- s5p_mfc_mem[0].base = rbase;
- s5p_mfc_mem[0].size = rsize;
-
- s5p_mfc_mem[1].dev = &s5p_device_mfc_l.dev;
- s5p_mfc_mem[1].base = lbase;
- s5p_mfc_mem[1].size = lsize;
-
- for (i = 0; i < ARRAY_SIZE(s5p_mfc_mem); i++) {
- struct s5p_mfc_reserved_mem *area = &s5p_mfc_mem[i];
- if (memblock_remove(area->base, area->size)) {
- printk(KERN_ERR "Failed to reserve memory for MFC device (%ld bytes at 0x%08lx)\n",
- area->size, (unsigned long) area->base);
- area->base = 0;
- }
- }
-}
-
-static int __init s5p_mfc_memory_init(void)
-{
- int i;
-
- for (i = 0; i < ARRAY_SIZE(s5p_mfc_mem); i++) {
- struct s5p_mfc_reserved_mem *area = &s5p_mfc_mem[i];
- if (!area->base)
- continue;
+ if (dma_declare_contiguous(&s5p_device_mfc_r.dev, rsize, rbase, 0))
+ printk(KERN_ERR "Failed to reserve memory for MFC device (%u bytes at 0x%08lx)\n",
+ rsize, (unsigned long) rbase);
- if (dma_declare_coherent_memory(area->dev, area->base,
- area->base, area->size,
- DMA_MEMORY_MAP | DMA_MEMORY_EXCLUSIVE) == 0)
- printk(KERN_ERR "Failed to declare coherent memory for MFC device (%ld bytes at 0x%08lx)\n",
- area->size, (unsigned long) area->base);
- }
- return 0;
+ if (dma_declare_contiguous(&s5p_device_mfc_l.dev, lsize, lbase, 0))
+ printk(KERN_ERR "Failed to reserve memory for MFC device (%u bytes at 0x%08lx)\n",
+ rsize, (unsigned long) rbase);
}
-device_initcall(s5p_mfc_memory_init);
--
1.7.1.569.g6f426
This patch adds support for CMA to dma-mapping subsystem for ARM
architecture. By default a global CMA area is used, but specific devices
are allowed to have their private memory areas if required (they can be
created with dma_declare_contiguous() function during board
initialization).
Contiguous memory areas reserved for DMA are remapped with 2-level page
tables on boot. Once a buffer is requested, a low memory kernel mapping
is updated to to match requested memory access type.
GFP_ATOMIC allocations are performed from special pool which is created
early during boot. This way remapping page attributes is not needed on
allocation time.
CMA has been enabled unconditionally for ARMv6+ systems.
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
CC: Michal Nazarewicz <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
Documentation/kernel-parameters.txt | 4 +
arch/arm/Kconfig | 2 +
arch/arm/include/asm/dma-contiguous.h | 16 ++
arch/arm/include/asm/mach/map.h | 1 +
arch/arm/kernel/setup.c | 9 +-
arch/arm/mm/dma-mapping.c | 368 +++++++++++++++++++++++++++------
arch/arm/mm/init.c | 22 ++-
arch/arm/mm/mm.h | 3 +
arch/arm/mm/mmu.c | 31 ++-
9 files changed, 368 insertions(+), 88 deletions(-)
create mode 100644 arch/arm/include/asm/dma-contiguous.h
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 84982e2..ff97085 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -520,6 +520,10 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
a hypervisor.
Default: yes
+ coherent_pool=nn[KMG] [ARM,KNL]
+ Sets the size of memory pool for coherent, atomic dma
+ allocations if Contiguous Memory Allocator (CMA) is used.
+
code_bytes [X86] How many bytes of object code to print
in an oops report.
Range: 0 - 8192
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index a48aecc..56192fe 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -4,6 +4,8 @@ config ARM
select HAVE_AOUT
select HAVE_DMA_API_DEBUG
select HAVE_IDE if PCI || ISA || PCMCIA
+ select HAVE_DMA_CONTIGUOUS if (CPU_V6 || CPU_V6K || CPU_V7)
+ select CMA if (CPU_V6 || CPU_V6K || CPU_V7)
select HAVE_MEMBLOCK
select RTC_LIB
select SYS_SUPPORTS_APM_EMULATION
diff --git a/arch/arm/include/asm/dma-contiguous.h b/arch/arm/include/asm/dma-contiguous.h
new file mode 100644
index 0000000..c7ba05e
--- /dev/null
+++ b/arch/arm/include/asm/dma-contiguous.h
@@ -0,0 +1,16 @@
+#ifndef ASMARM_DMA_CONTIGUOUS_H
+#define ASMARM_DMA_CONTIGUOUS_H
+
+#ifdef __KERNEL__
+
+#include <linux/device.h>
+#include <linux/dma-contiguous.h>
+#include <asm-generic/dma-contiguous.h>
+
+#ifdef CONFIG_CMA
+
+void dma_contiguous_early_fixup(phys_addr_t base, unsigned long size);
+
+#endif
+#endif
+#endif
diff --git a/arch/arm/include/asm/mach/map.h b/arch/arm/include/asm/mach/map.h
index b36f365..a6efcdd 100644
--- a/arch/arm/include/asm/mach/map.h
+++ b/arch/arm/include/asm/mach/map.h
@@ -30,6 +30,7 @@ struct map_desc {
#define MT_MEMORY_DTCM 12
#define MT_MEMORY_ITCM 13
#define MT_MEMORY_SO 14
+#define MT_MEMORY_DMA_READY 15
#ifdef CONFIG_MMU
extern void iotable_init(struct map_desc *, int);
diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c
index a255c39..a7d5fb7 100644
--- a/arch/arm/kernel/setup.c
+++ b/arch/arm/kernel/setup.c
@@ -79,6 +79,7 @@ __setup("fpe=", fpe_setup);
extern void paging_init(struct machine_desc *desc);
extern void sanity_check_meminfo(void);
extern void reboot_setup(char *str);
+extern void setup_dma_zone(struct machine_desc *desc);
unsigned int processor_id;
EXPORT_SYMBOL(processor_id);
@@ -923,12 +924,8 @@ void __init setup_arch(char **cmdline_p)
machine_desc = mdesc;
machine_name = mdesc->name;
-#ifdef CONFIG_ZONE_DMA
- if (mdesc->dma_zone_size) {
- extern unsigned long arm_dma_zone_size;
- arm_dma_zone_size = mdesc->dma_zone_size;
- }
-#endif
+ setup_dma_zone(mdesc);
+
if (mdesc->restart_mode)
reboot_setup(&mdesc->restart_mode);
diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
index 1aa664a..77e7755 100644
--- a/arch/arm/mm/dma-mapping.c
+++ b/arch/arm/mm/dma-mapping.c
@@ -17,7 +17,9 @@
#include <linux/init.h>
#include <linux/device.h>
#include <linux/dma-mapping.h>
+#include <linux/dma-contiguous.h>
#include <linux/highmem.h>
+#include <linux/memblock.h>
#include <linux/slab.h>
#include <asm/memory.h>
@@ -26,6 +28,8 @@
#include <asm/tlbflush.h>
#include <asm/sizes.h>
#include <asm/mach/arch.h>
+#include <asm/mach/map.h>
+#include <asm/dma-contiguous.h>
#include "mm.h"
@@ -56,6 +60,19 @@ static u64 get_coherent_dma_mask(struct device *dev)
return mask;
}
+static void __dma_clear_buffer(struct page *page, size_t size)
+{
+ void *ptr;
+ /*
+ * Ensure that the allocated pages are zeroed, and that any data
+ * lurking in the kernel direct-mapped region is invalidated.
+ */
+ ptr = page_address(page);
+ memset(ptr, 0, size);
+ dmac_flush_range(ptr, ptr + size);
+ outer_flush_range(__pa(ptr), __pa(ptr) + size);
+}
+
/*
* Allocate a DMA buffer for 'dev' of size 'size' using the
* specified gfp mask. Note that 'size' must be page aligned.
@@ -64,23 +81,6 @@ static struct page *__dma_alloc_buffer(struct device *dev, size_t size, gfp_t gf
{
unsigned long order = get_order(size);
struct page *page, *p, *e;
- void *ptr;
- u64 mask = get_coherent_dma_mask(dev);
-
-#ifdef CONFIG_DMA_API_DEBUG
- u64 limit = (mask + 1) & ~mask;
- if (limit && size >= limit) {
- dev_warn(dev, "coherent allocation too big (requested %#x mask %#llx)\n",
- size, mask);
- return NULL;
- }
-#endif
-
- if (!mask)
- return NULL;
-
- if (mask < 0xffffffffULL)
- gfp |= GFP_DMA;
page = alloc_pages(gfp, order);
if (!page)
@@ -93,14 +93,7 @@ static struct page *__dma_alloc_buffer(struct device *dev, size_t size, gfp_t gf
for (p = page + (size >> PAGE_SHIFT), e = page + (1 << order); p < e; p++)
__free_page(p);
- /*
- * Ensure that the allocated pages are zeroed, and that any data
- * lurking in the kernel direct-mapped region is invalidated.
- */
- ptr = page_address(page);
- memset(ptr, 0, size);
- dmac_flush_range(ptr, ptr + size);
- outer_flush_range(__pa(ptr), __pa(ptr) + size);
+ __dma_clear_buffer(page, size);
return page;
}
@@ -170,6 +163,9 @@ static int __init consistent_init(void)
unsigned long base = consistent_base;
unsigned long num_ptes = (CONSISTENT_END - base) >> PMD_SHIFT;
+ if (cpu_architecture() >= CPU_ARCH_ARMv6)
+ return 0;
+
consistent_pte = kmalloc(num_ptes * sizeof(pte_t), GFP_KERNEL);
if (!consistent_pte) {
pr_err("%s: no memory\n", __func__);
@@ -210,9 +206,101 @@ static int __init consistent_init(void)
return ret;
}
-
core_initcall(consistent_init);
+static void *__alloc_from_contiguous(struct device *dev, size_t size,
+ pgprot_t prot, struct page **ret_page);
+
+static struct arm_vmregion_head coherent_head = {
+ .vm_lock = __SPIN_LOCK_UNLOCKED(&coherent_head.vm_lock),
+ .vm_list = LIST_HEAD_INIT(coherent_head.vm_list),
+};
+
+size_t coherent_pool_size = DEFAULT_CONSISTENT_DMA_SIZE / 8;
+
+static int __init early_coherent_pool(char *p)
+{
+ coherent_pool_size = memparse(p, &p);
+ return 0;
+}
+early_param("coherent_pool", early_coherent_pool);
+
+/*
+ * Initialise the coherent pool for atomic allocations.
+ */
+static int __init coherent_init(void)
+{
+ pgprot_t prot = pgprot_dmacoherent(pgprot_kernel);
+ size_t size = coherent_pool_size;
+ struct page *page;
+ void *ptr;
+
+ if (cpu_architecture() < CPU_ARCH_ARMv6)
+ return 0;
+
+ ptr = __alloc_from_contiguous(NULL, size, prot, &page);
+ if (ptr) {
+ coherent_head.vm_start = (unsigned long) ptr;
+ coherent_head.vm_end = (unsigned long) ptr + size;
+ printk(KERN_INFO "DMA: preallocated %u KiB pool for atomic coherent allocations\n",
+ (unsigned)size / 1024);
+ return 0;
+ }
+ printk(KERN_ERR "DMA: failed to allocate %u KiB pool for atomic coherent allocation\n",
+ (unsigned)size / 1024);
+ return -ENOMEM;
+}
+/*
+ * CMA is activated by core_initcall, so we must be called after it
+ */
+postcore_initcall(coherent_init);
+
+struct dma_contig_early_reserve {
+ phys_addr_t base;
+ unsigned long size;
+};
+
+static struct dma_contig_early_reserve dma_mmu_remap[MAX_CMA_AREAS] __initdata;
+
+static int dma_mmu_remap_num __initdata;
+
+void __init dma_contiguous_early_fixup(phys_addr_t base, unsigned long size)
+{
+ dma_mmu_remap[dma_mmu_remap_num].base = base;
+ dma_mmu_remap[dma_mmu_remap_num].size = size;
+ dma_mmu_remap_num++;
+}
+
+void __init dma_contiguous_remap(void)
+{
+ int i;
+ for (i = 0; i < dma_mmu_remap_num; i++) {
+ phys_addr_t start = dma_mmu_remap[i].base;
+ phys_addr_t end = start + dma_mmu_remap[i].size;
+ struct map_desc map;
+ unsigned long addr;
+
+ if (end > arm_lowmem_limit)
+ end = arm_lowmem_limit;
+ if (start >= end)
+ return;
+
+ map.pfn = __phys_to_pfn(start);
+ map.virtual = __phys_to_virt(start);
+ map.length = end - start;
+ map.type = MT_MEMORY_DMA_READY;
+
+ /*
+ * Clear previous low-memory mapping
+ */
+ for (addr = __phys_to_virt(start); addr < __phys_to_virt(end);
+ addr += PGDIR_SIZE)
+ pmd_clear(pmd_off_k(addr));
+
+ iotable_init(&map, 1);
+ }
+}
+
static void *
__dma_alloc_remap(struct page *page, size_t size, gfp_t gfp, pgprot_t prot)
{
@@ -318,20 +406,172 @@ static void __dma_free_remap(void *cpu_addr, size_t size)
arm_vmregion_free(&consistent_head, c);
}
+static int __dma_update_pte(pte_t *pte, pgtable_t token, unsigned long addr,
+ void *data)
+{
+ struct page *page = virt_to_page(addr);
+ pgprot_t prot = *(pgprot_t *)data;
+
+ set_pte_ext(pte, mk_pte(page, prot), 0);
+ return 0;
+}
+
+static void __dma_remap(struct page *page, size_t size, pgprot_t prot)
+{
+ unsigned long start = (unsigned long) page_address(page);
+ unsigned end = start + size;
+
+ apply_to_page_range(&init_mm, start, size, __dma_update_pte, &prot);
+ dsb();
+ flush_tlb_kernel_range(start, end);
+}
+
+static void *__alloc_remap_buffer(struct device *dev, size_t size, gfp_t gfp,
+ pgprot_t prot, struct page **ret_page)
+{
+ struct page *page;
+ void *ptr;
+ page = __dma_alloc_buffer(dev, size, gfp);
+ if (!page)
+ return NULL;
+
+ ptr = __dma_alloc_remap(page, size, gfp, prot);
+ if (!ptr) {
+ __dma_free_buffer(page, size);
+ return NULL;
+ }
+
+ *ret_page = page;
+ return ptr;
+}
+
+static void *__alloc_from_pool(struct device *dev, size_t size,
+ struct page **ret_page)
+{
+ struct arm_vmregion *c;
+ size_t align;
+
+ if (!coherent_head.vm_start) {
+ printk(KERN_ERR "%s: coherent pool not initialised!\n",
+ __func__);
+ dump_stack();
+ return NULL;
+ }
+
+ /*
+ * Align the region allocation - allocations from pool are rather
+ * small, so align them to their order in pages, minimum is a page
+ * size. This helps reduce fragmentation of the DMA space.
