This patch set aims to clean up the linux/mm.h header and reduce
dependencies on it by moving parts out.
The goal was to eliminate dependencies on linux/mm.h from other
popular headers such as highmem.h and dma-mapping.h, and I started by
checking which symbols were really used and moved those declarations
to separate slim headers.
---
v1 -> v2: added more explanations to commit messages; renamed several
new headers from page_*.h to folio_*.h as suggested by Matthew
Wilcox; rebase on linux-next; fix build failures on architectures
um,nios2,hexagon by adding more missing includes
v2 -> v3: rebase on linux-next
v3 -> v4: rebase on linux-next; fix build failure on loongarch64
(reported by kernel test robot); add missing includes to
drivers/dma/bcm2835-dma.c and include/scsi/scsicam.h
Max Kellermann (15):
drivers: add missing includes on linux/mm.h (and others)
include/drm/drm_gem.h: add poll_table_struct forward declaration
include/scsi/scsicam.h: forward-declare struct block_device
linux/mm.h: move page_kasan_tag() to mm/page_kasan_tag.h
linux/mm.h: move section functions to mm/page_section.h
linux/mm.h: move page_address() and others to mm/page_address.h
linux/mm.h: move folio_size(), ... to mm/folio_size.h
linux/mm.h: move folio_next() to mm/folio_next.h
linux/mm.h: move devmap-related declarations to mm/devmap_managed.h
linux/mm.h: move usage count functions to mm/folio_usage.h
linux/mm.h: move page_zone_id() and more to mm/folio_zone.h
linux/mm.h: move pfmemalloc-related functions to pfmemalloc.h
linux/mm.h: move is_vmalloc_addr() to mm/vmalloc_addr.h
linux/mm.h: move high_memory to mm/high_memory.h
include: reduce dependencies on linux/mm.h
MAINTAINERS | 1 +
arch/arm/include/asm/memory.h | 4 +
arch/arm/include/asm/pgtable.h | 2 +
arch/arm/mm/iomap.c | 3 +
arch/csky/include/asm/page.h | 1 +
arch/hexagon/include/asm/mem-layout.h | 4 +
arch/m68k/include/asm/page_mm.h | 1 +
arch/m68k/include/asm/pgtable_mm.h | 1 +
arch/parisc/include/asm/floppy.h | 1 +
arch/powerpc/include/asm/book3s/32/pgtable.h | 4 +
arch/powerpc/include/asm/nohash/32/pgtable.h | 1 +
arch/powerpc/include/asm/page.h | 1 +
arch/x86/include/asm/floppy.h | 1 +
arch/x86/include/asm/pgtable_32_areas.h | 4 +
drivers/comedi/comedi_buf.c | 1 +
.../qat/qat_common/adf_gen4_pm_debugfs.c | 1 +
drivers/dma/bcm2835-dma.c | 1 +
drivers/dma/dma-axi-dmac.c | 1 +
drivers/dma/sh/rcar-dmac.c | 1 +
drivers/firmware/qcom/qcom_scm-legacy.c | 1 +
drivers/firmware/qcom/qcom_scm-smc.c | 1 +
drivers/firmware/raspberrypi.c | 1 +
drivers/iio/buffer/industrialio-buffer-dma.c | 1 +
drivers/iommu/iommufd/ioas.c | 2 +
drivers/iommu/iommufd/selftest.c | 1 +
drivers/media/platform/mediatek/vpu/mtk_vpu.c | 1 +
drivers/media/platform/ti/omap/omap_voutlib.c | 1 +
drivers/misc/bcm-vk/bcm_vk_dev.c | 1 +
drivers/misc/fastrpc.c | 1 +
drivers/misc/genwqe/card_dev.c | 1 +
drivers/misc/uacce/uacce.c | 1 +
drivers/mtd/nand/onenand/onenand_samsung.c | 1 +
drivers/mtd/spi-nor/core.h | 2 +
drivers/pci/p2pdma.c | 1 +
drivers/pci/pci.c | 1 +
drivers/remoteproc/remoteproc_core.c | 1 +
drivers/soc/qcom/rmtfs_mem.c | 1 +
drivers/spi/spi-aspeed-smc.c | 1 +
drivers/spi/spi-bcm2835.c | 2 +
drivers/spi/spi-intel.c | 1 +
drivers/virtio/virtio_ring.c | 1 +
include/drm/drm_file.h | 1 +
include/linux/bio.h | 2 +
include/linux/dma-mapping.h | 1 +
include/linux/highmem-internal.h | 2 +
include/linux/highmem.h | 4 +-
include/linux/huge_mm.h | 2 +
include/linux/iommu.h | 1 +
include/linux/mm.h | 583 +-----------------
include/linux/mm/devmap_managed.h | 37 ++
include/linux/mm/folio_next.h | 27 +
include/linux/mm/folio_size.h | 150 +++++
include/linux/mm/folio_usage.h | 182 ++++++
include/linux/mm/folio_zone.h | 36 ++
include/linux/mm/high_memory.h | 7 +
include/linux/mm/page_address.h | 71 +++
include/linux/mm/page_kasan_tag.h | 66 ++
include/linux/mm/page_section.h | 23 +
include/linux/mm/pfmemalloc.h | 52 ++
