diff --git a/board/raspberrypi/rpi/rpi.c b/board/raspberrypi/rpi/rpi.c
index d21750e2a0434d262a139681f805adc511abe847..6d7be11518ebf7131faac6f71386417b3ee7fcaa 100644
--- a/board/raspberrypi/rpi/rpi.c
+++ b/board/raspberrypi/rpi/rpi.c
@@ -10,6 +10,7 @@
#include <fdt_support.h>
#include <fdt_simplefb.h>
#include <lcd.h>
+#include <memalign.h>
#include <mmc.h>
#include <asm/gpio.h>
#include <asm/arch/mbox.h>
diff --git a/board/samsung/common/misc.c b/board/samsung/common/misc.c
index f0d69d4a48f0472e705f964b2a4522b31e879c95..e0e2c48632cf37907170f1888786a2f21361c5a1 100644
--- a/board/samsung/common/misc.c
+++ b/board/samsung/common/misc.c
@@ -12,6 +12,7 @@
#include <errno.h>
#include <version.h>
#include <malloc.h>
+#include <memalign.h>
#include <linux/sizes.h>
#include <asm/arch/cpu.h>
#include <asm/gpio.h>
diff --git a/common/cmd_usb.c b/common/cmd_usb.c
index 6874af7547148276f8b9564aa1e4b4e01ba8c7e1..6bdbbc5c05286e765abd1a901ef675499926d4ca 100644
--- a/common/cmd_usb.c
+++ b/common/cmd_usb.c
@@ -14,6 +14,7 @@
#include <common.h>
#include <command.h>
#include <dm.h>
+#include <memalign.h>
#include <asm/byteorder.h>
#include <asm/unaligned.h>
#include <part.h>
diff --git a/common/env_fat.c b/common/env_fat.c
index e4c848935ad19d7cac64d07102e7ebe39841b58b..d79d864a0c1216ee0e66a216b94d80a59e62de28 100644
--- a/common/env_fat.c
+++ b/common/env_fat.c
@@ -13,6 +13,7 @@
#include <environment.h>
#include <linux/stddef.h>
#include <malloc.h>
+#include <memalign.h>
#include <search.h>
#include <errno.h>
#include <fat.h>
diff --git a/common/env_mmc.c b/common/env_mmc.c
index 51e7707cb0f7b0bcc7ca9f377e4f42ca97255a9e..96398224cc849577a4150edf281337070e06d94b 100644
--- a/common/env_mmc.c
+++ b/common/env_mmc.c
@@ -12,6 +12,7 @@
#include <environment.h>
#include <linux/stddef.h>
#include <malloc.h>
+#include <memalign.h>
#include <mmc.h>
#include <search.h>
#include <errno.h>
diff --git a/common/env_nand.c b/common/env_nand.c
index 92e0e053dfe6b57abe49156bc246e780089a4af0..b32eeac9d7673c0137c3554dfd378f9ed2449bf4 100644
--- a/common/env_nand.c
+++ b/common/env_nand.c
@@ -19,6 +19,7 @@
#include <environment.h>
#include <linux/stddef.h>
#include <malloc.h>
+#include <memalign.h>
#include <nand.h>
#include <search.h>
#include <errno.h>
diff --git a/common/env_ubi.c b/common/env_ubi.c
index 77bbfa6ef43caf051ba5166dd8045944e6e7a821..e0dc5af8513e60d680080593a2a2fa91797964f0 100644
--- a/common/env_ubi.c
+++ b/common/env_ubi.c
@@ -11,6 +11,7 @@
#include <environment.h>
#include <errno.h>
#include <malloc.h>
+#include <memalign.h>
#include <search.h>
#include <ubi_uboot.h>
#undef crc32
diff --git a/common/usb.c b/common/usb.c
index fbaf8ecbe232ce63170228727df1ed7fa23b060f..700bfc315b67f9f58276ccdcb8a88006ac71321f 100644
--- a/common/usb.c
+++ b/common/usb.c
@@ -29,6 +29,7 @@
#include <common.h>
#include <command.h>
#include <dm.h>
+#include <memalign.h>
#include <asm/processor.h>
#include <linux/compiler.h>