+ */
+ align = PAGE_SIZE << get_order(size);
+ c = arm_vmregion_alloc(&coherent_head, align, size, 0);
+ if (c) {
+ void *ptr = (void *)c->vm_start;
+ struct page *page = virt_to_page(ptr);
+ *ret_page = page;
+ return ptr;
+ }
+ return NULL;
+}
+
+static int __free_from_pool(void *cpu_addr, size_t size)
+{
+ unsigned long start = (unsigned long)cpu_addr;
+ unsigned long end = start + size;
+ struct arm_vmregion *c;
+
+ if (start < coherent_head.vm_start || end > coherent_head.vm_end)
+ return 0;
+
+ c = arm_vmregion_find_remove(&coherent_head, (unsigned long)start);
+
+ if ((c->vm_end - c->vm_start) != size) {
+ printk(KERN_ERR "%s: freeing wrong coherent size (%ld != %d)\n",
+ __func__, c->vm_end - c->vm_start, size);
+ dump_stack();
+ size = c->vm_end - c->vm_start;
+ }
+
+ arm_vmregion_free(&coherent_head, c);
+ return 1;
+}
+
+static void *__alloc_from_contiguous(struct device *dev, size_t size,
+ pgprot_t prot, struct page **ret_page)
+{
+ unsigned long order = get_order(size);
+ size_t count = size >> PAGE_SHIFT;
+ struct page *page;
+
+ page = dma_alloc_from_contiguous(dev, count, order);
+ if (!page)
+ return NULL;
+
+ __dma_clear_buffer(page, size);
+ __dma_remap(page, size, prot);
+
+ *ret_page = page;
+ return page_address(page);
+}
+
+static void __free_from_contiguous(struct device *dev, struct page *page,
+ size_t size)
+{
+ __dma_remap(page, size, pgprot_kernel);
+ dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);
+}
+
+#define nommu() 0
+
#else /* !CONFIG_MMU */
-#define __dma_alloc_remap(page, size, gfp, prot) page_address(page)
-#define __dma_free_remap(addr, size) do { } while (0)
+#define nommu() 1
+
+#define __alloc_remap_buffer(dev, size, gfp, prot, ret) NULL
+#define __alloc_from_pool(dev, size, ret_page) NULL
+#define __alloc_from_contiguous(dev, size, prot, ret) NULL
+#define __free_from_pool(cpu_addr, size) 0
+#define __free_from_contiguous(dev, page, size) do { } while (0)
+#define __dma_free_remap(cpu_addr, size) do { } while (0)
#endif /* CONFIG_MMU */
-static void *
-__dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp,
- pgprot_t prot)
+static void *__alloc_simple_buffer(struct device *dev, size_t size, gfp_t gfp,
+ struct page **ret_page)
{
struct page *page;
+ page = __dma_alloc_buffer(dev, size, gfp);
+ if (!page)
+ return NULL;
+
+ *ret_page = page;
+ return page_address(page);
+}
+
+
+
+static void *__dma_alloc(struct device *dev, size_t size, dma_addr_t *handle,
+ gfp_t gfp, pgprot_t prot)
+{
+ u64 mask = get_coherent_dma_mask(dev);
+ struct page *page;
void *addr;
+#ifdef CONFIG_DMA_API_DEBUG
+ u64 limit = (mask + 1) & ~mask;
+ if (limit && size >= limit) {
+ dev_warn(dev, "coherent allocation too big (requested %#x mask %#llx)\n",
+ size, mask);
+ return NULL;
+ }
+#endif
+
+ if (!mask)
+ return NULL;
+
+ if (mask < 0xffffffffULL)
+ gfp |= GFP_DMA;
+
/*
* Following is a work-around (a.k.a. hack) to prevent pages
* with __GFP_COMP being passed to split_page() which cannot
@@ -344,19 +584,17 @@ __dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp,
*handle = ~0;
size = PAGE_ALIGN(size);
- page = __dma_alloc_buffer(dev, size, gfp);
- if (!page)
- return NULL;
-
- if (!arch_is_coherent())
- addr = __dma_alloc_remap(page, size, gfp, prot);
+ if (arch_is_coherent() || nommu())
+ addr = __alloc_simple_buffer(dev, size, gfp, &page);
+ else if (cpu_architecture() < CPU_ARCH_ARMv6)
+ addr = __alloc_remap_buffer(dev, size, gfp, prot, &page);
+ else if (gfp & GFP_ATOMIC)
+ addr = __alloc_from_pool(dev, size, &page);
else
- addr = page_address(page);
+ addr = __alloc_from_contiguous(dev, size, prot, &page);
if (addr)
*handle = pfn_to_dma(dev, page_to_pfn(page));
- else
- __dma_free_buffer(page, size);
return addr;
}
@@ -365,8 +603,8 @@ __dma_alloc(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp,
* Allocate DMA-coherent memory space and return both the kernel remapped
* virtual and bus address for that space.
*/
-void *
-dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *handle, gfp_t gfp)
+void *dma_alloc_coherent(struct device *dev, size_t size, dma_addr_t *handle,
+ gfp_t gfp)
{
void *memory;
@@ -395,25 +633,11 @@ static int dma_mmap(struct device *dev, struct vm_area_struct *vma,
{
int ret = -ENXIO;
#ifdef CONFIG_MMU
- unsigned long user_size, kern_size;
- struct arm_vmregion *c;
-
- user_size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-
- c = arm_vmregion_find(&consistent_head, (unsigned long)cpu_addr);
- if (c) {
- unsigned long off = vma->vm_pgoff;
-
- kern_size = (c->vm_end - c->vm_start) >> PAGE_SHIFT;
-
- if (off < kern_size &&
- user_size <= (kern_size - off)) {
- ret = remap_pfn_range(vma, vma->vm_start,
- page_to_pfn(c->vm_pages) + off,
- user_size << PAGE_SHIFT,
- vma->vm_page_prot);
- }
- }
+ unsigned long pfn = dma_to_pfn(dev, dma_addr);
+ ret = remap_pfn_range(vma, vma->vm_start,
+ pfn + vma->vm_pgoff,
+ vma->vm_end - vma->vm_start,
+ vma->vm_page_prot);
#endif /* CONFIG_MMU */
return ret;
@@ -435,23 +659,33 @@ int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma,
}
EXPORT_SYMBOL(dma_mmap_writecombine);
+
/*
- * free a page as defined by the above mapping.
- * Must not be called with IRQs disabled.
+ * Free a buffer as defined by the above mapping.
*/
void dma_free_coherent(struct device *dev, size_t size, void *cpu_addr, dma_addr_t handle)
{
- WARN_ON(irqs_disabled());
+ struct page *page = pfn_to_page(dma_to_pfn(dev, handle));
if (dma_release_from_coherent(dev, get_order(size), cpu_addr))
return;
size = PAGE_ALIGN(size);
- if (!arch_is_coherent())
+ if (arch_is_coherent() || nommu()) {
+ __dma_free_buffer(page, size);
+ } else if (cpu_architecture() < CPU_ARCH_ARMv6) {
__dma_free_remap(cpu_addr, size);
-
- __dma_free_buffer(pfn_to_page(dma_to_pfn(dev, handle)), size);
+ __dma_free_buffer(page, size);
+ } else {
+ if (__free_from_pool(cpu_addr, size))
+ return;
+ /*
+ * Non-atomic allocations cannot be freed with IRQs disabled
+ */
+ WARN_ON(irqs_disabled());
+ __free_from_contiguous(dev, page, size);
+ }
}
EXPORT_SYMBOL(dma_free_coherent);
diff --git a/arch/arm/mm/init.c b/arch/arm/mm/init.c
index 5dc7d12..34fa0f3 100644
--- a/arch/arm/mm/init.c
+++ b/arch/arm/mm/init.c
@@ -20,6 +20,7 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/memblock.h>
+#include <linux/dma-contiguous.h>
#include <asm/mach-types.h>
#include <asm/memblock.h>
@@ -227,6 +228,17 @@ static void __init arm_adjust_dma_zone(unsigned long *size, unsigned long *hole,
}
#endif
+void __init setup_dma_zone(struct machine_desc *mdesc)
+{
+#ifdef CONFIG_ZONE_DMA
+ if (mdesc->dma_zone_size) {
+ arm_dma_zone_size = mdesc->dma_zone_size;
+ arm_dma_limit = PHYS_OFFSET + arm_dma_zone_size - 1;
+ } else
+ arm_dma_limit = 0xffffffff;
+#endif
+}
+
static void __init arm_bootmem_free(unsigned long min, unsigned long max_low,
unsigned long max_high)
{
@@ -274,12 +286,9 @@ static void __init arm_bootmem_free(unsigned long min, unsigned long max_low,
* Adjust the sizes according to any special requirements for
* this machine type.
*/
- if (arm_dma_zone_size) {
+ if (arm_dma_zone_size)
arm_adjust_dma_zone(zone_size, zhole_size,
arm_dma_zone_size >> PAGE_SHIFT);
- arm_dma_limit = PHYS_OFFSET + arm_dma_zone_size - 1;
- } else
- arm_dma_limit = 0xffffffff;
#endif
free_area_init_node(0, zone_size, min, zhole_size);
@@ -365,6 +374,11 @@ void __init arm_memblock_init(struct meminfo *mi, struct machine_desc *mdesc)
if (mdesc->reserve)
mdesc->reserve();
+ /* reserve memory for DMA contigouos allocations,
+ must come from DMA area inside low memory */
+ dma_contiguous_reserve(arm_dma_limit < arm_lowmem_limit ?
+ arm_dma_limit : arm_lowmem_limit);
+
arm_memblock_steal_permitted = false;
memblock_allow_resize();
memblock_dump_all();
diff --git a/arch/arm/mm/mm.h b/arch/arm/mm/mm.h
index 70f6d3ea..398c438 100644
--- a/arch/arm/mm/mm.h
+++ b/arch/arm/mm/mm.h
@@ -43,5 +43,8 @@ extern u32 arm_dma_limit;
#define arm_dma_limit ((u32)~0)
#endif
+extern phys_addr_t arm_lowmem_limit;
+
void __init bootmem_init(void);
void arm_mm_memblock_reserve(void);
+void dma_contiguous_remap(void);
diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c
index 94c5a0c..b9fbec2 100644
--- a/arch/arm/mm/mmu.c
+++ b/arch/arm/mm/mmu.c
@@ -286,6 +286,11 @@ static struct mem_type mem_types[] = {
PMD_SECT_UNCACHED | PMD_SECT_XN,
.domain = DOMAIN_KERNEL,
},
+ [MT_MEMORY_DMA_READY] = {
+ .prot_pte = L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY,
+ .prot_l1 = PMD_TYPE_TABLE,
+ .domain = DOMAIN_KERNEL,
+ },
};
const struct mem_type *get_mem_type(unsigned int type)
@@ -427,6 +432,7 @@ static void __init build_mem_type_table(void)
if (arch_is_coherent() && cpu_is_xsc3()) {
mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
mem_types[MT_MEMORY].prot_pte |= L_PTE_SHARED;
+ mem_types[MT_MEMORY_DMA_READY].prot_pte |= L_PTE_SHARED;
mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_S;
mem_types[MT_MEMORY_NONCACHED].prot_pte |= L_PTE_SHARED;
}
@@ -458,6 +464,7 @@ static void __init build_mem_type_table(void)
mem_types[MT_DEVICE_CACHED].prot_pte |= L_PTE_SHARED;
mem_types[MT_MEMORY].prot_sect |= PMD_SECT_S;
mem_types[MT_MEMORY].prot_pte |= L_PTE_SHARED;
+ mem_types[MT_MEMORY_DMA_READY].prot_pte |= L_PTE_SHARED;
mem_types[MT_MEMORY_NONCACHED].prot_sect |= PMD_SECT_S;
mem_types[MT_MEMORY_NONCACHED].prot_pte |= L_PTE_SHARED;
}
@@ -509,6 +516,7 @@ static void __init build_mem_type_table(void)
mem_types[MT_HIGH_VECTORS].prot_l1 |= ecc_mask;
mem_types[MT_MEMORY].prot_sect |= ecc_mask | cp->pmd;
mem_types[MT_MEMORY].prot_pte |= kern_pgprot;
+ mem_types[MT_MEMORY_DMA_READY].prot_pte |= kern_pgprot;
mem_types[MT_MEMORY_NONCACHED].prot_sect |= ecc_mask;
mem_types[MT_ROM].prot_sect |= cp->pmd;
@@ -593,7 +601,7 @@ static void __init alloc_init_section(pud_t *pud, unsigned long addr,
* L1 entries, whereas PGDs refer to a group of L1 entries making
* up one logical pointer to an L2 table.
*/
- if (((addr | end | phys) & ~SECTION_MASK) == 0) {
+ if (type->prot_sect && ((addr | end | phys) & ~SECTION_MASK) == 0) {
pmd_t *p = pmd;
#ifndef CONFIG_ARM_LPAE
@@ -811,7 +819,7 @@ static int __init early_vmalloc(char *arg)
}
early_param("vmalloc", early_vmalloc);
-static phys_addr_t lowmem_limit __initdata = 0;
+phys_addr_t arm_lowmem_limit __initdata = 0;
void __init sanity_check_meminfo(void)
{
@@ -894,8 +902,8 @@ void __init sanity_check_meminfo(void)
bank->size = newsize;
}
#endif
- if (!bank->highmem && bank->start + bank->size > lowmem_limit)
- lowmem_limit = bank->start + bank->size;
+ if (!bank->highmem && bank->start + bank->size > arm_lowmem_limit)
+ arm_lowmem_limit = bank->start + bank->size;
j++;
}
@@ -920,8 +928,8 @@ void __init sanity_check_meminfo(void)
}
#endif
meminfo.nr_banks = j;
- high_memory = __va(lowmem_limit - 1) + 1;
- memblock_set_current_limit(lowmem_limit);
+ high_memory = __va(arm_lowmem_limit - 1) + 1;
+ memblock_set_current_limit(arm_lowmem_limit);
}
static inline void prepare_page_table(void)
@@ -946,8 +954,8 @@ static inline void prepare_page_table(void)
* Find the end of the first block of lowmem.
*/
end = memblock.memory.regions[0].base + memblock.memory.regions[0].size;
- if (end >= lowmem_limit)
- end = lowmem_limit;
+ if (end >= arm_lowmem_limit)
+ end = arm_lowmem_limit;
/*
* Clear out all the kernel space mappings, except for the first
@@ -1087,8 +1095,8 @@ static void __init map_lowmem(void)
phys_addr_t end = start + reg->size;
struct map_desc map;
- if (end > lowmem_limit)
- end = lowmem_limit;
+ if (end > arm_lowmem_limit)
+ end = arm_lowmem_limit;
if (start >= end)
break;
@@ -1109,11 +1117,12 @@ void __init paging_init(struct machine_desc *mdesc)
{
void *zero_page;
- memblock_set_current_limit(lowmem_limit);
+ memblock_set_current_limit(arm_lowmem_limit);
build_mem_type_table();
prepare_page_table();
map_lowmem();
+ dma_contiguous_remap();
devicemaps_init(mdesc);
kmap_init();
--
1.7.1.569.g6f426
This patch adds support for CMA to dma-mapping subsystem for x86
architecture that uses common pci-dma/pci-nommu implementation. This
allows to test CMA on KVM/QEMU and a lot of common x86 boxes.