include/linux/mm/vmalloc_addr.h | 33 +
include/linux/nvme-keyring.h | 2 +
include/linux/page-flags.h | 3 +
include/linux/scatterlist.h | 8 +-
include/linux/skbuff.h | 4 +
include/scsi/scsicam.h | 5 +
kernel/dma/ops_helpers.c | 1 +
kernel/dma/remap.c | 1 +
lib/scatterlist.c | 1 +
mm/dmapool.c | 1 +
69 files changed, 789 insertions(+), 576 deletions(-)
create mode 100644 include/linux/mm/devmap_managed.h
create mode 100644 include/linux/mm/folio_next.h
create mode 100644 include/linux/mm/folio_size.h
create mode 100644 include/linux/mm/folio_usage.h
create mode 100644 include/linux/mm/folio_zone.h
create mode 100644 include/linux/mm/high_memory.h
create mode 100644 include/linux/mm/page_address.h
create mode 100644 include/linux/mm/page_kasan_tag.h
create mode 100644 include/linux/mm/page_section.h
create mode 100644 include/linux/mm/pfmemalloc.h
create mode 100644 include/linux/mm/vmalloc_addr.h
--
2.39.2
After developing the other patches in this series, I had lots of build
failures because "#include <linux/mm.h>" was missing. This patch
fixes those build failures by adding the missing "#include" lines. To
allow bisects, it is ordered before the others.
Signed-off-by: Max Kellermann <[email protected]>
---
arch/arm/mm/iomap.c | 3 +++
drivers/comedi/comedi_buf.c | 1 +
drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c | 1 +
drivers/dma/bcm2835-dma.c | 1 +
drivers/dma/dma-axi-dmac.c | 1 +
drivers/dma/sh/rcar-dmac.c | 1 +
drivers/firmware/qcom/qcom_scm-legacy.c | 1 +
drivers/firmware/qcom/qcom_scm-smc.c | 1 +
drivers/firmware/raspberrypi.c | 1 +
drivers/iio/buffer/industrialio-buffer-dma.c | 1 +
drivers/iommu/iommufd/ioas.c | 2 ++
drivers/iommu/iommufd/selftest.c | 1 +
drivers/media/platform/mediatek/vpu/mtk_vpu.c | 1 +
drivers/media/platform/ti/omap/omap_voutlib.c | 1 +
drivers/misc/bcm-vk/bcm_vk_dev.c | 1 +
drivers/misc/fastrpc.c | 1 +
drivers/misc/genwqe/card_dev.c | 1 +
drivers/misc/uacce/uacce.c | 1 +
drivers/mtd/spi-nor/core.h | 2 ++
drivers/pci/p2pdma.c | 1 +
drivers/pci/pci.c | 1 +
drivers/remoteproc/remoteproc_core.c | 1 +
drivers/soc/qcom/rmtfs_mem.c | 1 +
drivers/spi/spi-aspeed-smc.c | 1 +
drivers/spi/spi-bcm2835.c | 2 ++
drivers/spi/spi-intel.c | 1 +
drivers/virtio/virtio_ring.c | 1 +
include/linux/huge_mm.h | 2 ++
include/linux/iommu.h | 1 +
include/linux/nvme-keyring.h | 2 ++
include/linux/page-flags.h | 3 +++
include/linux/scatterlist.h | 6 ++++++
include/scsi/scsicam.h | 3 +++
kernel/dma/ops_helpers.c | 1 +
kernel/dma/remap.c | 1 +
mm/dmapool.c | 1 +
36 files changed, 52 insertions(+)
diff --git a/arch/arm/mm/iomap.c b/arch/arm/mm/iomap.c
index 415d0a454237..9873d8156908 100644
--- a/arch/arm/mm/iomap.c
+++ b/arch/arm/mm/iomap.c
@@ -29,6 +29,9 @@ EXPORT_SYMBOL(ioport_unmap);
#endif
#ifdef CONFIG_PCI
+
+#include <asm/pgtable.h> // for VMALLOC_*
+
unsigned long pcibios_min_io = 0x1000;
EXPORT_SYMBOL(pcibios_min_io);
diff --git a/drivers/comedi/comedi_buf.c b/drivers/comedi/comedi_buf.c
index 393966c09740..23b07ebc97ca 100644
--- a/drivers/comedi/comedi_buf.c
+++ b/drivers/comedi/comedi_buf.c
@@ -10,6 +10,7 @@
#include <linux/vmalloc.h>
#include <linux/slab.h>
#include <linux/comedi/comedidev.h>
+#include <asm/pgtable.h> // for PAGE_KERNEL*
#include "comedi_internal.h"
#ifdef PAGE_KERNEL_NOCACHE
diff --git a/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c
index ee0b5079de3e..c9b2787baac7 100644
--- a/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c
+++ b/drivers/crypto/intel/qat/qat_common/adf_gen4_pm_debugfs.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0-only
/* Copyright(c) 2023 Intel Corporation */
#include <linux/dma-mapping.h>
+#include <linux/fs.h> // for simple_read_from_buffer()
#include <linux/kernel.h>
#include <linux/string_helpers.h>
#include <linux/stringify.h>
diff --git a/drivers/dma/bcm2835-dma.c b/drivers/dma/bcm2835-dma.c
index 9d74fe97452e..1d971999fd39 100644
--- a/drivers/dma/bcm2835-dma.c