#include <linux/ctype.h>
diff --git a/common/usb_hub.c b/common/usb_hub.c
index 652a104361f63ee715a99e924ccd1145a2c66224..415b45c1f139451c685fd450a6477bde34859cab 100644
--- a/common/usb_hub.c
+++ b/common/usb_hub.c
@@ -26,6 +26,7 @@
#include <command.h>
#include <dm.h>
#include <errno.h>
+#include <memalign.h>
#include <asm/processor.h>
#include <asm/unaligned.h>
#include <linux/ctype.h>
diff --git a/common/usb_kbd.c b/common/usb_kbd.c
index 0227024441710041a4cbcb2fd5c30920ad9078a6..95912f99c767ff70265d33e99ce336845ba67a35 100644
--- a/common/usb_kbd.c
+++ b/common/usb_kbd.c
@@ -11,6 +11,7 @@
#include <dm.h>
#include <errno.h>
#include <malloc.h>
+#include <memalign.h>
#include <stdio_dev.h>
#include <asm/byteorder.h>
diff --git a/common/usb_storage.c b/common/usb_storage.c
index b9784304086ecdc7ec7beaaeedb853af513824de..b390310ee35b554f3d152f39fbd1d9e77cc7132a 100644
--- a/common/usb_storage.c
+++ b/common/usb_storage.c
@@ -39,6 +39,7 @@
#include <errno.h>
#include <inttypes.h>
#include <mapmem.h>
+#include <memalign.h>
#include <asm/byteorder.h>
#include <asm/processor.h>
#include <dm/device-internal.h>
diff --git a/disk/part_dos.c b/disk/part_dos.c
index cf1a36ebb826699e9b391f6d68a9fad0c5806430..89263d38aaa03c609f4aa1ec3da83913832dee3d 100644
--- a/disk/part_dos.c
+++ b/disk/part_dos.c
@@ -17,6 +17,7 @@
#include <common.h>
#include <command.h>
#include <ide.h>
+#include <memalign.h>
#include "part_dos.h"
#ifdef HAVE_BLOCK_DEVICE
diff --git a/disk/part_efi.c b/disk/part_efi.c
index 5856f9321118da1be38015d4bce89e8e498f1683..15627f29e8bd4ad33ba1688ad846d131e9c0cb8c 100644
--- a/disk/part_efi.c
+++ b/disk/part_efi.c
@@ -16,6 +16,7 @@
#include <ide.h>
#include <inttypes.h>
#include <malloc.h>
+#include <memalign.h>
#include <part_efi.h>
#include <linux/ctype.h>
diff --git a/disk/part_mac.c b/disk/part_mac.c
index cbcb995c48c715b67ff4ab33826113c3f3d8daac..099e0a0035be07303ac6b69a646ab59ba6260819 100644
--- a/disk/part_mac.c
+++ b/disk/part_mac.c
@@ -15,6 +15,7 @@
#include <common.h>
#include <command.h>
+#include <memalign.h>
#include <ide.h>
#include "part_mac.h"
diff --git a/drivers/block/ahci.c b/drivers/block/ahci.c
index 0d19dd25a3441c95d1e159dfeacacdd4a2960cc3..82c68439f8630d0b6342676bf43910f33d5847b1 100644
--- a/drivers/block/ahci.c
+++ b/drivers/block/ahci.c
@@ -15,6 +15,7 @@
#include <asm/errno.h>
#include <asm/io.h>
#include <malloc.h>
+#include <memalign.h>
#include <scsi.h>
#include <libata.h>
#include <linux/ctype.h>
diff --git a/drivers/mmc/dw_mmc.c b/drivers/mmc/dw_mmc.c
index 1117fedefeb19a6fac62e8792db1b0c96136c4ff..a84c1e16d8b706332c86db4a439227d3601027cb 100644
--- a/drivers/mmc/dw_mmc.c
+++ b/drivers/mmc/dw_mmc.c
@@ -10,6 +10,7 @@
#include <common.h>
#include <errno.h>
#include <malloc.h>
+#include <memalign.h>
#include <mmc.h>
#include <dwmmc.h>
#include <asm-generic/errno.h>
diff --git a/drivers/mmc/mmc.c b/drivers/mmc/mmc.c
index 371c1ec21230ba34ae42c822369628b0810bb6c6..2a58702848b7f977bd3d10b12bd9138287fc1df5 100644