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
CC: Michal Nazarewicz <[email protected]>
Acked-by: Arnd Bergmann <[email protected]>
---
arch/x86/Kconfig | 1 +
arch/x86/include/asm/dma-contiguous.h | 13 +++++++++++++
arch/x86/include/asm/dma-mapping.h | 4 ++++
arch/x86/kernel/pci-dma.c | 18 ++++++++++++++++--
arch/x86/kernel/pci-nommu.c | 8 +-------
arch/x86/kernel/setup.c | 2 ++
6 files changed, 37 insertions(+), 9 deletions(-)
create mode 100644 arch/x86/include/asm/dma-contiguous.h
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 5bed94e..de6e069 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -31,6 +31,7 @@ config X86
select ARCH_WANT_OPTIONAL_GPIOLIB
select ARCH_WANT_FRAME_POINTERS
select HAVE_DMA_ATTRS
+ select HAVE_DMA_CONTIGUOUS if !SWIOTLB
select HAVE_KRETPROBES
select HAVE_OPTPROBES
select HAVE_FTRACE_MCOUNT_RECORD
diff --git a/arch/x86/include/asm/dma-contiguous.h b/arch/x86/include/asm/dma-contiguous.h
new file mode 100644
index 0000000..8fb117d
--- /dev/null
+++ b/arch/x86/include/asm/dma-contiguous.h
@@ -0,0 +1,13 @@
+#ifndef ASMX86_DMA_CONTIGUOUS_H
+#define ASMX86_DMA_CONTIGUOUS_H
+
+#ifdef __KERNEL__
+
+#include <linux/device.h>
+#include <linux/dma-contiguous.h>
+#include <asm-generic/dma-contiguous.h>
+
+static inline void dma_contiguous_early_fixup(phys_addr_t base, unsigned long size) { }
+
+#endif
+#endif
diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h
index ed3065f..90ac6f0 100644
--- a/arch/x86/include/asm/dma-mapping.h
+++ b/arch/x86/include/asm/dma-mapping.h
@@ -13,6 +13,7 @@
#include <asm/io.h>
#include <asm/swiotlb.h>
#include <asm-generic/dma-coherent.h>
+#include <linux/dma-contiguous.h>
#ifdef CONFIG_ISA
# define ISA_DMA_BIT_MASK DMA_BIT_MASK(24)
@@ -61,6 +62,9 @@ extern int dma_set_mask(struct device *dev, u64 mask);
extern void *dma_generic_alloc_coherent(struct device *dev, size_t size,
dma_addr_t *dma_addr, gfp_t flag);
+extern void dma_generic_free_coherent(struct device *dev, size_t size,
+ void *vaddr, dma_addr_t dma_addr);
+
static inline bool dma_capable(struct device *dev, dma_addr_t addr, size_t size)
{
if (!dev->dma_mask)
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index 1c4d769..d3c3723 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -99,14 +99,18 @@ void *dma_generic_alloc_coherent(struct device *dev, size_t size,
dma_addr_t *dma_addr, gfp_t flag)
{
unsigned long dma_mask;
- struct page *page;
+ struct page *page = NULL;
+ unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
dma_addr_t addr;
dma_mask = dma_alloc_coherent_mask(dev, flag);
flag |= __GFP_ZERO;
again:
- page = alloc_pages_node(dev_to_node(dev), flag, get_order(size));
+ if (!(flag & GFP_ATOMIC))
+ page = dma_alloc_from_contiguous(dev, count, get_order(size));
+ if (!page)
+ page = alloc_pages_node(dev_to_node(dev), flag, get_order(size));
if (!page)
return NULL;
@@ -126,6 +130,16 @@ again:
return page_address(page);
}
+void dma_generic_free_coherent(struct device *dev, size_t size, void *vaddr,
+ dma_addr_t dma_addr)
+{
+ unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
+ struct page *page = virt_to_page(vaddr);
+
+ if (!dma_release_from_contiguous(dev, page, count))
+ free_pages((unsigned long)vaddr, get_order(size));
+}
+
/*
* See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
* parameter documentation.
diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c
index 3af4af8..656566f 100644
--- a/arch/x86/kernel/pci-nommu.c
+++ b/arch/x86/kernel/pci-nommu.c
@@ -74,12 +74,6 @@ static int nommu_map_sg(struct device *hwdev, struct scatterlist *sg,
return nents;
}
-static void nommu_free_coherent(struct device *dev, size_t size, void *vaddr,
- dma_addr_t dma_addr)
-{
- free_pages((unsigned long)vaddr, get_order(size));
-}
-
static void nommu_sync_single_for_device(struct device *dev,
dma_addr_t addr, size_t size,
enum dma_data_direction dir)
@@ -97,7 +91,7 @@ static void nommu_sync_sg_for_device(struct device *dev,
struct dma_map_ops nommu_dma_ops = {
.alloc_coherent = dma_generic_alloc_coherent,
- .free_coherent = nommu_free_coherent,
+ .free_coherent = dma_generic_free_coherent,
.map_sg = nommu_map_sg,
.map_page = nommu_map_page,
.sync_single_for_device = nommu_sync_single_for_device,
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index d7d5099..be6795f 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -50,6 +50,7 @@
#include <asm/pci-direct.h>
#include <linux/init_ohci1394_dma.h>
#include <linux/kvm_para.h>
+#include <linux/dma-contiguous.h>
#include <linux/errno.h>
#include <linux/kernel.h>
@@ -938,6 +939,7 @@ void __init setup_arch(char **cmdline_p)
}
#endif
memblock.current_limit = get_max_mapped();
+ dma_contiguous_reserve(0);
/*
* NOTE: On x86-32, only from this point on, fixmaps are ready for use.
--
1.7.1.569.g6f426
This patch extracts common reclaim code from __alloc_pages_direct_reclaim()
function to separate function: __perform_reclaim() which can be later used
by alloc_contig_range().
Signed-off-by: Marek Szyprowski <[email protected]>
Signed-off-by: Kyungmin Park <[email protected]>
Signed-off-by: Michal Nazarewicz <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/page_alloc.c | 30 +++++++++++++++++++++---------
1 files changed, 21 insertions(+), 9 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 947879c..983ccba 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2094,16 +2094,13 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
}
#endif /* CONFIG_COMPACTION */
-/* The really slow allocator path where we enter direct reclaim */
-static inline struct page *
-__alloc_pages_direct_reclaim(gfp_t gfp_mask, unsigned int order,
- struct zonelist *zonelist, enum zone_type high_zoneidx,
- nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
- int migratetype, unsigned long *did_some_progress)
+/* Perform direct synchronous page reclaim */
+static int
+__perform_reclaim(gfp_t gfp_mask, unsigned int order, struct zonelist *zonelist,
+ nodemask_t *nodemask)
{
- struct page *page = NULL;
struct reclaim_state reclaim_state;
- bool drained = false;
+ int progress;
cond_resched();
@@ -2114,7 +2111,7 @@ __alloc_pages_direct_reclaim(gfp_t gfp_mask, unsigned int order,
reclaim_state.reclaimed_slab = 0;
current->reclaim_state = &reclaim_state;
- *did_some_progress = try_to_free_pages(zonelist, order, gfp_mask, nodemask);
+ progress = try_to_free_pages(zonelist, order, gfp_mask, nodemask);
current->reclaim_state = NULL;
lockdep_clear_current_reclaim_state();
@@ -2122,6 +2119,21 @@ __alloc_pages_direct_reclaim(gfp_t gfp_mask, unsigned int order,
cond_resched();
+ return progress;
+}
+
+/* The really slow allocator path where we enter direct reclaim */
+static inline struct page *
+__alloc_pages_direct_reclaim(gfp_t gfp_mask, unsigned int order,
+ struct zonelist *zonelist, enum zone_type high_zoneidx,
+ nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
+ int migratetype, unsigned long *did_some_progress)
+{
+ struct page *page = NULL;
+ bool drained = false;
+
+ *did_some_progress = __perform_reclaim(gfp_mask, order, zonelist,
+ nodemask);
if (unlikely(!(*did_some_progress)))
return NULL;
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit exports some of the functions from compaction.c file
outside of it adding their declaration into internal.h header
file so that other mm related code can use them.
This forced compaction.c to always be compiled (as opposed to being
compiled only if CONFIG_COMPACTION is defined) but as to avoid
introducing code that user did not ask for, part of the compaction.c
is now wrapped in on #ifdef.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/Makefile | 3 +-
mm/compaction.c | 328 ++++++++++++++++++++++++++-----------------------------
mm/internal.h | 33 ++++++
3 files changed, 191 insertions(+), 173 deletions(-)
diff --git a/mm/Makefile b/mm/Makefile
index 50ec00e..8aada89 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -13,7 +13,7 @@ obj-y := filemap.o mempool.o oom_kill.o fadvise.o \
readahead.o swap.o truncate.o vmscan.o shmem.o \
prio_tree.o util.o mmzone.o vmstat.o backing-dev.o \
page_isolation.o mm_init.o mmu_context.o percpu.o \
- $(mmu-y)
+ compaction.o $(mmu-y)
obj-y += init-mm.o
ifdef CONFIG_NO_BOOTMEM
@@ -32,7 +32,6 @@ obj-$(CONFIG_NUMA) += mempolicy.o
obj-$(CONFIG_SPARSEMEM) += sparse.o
obj-$(CONFIG_SPARSEMEM_VMEMMAP) += sparse-vmemmap.o
obj-$(CONFIG_SLOB) += slob.o
-obj-$(CONFIG_COMPACTION) += compaction.o
obj-$(CONFIG_MMU_NOTIFIER) += mmu_notifier.o
obj-$(CONFIG_KSM) += ksm.o
obj-$(CONFIG_PAGE_POISONING) += debug-pagealloc.o
diff --git a/mm/compaction.c b/mm/compaction.c
index 9fef891..d5174c4 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -16,30 +16,11 @@
#include <linux/sysfs.h>
#include "internal.h"
+#if defined CONFIG_COMPACTION || defined CONFIG_CMA
+
#define CREATE_TRACE_POINTS
#include <trace/events/compaction.h>
-/*
- * compact_control is used to track pages being migrated and the free pages
- * they are being migrated to during memory compaction. The free_pfn starts
- * at the end of a zone and migrate_pfn begins at the start. Movable pages
- * are moved to the end of a zone during a compaction run and the run
- * completes when free_pfn <= migrate_pfn
- */
-struct compact_control {
- struct list_head freepages; /* List of free pages to migrate to */
- struct list_head migratepages; /* List of pages being migrated */
- unsigned long nr_freepages; /* Number of isolated free pages */
- unsigned long nr_migratepages; /* Number of pages to migrate */
- unsigned long free_pfn; /* isolate_freepages search base */
- unsigned long migrate_pfn; /* isolate_migratepages search base */
- bool sync; /* Synchronous migration */
-
- unsigned int order; /* order a direct compactor needs */
- int migratetype; /* MOVABLE, RECLAIMABLE etc */
- struct zone *zone;
-};
-
static unsigned long release_freepages(struct list_head *freelist)
{
struct page *page, *next;
@@ -54,6 +35,16 @@ static unsigned long release_freepages(struct list_head *freelist)
return count;
}
+static void map_pages(struct list_head *list)
+{
+ struct page *page;
+
+ list_for_each_entry(page, list, lru) {
+ arch_alloc_page(page, 0);
+ kernel_map_pages(page, 1, 1);
+ }
+}
+
/*
* Isolate free pages onto a private freelist. Caller must hold zone->lock.
* If @strict is true, will abort returning 0 on any invalid PFNs or non-free
@@ -122,7 +113,7 @@ static unsigned long isolate_freepages_block(unsigned long blockpfn,
* (which may be greater then end_pfn if end fell in a middle of
* a free page).
*/
-static unsigned long
+unsigned long
isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn)
{
unsigned long isolated, pfn, block_end_pfn, flags;
@@ -176,127 +167,6 @@ isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn)
return pfn;
}
-/* Returns true if the page is within a block suitable for migration to */
-static bool suitable_migration_target(struct page *page)
-{
-
- int migratetype = get_pageblock_migratetype(page);
-
- /* Don't interfere with memory hot-remove or the min_free_kbytes blocks */
- if (migratetype == MIGRATE_ISOLATE || migratetype == MIGRATE_RESERVE)
- return false;
-
- /* If the page is a large free page, then allow migration */
- if (PageBuddy(page) && page_order(page) >= pageblock_order)
- return true;
-
- /* If the block is MIGRATE_MOVABLE, allow migration */
- if (migratetype == MIGRATE_MOVABLE)
- return true;
-
- /* Otherwise skip the block */
- return false;
-}
-
-static void map_pages(struct list_head *list)
-{
- struct page *page;
-
- list_for_each_entry(page, list, lru) {
- arch_alloc_page(page, 0);
- kernel_map_pages(page, 1, 1);
- }
-}
-
-/*
- * Based on information in the current compact_control, find blocks
- * suitable for isolating free pages from and then isolate them.
- */
-static void isolate_freepages(struct zone *zone,
- struct compact_control *cc)
-{
- struct page *page;
- unsigned long high_pfn, low_pfn, pfn, zone_end_pfn, end_pfn;
- unsigned long flags;
- int nr_freepages = cc->nr_freepages;
- struct list_head *freelist = &cc->freepages;
-
- /*
- * Initialise the free scanner. The starting point is where we last
- * scanned from (or the end of the zone if starting). The low point
- * is the end of the pageblock the migration scanner is using.
- */
- pfn = cc->free_pfn;
- low_pfn = cc->migrate_pfn + pageblock_nr_pages;
-
- /*
- * Take care that if the migration scanner is at the end of the zone
- * that the free scanner does not accidentally move to the next zone
- * in the next isolation cycle.
- */
- high_pfn = min(low_pfn, pfn);
-
- zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
-
- /*
- * Isolate free pages until enough are available to migrate the
- * pages on cc->migratepages. We stop searching if the migrate
- * and free page scanners meet or enough free pages are isolated.
- */
- for (; pfn > low_pfn && cc->nr_migratepages > nr_freepages;
- pfn -= pageblock_nr_pages) {
- unsigned long isolated;
-
- if (!pfn_valid(pfn))
- continue;
-
- /*
- * Check for overlapping nodes/zones. It's possible on some
- * configurations to have a setup like
- * node0 node1 node0
- * i.e. it's possible that all pages within a zones range of
- * pages do not belong to a single zone.
- */
- page = pfn_to_page(pfn);
- if (page_zone(page) != zone)
- continue;
-
- /* Check the block is suitable for migration */
- if (!suitable_migration_target(page))
- continue;
-
- /*
- * Found a block suitable for isolating free pages from. Now
- * we disabled interrupts, double check things are ok and
- * isolate the pages. This is to minimise the time IRQs
- * are disabled
- */
- isolated = 0;
- spin_lock_irqsave(&zone->lock, flags);
- if (suitable_migration_target(page)) {
- end_pfn = min(pfn + pageblock_nr_pages, zone_end_pfn);
- isolated = isolate_freepages_block(pfn, end_pfn,
- freelist, false);
- nr_freepages += isolated;
- }
- spin_unlock_irqrestore(&zone->lock, flags);
-
- /*
- * Record the highest PFN we isolated pages from. When next
- * looking for free pages, the search will restart here as
- * page migration may have returned some pages to the allocator
- */
- if (isolated)
- high_pfn = max(high_pfn, pfn);
- }
-
- /* split_free_page does not map the pages */
- map_pages(freelist);
-
- cc->free_pfn = high_pfn;
- cc->nr_freepages = nr_freepages;
-}
-
/* Update the number of anon and file isolated pages in the zone */
static void acct_isolated(struct zone *zone, struct compact_control *cc)
{
@@ -325,13 +195,6 @@ static bool too_many_isolated(struct zone *zone)
return isolated > (inactive + active) / 2;
}
-/* possible outcome of isolate_migratepages */
-typedef enum {
- ISOLATE_ABORT, /* Abort compaction now */
- ISOLATE_NONE, /* No pages isolated, continue scanning */
- ISOLATE_SUCCESS, /* Pages isolated, migrate */
-} isolate_migrate_t;
-
/**
* isolate_migratepages_range() - isolate all migrate-able pages in range.
* @zone: Zone pages are in.
@@ -351,7 +214,7 @@ typedef enum {
* does not modify any cc's fields, in particular it does not modify
* (or read for that matter) cc->migrate_pfn.
*/
-static unsigned long
+unsigned long
isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
unsigned long low_pfn, unsigned long end_pfn)
{
@@ -465,35 +328,118 @@ isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
return low_pfn;
}
+#endif /* CONFIG_COMPACTION || CONFIG_CMA */
+#ifdef CONFIG_COMPACTION
+
+/* Returns true if the page is within a block suitable for migration to */
+static bool suitable_migration_target(struct page *page)
+{
+
+ int migratetype = get_pageblock_migratetype(page);
+
+ /* Don't interfere with memory hot-remove or the min_free_kbytes blocks */
+ if (migratetype == MIGRATE_ISOLATE || migratetype == MIGRATE_RESERVE)
+ return false;
+
+ /* If the page is a large free page, then allow migration */
+ if (PageBuddy(page) && page_order(page) >= pageblock_order)
+ return true;
+
+ /* If the block is MIGRATE_MOVABLE, allow migration */
+ if (migratetype == MIGRATE_MOVABLE)
+ return true;
+
+ /* Otherwise skip the block */
+ return false;
+}
+
/*
- * Isolate all pages that can be migrated from the block pointed to by
- * the migrate scanner within compact_control.