+++ b/drivers/dma/bcm2835-dma.c
@@ -31,6 +31,7 @@
#include <linux/spinlock.h>
#include <linux/of.h>
#include <linux/of_dma.h>
+#include <asm/pgtable.h> // for ZERO_PAGE()
#include "virt-dma.h"
diff --git a/drivers/dma/dma-axi-dmac.c b/drivers/dma/dma-axi-dmac.c
index 4e339c04fc1e..0c98bf0e4b5d 100644
--- a/drivers/dma/dma-axi-dmac.c
+++ b/drivers/dma/dma-axi-dmac.c
@@ -15,6 +15,7 @@
#include <linux/interrupt.h>
#include <linux/io.h>
#include <linux/kernel.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/of.h>
#include <linux/of_dma.h>
diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c
index 40482cb73d79..784da367665c 100644
--- a/drivers/dma/sh/rcar-dmac.c
+++ b/drivers/dma/sh/rcar-dmac.c
@@ -12,6 +12,7 @@
#include <linux/dmaengine.h>
#include <linux/interrupt.h>
#include <linux/list.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/of.h>
diff --git a/drivers/firmware/qcom/qcom_scm-legacy.c b/drivers/firmware/qcom/qcom_scm-legacy.c
index 029e6d117cb8..01e082c7163c 100644
--- a/drivers/firmware/qcom/qcom_scm-legacy.c
+++ b/drivers/firmware/qcom/qcom_scm-legacy.c
@@ -5,6 +5,7 @@
#include <linux/slab.h>
#include <linux/io.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/errno.h>
diff --git a/drivers/firmware/qcom/qcom_scm-smc.c b/drivers/firmware/qcom/qcom_scm-smc.c
index 16cf88acfa8e..a5c74d8996fe 100644
--- a/drivers/firmware/qcom/qcom_scm-smc.c
+++ b/drivers/firmware/qcom/qcom_scm-smc.c
@@ -5,6 +5,7 @@
#include <linux/io.h>
#include <linux/errno.h>
#include <linux/delay.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/mutex.h>
#include <linux/slab.h>
#include <linux/types.h>
diff --git a/drivers/firmware/raspberrypi.c b/drivers/firmware/raspberrypi.c
index 322aada20f74..beef5c4afef2 100644
--- a/drivers/firmware/raspberrypi.c
+++ b/drivers/firmware/raspberrypi.c
@@ -9,6 +9,7 @@
#include <linux/dma-mapping.h>
#include <linux/kref.h>
#include <linux/mailbox_client.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/of.h>
#include <linux/of_platform.h>
diff --git a/drivers/iio/buffer/industrialio-buffer-dma.c b/drivers/iio/buffer/industrialio-buffer-dma.c
index 5610ba67925e..b3ce8c4b101a 100644
--- a/drivers/iio/buffer/industrialio-buffer-dma.c
+++ b/drivers/iio/buffer/industrialio-buffer-dma.c
@@ -6,6 +6,7 @@
#include <linux/slab.h>
#include <linux/kernel.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/device.h>
#include <linux/workqueue.h>
diff --git a/drivers/iommu/iommufd/ioas.c b/drivers/iommu/iommufd/ioas.c
index 742248276548..40c4942406cd 100644
--- a/drivers/iommu/iommufd/ioas.c
+++ b/drivers/iommu/iommufd/ioas.c
@@ -5,6 +5,8 @@
#include <linux/interval_tree.h>
#include <linux/iommufd.h>
#include <linux/iommu.h>
+#include <linux/capability.h>
+#include <linux/slab.h>
#include <uapi/linux/iommufd.h>
#include "io_pagetable.h"
diff --git a/drivers/iommu/iommufd/selftest.c b/drivers/iommu/iommufd/selftest.c
index 7a2199470f31..bcecf44681fe 100644
--- a/drivers/iommu/iommufd/selftest.c
+++ b/drivers/iommu/iommufd/selftest.c
@@ -5,6 +5,7 @@
*/
#include <linux/slab.h>
#include <linux/iommu.h>
+#include <linux/mm.h> // for GUP_PIN_COUNTING_BIAS
#include <linux/xarray.h>
#include <linux/file.h>
#include <linux/anon_inodes.h>
diff --git a/drivers/media/platform/mediatek/vpu/mtk_vpu.c b/drivers/media/platform/mediatek/vpu/mtk_vpu.c
index 724ae7c2ab3b..f248e4194b49 100644
--- a/drivers/media/platform/mediatek/vpu/mtk_vpu.c
+++ b/drivers/media/platform/mediatek/vpu/mtk_vpu.c
@@ -8,6 +8,7 @@
#include <linux/firmware.h>
#include <linux/interrupt.h>
#include <linux/iommu.h>
+#include <linux/mm.h> // for totalram_pages()
#include <linux/module.h>
#include <linux/of.h>
#include <linux/of_platform.h>
diff --git a/drivers/media/platform/ti/omap/omap_voutlib.c b/drivers/media/platform/ti/omap/omap_voutlib.c
index 0ac46458e41c..3b653b49cc48 100644
--- a/drivers/media/platform/ti/omap/omap_voutlib.c