--- a/drivers/mmc/mmc.c
+++ b/drivers/mmc/mmc.c
@@ -16,6 +16,7 @@
#include <mmc.h>
#include <part.h>
#include <malloc.h>
+#include <memalign.h>
#include <linux/list.h>
#include <div64.h>
#include "mmc_private.h"
diff --git a/drivers/mmc/rpmb.c b/drivers/mmc/rpmb.c
index 9d0b8bc0c8f9b41fd91185729e87c3777e8c7a23..1c6888fc48f9370a2befa194743ea69d3be6926b 100644
--- a/drivers/mmc/rpmb.c
+++ b/drivers/mmc/rpmb.c
@@ -10,6 +10,7 @@
#include <config.h>
#include <common.h>
+#include <memalign.h>
#include <mmc.h>
#include <u-boot/sha256.h>
#include "mmc_private.h"
diff --git a/drivers/mtd/nand/tegra_nand.c b/drivers/mtd/nand/tegra_nand.c
index debad4f791f97ee81c172b243f14d204a07b666d..a77db7b65dfa7d8519f555ba19845431c582c223 100644
--- a/drivers/mtd/nand/tegra_nand.c
+++ b/drivers/mtd/nand/tegra_nand.c
@@ -9,6 +9,7 @@
#include <common.h>
#include <asm/io.h>
+#include <memalign.h>
#include <nand.h>
#include <asm/arch/clock.h>
#include <asm/arch/funcmux.h>
diff --git a/drivers/net/e1000.c b/drivers/net/e1000.c
index 7b830ff8c0515a66f763ff4c4270965f39a2e65a..2ba03ed73e45e40b177221cc47483465dba7a3cc 100644
--- a/drivers/net/e1000.c
+++ b/drivers/net/e1000.c
@@ -32,6 +32,7 @@ tested on both gig copper and gig fiber boards
#include <common.h>
#include <dm.h>
#include <errno.h>
+#include <memalign.h>
#include <pci.h>
#include "e1000.h"
diff --git a/drivers/net/fec_mxc.c b/drivers/net/fec_mxc.c
index bff5fd111971dc7e7c50f1ae3801d8b8eefe94ce..79f6737e8e9c9727c46591cfe543643c83b17d4e 100644
--- a/drivers/net/fec_mxc.c
+++ b/drivers/net/fec_mxc.c
@@ -10,6 +10,7 @@
#include <common.h>
#include <malloc.h>
+#include <memalign.h>
#include <net.h>
#include <netdev.h>
#include <miiphy.h>
diff --git a/drivers/net/rtl8169.c b/drivers/net/rtl8169.c
index 7b6e20f30fbfe3ee4f0413be7dce4f4f645be219..ebd46b27e5fdeda6e2e543b2e652802d745cd1ec 100644
--- a/drivers/net/rtl8169.c
+++ b/drivers/net/rtl8169.c
@@ -44,6 +44,7 @@
#include <dm.h>
#include <errno.h>
#include <malloc.h>
+#include <memalign.h>
#include <net.h>
#ifndef CONFIG_DM_ETH
#include <netdev.h>
diff --git a/drivers/spi/mxs_spi.c b/drivers/spi/mxs_spi.c
index 2b9f395a97821568dc12c5e28feaaf42b42ffd50..627644b56bdf13f94fbdbae4d10f80cbbe06f0d1 100644
--- a/drivers/spi/mxs_spi.c
+++ b/drivers/spi/mxs_spi.c
@@ -12,6 +12,7 @@
#include <common.h>
#include <malloc.h>
+#include <memalign.h>
#include <spi.h>
#include <asm/errno.h>
#include <asm/io.h>
diff --git a/drivers/usb/eth/asix.c b/drivers/usb/eth/asix.c
index 8a43e7c27d6fe263f1635206f30570ad92852ff8..ad083cf8ae4a28b4a6e56202f26dab0fcd56e592 100644
--- a/drivers/usb/eth/asix.c
+++ b/drivers/usb/eth/asix.c
@@ -10,6 +10,7 @@
#include <dm.h>
#include <usb.h>
#include <malloc.h>
+#include <memalign.h>
#include <linux/mii.h>
#include "usb_ether.h"
diff --git a/drivers/usb/eth/asix88179.c b/drivers/usb/eth/asix88179.c
index 94dfe85eff3aa8b89aa5b85163e742f46e35539c..cf4085d76572c986c268ce06f3ab97e8b30dab9b 100644