+ * Based on information in the current compact_control, find blocks
+ * suitable for isolating free pages from and then isolate them.
*/
-static isolate_migrate_t isolate_migratepages(struct zone *zone,
- struct compact_control *cc)
+static void isolate_freepages(struct zone *zone,
+ struct compact_control *cc)
{
- unsigned long low_pfn, end_pfn;
+ struct page *page;
+ unsigned long high_pfn, low_pfn, pfn, zone_end_pfn, end_pfn;
+ unsigned long flags;
+ int nr_freepages = cc->nr_freepages;
+ struct list_head *freelist = &cc->freepages;
- /* Do not scan outside zone boundaries */
- low_pfn = max(cc->migrate_pfn, zone->zone_start_pfn);
+ /*
+ * Initialise the free scanner. The starting point is where we last
+ * scanned from (or the end of the zone if starting). The low point
+ * is the end of the pageblock the migration scanner is using.
+ */
+ pfn = cc->free_pfn;
+ low_pfn = cc->migrate_pfn + pageblock_nr_pages;
- /* Only scan within a pageblock boundary */
- end_pfn = ALIGN(low_pfn + pageblock_nr_pages, pageblock_nr_pages);
+ /*
+ * Take care that if the migration scanner is at the end of the zone
+ * that the free scanner does not accidentally move to the next zone
+ * in the next isolation cycle.
+ */
+ high_pfn = min(low_pfn, pfn);
- /* Do not cross the free scanner or scan within a memory hole */
- if (end_pfn > cc->free_pfn || !pfn_valid(low_pfn)) {
- cc->migrate_pfn = end_pfn;
- return ISOLATE_NONE;
- }
+ zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
- /* Perform the isolation */
- low_pfn = isolate_migratepages_range(zone, cc, low_pfn, end_pfn);
- if (!low_pfn)
- return ISOLATE_ABORT;
+ /*
+ * Isolate free pages until enough are available to migrate the
+ * pages on cc->migratepages. We stop searching if the migrate
+ * and free page scanners meet or enough free pages are isolated.
+ */
+ for (; pfn > low_pfn && cc->nr_migratepages > nr_freepages;
+ pfn -= pageblock_nr_pages) {
+ unsigned long isolated;
- cc->migrate_pfn = low_pfn;
+ if (!pfn_valid(pfn))
+ continue;
- return ISOLATE_SUCCESS;
+ /*
+ * Check for overlapping nodes/zones. It's possible on some
+ * configurations to have a setup like
+ * node0 node1 node0
+ * i.e. it's possible that all pages within a zones range of
+ * pages do not belong to a single zone.
+ */
+ page = pfn_to_page(pfn);
+ if (page_zone(page) != zone)
+ continue;
+
+ /* Check the block is suitable for migration */
+ if (!suitable_migration_target(page))
+ continue;
+
+ /*
+ * Found a block suitable for isolating free pages from. Now
+ * we disabled interrupts, double check things are ok and
+ * isolate the pages. This is to minimise the time IRQs
+ * are disabled
+ */
+ isolated = 0;
+ spin_lock_irqsave(&zone->lock, flags);
+ if (suitable_migration_target(page)) {
+ end_pfn = min(pfn + pageblock_nr_pages, zone_end_pfn);
+ isolated = isolate_freepages_block(pfn, end_pfn,
+ freelist, false);
+ nr_freepages += isolated;
+ }
+ spin_unlock_irqrestore(&zone->lock, flags);
+
+ /*
+ * Record the highest PFN we isolated pages from. When next
+ * looking for free pages, the search will restart here as
+ * page migration may have returned some pages to the allocator
+ */
+ if (isolated)
+ high_pfn = max(high_pfn, pfn);
+ }
+
+ /* split_free_page does not map the pages */
+ map_pages(freelist);
+
+ cc->free_pfn = high_pfn;
+ cc->nr_freepages = nr_freepages;
}
/*
@@ -542,6 +488,44 @@ static void update_nr_listpages(struct compact_control *cc)
cc->nr_freepages = nr_freepages;
}
+/* possible outcome of isolate_migratepages */
+typedef enum {
+ ISOLATE_ABORT, /* Abort compaction now */
+ ISOLATE_NONE, /* No pages isolated, continue scanning */
+ ISOLATE_SUCCESS, /* Pages isolated, migrate */
+} isolate_migrate_t;
+
+/*
+ * Isolate all pages that can be migrated from the block pointed to by
+ * the migrate scanner within compact_control.
+ */
+static isolate_migrate_t isolate_migratepages(struct zone *zone,
+ struct compact_control *cc)
+{
+ unsigned long low_pfn, end_pfn;
+
+ /* Do not scan outside zone boundaries */
+ low_pfn = max(cc->migrate_pfn, zone->zone_start_pfn);
+
+ /* Only scan within a pageblock boundary */
+ end_pfn = ALIGN(low_pfn + pageblock_nr_pages, pageblock_nr_pages);
+
+ /* Do not cross the free scanner or scan within a memory hole */
+ if (end_pfn > cc->free_pfn || !pfn_valid(low_pfn)) {
+ cc->migrate_pfn = end_pfn;
+ return ISOLATE_NONE;
+ }
+
+ /* Perform the isolation */
+ low_pfn = isolate_migratepages_range(zone, cc, low_pfn, end_pfn);
+ if (!low_pfn)
+ return ISOLATE_ABORT;
+
+ cc->migrate_pfn = low_pfn;
+
+ return ISOLATE_SUCCESS;
+}
+
static int compact_finished(struct zone *zone,
struct compact_control *cc)
{
@@ -859,3 +843,5 @@ void compaction_unregister_node(struct node *node)
return device_remove_file(&node->dev, &dev_attr_compact);
}
#endif /* CONFIG_SYSFS && CONFIG_NUMA */
+
+#endif /* CONFIG_COMPACTION */
diff --git a/mm/internal.h b/mm/internal.h
index 2189af4..55e7eed 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -100,6 +100,39 @@ extern void prep_compound_page(struct page *page, unsigned long order);
extern bool is_free_buddy_page(struct page *page);
#endif
+#if defined CONFIG_COMPACTION || defined CONFIG_CMA
+
+/*
+ * in mm/compaction.c
+ */
+/*
+ * compact_control is used to track pages being migrated and the free pages
+ * they are being migrated to during memory compaction. The free_pfn starts
+ * at the end of a zone and migrate_pfn begins at the start. Movable pages
+ * are moved to the end of a zone during a compaction run and the run
+ * completes when free_pfn <= migrate_pfn
+ */
+struct compact_control {
+ struct list_head freepages; /* List of free pages to migrate to */
+ struct list_head migratepages; /* List of pages being migrated */
+ unsigned long nr_freepages; /* Number of isolated free pages */
+ unsigned long nr_migratepages; /* Number of pages to migrate */
+ unsigned long free_pfn; /* isolate_freepages search base */
+ unsigned long migrate_pfn; /* isolate_migratepages search base */
+ bool sync; /* Synchronous migration */
+
+ unsigned int order; /* order a direct compactor needs */
+ int migratetype; /* MOVABLE, RECLAIMABLE etc */
+ struct zone *zone;
+};
+
+unsigned long
+isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn);
+unsigned long
+isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
+ unsigned long low_pfn, unsigned long end_pfn);
+
+#endif
/*
* function for dealing with page's order in buddy system.
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit changes various functions that change pages and
pageblocks migrate type between MIGRATE_ISOLATE and
MIGRATE_MOVABLE in such a way as to allow to work with
MIGRATE_CMA migrate type.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
include/linux/gfp.h | 3 ++-
include/linux/page-isolation.h | 18 +++++++++---------
mm/memory-failure.c | 2 +-
mm/memory_hotplug.c | 6 +++---
mm/page_alloc.c | 18 ++++++++++++------
mm/page_isolation.c | 15 ++++++++-------
6 files changed, 35 insertions(+), 27 deletions(-)
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 78d32a7..1e49be4 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -394,7 +394,8 @@ static inline bool pm_suspended_storage(void)
#ifdef CONFIG_CMA
/* The below functions must be run on a range from a single zone. */
-extern int alloc_contig_range(unsigned long start, unsigned long end);
+extern int alloc_contig_range(unsigned long start, unsigned long end,
+ unsigned migratetype);
extern void free_contig_range(unsigned long pfn, unsigned nr_pages);
/* CMA stuff */
diff --git a/include/linux/page-isolation.h b/include/linux/page-isolation.h
index 051c1b1..3bdcab3 100644
--- a/include/linux/page-isolation.h
+++ b/include/linux/page-isolation.h
@@ -3,7 +3,7 @@
/*
* Changes migrate type in [start_pfn, end_pfn) to be MIGRATE_ISOLATE.
- * If specified range includes migrate types other than MOVABLE,
+ * If specified range includes migrate types other than MOVABLE or CMA,
* this will fail with -EBUSY.
*
* For isolating all pages in the range finally, the caller have to
@@ -11,27 +11,27 @@
* test it.
*/
extern int
-start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn);
+start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
+ unsigned migratetype);
/*
* Changes MIGRATE_ISOLATE to MIGRATE_MOVABLE.
* target range is [start_pfn, end_pfn)
*/
extern int
-undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn);
+undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
+ unsigned migratetype);
/*
- * test all pages in [start_pfn, end_pfn)are isolated or not.
+ * Test all pages in [start_pfn, end_pfn) are isolated or not.
*/
-extern int
-test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn);
+int test_pages_isolated(unsigned long start_pfn, unsigned long end_pfn);
/*
- * Internal funcs.Changes pageblock's migrate type.
- * Please use make_pagetype_isolated()/make_pagetype_movable().
+ * Internal functions. Changes pageblock's migrate type.
*/
extern int set_migratetype_isolate(struct page *page);
-extern void unset_migratetype_isolate(struct page *page);
+extern void unset_migratetype_isolate(struct page *page, unsigned migratetype);
#endif
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 56080ea..76b01bf 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1400,7 +1400,7 @@ static int get_any_page(struct page *p, unsigned long pfn, int flags)
/* Not a free page */
ret = 1;
}
- unset_migratetype_isolate(p);
+ unset_migratetype_isolate(p, MIGRATE_MOVABLE);
unlock_memory_hotplug();
return ret;
}
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 6629faf..fc898cb 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -891,7 +891,7 @@ static int __ref offline_pages(unsigned long start_pfn,
nr_pages = end_pfn - start_pfn;
/* set above range as isolated */
- ret = start_isolate_page_range(start_pfn, end_pfn);
+ ret = start_isolate_page_range(start_pfn, end_pfn, MIGRATE_MOVABLE);
if (ret)
goto out;
@@ -956,7 +956,7 @@ repeat:
We cannot do rollback at this point. */
offline_isolated_pages(start_pfn, end_pfn);
/* reset pagetype flags and makes migrate type to be MOVABLE */
- undo_isolate_page_range(start_pfn, end_pfn);
+ undo_isolate_page_range(start_pfn, end_pfn, MIGRATE_MOVABLE);
/* removal success */
zone->present_pages -= offlined_pages;
zone->zone_pgdat->node_present_pages -= offlined_pages;
@@ -981,7 +981,7 @@ failed_removal:
start_pfn, end_pfn);
memory_notify(MEM_CANCEL_OFFLINE, &arg);
/* pushback to free area */
- undo_isolate_page_range(start_pfn, end_pfn);
+ undo_isolate_page_range(start_pfn, end_pfn, MIGRATE_MOVABLE);
out:
unlock_memory_hotplug();
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 993c375..947879c 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5539,7 +5539,7 @@ out:
return ret;
}
-void unset_migratetype_isolate(struct page *page)
+void unset_migratetype_isolate(struct page *page, unsigned migratetype)
{
struct zone *zone;
unsigned long flags;
@@ -5547,8 +5547,8 @@ void unset_migratetype_isolate(struct page *page)
spin_lock_irqsave(&zone->lock, flags);
if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
goto out;
- set_pageblock_migratetype(page, MIGRATE_MOVABLE);
- move_freepages_block(zone, page, MIGRATE_MOVABLE);
+ set_pageblock_migratetype(page, migratetype);
+ move_freepages_block(zone, page, migratetype);
out:
spin_unlock_irqrestore(&zone->lock, flags);
}
@@ -5624,6 +5624,10 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
* alloc_contig_range() -- tries to allocate given range of pages
* @start: start PFN to allocate
* @end: one-past-the-last PFN to allocate
+ * @migratetype: migratetype of the underlaying pageblocks (either
+ * #MIGRATE_MOVABLE or #MIGRATE_CMA). All pageblocks
+ * in range must have the same migratetype and it must
+ * be either of the two.
*
* The PFN range does not have to be pageblock or MAX_ORDER_NR_PAGES
* aligned, however it's the caller's responsibility to guarantee that
@@ -5636,7 +5640,8 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
* pages which PFN is in [start, end) are allocated for the caller and
* need to be freed with free_contig_range().
*/
-int alloc_contig_range(unsigned long start, unsigned long end)
+int alloc_contig_range(unsigned long start, unsigned long end,
+ unsigned migratetype)
{
struct zone *zone = page_zone(pfn_to_page(start));
unsigned long outer_start, outer_end;
@@ -5666,7 +5671,8 @@ int alloc_contig_range(unsigned long start, unsigned long end)
*/
ret = start_isolate_page_range(pfn_align_to_maxpage_down(start),
- pfn_align_to_maxpage_up(end));
+ pfn_align_to_maxpage_up(end),
+ migratetype);
if (ret)
goto done;
@@ -5726,7 +5732,7 @@ int alloc_contig_range(unsigned long start, unsigned long end)
done:
undo_isolate_page_range(pfn_align_to_maxpage_down(start),
- pfn_align_to_maxpage_up(end));
+ pfn_align_to_maxpage_up(end), migratetype);
return ret;
}
diff --git a/mm/page_isolation.c b/mm/page_isolation.c
index 4ae42bb..c9f0477 100644
--- a/mm/page_isolation.c
+++ b/mm/page_isolation.c
@@ -24,6 +24,7 @@ __first_valid_page(unsigned long pfn, unsigned long nr_pages)
* to be MIGRATE_ISOLATE.
* @start_pfn: The lower PFN of the range to be isolated.
* @end_pfn: The upper PFN of the range to be isolated.
+ * @migratetype: migrate type to set in error recovery.
*
* Making page-allocation-type to be MIGRATE_ISOLATE means free pages in
* the range will never be allocated. Any free pages and pages freed in the
@@ -32,8 +33,8 @@ __first_valid_page(unsigned long pfn, unsigned long nr_pages)
* start_pfn/end_pfn must be aligned to pageblock_order.
* Returns 0 on success and -EBUSY if any part of range cannot be isolated.
*/
-int
-start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn)
+int start_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
+ unsigned migratetype)
{
unsigned long pfn;
unsigned long undo_pfn;
@@ -56,7 +57,7 @@ undo:
for (pfn = start_pfn;
pfn < undo_pfn;
pfn += pageblock_nr_pages)
- unset_migratetype_isolate(pfn_to_page(pfn));
+ unset_migratetype_isolate(pfn_to_page(pfn), migratetype);
return -EBUSY;
}
@@ -64,8 +65,8 @@ undo:
/*
* Make isolated pages available again.
*/
-int
-undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn)
+int undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn,
+ unsigned migratetype)
{
unsigned long pfn;
struct page *page;
@@ -77,7 +78,7 @@ undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn)
page = __first_valid_page(pfn, pageblock_nr_pages);
if (!page || get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
continue;
- unset_migratetype_isolate(page);
+ unset_migratetype_isolate(page, migratetype);
}
return 0;
}
@@ -86,7 +87,7 @@ undo_isolate_page_range(unsigned long start_pfn, unsigned long end_pfn)
* all pages in [start_pfn...end_pfn) must be in the same zone.