+++ b/drivers/media/platform/ti/omap/omap_voutlib.c
@@ -18,6 +18,7 @@
*
*/
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/kernel.h>
diff --git a/drivers/misc/bcm-vk/bcm_vk_dev.c b/drivers/misc/bcm-vk/bcm_vk_dev.c
index d4a96137728d..c5a39a8189bf 100644
--- a/drivers/misc/bcm-vk/bcm_vk_dev.c
+++ b/drivers/misc/bcm-vk/bcm_vk_dev.c
@@ -11,6 +11,7 @@
#include <linux/interrupt.h>
#include <linux/panic_notifier.h>
#include <linux/kref.h>
+#include <linux/mm.h> // for io_remap_pfn_range()
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/pci.h>
diff --git a/drivers/misc/fastrpc.c b/drivers/misc/fastrpc.c
index 4c67e2c5a82e..e0549b4b1c33 100644
--- a/drivers/misc/fastrpc.c
+++ b/drivers/misc/fastrpc.c
@@ -10,6 +10,7 @@
#include <linux/idr.h>
#include <linux/list.h>
#include <linux/miscdevice.h>
+#include <linux/mm.h> // for find_vma()
#include <linux/module.h>
#include <linux/of_address.h>
#include <linux/of.h>
diff --git a/drivers/misc/genwqe/card_dev.c b/drivers/misc/genwqe/card_dev.c
index 4441aca2280a..ce3acb938ca6 100644
--- a/drivers/misc/genwqe/card_dev.c
+++ b/drivers/misc/genwqe/card_dev.c
@@ -17,6 +17,7 @@
#include <linux/kernel.h>
#include <linux/types.h>
+#include <linux/mm.h> // for struct vm_operations_struct
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/string.h>
diff --git a/drivers/misc/uacce/uacce.c b/drivers/misc/uacce/uacce.c
index bdc2e6fda782..af815b8a718e 100644
--- a/drivers/misc/uacce/uacce.c
+++ b/drivers/misc/uacce/uacce.c
@@ -2,6 +2,7 @@
#include <linux/compat.h>
#include <linux/dma-mapping.h>
#include <linux/iommu.h>
+#include <linux/mm.h> // for struct vm_operations_struct
#include <linux/module.h>
#include <linux/poll.h>
#include <linux/slab.h>
diff --git a/drivers/mtd/spi-nor/core.h b/drivers/mtd/spi-nor/core.h
index 442786685515..c113ed8e8751 100644
--- a/drivers/mtd/spi-nor/core.h
+++ b/drivers/mtd/spi-nor/core.h
@@ -9,6 +9,8 @@
#include "sfdp.h"
+#include <linux/sizes.h> // for SZ_*
+
#define SPI_NOR_MAX_ID_LEN 6
/*
* 256 bytes is a sane default for most older flashes. Newer flashes will
diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c
index 4f47a13cb500..ef37ea6c09fc 100644
--- a/drivers/pci/p2pdma.c
+++ b/drivers/pci/p2pdma.c
@@ -16,6 +16,7 @@
#include <linux/slab.h>
#include <linux/genalloc.h>
#include <linux/memremap.h>
+#include <linux/mm.h> // for VM_MAYSHARE
#include <linux/percpu-refcount.h>
#include <linux/random.h>
#include <linux/seq_buf.h>
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index e5f243dd4288..75e05c3518f4 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -25,6 +25,7 @@
#include <linux/logic_pio.h>
#include <linux/pm_wakeup.h>
#include <linux/device.h>
+#include <linux/pgtable.h> // for pgprot_device()
#include <linux/pm_runtime.h>
#include <linux/pci_hotplug.h>
#include <linux/vmalloc.h>
diff --git a/drivers/remoteproc/remoteproc_core.c b/drivers/remoteproc/remoteproc_core.c
index f276956f2c5c..938220fe29f5 100644
--- a/drivers/remoteproc/remoteproc_core.c
+++ b/drivers/remoteproc/remoteproc_core.c
@@ -18,6 +18,7 @@
#include <linux/delay.h>
#include <linux/kernel.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/module.h>
#include <linux/device.h>
#include <linux/panic_notifier.h>
diff --git a/drivers/soc/qcom/rmtfs_mem.c b/drivers/soc/qcom/rmtfs_mem.c
index df850d073102..15c2f6e5eaff 100644
--- a/drivers/soc/qcom/rmtfs_mem.c
+++ b/drivers/soc/qcom/rmtfs_mem.c
@@ -6,6 +6,7 @@
#include <linux/kernel.h>
#include <linux/cdev.h>
#include <linux/err.h>
+#include <linux/mm.h> // for remap_pfn_range()
#include <linux/module.h>
#include <linux/platform_device.h>
#include <linux/of.h>
diff --git a/drivers/spi/spi-aspeed-smc.c b/drivers/spi/spi-aspeed-smc.c
index bbd417c55e7f..e45f64761e5e 100644
--- a/drivers/spi/spi-aspeed-smc.c
+++ b/drivers/spi/spi-aspeed-smc.c
@@ -11,6 +11,7 @@
#include <linux/of.h>
#include <linux/of_platform.h>