--- a/drivers/usb/eth/asix88179.c
+++ b/drivers/usb/eth/asix88179.c
@@ -12,6 +12,7 @@
#include <linux/mii.h>
#include "usb_ether.h"
#include <malloc.h>
+#include <memalign.h>
#include <errno.h>
/* ASIX AX88179 based USB 3.0 Ethernet Devices */
diff --git a/drivers/usb/eth/mcs7830.c b/drivers/usb/eth/mcs7830.c
index c1b708600e891e553884ddbe4d91910ffe0be2b4..bbdad8b79ae49eace0e24acd36baa6928873b1ab 100644
--- a/drivers/usb/eth/mcs7830.c
+++ b/drivers/usb/eth/mcs7830.c
@@ -14,6 +14,7 @@
#include <errno.h>
#include <linux/mii.h>
#include <malloc.h>
+#include <memalign.h>
#include <usb.h>
#include "usb_ether.h"
diff --git a/drivers/usb/eth/smsc95xx.c b/drivers/usb/eth/smsc95xx.c
index 1dcd088b8dfc4772ff2d621774aa04767fc13eff..dc8fa8891be6b95200d95ca696215bd04e36bdd8 100644
--- a/drivers/usb/eth/smsc95xx.c
+++ b/drivers/usb/eth/smsc95xx.c
@@ -11,6 +11,7 @@
#include <dm.h>
#include <errno.h>
#include <malloc.h>
+#include <memalign.h>
#include <usb.h>
#include <asm/unaligned.h>
#include <linux/mii.h>
diff --git a/drivers/usb/gadget/ether.c b/drivers/usb/gadget/ether.c
index 53f4672291f3d83a32cae43f4bd99caedba6c502..c5e35ee35077a37cca9a623c77c835dc75542bb2 100644
--- a/drivers/usb/gadget/ether.c
+++ b/drivers/usb/gadget/ether.c
@@ -17,6 +17,7 @@
#include <net.h>
#include <usb.h>
#include <malloc.h>
+#include <memalign.h>
#include <linux/ctype.h>
#include "gadget_chips.h"
diff --git a/drivers/usb/gadget/f_thor.c b/drivers/usb/gadget/f_thor.c
index 06139ee04d2123f15d44a7613094e168962acb95..ff1481ba372304f3f8f5af5be18ea2185a6d57b2 100644
--- a/drivers/usb/gadget/f_thor.c
+++ b/drivers/usb/gadget/f_thor.c
@@ -18,6 +18,7 @@
#include <errno.h>
#include <common.h>
#include <malloc.h>
+#include <memalign.h>
#include <version.h>
#include <linux/usb/ch9.h>
#include <linux/usb/gadget.h>
diff --git a/drivers/usb/host/dwc2.c b/drivers/usb/host/dwc2.c
index 85236aeb2a48bc135248a46a12f4a9ae2d4c0a20..541c0f968701ef9ed6d9e290b54bd8e927b51fce 100644
--- a/drivers/usb/host/dwc2.c
+++ b/drivers/usb/host/dwc2.c
@@ -10,6 +10,7 @@
#include <errno.h>
#include <usb.h>
#include <malloc.h>
+#include <memalign.h>
#include <phys2bus.h>
#include <usbroothubdes.h>
#include <asm/io.h>
diff --git a/drivers/usb/host/ehci-hcd.c b/drivers/usb/host/ehci-hcd.c
index 88b670b7f8a76511fdc7c127e685b4c99b004270..135ef7262278c67f9c39f171560cd29b338963d1 100644
--- a/drivers/usb/host/ehci-hcd.c
+++ b/drivers/usb/host/ehci-hcd.c
@@ -15,6 +15,7 @@
#include <usb.h>
#include <asm/io.h>
#include <malloc.h>
+#include <memalign.h>
#include <watchdog.h>
#include <linux/compiler.h>
diff --git a/drivers/usb/host/ohci-hcd.c b/drivers/usb/host/ohci-hcd.c
index 0ffd838db29e2535ba064f73b5add9973af8d240..9bde2b252c03783356a83ffbb6d1769e4fbdb11d 100644
--- a/drivers/usb/host/ohci-hcd.c
+++ b/drivers/usb/host/ohci-hcd.c
@@ -41,6 +41,7 @@
#endif
#include <malloc.h>
+#include <memalign.h>
#include <usb.h>
#include "ohci.h"