* zone->lock must be held before call this.
*
- * Returns 1 if all pages in the range is isolated.
+ * Returns 1 if all pages in the range are isolated.
*/
static int
__test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn)
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit creates a map_pages() function which map pages freed
using split_free_pages(). This merely moves some code from
isolate_freepages() so that it can be reused in other places.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
---
mm/compaction.c | 15 +++++++++++----
1 files changed, 11 insertions(+), 4 deletions(-)
diff --git a/mm/compaction.c b/mm/compaction.c
index 62902b6..9bbcc53 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -127,6 +127,16 @@ static bool suitable_migration_target(struct page *page)
return false;
}
+static void map_pages(struct list_head *list)
+{
+ struct page *page;
+
+ list_for_each_entry(page, list, lru) {
+ arch_alloc_page(page, 0);
+ kernel_map_pages(page, 1, 1);
+ }
+}
+
/*
* Based on information in the current compact_control, find blocks
* suitable for isolating free pages from and then isolate them.
@@ -206,10 +216,7 @@ static void isolate_freepages(struct zone *zone,
}
/* split_free_page does not map the pages */
- list_for_each_entry(page, freelist, lru) {
- arch_alloc_page(page, 0);
- kernel_map_pages(page, 1, 1);
- }
+ map_pages(freelist);
cc->free_pfn = high_pfn;
cc->nr_freepages = nr_freepages;
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit introduces isolate_freepages_range() function which
generalises isolate_freepages_block() so that it can be used on
arbitrary PFN ranges.
isolate_freepages_block() is left with only minor changes.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/compaction.c | 111 ++++++++++++++++++++++++++++++++++++++++++++++---------
1 files changed, 93 insertions(+), 18 deletions(-)
diff --git a/mm/compaction.c b/mm/compaction.c
index 9bbcc53..9fef891 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -54,24 +54,20 @@ static unsigned long release_freepages(struct list_head *freelist)
return count;
}
-/* Isolate free pages onto a private freelist. Must hold zone->lock */
-static unsigned long isolate_freepages_block(struct zone *zone,
- unsigned long blockpfn,
- struct list_head *freelist)
+/*
+ * Isolate free pages onto a private freelist. Caller must hold zone->lock.
+ * If @strict is true, will abort returning 0 on any invalid PFNs or non-free
+ * pages inside of the pageblock (even though it may still end up isolating
+ * some pages).
+ */
+static unsigned long isolate_freepages_block(unsigned long blockpfn,
+ unsigned long end_pfn,
+ struct list_head *freelist,
+ bool strict)
{
- unsigned long zone_end_pfn, end_pfn;
int nr_scanned = 0, total_isolated = 0;
struct page *cursor;
- /* Get the last PFN we should scan for free pages at */
- zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
- end_pfn = min(blockpfn + pageblock_nr_pages, zone_end_pfn);
-
- /* Find the first usable PFN in the block to initialse page cursor */
- for (; blockpfn < end_pfn; blockpfn++) {
- if (pfn_valid_within(blockpfn))
- break;
- }
cursor = pfn_to_page(blockpfn);
/* Isolate free pages. This assumes the block is valid */
@@ -79,15 +75,23 @@ static unsigned long isolate_freepages_block(struct zone *zone,
int isolated, i;
struct page *page = cursor;
- if (!pfn_valid_within(blockpfn))
+ if (!pfn_valid_within(blockpfn)) {
+ if (strict)
+ return 0;
continue;
+ }
nr_scanned++;
- if (!PageBuddy(page))
+ if (!PageBuddy(page)) {
+ if (strict)
+ return 0;
continue;
+ }
/* Found a free page, break it into order-0 pages */
isolated = split_free_page(page);
+ if (!isolated && strict)
+ return 0;
total_isolated += isolated;
for (i = 0; i < isolated; i++) {
list_add(&page->lru, freelist);
@@ -105,6 +109,73 @@ static unsigned long isolate_freepages_block(struct zone *zone,
return total_isolated;
}
+/**
+ * isolate_freepages_range() - isolate free pages.
+ * @start_pfn: The first PFN to start isolating.
+ * @end_pfn: The one-past-last PFN.
+ *
+ * Non-free pages, invalid PFNs, or zone boundaries within the
+ * [start_pfn, end_pfn) range are considered errors, cause function to
+ * undo its actions and return zero.
+ *
+ * Otherwise, function returns one-past-the-last PFN of isolated page
+ * (which may be greater then end_pfn if end fell in a middle of
+ * a free page).
+ */
+static unsigned long
+isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn)
+{
+ unsigned long isolated, pfn, block_end_pfn, flags;
+ struct zone *zone = NULL;
+ LIST_HEAD(freelist);
+
+ if (pfn_valid(start_pfn))
+ zone = page_zone(pfn_to_page(start_pfn));
+
+ for (pfn = start_pfn; pfn < end_pfn; pfn += isolated) {
+ if (!pfn_valid(pfn) || zone != page_zone(pfn_to_page(pfn)))
+ break;
+
+ /*
+ * On subsequent iterations ALIGN() is actually not needed,
+ * but we keep it that we not to complicate the code.
+ */
+ block_end_pfn = ALIGN(pfn + 1, pageblock_nr_pages);
+ block_end_pfn = min(block_end_pfn, end_pfn);
+
+ spin_lock_irqsave(&zone->lock, flags);
+ isolated = isolate_freepages_block(pfn, block_end_pfn,
+ &freelist, true);
+ spin_unlock_irqrestore(&zone->lock, flags);
+
+ /*
+ * In strict mode, isolate_freepages_block() returns 0 if
+ * there are any holes in the block (ie. invalid PFNs or
+ * non-free pages).
+ */
+ if (!isolated)
+ break;
+
+ /*
+ * If we managed to isolate pages, it is always (1 << n) *
+ * pageblock_nr_pages for some non-negative n. (Max order
+ * page may span two pageblocks).
+ */
+ }
+
+ /* split_free_page does not map the pages */
+ map_pages(&freelist);
+
+ if (pfn < end_pfn) {
+ /* Loop terminated early, cleanup. */
+ release_freepages(&freelist);
+ return 0;
+ }
+
+ /* We don't use freelists for anything. */
+ return pfn;
+}
+
/* Returns true if the page is within a block suitable for migration to */
static bool suitable_migration_target(struct page *page)
{
@@ -145,7 +216,7 @@ static void isolate_freepages(struct zone *zone,
struct compact_control *cc)
{
struct page *page;
- unsigned long high_pfn, low_pfn, pfn;
+ unsigned long high_pfn, low_pfn, pfn, zone_end_pfn, end_pfn;
unsigned long flags;
int nr_freepages = cc->nr_freepages;
struct list_head *freelist = &cc->freepages;
@@ -165,6 +236,8 @@ static void isolate_freepages(struct zone *zone,
*/
high_pfn = min(low_pfn, pfn);
+ zone_end_pfn = zone->zone_start_pfn + zone->spanned_pages;
+
/*
* Isolate free pages until enough are available to migrate the
* pages on cc->migratepages. We stop searching if the migrate
@@ -201,7 +274,9 @@ static void isolate_freepages(struct zone *zone,
isolated = 0;
spin_lock_irqsave(&zone->lock, flags);
if (suitable_migration_target(page)) {
- isolated = isolate_freepages_block(zone, pfn, freelist);
+ end_pfn = min(pfn + pageblock_nr_pages, zone_end_pfn);
+ isolated = isolate_freepages_block(pfn, end_pfn,
+ freelist, false);
nr_freepages += isolated;
}
spin_unlock_irqrestore(&zone->lock, flags);
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit adds a row for MIGRATE_ISOLATE type to the fallbacks array
which was missing from it. It also, changes the array traversal logic
a little making MIGRATE_RESERVE an end marker. The letter change,
removes the implicit MIGRATE_UNMOVABLE from the end of each row which
was read by __rmqueue_fallback() function.
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/page_alloc.c | 9 +++++----
1 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 9006e69..238fcec 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -875,11 +875,12 @@ struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
* This array describes the order lists are fallen back to when
* the free lists for the desirable migrate type are depleted
*/
-static int fallbacks[MIGRATE_TYPES][MIGRATE_TYPES-1] = {
+static int fallbacks[MIGRATE_TYPES][3] = {
[MIGRATE_UNMOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
[MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
[MIGRATE_MOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
- [MIGRATE_RESERVE] = { MIGRATE_RESERVE, MIGRATE_RESERVE, MIGRATE_RESERVE }, /* Never used */
+ [MIGRATE_RESERVE] = { MIGRATE_RESERVE }, /* Never used */
+ [MIGRATE_ISOLATE] = { MIGRATE_RESERVE }, /* Never used */
};
/*
@@ -974,12 +975,12 @@ __rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
/* Find the largest possible block of pages in the other list */
for (current_order = MAX_ORDER-1; current_order >= order;
--current_order) {
- for (i = 0; i < MIGRATE_TYPES - 1; i++) {
+ for (i = 0;; i++) {
migratetype = fallbacks[start_migratetype][i];
/* MIGRATE_RESERVE handled later if necessary */
if (migratetype == MIGRATE_RESERVE)
- continue;
+ break;
area = &(zone->free_area[current_order]);
if (list_empty(&area->free_list[migratetype]))
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
This commit introduces isolate_migratepages_range() function which
extracts functionality from isolate_migratepages() so that it can be
used on arbitrary PFN ranges.
isolate_migratepages() function is implemented as a simple wrapper
around isolate_migratepages_range().
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
Tested-by: Rob Clark <[email protected]>
Tested-by: Ohad Ben-Cohen <[email protected]>
Tested-by: Benjamin Gaignard <[email protected]>
---
mm/compaction.c | 75 +++++++++++++++++++++++++++++++++++++++---------------
1 files changed, 54 insertions(+), 21 deletions(-)
diff --git a/mm/compaction.c b/mm/compaction.c
index 71a58f6..62902b6 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -250,31 +250,34 @@ typedef enum {
ISOLATE_SUCCESS, /* Pages isolated, migrate */
} isolate_migrate_t;
-/*
- * Isolate all pages that can be migrated from the block pointed to by
- * the migrate scanner within compact_control.
+/**
+ * isolate_migratepages_range() - isolate all migrate-able pages in range.
+ * @zone: Zone pages are in.
+ * @cc: Compaction control structure.
+ * @low_pfn: The first PFN of the range.
+ * @end_pfn: The one-past-the-last PFN of the range.
+ *
+ * Isolate all pages that can be migrated from the range specified by
+ * [low_pfn, end_pfn). Returns zero if there is a fatal signal
+ * pending), otherwise PFN of the first page that was not scanned
+ * (which may be both less, equal to or more then end_pfn).
+ *
+ * Assumes that cc->migratepages is empty and cc->nr_migratepages is
+ * zero.
+ *
+ * Apart from cc->migratepages and cc->nr_migratetypes this function
+ * does not modify any cc's fields, in particular it does not modify
+ * (or read for that matter) cc->migrate_pfn.
*/
-static isolate_migrate_t isolate_migratepages(struct zone *zone,
- struct compact_control *cc)
+static unsigned long
+isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
+ unsigned long low_pfn, unsigned long end_pfn)
{
- unsigned long low_pfn, end_pfn;
unsigned long last_pageblock_nr = 0, pageblock_nr;
unsigned long nr_scanned = 0, nr_isolated = 0;
struct list_head *migratelist = &cc->migratepages;
isolate_mode_t mode = ISOLATE_ACTIVE|ISOLATE_INACTIVE;
- /* Do not scan outside zone boundaries */
- low_pfn = max(cc->migrate_pfn, zone->zone_start_pfn);
-
- /* Only scan within a pageblock boundary */
- end_pfn = ALIGN(low_pfn + pageblock_nr_pages, pageblock_nr_pages);
-
- /* Do not cross the free scanner or scan within a memory hole */
- if (end_pfn > cc->free_pfn || !pfn_valid(low_pfn)) {
- cc->migrate_pfn = end_pfn;
- return ISOLATE_NONE;
- }
-
/*
* Ensure that there are not too many pages isolated from the LRU
* list by either parallel reclaimers or compaction. If there are,
@@ -283,12 +286,12 @@ static isolate_migrate_t isolate_migratepages(struct zone *zone,
while (unlikely(too_many_isolated(zone))) {
/* async migration should just abort */
if (!cc->sync)
- return ISOLATE_ABORT;
+ return 0;
congestion_wait(BLK_RW_ASYNC, HZ/10);
if (fatal_signal_pending(current))
- return ISOLATE_ABORT;
+ return 0;
}
/* Time to isolate some pages for migration */
@@ -374,10 +377,40 @@ static isolate_migrate_t isolate_migratepages(struct zone *zone,
acct_isolated(zone, cc);
spin_unlock_irq(&zone->lru_lock);
- cc->migrate_pfn = low_pfn;
trace_mm_compaction_isolate_migratepages(nr_scanned, nr_isolated);
+ return low_pfn;
+}
+
+/*
+ * Isolate all pages that can be migrated from the block pointed to by
+ * the migrate scanner within compact_control.
+ */
+static isolate_migrate_t isolate_migratepages(struct zone *zone,
+ struct compact_control *cc)
+{
+ unsigned long low_pfn, end_pfn;
+
+ /* Do not scan outside zone boundaries */
+ low_pfn = max(cc->migrate_pfn, zone->zone_start_pfn);
+
+ /* Only scan within a pageblock boundary */
+ end_pfn = ALIGN(low_pfn + pageblock_nr_pages, pageblock_nr_pages);
+
+ /* Do not cross the free scanner or scan within a memory hole */
+ if (end_pfn > cc->free_pfn || !pfn_valid(low_pfn)) {
+ cc->migrate_pfn = end_pfn;
+ return ISOLATE_NONE;
+ }
+
+ /* Perform the isolation */
+ low_pfn = isolate_migratepages_range(zone, cc, low_pfn, end_pfn);
+ if (!low_pfn)
+ return ISOLATE_ABORT;
+
+ cc->migrate_pfn = low_pfn;
+
return ISOLATE_SUCCESS;
}
--
1.7.1.569.g6f426
From: Michal Nazarewicz <[email protected]>
Signed-off-by: Michal Nazarewicz <[email protected]>
Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
---
mm/page_alloc.c | 18 +++++++++---------
1 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index d2186ec..7fe7697 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -513,10 +513,10 @@ static inline int page_is_buddy(struct page *page, struct page *buddy,
* free pages of length of (1 << order) and marked with _mapcount -2. Page's
* order is recorded in page_private(page) field.
* So when we are allocating or freeing one, we can derive the state of the
- * other. That is, if we allocate a small block, and both were
- * free, the remainder of the region must be split into blocks.
+ * other. That is, if we allocate a small block, and both were
+ * free, the remainder of the region must be split into blocks.
* If a block is freed, and its buddy is also free, then this
- * triggers coalescing into a block of larger size.
+ * triggers coalescing into a block of larger size.
*
* -- wli
*/
@@ -1061,17 +1061,17 @@ retry_reserve:
return page;
}
-/*
+/*
* Obtain a specified number of elements from the buddy allocator, all under
* a single hold of the lock, for efficiency. Add them to the supplied list.
* Returns the number of new pages which were placed at *list.
*/
-static int rmqueue_bulk(struct zone *zone, unsigned int order,
+static int rmqueue_bulk(struct zone *zone, unsigned int order,
unsigned long count, struct list_head *list,
int migratetype, int cold)
{
int i;
-
+
spin_lock(&zone->lock);
for (i = 0; i < count; ++i) {
struct page *page = __rmqueue(zone, order, migratetype);
@@ -4258,7 +4258,7 @@ static void __paginginit free_area_init_core(struct pglist_data *pgdat,
init_waitqueue_head(&pgdat->kswapd_wait);
pgdat->kswapd_max_order = 0;
pgdat_page_cgroup_init(pgdat);
-
+
for (j = 0; j < MAX_NR_ZONES; j++) {
struct zone *zone = pgdat->node_zones + j;
unsigned long size, realsize, memmap_pages;
@@ -5081,11 +5081,11 @@ int __meminit init_per_zone_wmark_min(void)
module_init(init_per_zone_wmark_min)
/*
- * min_free_kbytes_sysctl_handler - just a wrapper around proc_dointvec() so
+ * min_free_kbytes_sysctl_handler - just a wrapper around proc_dointvec() so
* that we can call two helper functions whenever min_free_kbytes
* changes.