#include <linux/platform_device.h>
+#include <linux/sizes.h> // for SZ_*
#include <linux/spi/spi.h>
#include <linux/spi/spi-mem.h>
diff --git a/drivers/spi/spi-bcm2835.c b/drivers/spi/spi-bcm2835.c
index e1b9b1235787..d8ed5575a9c6 100644
--- a/drivers/spi/spi-bcm2835.c
+++ b/drivers/spi/spi-bcm2835.c
@@ -33,6 +33,8 @@
#include <linux/slab.h>
#include <linux/spi/spi.h>
+#include <asm/pgtable.h> // for ZERO_PAGE()
+
/* SPI register offsets */
#define BCM2835_SPI_CS 0x00
#define BCM2835_SPI_FIFO 0x04
diff --git a/drivers/spi/spi-intel.c b/drivers/spi/spi-intel.c
index 3e5dcf2b3c8a..cf8e4f1c4d93 100644
--- a/drivers/spi/spi-intel.c
+++ b/drivers/spi/spi-intel.c
@@ -8,6 +8,7 @@
#include <linux/iopoll.h>
#include <linux/module.h>
+#include <linux/sizes.h> // for SZ_*
#include <linux/mtd/partitions.h>
#include <linux/mtd/spi-nor.h>
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index 6f7e5010a673..48f694d1bc46 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -12,6 +12,7 @@
#include <linux/hrtimer.h>
#include <linux/dma-mapping.h>
#include <linux/kmsan.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
#include <linux/spinlock.h>
#include <xen/xen.h>
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index de0c89105076..55b94db3bd5c 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -7,6 +7,8 @@
#include <linux/fs.h> /* only for vma_is_dax() */
+#include <asm/page.h> // for pmd_t
+
vm_fault_t do_huge_pmd_anonymous_page(struct vm_fault *vmf);
int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
pmd_t *dst_pmd, pmd_t *src_pmd, unsigned long addr,
diff --git a/include/linux/iommu.h b/include/linux/iommu.h
index 2e925b5eba53..4e72c2429342 100644
--- a/include/linux/iommu.h
+++ b/include/linux/iommu.h
@@ -14,6 +14,7 @@
#include <linux/err.h>
#include <linux/of.h>
#include <linux/iova_bitmap.h>
+#include <linux/uaccess.h> // for copy_struct_from_user()
#define IOMMU_READ (1 << 0)
#define IOMMU_WRITE (1 << 1)
diff --git a/include/linux/nvme-keyring.h b/include/linux/nvme-keyring.h
index e10333d78dbb..9e033850b967 100644
--- a/include/linux/nvme-keyring.h
+++ b/include/linux/nvme-keyring.h
@@ -6,6 +6,8 @@
#ifndef _NVME_KEYRING_H
#define _NVME_KEYRING_H
+#include <linux/key.h>
+
#if IS_ENABLED(CONFIG_NVME_KEYRING)
key_serial_t nvme_tls_psk_default(struct key *keyring,
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 652d77805e99..c7243b85328b 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -12,6 +12,9 @@
#ifndef __GENERATING_BOUNDS_H
#include <linux/mm_types.h>
#include <generated/bounds.h>
+#ifdef CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
+#include <linux/jump_label.h> // for DECLARE_STATIC_KEY_FALSE
+#endif
#endif /* !__GENERATING_BOUNDS_H */
/*
diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h
index 77df3d7b18a6..0516e64dc03e 100644
--- a/include/linux/scatterlist.h
+++ b/include/linux/scatterlist.h
@@ -8,6 +8,12 @@
#include <linux/mm.h>
#include <asm/io.h>
+#ifdef CONFIG_UML
+#include <asm/pgtable.h> // for virt_to_page(), page_to_phys()
+#else
+#include <asm/page.h> // for virt_to_page(), page_to_phys()
+#endif
+
struct scatterlist {
unsigned long page_link;
unsigned int offset;
diff --git a/include/scsi/scsicam.h b/include/scsi/scsicam.h
index 08edd603e521..6eb9fb7a57d0 100644
--- a/include/scsi/scsicam.h
+++ b/include/scsi/scsicam.h
@@ -13,6 +13,9 @@
#ifndef SCSICAM_H
#define SCSICAM_H
+
+#include <linux/types.h> // for sector_t
+
int scsicam_bios_param(struct block_device *bdev, sector_t capacity, int *ip);
bool scsi_partsize(struct block_device *bdev, sector_t capacity, int geom[3]);
unsigned char *scsi_bios_ptable(struct block_device *bdev);
diff --git a/kernel/dma/ops_helpers.c b/kernel/dma/ops_helpers.c
index af4a6ef48ce0..641363b13bb9 100644
--- a/kernel/dma/ops_helpers.c
+++ b/kernel/dma/ops_helpers.c
@@ -4,6 +4,7 @@
* the allocated memory contains normal pages in the direct kernel mapping.