diff --git a/drivers/usb/host/usb-uclass.c b/drivers/usb/host/usb-uclass.c
index b17a7d762e03350fd01bb1225d59a5ce4f4a5701..c66ebb6678dee1d1340a7b9010fc4d96ae0c642c 100644
--- a/drivers/usb/host/usb-uclass.c
+++ b/drivers/usb/host/usb-uclass.c
@@ -10,6 +10,7 @@
#include <common.h>
#include <dm.h>
#include <errno.h>
+#include <memalign.h>
#include <usb.h>
#include <dm/device-internal.h>
#include <dm/lists.h>
diff --git a/drivers/video/bcm2835.c b/drivers/video/bcm2835.c
index 61d054dd8923d7423df19aa5ee13366c6e084d32..7867fe3895d00a763ff7b841554f46de1fd2ecd1 100644
--- a/drivers/video/bcm2835.c
+++ b/drivers/video/bcm2835.c
@@ -6,6 +6,7 @@
#include <common.h>
#include <lcd.h>
+#include <memalign.h>
#include <asm/arch/mbox.h>
#include <asm/global_data.h>
diff --git a/fs/ext4/dev.c b/fs/ext4/dev.c
index c77c02cdfce262cdb6910a4746b149f2b9f0d07f..20f52566f09c03bfeaa869855aa81649e109dc2c 100644
--- a/fs/ext4/dev.c
+++ b/fs/ext4/dev.c
@@ -25,6 +25,7 @@
#include <common.h>
#include <config.h>
+#include <memalign.h>
#include <ext4fs.h>
#include <ext_common.h>
#include "ext4_common.h"
diff --git a/fs/ext4/ext4_common.c b/fs/ext4/ext4_common.c
index cab5465b9d4f9e99158dbdc8cd2f704e090c9b4b..c77e22d477a22576f484cb03b8fc72a6d0ef0556 100644
--- a/fs/ext4/ext4_common.c
+++ b/fs/ext4/ext4_common.c
@@ -24,6 +24,7 @@
#include <ext4fs.h>
#include <inttypes.h>
#include <malloc.h>
+#include <memalign.h>
#include <stddef.h>
#include <linux/stat.h>
#include <linux/time.h>
diff --git a/fs/ext4/ext4_write.c b/fs/ext4/ext4_write.c
index fbc4c4b1cc1a48e2f458a0ef30df97b451d40675..d346c065cd8a75e7f9dc3d1440b6aad951b50ebf 100644
--- a/fs/ext4/ext4_write.c
+++ b/fs/ext4/ext4_write.c
@@ -23,6 +23,7 @@
#include <common.h>
+#include <memalign.h>
#include <linux/stat.h>
#include <div64.h>
#include "ext4_common.h"
diff --git a/fs/fat/fat.c b/fs/fat/fat.c
index a863644d648068749a5803276d82ed5feca17604..06dcbb70057488b46b4a2d85afb08bbfeb7e2b4c 100644
--- a/fs/fat/fat.c
+++ b/fs/fat/fat.c
@@ -16,6 +16,7 @@
#include <asm/byteorder.h>
#include <part.h>
#include <malloc.h>
+#include <memalign.h>
#include <linux/compiler.h>
#include <linux/ctype.h>
diff --git a/include/common.h b/include/common.h
index c48e5bc11bf5b7ffe1d6f4314a17aa431b247b9a..68b24d0778d4c2f64b2d608c14293387d130e5d3 100644
--- a/include/common.h
+++ b/include/common.h
@@ -973,93 +973,6 @@ int cpu_release(int nr, int argc, char * const argv[]);
#define ROUND(a,b) (((a) + (b) - 1) & ~((b) - 1))
-/*
- * ARCH_DMA_MINALIGN is defined in asm/cache.h for each architecture. It
- * is used to align DMA buffers.
- */
-#ifndef __ASSEMBLY__
-#include <asm/cache.h>
-#endif
-
-/*
- * The ALLOC_CACHE_ALIGN_BUFFER macro is used to allocate a buffer on the
- * stack that meets the minimum architecture alignment requirements for DMA.
- * Such a buffer is useful for DMA operations where flushing and invalidating
- * the cache before and after a read and/or write operation is required for
- * correct operations.