*/
-int min_free_kbytes_sysctl_handler(ctl_table *table, int write,
+int min_free_kbytes_sysctl_handler(ctl_table *table, int write,
void __user *buffer, size_t *length, loff_t *ppos)
{
proc_dointvec(table, write, buffer, length, ppos);
--
1.7.1.569.g6f426
On Fri, Feb 03, 2012 at 01:18:46PM +0100, Marek Szyprowski wrote:
> From: Michal Nazarewicz <[email protected]>
>
> This commit creates a map_pages() function which map pages freed
> using split_free_pages(). This merely moves some code from
> isolate_freepages() so that it can be reused in other places.
>
> Signed-off-by: Michal Nazarewicz <[email protected]>
> Signed-off-by: Marek Szyprowski <[email protected]>
Acked-by: Mel Gorman <[email protected]>
--
Mel Gorman
SUSE Labs
On Fri, Feb 03, 2012 at 01:18:51PM +0100, Marek Szyprowski wrote:
> From: Michal Nazarewicz <[email protected]>
>
> The MIGRATE_CMA migration type has two main characteristics:
> (i) only movable pages can be allocated from MIGRATE_CMA
> pageblocks and (ii) page allocator will never change migration
> type of MIGRATE_CMA pageblocks.
>
> This guarantees (to some degree) that page in a MIGRATE_CMA page
> block can always be migrated somewhere else (unless there's no
> memory left in the system).
>
> It is designed to be used for allocating big chunks (eg. 10MiB)
> of physically contiguous memory. Once driver requests
> contiguous memory, pages from MIGRATE_CMA pageblocks may be
> migrated away to create a contiguous block.
>
> To minimise number of migrations, MIGRATE_CMA migration type
> is the last type tried when page allocator falls back to other
> migration types then requested.
>
> Signed-off-by: Michal Nazarewicz <[email protected]>
> Signed-off-by: Marek Szyprowski <[email protected]>
> Signed-off-by: Kyungmin Park <[email protected]>
> Tested-by: Rob Clark <[email protected]>
> Tested-by: Ohad Ben-Cohen <[email protected]>
> Tested-by: Benjamin Gaignard <[email protected]>
Minor comments that can be handled as a follow-up but otherwise
Acked-by: Mel Gorman <[email protected]>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 238fcec..993c375 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -750,6 +750,26 @@ void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
> __free_pages(page, order);
> }
>
> +#ifdef CONFIG_CMA
> +/*
> + * Free whole pageblock and set it's migration type to MIGRATE_CMA.
> + */
> +void __init init_cma_reserved_pageblock(struct page *page)
> +{
> + unsigned i = pageblock_nr_pages;
> + struct page *p = page;
> +
> + do {
> + __ClearPageReserved(p);
> + set_page_count(p, 0);
> + } while (++p, --i);
> +
> + set_page_refcounted(page);
> + set_pageblock_migratetype(page, MIGRATE_CMA);
> + __free_pages(page, pageblock_order);
> + totalram_pages += pageblock_nr_pages;
> +}
> +#endif
>
This chunk is not used with the patch. Usually a hunk like this would
be part of the patch that used it. Functionally it looks ok but
I see that the function that calls it is *not* __init. That should
trigger a section warning. Do a make CONFIG_DEBUG_SECTION_MISMATCH=y
and clean it up if necessary.
> /*
> * The order of subdivision here is critical for the IO subsystem.
> @@ -875,10 +895,15 @@ struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
> * This array describes the order lists are fallen back to when
> * the free lists for the desirable migrate type are depleted
> */
> -static int fallbacks[MIGRATE_TYPES][3] = {
> - [MIGRATE_UNMOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
> - [MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
> - [MIGRATE_MOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
> +static int fallbacks[MIGRATE_TYPES][4] = {
> + [MIGRATE_UNMOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
> + [MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE, MIGRATE_MOVABLE, MIGRATE_RESERVE },
> +#ifdef CONFIG_CMA
> + [MIGRATE_MOVABLE] = { MIGRATE_CMA, MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
> + [MIGRATE_CMA] = { MIGRATE_RESERVE }, /* Never used */
> +#else
> + [MIGRATE_MOVABLE] = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
> +#endif
> [MIGRATE_RESERVE] = { MIGRATE_RESERVE }, /* Never used */
> [MIGRATE_ISOLATE] = { MIGRATE_RESERVE }, /* Never used */
> };
> @@ -995,11 +1020,18 @@ __rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
> * pages to the preferred allocation list. If falling
> * back for a reclaimable kernel allocation, be more
> * aggressive about taking ownership of free pages
> + *
> + * On the other hand, never change migration
> + * type of MIGRATE_CMA pageblocks nor move CMA
> + * pages on different free lists. We don't
> + * want unmovable pages to be allocated from
> + * MIGRATE_CMA areas.
> */
> - if (unlikely(current_order >= (pageblock_order >> 1)) ||
> - start_migratetype == MIGRATE_RECLAIMABLE ||
> - page_group_by_mobility_disabled) {
> - unsigned long pages;
> + if (!is_migrate_cma(migratetype) &&
> + (unlikely(current_order >= pageblock_order / 2) ||
> + start_migratetype == MIGRATE_RECLAIMABLE ||
> + page_group_by_mobility_disabled)) {
> + int pages;
> pages = move_freepages_block(zone, page,
> start_migratetype);
>
> @@ -1017,11 +1049,14 @@ __rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
> rmv_page_order(page);
>
> /* Take ownership for orders >= pageblock_order */
> - if (current_order >= pageblock_order)
> + if (current_order >= pageblock_order &&
> + !is_migrate_cma(migratetype))
> change_pageblock_range(page, current_order,
> start_migratetype);
>
> - expand(zone, page, order, current_order, area, migratetype);
> + expand(zone, page, order, current_order, area,
> + is_migrate_cma(migratetype)
> + ? migratetype : start_migratetype);
>
> trace_mm_page_alloc_extfrag(page, order, current_order,
> start_migratetype, migratetype);
> @@ -1072,7 +1107,7 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
> unsigned long count, struct list_head *list,
> int migratetype, int cold)
> {
> - int i;
> + int mt = migratetype, i;
>
> spin_lock(&zone->lock);
> for (i = 0; i < count; ++i) {
> @@ -1093,7 +1128,12 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order,
> list_add(&page->lru, list);
> else
> list_add_tail(&page->lru, list);
> - set_page_private(page, migratetype);
> +#ifdef CONFIG_CMA
> + mt = get_pageblock_migratetype(page);
> + if (!is_migrate_cma(mt) && mt != MIGRATE_ISOLATE)
> + mt = migratetype;
> +#endif
> + set_page_private(page, mt);
Ok, so while I'm not happy with the CONFIG_CMA bit here, so be it for
now. There are a few things I would like to see with it though in the
future
1. Add a comment explaning why it is necessary only for CONFIG_CMA.
Put all the ugliness in the changelog if you like, it's already
been written up in a mail so you can cut and paste the changelog
2. If there exists a second hunk that has this sort of ugliness,
consider doing something like
#ifdef CONFIG_CMA
#define CMA_BUILD 1
#else
#define CMA_BUILD 0
#endif
if (CONFIG_CMA) {
int mt = get_pageblock_migratetype(page);
if (!is_migrate_cma(mt) && mt != MIGRATE_ISOLATE)
mt = migratetype;
}
That can be slightly tidier and easier to follow while still getting
optimised out for !CONFIG_CMA
3. Consider trying to get rid of the CONFIG_CMA part entirely. Do this
by having a readmostly static variable that is *only* set
while MIGRATE_ISOLATE pageblocks exist. If they exist, then
unconditionally do this paranoid check documenting that both memory
hotplug and CMA benefit from it. The advantage is that you get
the careful checking that you want but incur the cost in the page
allocator *only* when you are actively trying to allocate with CMA.
e.g.
static int nr_migrate_isolate __read_mostly;
int set_migratetype_isolate(struct page *page)
{
....
nr_migrate_isolate++;
....
}
void unset_migratetype_isolate(struct page *page)
{
...
nr_migrate_isolate--;
...
}
Be careful to get the accounting of nr_migrate_isolate right during
memory hot-remove if unset_migratetype_isolate is not called because
the memory is offlined.In rmqueue_bulk() then do
/*
* During memory hot-remove and during CMA allocation, be
* careful that pages that were added to the per-cpu
* lists before the pageblock was marked MIGRATE_ISOLATE and
* not moved properly are accounted for properly
*/
if (nr_migrate_isolate) {
int mt = get_pageblock_migratetype(page);
if (!is_migrate_cma(mt) && mt != MIGRATE_ISOLATE)
mt = migratetype;
}
Make sense?
--
Mel Gorman
SUSE Labs
On Fri, Feb 03, 2012 at 01:18:54PM +0100, Marek Szyprowski wrote:
> alloc_contig_range() performs memory allocation so it also should keep
> track on keeping the correct level of memory watermarks. This commit adds
> a call to *_slowpath style reclaim to grab enough pages to make sure that
> the final collection of contiguous pages from freelists will not starve
> the system.
>
> Signed-off-by: Marek Szyprowski <[email protected]>
> Signed-off-by: Kyungmin Park <[email protected]>
> CC: Michal Nazarewicz <[email protected]>
> Tested-by: Rob Clark <[email protected]>
> Tested-by: Ohad Ben-Cohen <[email protected]>
> Tested-by: Benjamin Gaignard <[email protected]>
I still do not intend to ack this patch and any damage is confined to
CMA but I have a few comments anyway.
> ---
> mm/page_alloc.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
> 1 files changed, 47 insertions(+), 0 deletions(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 983ccba..371a79f 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5632,6 +5632,46 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
> return ret > 0 ? 0 : ret;
> }
>
> +/*
> + * Trigger memory pressure bump to reclaim some pages in order to be able to
> + * allocate 'count' pages in single page units. Does similar work as
> + *__alloc_pages_slowpath() function.
> + */
> +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
> +{
> + enum zone_type high_zoneidx = gfp_zone(gfp_mask);
> + struct zonelist *zonelist = node_zonelist(0, gfp_mask);
> + int did_some_progress = 0;
> + int order = 1;
> + unsigned long watermark;
> +
> + /*
> + * Increase level of watermarks to force kswapd do his job
> + * to stabilize at new watermark level.
> + */
> + min_free_kbytes += count * PAGE_SIZE / 1024;
There is a risk of overflow here although it is incredibly
small. Still, a potentially nicer way of doing this was
count << (PAGE_SHIFT - 10)
> + setup_per_zone_wmarks();
> +
Nothing prevents two or more processes updating the wmarks at the same
time which is racy and unpredictable. Today it is not much of a problem
but CMA makes this path hotter than it was and you may see weirdness
if two processes are updating zonelists at the same time. Swap-over-NFS
actually starts with a patch that serialises setup_per_zone_wmarks()
You also potentially have a BIG problem here if this happens
min_free_kbytes = 32768
Process a: min_free_kbytes += 65536
Process a: start direct reclaim
echo 16374 > /proc/sys/vm/min_free_kbytes
Process a: exit direct_reclaim
Process a: min_free_kbytes -= 65536
min_free_kbytes now wraps negative and the machine hangs.
The damage is confined to CMA though so I am not going to lose sleep
over it but you might want to consider at least preventing parallel
updates to min_free_kbytes from proc.
--
Mel Gorman
SUSE Labs
On Fri, Feb 03, 2012 at 01:18:43PM +0100, Marek Szyprowski wrote:
> Welcome everyone again!
>
> This is yet another quick update on Contiguous Memory Allocator patches.
> This version includes another set of code cleanups requested by Mel
> Gorman and a few minor bug fixes. I really hope that this version will
> be accepted for merging and future development will be handled by
> incremental patches.
FWIW, I've acked all I'm going to ack of this series and made some
suggestions on follow-ups on the core MM parts that could be done
in-tree. I think the current reclaim logic is going to burn CMA with
race conditions but it is a CMA-specific problem so watch out for
that :)
As before, I did not even look at the CMA driver itself or the
arch-specific parts. I'm assuming Arnd has that side of things covered.
Thanks Marek.
--
Mel Gorman
SUSE Labs
Hello Marek
On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
<[email protected]> wrote:
> From: Michal Nazarewicz <[email protected]>
>
> The MIGRATE_CMA migration type has two main characteristics:
> (i) only movable pages can be allocated from MIGRATE_CMA
> pageblocks and (ii) page allocator will never change migration
> type of MIGRATE_CMA pageblocks.
>
> This guarantees (to some degree) that page in a MIGRATE_CMA page
> block can always be migrated somewhere else (unless there's no
> memory left in the system).
>
> It is designed to be used for allocating big chunks (eg. 10MiB)
> of physically contiguous memory. Once driver requests
> contiguous memory, pages from MIGRATE_CMA pageblocks may be
> migrated away to create a contiguous block.
>
> To minimise number of migrations, MIGRATE_CMA migration type
> is the last type tried when page allocator falls back to other
> migration types then requested.
>
> Signed-off-by: Michal Nazarewicz <[email protected]>
> Signed-off-by: Marek Szyprowski <[email protected]>
> Signed-off-by: Kyungmin Park <[email protected]>
> Tested-by: Rob Clark <[email protected]>
> Tested-by: Ohad Ben-Cohen <[email protected]>
> Tested-by: Benjamin Gaignard <[email protected]>
> ---
> include/linux/gfp.h | 3 ++
> include/linux/mmzone.h | 38 +++++++++++++++++++----
> mm/Kconfig | 2 +-
> mm/compaction.c | 11 +++++--
> mm/page_alloc.c | 78 ++++++++++++++++++++++++++++++++++++++----------
> mm/vmstat.c | 3 ++
> 6 files changed, 108 insertions(+), 27 deletions(-)
>
> diff --git a/include/linux/gfp.h b/include/linux/gfp.h
> index 052a5b6..78d32a7 100644
> --- a/include/linux/gfp.h
> +++ b/include/linux/gfp.h
> @@ -397,6 +397,9 @@ static inline bool pm_suspended_storage(void)
> extern int alloc_contig_range(unsigned long start, unsigned long end);
> extern void free_contig_range(unsigned long pfn, unsigned nr_pages);
>
> +/* CMA stuff */
> +extern void init_cma_reserved_pageblock(struct page *page);
> +
> #endif
>
> #endif /* __LINUX_GFP_H */
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 650ba2f..82f4fa5 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -35,13 +35,37 @@
> */
> #define PAGE_ALLOC_COSTLY_ORDER 3
>
> -#define MIGRATE_UNMOVABLE 0
> -#define MIGRATE_RECLAIMABLE 1
> -#define MIGRATE_MOVABLE 2
> -#define MIGRATE_PCPTYPES 3 /* the number of types on the pcp lists */
> -#define MIGRATE_RESERVE 3
> -#define MIGRATE_ISOLATE 4 /* can't allocate from here */
> -#define MIGRATE_TYPES 5
> +enum {
> + MIGRATE_UNMOVABLE,
> + MIGRATE_RECLAIMABLE,
> + MIGRATE_MOVABLE,
> + MIGRATE_PCPTYPES, /* the number of types on the pcp lists */
> + MIGRATE_RESERVE = MIGRATE_PCPTYPES,
> +#ifdef CONFIG_CMA
> + /*
> + * MIGRATE_CMA migration type is designed to mimic the way
> + * ZONE_MOVABLE works. Only movable pages can be allocated
> + * from MIGRATE_CMA pageblocks and page allocator never
> + * implicitly change migration type of MIGRATE_CMA pageblock.