*/
#include <linux/dma-map-ops.h>
+#include <linux/mm.h> // for PAGE_ALIGN()
static struct page *dma_common_vaddr_to_page(void *cpu_addr)
{
diff --git a/kernel/dma/remap.c b/kernel/dma/remap.c
index 27596f3b4aef..87b3c874d370 100644
--- a/kernel/dma/remap.c
+++ b/kernel/dma/remap.c
@@ -3,6 +3,7 @@
* Copyright (c) 2014 The Linux Foundation
*/
#include <linux/dma-map-ops.h>
+#include <linux/mm.h> // for vmap(), PAGE_ALIGN()
#include <linux/slab.h>
#include <linux/vmalloc.h>
diff --git a/mm/dmapool.c b/mm/dmapool.c
index f0bfc6c490f4..b8e23fdc81d7 100644
--- a/mm/dmapool.c
+++ b/mm/dmapool.c
@@ -25,6 +25,7 @@
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/export.h>
+#include <linux/mm.h> // for want_init_on_alloc()
#include <linux/mutex.h>
#include <linux/poison.h>
#include <linux/sched.h>
--
2.39.2
After eliminating includes of linux/mm.h, the following build failure
occurred:
In file included from drivers/scsi/fdomain.c:87:
./include/scsi/scsicam.h:16:31: error: 'struct block_device' declared inside parameter list will not be visible outside of this definition or declaration [-Werror]
16 | int scsicam_bios_param(struct block_device *bdev, sector_t capacity, int *ip);
| ^~~~~~~~~~~~
./include/scsi/scsicam.h:17:27: error: 'struct block_device' declared inside parameter list will not be visible outside of this definition or declaration [-Werror]
17 | bool scsi_partsize(struct block_device *bdev, sector_t capacity, int geom[3]);
| ^~~~~~~~~~~~
./include/scsi/scsicam.h:18:40: error: 'struct block_device' declared inside parameter list will not be visible outside of this definition or declaration [-Werror]
18 | unsigned char *scsi_bios_ptable(struct block_device *bdev);
| ^~~~~~~~~~~~
drivers/scsi/fdomain.c: In function 'fdomain_biosparam':
drivers/scsi/fdomain.c:475:45: error: passing argument 1 of 'scsi_bios_ptable' from incompatible pointer type [-Werror=incompatible-pointer-types]
475 | unsigned char *p = scsi_bios_ptable(bdev);
| ^~~~
| |
| struct block_device *
./include/scsi/scsicam.h:18:54: note: expected 'struct block_device *' but argument is of type 'struct block_device *'
18 | unsigned char *scsi_bios_ptable(struct block_device *bdev);
| ~~~~~~~~~~~~~~~~~~~~~^~~~
Signed-off-by: Max Kellermann <[email protected]>
---
include/scsi/scsicam.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/include/scsi/scsicam.h b/include/scsi/scsicam.h
index 6eb9fb7a57d0..c0de038b76b5 100644
--- a/include/scsi/scsicam.h
+++ b/include/scsi/scsicam.h
@@ -16,6 +16,8 @@
#include <linux/types.h> // for sector_t
+struct block_device;
+
int scsicam_bios_param(struct block_device *bdev, sector_t capacity, int *ip);
bool scsi_partsize(struct block_device *bdev, sector_t capacity, int geom[3]);
unsigned char *scsi_bios_ptable(struct block_device *bdev);
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
page_to_section() is used by __page_to_pfn() (but
asm-generic/memory_model.h does not include linux/mm.h). Thus, in
order to be able to use anything that calls page_to_pfn(), linux/mm.h
needs to be included.
This would prevent us from moving page_address() to a separate header,
because it calls lowmem_page_address() which in turn calls
page_to_virt() and then page_to_pfn(). To prepare for this move, we
move page_to_section() out of linux/mm.h into a separate lean header.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 18 +-----------------
include/linux/mm/page_section.h | 23 +++++++++++++++++++++++
2 files changed, 24 insertions(+), 17 deletions(-)
create mode 100644 include/linux/mm/page_section.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index e140ee45f49c..79c1f924d4b5 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3,6 +3,7 @@
#define _LINUX_MM_H
#include <linux/mm/page_kasan_tag.h>
+#include <linux/mm/page_section.h>
#include <linux/errno.h>
#include <linux/mmdebug.h>
#include <linux/gfp.h>
@@ -1637,10 +1638,6 @@ static inline bool is_nommu_shared_mapping(vm_flags_t flags)
}
#endif
-#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
-#define SECTION_IN_PAGE_FLAGS
-#endif
-
/*
* The identification function is mainly used by the buddy allocator for
* determining if two pages could be buddies. We are not really identifying
@@ -1838,19 +1835,6 @@ static inline pg_data_t *folio_pgdat(const struct folio *folio)
return page_pgdat(&folio->page);
}
-#ifdef SECTION_IN_PAGE_FLAGS
-static inline void set_page_section(struct page *page, unsigned long section)
-{
- page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
- page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
-}
-
-static inline unsigned long page_to_section(const struct page *page)
-{
- return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
-}
-#endif
-
/**
* folio_pfn - Return the Page Frame Number of a folio.