- *
- * When called the macro creates an array on the stack that is sized such
- * that:
- *
- * 1) The beginning of the array can be advanced enough to be aligned.
- *
- * 2) The size of the aligned portion of the array is a multiple of the minimum
- * architecture alignment required for DMA.
- *
- * 3) The aligned portion contains enough space for the original number of
- * elements requested.
- *
- * The macro then creates a pointer to the aligned portion of this array and
- * assigns to the pointer the address of the first element in the aligned
- * portion of the array.
- *
- * Calling the macro as:
- *
- * ALLOC_CACHE_ALIGN_BUFFER(uint32_t, buffer, 1024);
- *
- * Will result in something similar to saying:
- *
- * uint32_t buffer[1024];
- *
- * The following differences exist:
- *
- * 1) The resulting buffer is guaranteed to be aligned to the value of
- * ARCH_DMA_MINALIGN.
- *
- * 2) The buffer variable created by the macro is a pointer to the specified
- * type, and NOT an array of the specified type. This can be very important
- * if you want the address of the buffer, which you probably do, to pass it
- * to the DMA hardware. The value of &buffer is different in the two cases.
- * In the macro case it will be the address of the pointer, not the address
- * of the space reserved for the buffer. However, in the second case it
- * would be the address of the buffer. So if you are replacing hard coded
- * stack buffers with this macro you need to make sure you remove the & from
- * the locations where you are taking the address of the buffer.
- *
- * Note that the size parameter is the number of array elements to allocate,
- * not the number of bytes.
- *
- * This macro can not be used outside of function scope, or for the creation
- * of a function scoped static buffer. It can not be used to create a cache
- * line aligned global buffer.
- */
-#define PAD_COUNT(s, pad) (((s) - 1) / (pad) + 1)
-#define PAD_SIZE(s, pad) (PAD_COUNT(s, pad) * pad)
-#define ALLOC_ALIGN_BUFFER_PAD(type, name, size, align, pad) \
- char __##name[ROUND(PAD_SIZE((size) * sizeof(type), pad), align) \
- + (align - 1)]; \
- \
- type *name = (type *) ALIGN((uintptr_t)__##name, align)
-#define ALLOC_ALIGN_BUFFER(type, name, size, align) \
- ALLOC_ALIGN_BUFFER_PAD(type, name, size, align, 1)
-#define ALLOC_CACHE_ALIGN_BUFFER_PAD(type, name, size, pad) \
- ALLOC_ALIGN_BUFFER_PAD(type, name, size, ARCH_DMA_MINALIGN, pad)
-#define ALLOC_CACHE_ALIGN_BUFFER(type, name, size) \
- ALLOC_ALIGN_BUFFER(type, name, size, ARCH_DMA_MINALIGN)
-
-/*
- * DEFINE_CACHE_ALIGN_BUFFER() is similar to ALLOC_CACHE_ALIGN_BUFFER, but it's
- * purpose is to allow allocating aligned buffers outside of function scope.
- * Usage of this macro shall be avoided or used with extreme care!
- */
-#define DEFINE_ALIGN_BUFFER(type, name, size, align) \
- static char __##name[ALIGN(size * sizeof(type), align)] \
- __aligned(align); \
- \
- static type *name = (type *)__##name
-#define DEFINE_CACHE_ALIGN_BUFFER(type, name, size) \
- DEFINE_ALIGN_BUFFER(type, name, size, ARCH_DMA_MINALIGN)
-
/*
* check_member() - Check the offset of a structure member
*
diff --git a/include/memalign.h b/include/memalign.h
index f78b9dd9dc7c31b16a0a97d97b9170bd27924f7a..a96003908397bda3e6b78c4ca08bc1872cbe5da3 100644
--- a/include/memalign.h
+++ b/include/memalign.h
@@ -13,9 +13,100 @@
*/
#ifndef __ASSEMBLY__
#include <asm/cache.h>
-
#include <malloc.h>
+/*
+ * The ALLOC_CACHE_ALIGN_BUFFER macro is used to allocate a buffer on the
+ * stack that meets the minimum architecture alignment requirements for DMA.
+ * Such a buffer is useful for DMA operations where flushing and invalidating
+ * the cache before and after a read and/or write operation is required for
+ * correct operations.