> + *
> + * The way to use it is to change migratetype of a range of
> + * pageblocks to MIGRATE_CMA which can be done by
> + * __free_pageblock_cma() function. What is important though
> + * is that a range of pageblocks must be aligned to
> + * MAX_ORDER_NR_PAGES should biggest page be bigger then
> + * a single pageblock.
> + */
> + MIGRATE_CMA,
> +#endif
> + MIGRATE_ISOLATE, /* can't allocate from here */
> + MIGRATE_TYPES
> +};
> +
> +#ifdef CONFIG_CMA
> +# define is_migrate_cma(migratetype) unlikely((migratetype) == MIGRATE_CMA)
> +#else
> +# define is_migrate_cma(migratetype) false
> +#endif
>
> #define for_each_migratetype_order(order, type) \
> for (order = 0; order < MAX_ORDER; order++) \
> diff --git a/mm/Kconfig b/mm/Kconfig
> index e338407..3922002 100644
> --- a/mm/Kconfig
> +++ b/mm/Kconfig
> @@ -198,7 +198,7 @@ config COMPACTION
> config MIGRATION
> bool "Page migration"
> def_bool y
> - depends on NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION
> + depends on NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA
> help
> Allows the migration of the physical location of pages of processes
> while the virtual addresses are not changed. This is useful in
> diff --git a/mm/compaction.c b/mm/compaction.c
> index d5174c4..a6e7c64 100644
> --- a/mm/compaction.c
> +++ b/mm/compaction.c
> @@ -45,6 +45,11 @@ static void map_pages(struct list_head *list)
> }
> }
>
> +static inline bool migrate_async_suitable(int migratetype)
Just nitpick, since the helper is not directly related to what async means,
how about migrate_suitable(int migrate_type) ?
> +{
> + return is_migrate_cma(migratetype) || migratetype == MIGRATE_MOVABLE;
> +}
> +
> /*
> * Isolate free pages onto a private freelist. Caller must hold zone->lock.
> * If @strict is true, will abort returning 0 on any invalid PFNs or non-free
> @@ -277,7 +282,7 @@ isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
> */
> pageblock_nr = low_pfn >> pageblock_order;
> if (!cc->sync && last_pageblock_nr != pageblock_nr &&
> - get_pageblock_migratetype(page) != MIGRATE_MOVABLE) {
> + migrate_async_suitable(get_pageblock_migratetype(page))) {
Here compaction looks corrupted if CMA not enabled, Mel?
btw, Kame-san is not Cced correctly 8;/
Hillf
> On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
> <[email protected]> wrote:
>> From: Michal Nazarewicz <[email protected]>
>> diff --git a/mm/compaction.c b/mm/compaction.c
>> index d5174c4..a6e7c64 100644
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -45,6 +45,11 @@ static void map_pages(struct list_head *list)
>> }
>> }
>>
>> +static inline bool migrate_async_suitable(int migratetype)
On Fri, 03 Feb 2012 15:19:54 +0100, Hillf Danton <[email protected]> wrote:
> Just nitpick, since the helper is not directly related to what async means,
> how about migrate_suitable(int migrate_type) ?
I feel current name is better suited since it says that it's OK to scan this
block if it's an asynchronous compaction run.
>> +{
>> + return is_migrate_cma(migratetype) || migratetype == MIGRATE_MOVABLE;
>> +}
>> +
>> /*
>> * Isolate free pages onto a private freelist. Caller must hold zone->lock.
>> * If @strict is true, will abort returning 0 on any invalid PFNs or non-free
>> @@ -277,7 +282,7 @@ isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
>> */
>> pageblock_nr = low_pfn >> pageblock_order;
>> if (!cc->sync && last_pageblock_nr != pageblock_nr &&
>> - get_pageblock_migratetype(page) != MIGRATE_MOVABLE) {
>> + migrate_async_suitable(get_pageblock_migratetype(page))) {
>
> Here compaction looks corrupted if CMA not enabled, Mel?
Damn, yes, this should be !migrate_async_suitable(...). Sorry about that.
--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
2012/2/3 Michal Nazarewicz <[email protected]>:
>>> +static inline bool migrate_async_suitable(int migratetype)
>
> On Fri, 03 Feb 2012 15:19:54 +0100, Hillf Danton <[email protected]> wrote:
>>
>> Just nitpick, since the helper is not directly related to what async
>> means,
>> how about migrate_suitable(int migrate_type) ?
>
>
> I feel current name is better suited since it says that it's OK to scan this
> block if it's an asynchronous compaction run.
>
The input is the migrate type of page considered, and the async is only one
of the modes that compaction should be carried out. Plus the helper is
also used in other cases where async is entirely not concerned.
That said, the naming is not clear, if not misleading.
On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
<[email protected]> wrote:
> The Contiguous Memory Allocator is a set of helper functions for DMA
> mapping framework that improves allocations of contiguous memory chunks.
>
> CMA grabs memory on system boot, marks it with CMA_MIGRATE_TYPE and
> gives back to the system. Kernel is allowed to allocate movable pages
> within CMA's managed memory so that it can be used for example for page
> cache when DMA mapping do not use it. On dma_alloc_from_contiguous()
> request such pages are migrated out of CMA area to free required
> contiguous block and fulfill the request. This allows to allocate large
> contiguous chunks of memory at any time assuming that there is enough
> free memory available in the system.
>
> This code is heavily based on earlier works by Michal Nazarewicz.
>
> Signed-off-by: Marek Szyprowski <[email protected]>
> Signed-off-by: Kyungmin Park <[email protected]>
> CC: Michal Nazarewicz <[email protected]>
> Acked-by: Arnd Bergmann <[email protected]>
> Tested-by: Rob Clark <[email protected]>
> Tested-by: Ohad Ben-Cohen <[email protected]>
> Tested-by: Benjamin Gaignard <[email protected]>
> ---
[...]
> +/*
> + * Contiguous Memory Allocator
> + *
> + * The Contiguous Memory Allocator (CMA) makes it possible to
> + * allocate big contiguous chunks of memory after the system has
> + * booted.
> + *
> + * Why is it needed?
> + *
> + * Various devices on embedded systems have no scatter-getter and/or
> + * IO map support and require contiguous blocks of memory to
> + * operate. They include devices such as cameras, hardware video
> + * coders, etc.
> + *
> + * Such devices often require big memory buffers (a full HD frame
> + * is, for instance, more then 2 mega pixels large, i.e. more than 6
> + * MB of memory), which makes mechanisms such as kmalloc() or
> + * alloc_page() ineffective.
> + *
> + * At the same time, a solution where a big memory region is
> + * reserved for a device is suboptimal since often more memory is
> + * reserved then strictly required and, moreover, the memory is
> + * inaccessible to page system even if device drivers don't use it.
> + *
> + * CMA tries to solve this issue by operating on memory regions
> + * where only movable pages can be allocated from. This way, kernel
> + * can use the memory for pagecache and when device driver requests
> + * it, allocated pages can be migrated.
> + *
Without boot mem reservation, what is the successful rate of CMA to
serve requests of 1MiB, 2MiB, 4MiB and 8MiB chunks?
On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
<[email protected]> wrote:
> From: Michal Nazarewicz <[email protected]>
>
> This commit exports some of the functions from compaction.c file
> outside of it adding their declaration into internal.h header
> file so that other mm related code can use them.
>
> This forced compaction.c to always be compiled (as opposed to being
> compiled only if CONFIG_COMPACTION is defined) but as to avoid
> introducing code that user did not ask for, part of the compaction.c
> is now wrapped in on #ifdef.
>
What if both compaction and CMA are not enabled?
Good weekend
Hillf
On Sun, 05 Feb 2012 05:25:40 +0100, Hillf Danton <[email protected]> wrote:
> Without boot mem reservation, what is the successful rate of CMA to
> serve requests of 1MiB, 2MiB, 4MiB and 8MiB chunks?
CMA will work as long as you manage to get some pageblocks marked as
MIGRATE_CMA and move all non-movable pages away. You might try and get it
done after system has booted but we have not tried nor tested it.
Reservation at boot time lets us make sure that the portion of memory we
are grabbing has no unmovable pages.
You might still and use alloc_contig_pages() on its own (even without
MIGRATE_CMA) but that would require additional code which would look for
a region of memory that could be used (ie. that does not have unmovable
pages in it). That in fact was what Kamezawa's code was doing.
--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
> On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
> <[email protected]> wrote:
>> From: Michal Nazarewicz <[email protected]>
>>
>> This commit exports some of the functions from compaction.c file
>> outside of it adding their declaration into internal.h header
>> file so that other mm related code can use them.
>>
>> This forced compaction.c to always be compiled (as opposed to being
>> compiled only if CONFIG_COMPACTION is defined) but as to avoid
>> introducing code that user did not ask for, part of the compaction.c
>> is now wrapped in on #ifdef.
On Sun, 05 Feb 2012 08:40:08 +0100, Hillf Danton <[email protected]> wrote:
> What if both compaction and CMA are not enabled?
What about it? If both are enabled, both will be compiled and usable.
--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
>>>> +static inline bool migrate_async_suitable(int migratetype)
>> On Fri, 03 Feb 2012 15:19:54 +0100, Hillf Danton <[email protected]> wrote:
>>> Just nitpick, since the helper is not directly related to what async
>>> means, how about migrate_suitable(int migrate_type) ?
> 2012/2/3 Michal Nazarewicz <[email protected]>:
>> I feel current name is better suited since it says that it's OK to scan this
>> block if it's an asynchronous compaction run.
On Sat, 04 Feb 2012 10:09:02 +0100, Hillf Danton <[email protected]> wrote:
> The input is the migrate type of page considered, and the async is only one
> of the modes that compaction should be carried out. Plus the helper is
> also used in other cases where async is entirely not concerned.
>
> That said, the naming is not clear, if not misleading.
In the first version the function was called is_migrate_cma_or_movable() which
described what the function checked. Mel did not like it though, hence the
change to migrate_async_suitable(). Honestly, I'm not sure what would be the
best name for function.
--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
2012/2/5 Michal Nazarewicz <[email protected]>:
>> On Fri, Feb 3, 2012 at 8:18 PM, Marek Szyprowski
>> <[email protected]> wrote:
>>>
>>> From: Michal Nazarewicz <[email protected]>
>>>
>>> This commit exports some of the functions from compaction.c file
>>> outside of it adding their declaration into internal.h header
>>> file so that other mm related code can use them.
>>>
>>> This forced compaction.c to always be compiled (as opposed to being
>>> compiled only if CONFIG_COMPACTION is defined) but as to avoid
>>> introducing code that user did not ask for, part of the compaction.c
>>> is now wrapped in on #ifdef.
>
>
> On Sun, 05 Feb 2012 08:40:08 +0100, Hillf Danton <[email protected]> wrote:
>>
>> What if both compaction and CMA are not enabled?
>
>
> What about it? If both are enabled, both will be compiled and usable.
>
Better if enforced compilation of compaction is addressed in separate
patch in the patchset, according to the rule that one patch is delivered
with one issue concerned.
2012/2/5 Michal Nazarewicz <[email protected]>:
> On Sun, 05 Feb 2012 05:25:40 +0100, Hillf Danton <[email protected]> wrote:
>>
>> Without boot mem reservation, what is the successful rate of CMA to
>> serve requests of 1MiB, 2MiB, 4MiB and 8MiB chunks?
>
>
> CMA will work as long as you manage to get some pageblocks marked as
> MIGRATE_CMA and move all non-movable pages away. You might try and get it
> done after system has booted but we have not tried nor tested it.
Better to include whatever test results in change log.
And no more questions ;)
Hello everyone!
I'm recently learning CMA, and I want to implement support for
contiguous memory areas placed in HIGHMEM zone,do you have any suggestions?
--
Regards,
cary.zou
Hello,
On Tuesday, February 07, 2012 10:06 AM cp.zou wrote:
> Hello everyone!
>
> I'm recently learning CMA, and I want to implement support for
> contiguous memory areas placed in HIGHMEM zone,do you have any suggestions?
CMA memory management core (migration and allocation of pages) should support
areas placed in HIGHMEM without any additional works. The main limitation is
in the DMA-mapping framework. Right now it makes certain assumptions about page
mappings to simplify allocation process. You need to add support for dynamic
mappings there.
Best regards
--
Marek Szyprowski
Samsung Poland R&D Center
On Fri, Feb 3, 2012 at 6:04 AM, Mel Gorman <[email protected]> wrote:
> On Fri, Feb 03, 2012 at 01:18:54PM +0100, Marek Szyprowski wrote:
>
> Nothing prevents two or more processes updating the wmarks at the same
> time which is racy and unpredictable. Today it is not much of a problem
> but CMA makes this path hotter than it was and you may see weirdness
> if two processes are updating zonelists at the same time. Swap-over-NFS
> actually starts with a patch that serialises setup_per_zone_wmarks()
>
> You also potentially have a BIG problem here if this happens
>
> min_free_kbytes = 32768
> Process a: min_free_kbytes ?+= 65536
> Process a: start direct reclaim
> echo 16374 > /proc/sys/vm/min_free_kbytes
> Process a: exit direct_reclaim
> Process a: min_free_kbytes -= 65536
>
> min_free_kbytes now wraps negative and the machine hangs.
>
There's another problem I am facing with zone watermarks and CMA.
Test details:
Memory : 480 MB of total memory, 128 MB CMA region
Test case : around 600 MB of file transfer over USB RNDIS onto target
System Load : ftpd with console running on target.
No one is doing CMA allocations except for the DMA allocations done by the
drivers.
Result : After about 300MB transfer, I start getting GFP_ATOMIC
allocation failures.
This only happens if CMA region is reserved.
Here's the free_list before I start the test
Free pages count per migrate type at order 0 1 2
3 4 5 6 7 8 9 10
Node 0, zone Normal, type Unmovable 2 9 6
7 3 3 3 4 2 1 0
Node 0, zone Normal, type Reclaimable 31 4 1
2 1 1 0 1 1 0 0
Node 0, zone Normal, type Movable 22 20 23
14 3 4 4 3 1 0 70
Node 0, zone Normal, type Reserve 0 0 0
0 0 0 0 0 0 0 1
Node 0, zone Normal, type CMA 2 0 0
2 1 1 1 1 1 1 34
Node 0, zone Normal, type Isolate 0 0 0
0 0 0 0 0 0 0 0
and here's what I get when I print the same when allocation fails.
Normal: Free pages count per migrate type at order 0 1
2 3 4 5 6 7 8 9 10
[ 401.887634] zone Normal, type Unmovable
0 0 0 0 0 0 0 0 0 0
0
[ 401.901916] zone Normal, type Reclaimable
0 0 0 0 0 0 0 0 0 0
0
[ 401.916229] zone Normal, type Movable
0 0 0 0 0 0 0 0 0 0
0
[ 401.930541] zone Normal, type Reserve
0 0 0 0 0 0 0 0 0 0
0
[ 401.944824] zone Normal, type CMA
6582 6580 2380 0 0 0 0 0 0 0
0
[ 401.961486] zone Normal, type Isolate
0 0 0 0 0 0 0 0 0 0
0
Total memory available is way above the zone watermarks. So, we ended
up starving
UNMOVABLE/RECLAIMABLE atomic allocations that cannot fallback on CMA region.
I know the CMA region is big, but I think reducing the region size
will only delay the problem.
it walso on't recover as long as most of the CMA region pages get
allocated and the zone
watermark is hit
To check my theory, I changed __zone_watermark_ok() to ignore free CMA pages
With this change, the transfer succeeds w/o any failures.
The patch does make things slow of course. Ideally, I would have liked
to do this only if
the watermark is being checked for non-Movable allocations, but I couldn't find
an easy way to do that.