* @folio: The folio.
diff --git a/include/linux/mm/page_section.h b/include/linux/mm/page_section.h
new file mode 100644
index 000000000000..e4558c2691b8
--- /dev/null
+++ b/include/linux/mm/page_section.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_PAGE_SECTION_H
+#define _LINUX_MM_PAGE_SECTION_H
+
+#include <linux/mm_types.h> // for struct page
+#include <linux/mmzone.h> // for SECTIONS_*
+
+#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
+#define SECTION_IN_PAGE_FLAGS
+
+static inline void set_page_section(struct page *page, unsigned long section)
+{
+ page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
+ page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
+}
+
+static inline unsigned long page_to_section(const struct page *page)
+{
+ return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
+}
+#endif
+
+#endif /* _LINUX_MM_PAGE_SECTION_H */
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
page_address() is used by the following popular headers:
- linux/bio.h
- linux/bvec.h
- linux/highmem.h
- linux/scatterlist.h
- linux/skbuff.h
Moving it to a separate lean header will allow us to avoid the
dependency on linux/mm.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 56 +-------------------------
include/linux/mm/page_address.h | 71 +++++++++++++++++++++++++++++++++
2 files changed, 72 insertions(+), 55 deletions(-)
create mode 100644 include/linux/mm/page_address.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 79c1f924d4b5..713cedc03b88 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,7 +2,7 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
-#include <linux/mm/page_kasan_tag.h>
+#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/errno.h>
#include <linux/mmdebug.h>
@@ -104,10 +104,6 @@ extern int mmap_rnd_compat_bits __read_mostly;
#define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
#endif
-#ifndef page_to_virt
-#define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
-#endif
-
#ifndef lm_alias
#define lm_alias(x) __va(__pa_symbol(x))
#endif
@@ -211,14 +207,6 @@ int overcommit_kbytes_handler(struct ctl_table *, int, void *, size_t *,
int overcommit_policy_handler(struct ctl_table *, int, void *, size_t *,
loff_t *);
-#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
-#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
-#define folio_page_idx(folio, p) (page_to_pfn(p) - folio_pfn(folio))
-#else
-#define nth_page(page,n) ((page) + (n))
-#define folio_page_idx(folio, p) ((p) - &(folio)->page)
-#endif
-
/* to align the pointer to the (next) page boundary */
#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
@@ -2137,44 +2125,6 @@ static inline int arch_make_folio_accessible(struct folio *folio)
*/
#include <linux/vmstat.h>
-static __always_inline void *lowmem_page_address(const struct page *page)
-{
- return page_to_virt(page);
-}
-
-#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
-#define HASHED_PAGE_VIRTUAL
-#endif
-
-#if defined(WANT_PAGE_VIRTUAL)
-static inline void *page_address(const struct page *page)
-{
- return page->virtual;
-}
-static inline void set_page_address(struct page *page, void *address)
-{
- page->virtual = address;
-}
-#define page_address_init() do { } while(0)
-#endif
-
-#if defined(HASHED_PAGE_VIRTUAL)
-void *page_address(const struct page *page);
-void set_page_address(struct page *page, void *virtual);
-void page_address_init(void);
-#endif
-
-#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
-#define page_address(page) lowmem_page_address(page)
-#define set_page_address(page, address) do { } while(0)
-#define page_address_init() do { } while(0)
-#endif
-
-static inline void *folio_address(const struct folio *folio)
-{
- return page_address(&folio->page);
-}
-
extern pgoff_t __page_file_index(struct page *page);
/*
@@ -2237,10 +2187,6 @@ static inline void clear_page_pfmemalloc(struct page *page)
*/
extern void pagefault_out_of_memory(void);
-#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
-#define offset_in_thp(page, p) ((unsigned long)(p) & (thp_size(page) - 1))
-#define offset_in_folio(folio, p) ((unsigned long)(p) & (folio_size(folio) - 1))
-
/*
* Parameter block passed down to zap_pte_range in exceptional cases.