+ *
+ * When called the macro creates an array on the stack that is sized such
+ * that:
+ *
+ * 1) The beginning of the array can be advanced enough to be aligned.
+ *
+ * 2) The size of the aligned portion of the array is a multiple of the minimum
+ * architecture alignment required for DMA.
+ *
+ * 3) The aligned portion contains enough space for the original number of
+ * elements requested.
+ *
+ * The macro then creates a pointer to the aligned portion of this array and
+ * assigns to the pointer the address of the first element in the aligned
+ * portion of the array.
+ *
+ * Calling the macro as:
+ *
+ * ALLOC_CACHE_ALIGN_BUFFER(uint32_t, buffer, 1024);
+ *
+ * Will result in something similar to saying:
+ *
+ * uint32_t buffer[1024];
+ *
+ * The following differences exist:
+ *
+ * 1) The resulting buffer is guaranteed to be aligned to the value of
+ * ARCH_DMA_MINALIGN.
+ *
+ * 2) The buffer variable created by the macro is a pointer to the specified
+ * type, and NOT an array of the specified type. This can be very important
+ * if you want the address of the buffer, which you probably do, to pass it
+ * to the DMA hardware. The value of &buffer is different in the two cases.
+ * In the macro case it will be the address of the pointer, not the address
+ * of the space reserved for the buffer. However, in the second case it
+ * would be the address of the buffer. So if you are replacing hard coded
+ * stack buffers with this macro you need to make sure you remove the & from
+ * the locations where you are taking the address of the buffer.
+ *
+ * Note that the size parameter is the number of array elements to allocate,
+ * not the number of bytes.
+ *
+ * This macro can not be used outside of function scope, or for the creation
+ * of a function scoped static buffer. It can not be used to create a cache
+ * line aligned global buffer.
+ */
+#define PAD_COUNT(s, pad) (((s) - 1) / (pad) + 1)
+#define PAD_SIZE(s, pad) (PAD_COUNT(s, pad) * pad)
+#define ALLOC_ALIGN_BUFFER_PAD(type, name, size, align, pad) \
+ char __##name[ROUND(PAD_SIZE((size) * sizeof(type), pad), align) \
+ + (align - 1)]; \
+ \
+ type *name = (type *)ALIGN((uintptr_t)__##name, align)
+#define ALLOC_ALIGN_BUFFER(type, name, size, align) \
+ ALLOC_ALIGN_BUFFER_PAD(type, name, size, align, 1)
+#define ALLOC_CACHE_ALIGN_BUFFER_PAD(type, name, size, pad) \
+ ALLOC_ALIGN_BUFFER_PAD(type, name, size, ARCH_DMA_MINALIGN, pad)
+#define ALLOC_CACHE_ALIGN_BUFFER(type, name, size) \
+ ALLOC_ALIGN_BUFFER(type, name, size, ARCH_DMA_MINALIGN)
+
+/*
+ * DEFINE_CACHE_ALIGN_BUFFER() is similar to ALLOC_CACHE_ALIGN_BUFFER, but it's
+ * purpose is to allow allocating aligned buffers outside of function scope.
+ * Usage of this macro shall be avoided or used with extreme care!
+ */
+#define DEFINE_ALIGN_BUFFER(type, name, size, align) \
+ static char __##name[ALIGN(size * sizeof(type), align)] \
+ __aligned(align); \
+ \
+ static type *name = (type *)__##name
+#define DEFINE_CACHE_ALIGN_BUFFER(type, name, size) \
+ DEFINE_ALIGN_BUFFER(type, name, size, ARCH_DMA_MINALIGN)
+
+/**
+ * malloc_cache_aligned() - allocate a memory region aligned to cache line size
+ *
+ * This allocates memory at a cache-line boundary. The amount allocated may
+ * be larger than requested as it is rounded up to the nearest multiple of the
+ * cache-line size. This ensured that subsequent cache operations on this
+ * memory (flush, invalidate) will not affect subsequently allocated regions.
+ *
+ * @size: Minimum number of bytes to allocate
+ *
+ * @return pointer to new memory region, or NULL if there is no more memory
+ * available.
+ */
static inline void *malloc_cache_aligned(size_t size)
{
return memalign(ARCH_DMA_MINALIGN, ALIGN(size, ARCH_DMA_MINALIGN));