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 371a79f..b672d97 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1580,6 +1580,21 @@ static bool __zone_watermark_ok(struct zone *z,
int order, unsigned long mark,
if (free_pages <= min)
return false;
}
+
+#ifdef CONFIG_CMA
+ /* If cma is enabled, ignore free pages from MIGRATE_CMA list
+ * for watermark checks
+ */
+ for (o = order; o < MAX_ORDER; o++) {
+ struct list_head *curr;
+ list_for_each(curr, &z->free_area[o].free_list[MIGRATE_CMA]) {
+ free_pages -= (1 << o);
+ if (free_pages <= min)
+ return false;
+ }
+ }
+#endif
+
return true;
}
Sandeep
On Wed, 08 Feb 2012 03:04:18 +0100, sandeep patil <[email protected]> wrote:
> There's another problem I am facing with zone watermarks and CMA.
>
> Test details:
> Memory : 480 MB of total memory, 128 MB CMA region
> Test case : around 600 MB of file transfer over USB RNDIS onto target
> System Load : ftpd with console running on target.
> No one is doing CMA allocations except for the DMA allocations done by the
> drivers.
>
> Result : After about 300MB transfer, I start getting GFP_ATOMIC
> allocation failures. This only happens if CMA region is reserved.
[...]
> Total memory available is way above the zone watermarks. So, we ended
> up starving
> UNMOVABLE/RECLAIMABLE atomic allocations that cannot fallback on CMA region.
This looks like something Mel warned me about. I don't really have a good
solution for that yet. ;/
--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +----<email/xmpp: [email protected]>--------------ooO--(_)--Ooo--
Hello,
On Friday, February 03, 2012 3:04 PM Mel Gorman wrote:
> On Fri, Feb 03, 2012 at 01:18:54PM +0100, Marek Szyprowski wrote:
> > alloc_contig_range() performs memory allocation so it also should keep
> > track on keeping the correct level of memory watermarks. This commit adds
> > a call to *_slowpath style reclaim to grab enough pages to make sure that
> > the final collection of contiguous pages from freelists will not starve
> > the system.
> >
> > Signed-off-by: Marek Szyprowski <[email protected]>
> > Signed-off-by: Kyungmin Park <[email protected]>
> > CC: Michal Nazarewicz <[email protected]>
> > Tested-by: Rob Clark <[email protected]>
> > Tested-by: Ohad Ben-Cohen <[email protected]>
> > Tested-by: Benjamin Gaignard <[email protected]>
>
> I still do not intend to ack this patch and any damage is confined to
> CMA but I have a few comments anyway.
>
> > ---
> > mm/page_alloc.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
> > 1 files changed, 47 insertions(+), 0 deletions(-)
> >
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index 983ccba..371a79f 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -5632,6 +5632,46 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned
> long end)
> > return ret > 0 ? 0 : ret;
> > }
> >
> > +/*
> > + * Trigger memory pressure bump to reclaim some pages in order to be able to
> > + * allocate 'count' pages in single page units. Does similar work as
> > + *__alloc_pages_slowpath() function.
> > + */
> > +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
> > +{
> > + enum zone_type high_zoneidx = gfp_zone(gfp_mask);
> > + struct zonelist *zonelist = node_zonelist(0, gfp_mask);
> > + int did_some_progress = 0;
> > + int order = 1;
> > + unsigned long watermark;
> > +
> > + /*
> > + * Increase level of watermarks to force kswapd do his job
> > + * to stabilize at new watermark level.
> > + */
> > + min_free_kbytes += count * PAGE_SIZE / 1024;
>
> There is a risk of overflow here although it is incredibly
> small. Still, a potentially nicer way of doing this was
>
> count << (PAGE_SHIFT - 10)
>
> > + setup_per_zone_wmarks();
> > +
>
> Nothing prevents two or more processes updating the wmarks at the same
> time which is racy and unpredictable. Today it is not much of a problem
> but CMA makes this path hotter than it was and you may see weirdness
> if two processes are updating zonelists at the same time. Swap-over-NFS
> actually starts with a patch that serialises setup_per_zone_wmarks()
>
> You also potentially have a BIG problem here if this happens
>
> min_free_kbytes = 32768
> Process a: min_free_kbytes += 65536
> Process a: start direct reclaim
> echo 16374 > /proc/sys/vm/min_free_kbytes
> Process a: exit direct_reclaim
> Process a: min_free_kbytes -= 65536
>
> min_free_kbytes now wraps negative and the machine hangs.
>
> The damage is confined to CMA though so I am not going to lose sleep
> over it but you might want to consider at least preventing parallel
> updates to min_free_kbytes from proc.
Right. This approach was definitely too hacky. What do you think about replacing
it with the following code (I assume that setup_per_zone_wmarks() serialization
patch will be merged anyway so I skipped it here):
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 82f4fa5..bb9ae41 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -371,6 +371,13 @@ struct zone {
/* see spanned/present_pages for more description */
seqlock_t span_seqlock;
#endif
+#ifdef CONFIG_CMA
+ /*
+ * CMA needs to increase watermark levels during the allocation
+ * process to make sure that the system is not starved.
+ */
+ unsigned long min_cma_pages;
+#endif
struct free_area free_area[MAX_ORDER];
#ifndef CONFIG_SPARSEMEM
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 824fb37..1ca52f0 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5044,6 +5044,11 @@ void setup_per_zone_wmarks(void)
zone->watermark[WMARK_LOW] = min_wmark_pages(zone) + (tmp >> 2);
zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
+#ifdef CONFIG_CMA
+ zone->watermark[WMARK_MIN] += zone->min_cma_pages;
+ zone->watermark[WMARK_LOW] += zone->min_cma_pages;
+ zone->watermark[WMARK_HIGH] += zone->min_cma_pages;
+#endif
setup_zone_migrate_reserve(zone);
spin_unlock_irqrestore(&zone->lock, flags);
}
@@ -5625,13 +5630,15 @@ static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
struct zonelist *zonelist = node_zonelist(0, gfp_mask);
int did_some_progress = 0;
int order = 1;
- unsigned long watermark;
+ unsigned long watermark, flags;
/*
* Increase level of watermarks to force kswapd do his job
* to stabilize at new watermark level.
*/
- min_free_kbytes += count * PAGE_SIZE / 1024;
+ spin_lock_irqsave(&zone->lock, flags);
+ zone->min_cma_pages += count;
+ spin_unlock_irqrestore(&zone->lock, flags);
setup_per_zone_wmarks();
/* Obey watermarks as if the page was being allocated */
@@ -5648,7 +5655,9 @@ static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
}
/* Restore original watermark levels. */
- min_free_kbytes -= count * PAGE_SIZE / 1024;
+ spin_lock_irqsave(&zone->lock, flags);
+ zone->min_cma_pages -= count;
+ spin_unlock_irqrestore(&zone->lock, flags);
setup_per_zone_wmarks();
return count;
Best regards
--
Marek Szyprowski
Samsung Poland R&D Center
2012/2/8 Michal Nazarewicz <[email protected]>:
> On Wed, 08 Feb 2012 03:04:18 +0100, sandeep patil <[email protected]>
> wrote:
>>
>> There's another problem I am facing with zone watermarks and CMA.
>>
>> Test details:
>> Memory ?: 480 MB of total memory, 128 MB CMA region
>> Test case : around 600 MB of file transfer over USB RNDIS onto target
>> System Load : ftpd with console running on target.
>> No one is doing CMA allocations except for the DMA allocations done by the
>> drivers.
>>
>> Result : After about 300MB transfer, I start getting GFP_ATOMIC
>> allocation failures. ?This only happens if CMA region is reserved.
>> Total memory available is way above the zone watermarks. So, we ended
>> up starving
>> UNMOVABLE/RECLAIMABLE atomic allocations that cannot fallback on CMA
>> region.
>
>
> This looks like something Mel warned me about. ?I don't really have a good
> solution for that yet. ;/
What if we have NR_FREE_CMA_PAGES in vmstat and use them to calculate
__zone_watermark_ok()?
However, it still doesn't solve the problem when we DON'T want to use
NR_FREE_CMA_PAGES in case of MOVABLE allocations.
Sandeep
On Wed, Feb 08, 2012 at 04:14:46PM +0100, Marek Szyprowski wrote:
> > > <SNIP>
> > > +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
> > > +{
> > > + enum zone_type high_zoneidx = gfp_zone(gfp_mask);
> > > + struct zonelist *zonelist = node_zonelist(0, gfp_mask);
> > > + int did_some_progress = 0;
> > > + int order = 1;
> > > + unsigned long watermark;
> > > +
> > > + /*
> > > + * Increase level of watermarks to force kswapd do his job
> > > + * to stabilize at new watermark level.
> > > + */
> > > + min_free_kbytes += count * PAGE_SIZE / 1024;
> >
> > There is a risk of overflow here although it is incredibly
> > small. Still, a potentially nicer way of doing this was
> >
> > count << (PAGE_SHIFT - 10)
> >
> > > + setup_per_zone_wmarks();
> > > +
> >
> > Nothing prevents two or more processes updating the wmarks at the same
> > time which is racy and unpredictable. Today it is not much of a problem
> > but CMA makes this path hotter than it was and you may see weirdness
> > if two processes are updating zonelists at the same time. Swap-over-NFS
> > actually starts with a patch that serialises setup_per_zone_wmarks()
> >
> > You also potentially have a BIG problem here if this happens
> >
> > min_free_kbytes = 32768
> > Process a: min_free_kbytes += 65536
> > Process a: start direct reclaim
> > echo 16374 > /proc/sys/vm/min_free_kbytes
> > Process a: exit direct_reclaim
> > Process a: min_free_kbytes -= 65536
> >
> > min_free_kbytes now wraps negative and the machine hangs.
> >
> > The damage is confined to CMA though so I am not going to lose sleep
> > over it but you might want to consider at least preventing parallel
> > updates to min_free_kbytes from proc.
>
> Right. This approach was definitely too hacky. What do you think about replacing
> it with the following code (I assume that setup_per_zone_wmarks() serialization
> patch will be merged anyway so I skipped it here):
>
It's part of a larger series and the rest of that series is
controversial. That single patch can be split out obviously so feel free
to add it to your series and stick your Signed-off-by on the end of it.
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 82f4fa5..bb9ae41 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -371,6 +371,13 @@ struct zone {
> /* see spanned/present_pages for more description */
> seqlock_t span_seqlock;
> #endif
> +#ifdef CONFIG_CMA
> + /*
> + * CMA needs to increase watermark levels during the allocation
> + * process to make sure that the system is not starved.
> + */
> + unsigned long min_cma_pages;
> +#endif
> struct free_area free_area[MAX_ORDER];
>
> #ifndef CONFIG_SPARSEMEM
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 824fb37..1ca52f0 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5044,6 +5044,11 @@ void setup_per_zone_wmarks(void)
>
> zone->watermark[WMARK_LOW] = min_wmark_pages(zone) + (tmp >> 2);
> zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
> +#ifdef CONFIG_CMA
> + zone->watermark[WMARK_MIN] += zone->min_cma_pages;
> + zone->watermark[WMARK_LOW] += zone->min_cma_pages;
> + zone->watermark[WMARK_HIGH] += zone->min_cma_pages;
> +#endif
> setup_zone_migrate_reserve(zone);
> spin_unlock_irqrestore(&zone->lock, flags);
> }
This is better in that it is not vunerable to parallel updates of
min_free_kbytes. It would be slightly tidier to introduce something
like cma_wmark_pages() that returns min_cma_pages if CONFIG_CMA and 0
otherwise. Use the helper to get right of this ifdef CONFIG_CMA within
setup_per_zone_wmarks().
You'll still have the problem of kswapd not taking CMA pages properly into
account when deciding whether to reclaim or not though.
--
Mel Gorman
SUSE Labs
Hello,
On Friday, February 10, 2012 12:19 PM Mel Gorman wrote:
> On Wed, Feb 08, 2012 at 04:14:46PM +0100, Marek Szyprowski wrote:
> > > > <SNIP>
> > > > +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
> > > > +{
> > > > + enum zone_type high_zoneidx = gfp_zone(gfp_mask);
> > > > + struct zonelist *zonelist = node_zonelist(0, gfp_mask);
> > > > + int did_some_progress = 0;
> > > > + int order = 1;
> > > > + unsigned long watermark;
> > > > +
> > > > + /*
> > > > + * Increase level of watermarks to force kswapd do his job
> > > > + * to stabilize at new watermark level.
> > > > + */
> > > > + min_free_kbytes += count * PAGE_SIZE / 1024;
> > >
> > > There is a risk of overflow here although it is incredibly
> > > small. Still, a potentially nicer way of doing this was
> > >
> > > count << (PAGE_SHIFT - 10)
> > >
> > > > + setup_per_zone_wmarks();
> > > > +
> > >
> > > Nothing prevents two or more processes updating the wmarks at the same
> > > time which is racy and unpredictable. Today it is not much of a problem
> > > but CMA makes this path hotter than it was and you may see weirdness
> > > if two processes are updating zonelists at the same time. Swap-over-NFS
> > > actually starts with a patch that serialises setup_per_zone_wmarks()
> > >
> > > You also potentially have a BIG problem here if this happens
> > >
> > > min_free_kbytes = 32768
> > > Process a: min_free_kbytes += 65536
> > > Process a: start direct reclaim
> > > echo 16374 > /proc/sys/vm/min_free_kbytes
> > > Process a: exit direct_reclaim
> > > Process a: min_free_kbytes -= 65536
> > >
> > > min_free_kbytes now wraps negative and the machine hangs.
> > >
> > > The damage is confined to CMA though so I am not going to lose sleep
> > > over it but you might want to consider at least preventing parallel
> > > updates to min_free_kbytes from proc.
> >
> > Right. This approach was definitely too hacky. What do you think about replacing
> > it with the following code (I assume that setup_per_zone_wmarks() serialization
> > patch will be merged anyway so I skipped it here):
> >
>
> It's part of a larger series and the rest of that series is
> controversial. That single patch can be split out obviously so feel free
> to add it to your series and stick your Signed-off-by on the end of it.
Ok, I will grab it from that series.
> > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> > index 82f4fa5..bb9ae41 100644
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -371,6 +371,13 @@ struct zone {
> > /* see spanned/present_pages for more description */
> > seqlock_t span_seqlock;
> > #endif
> > +#ifdef CONFIG_CMA
> > + /*
> > + * CMA needs to increase watermark levels during the allocation
> > + * process to make sure that the system is not starved.
> > + */
> > + unsigned long min_cma_pages;
> > +#endif
> > struct free_area free_area[MAX_ORDER];
> >
> > #ifndef CONFIG_SPARSEMEM
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index 824fb37..1ca52f0 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -5044,6 +5044,11 @@ void setup_per_zone_wmarks(void)
> >
> > zone->watermark[WMARK_LOW] = min_wmark_pages(zone) + (tmp >> 2);
> > zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
> > +#ifdef CONFIG_CMA
> > + zone->watermark[WMARK_MIN] += zone->min_cma_pages;
> > + zone->watermark[WMARK_LOW] += zone->min_cma_pages;
> > + zone->watermark[WMARK_HIGH] += zone->min_cma_pages;
> > +#endif
> > setup_zone_migrate_reserve(zone);
> > spin_unlock_irqrestore(&zone->lock, flags);
> > }
>
> This is better in that it is not vunerable to parallel updates of
> min_free_kbytes. It would be slightly tidier to introduce something
> like cma_wmark_pages() that returns min_cma_pages if CONFIG_CMA and 0
> otherwise. Use the helper to get right of this ifdef CONFIG_CMA within
> setup_per_zone_wmarks().
Thanks for the hint. I will submit this change soon.
> You'll still have the problem of kswapd not taking CMA pages properly into
> account when deciding whether to reclaim or not though.
This will only affect CMA clients, I hope we can fix this later once the patches
get merged. I would really like to get it into v3.4, what means that it should be
tested in next asap.
Best regards
--
Marek Szyprowski
Samsung Poland R&D Center