*/
diff --git a/include/linux/mm/page_address.h b/include/linux/mm/page_address.h
new file mode 100644
index 000000000000..e1aaacc5003f
--- /dev/null
+++ b/include/linux/mm/page_address.h
@@ -0,0 +1,71 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_PAGE_ADDRESS_H
+#define _LINUX_MM_PAGE_ADDRESS_H
+
+#include <linux/mm_types.h> // for struct page
+#include <linux/mm/page_kasan_tag.h> // needed by the page_to_virt() macro on some architectures (e.g. arm64)
+#include <asm/page.h> // for PAGE_MASK, page_to_virt()
+
+#if defined(CONFIG_FLATMEM)
+#include <linux/mmzone.h> // for memmap (used by __pfn_to_page())
+#elif defined(CONFIG_SPARSEMEM_VMEMMAP)
+#include <asm/pgtable.h> // for vmemmap (used by __pfn_to_page())
+#elif defined(CONFIG_SPARSEMEM)
+#include <linux/mm/page_section.h> // for page_to_section() (used by __page_to_pfn())
+#endif
+
+#ifndef page_to_virt
+#define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
+#endif
+
+#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
+#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
+#define folio_page_idx(folio, p) (page_to_pfn(p) - folio_pfn(folio))
+#else
+#define nth_page(page,n) ((page) + (n))
+#define folio_page_idx(folio, p) ((p) - &(folio)->page)
+#endif
+
+static __always_inline void *lowmem_page_address(const struct page *page)
+{
+ return page_to_virt(page);
+}
+
+#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
+#define HASHED_PAGE_VIRTUAL
+#endif
+
+#if defined(WANT_PAGE_VIRTUAL)
+static inline void *page_address(const struct page *page)
+{
+ return page->virtual;
+}
+static inline void set_page_address(struct page *page, void *address)
+{
+ page->virtual = address;
+}
+#define page_address_init() do { } while(0)
+#endif
+
+#if defined(HASHED_PAGE_VIRTUAL)
+void *page_address(const struct page *page);
+void set_page_address(struct page *page, void *virtual);
+void page_address_init(void);
+#endif
+
+#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
+#define page_address(page) lowmem_page_address(page)
+#define set_page_address(page, address) do { } while(0)
+#define page_address_init() do { } while(0)
+#endif
+
+static inline void *folio_address(const struct folio *folio)
+{
+ return page_address(&folio->page);
+}
+
+#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
+#define offset_in_thp(page, p) ((unsigned long)(p) & (thp_size(page) - 1))
+#define offset_in_folio(folio, p) ((unsigned long)(p) & (folio_size(folio) - 1))
+
+#endif /* _LINUX_MM_PAGE_ADDRESS_H */
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
folio_next() is used by linux/bio.h. Moving it to a separate lean
header will allow us to avoid the dependency on linux/mm.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 20 +-------------------
include/linux/mm/folio_next.h | 27 +++++++++++++++++++++++++++
2 files changed, 28 insertions(+), 19 deletions(-)
create mode 100644 include/linux/mm/folio_next.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 0d291ff61db3..e6c2df977abc 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,6 +2,7 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
+#include <linux/mm/folio_next.h>
#include <linux/mm/folio_size.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
@@ -1922,25 +1923,6 @@ static inline void set_page_links(struct page *page, enum zone_type zone,
#define MAX_FOLIO_NR_PAGES MAX_ORDER_NR_PAGES
#endif
-/**
- * folio_next - Move to the next physical folio.
- * @folio: The folio we're currently operating on.
- *
- * If you have physically contiguous memory which may span more than
- * one folio (eg a &struct bio_vec), use this function to move from one
- * folio to the next. Do not use it if the memory is only virtually
- * contiguous as the folios are almost certainly not adjacent to each
- * other. This is the folio equivalent to writing ``page++``.
- *
- * Context: We assume that the folios are refcounted and/or locked at a
- * higher level and do not adjust the reference counts.
- * Return: The next struct folio.
- */
-static inline struct folio *folio_next(struct folio *folio)
-{
- return (struct folio *)folio_page(folio, folio_nr_pages(folio));
-}
-
/**
* folio_estimated_sharers - Estimate the number of sharers of a folio.
* @folio: The folio.
diff --git a/include/linux/mm/folio_next.h b/include/linux/mm/folio_next.h
new file mode 100644
index 000000000000..7016e303439c
--- /dev/null
+++ b/include/linux/mm/folio_next.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_FOLIO_NEXT_H
+#define _LINUX_MM_FOLIO_NEXT_H
+
+#include <linux/mm/folio_size.h> // for folio_nr_pages()
+#include <linux/mm/page_address.h> // for nth_page(), needed by folio_page()
+
+/**
+ * folio_next - Move to the next physical folio.
+ * @folio: The folio we're currently operating on.
+ *
+ * If you have physically contiguous memory which may span more than
+ * one folio (eg a &struct bio_vec), use this function to move from one
+ * folio to the next. Do not use it if the memory is only virtually
+ * contiguous as the folios are almost certainly not adjacent to each
+ * other. This is the folio equivalent to writing ``page++``.
+ *
+ * Context: We assume that the folios are refcounted and/or locked at a
+ * higher level and do not adjust the reference counts.
+ * Return: The next struct folio.
+ */
+static inline struct folio *folio_next(struct folio *folio)
+{
+ return (struct folio *)folio_page(folio, folio_nr_pages(folio));
+}
+
+#endif /* _LINUX_MM_FOLIO_NEXT_H */
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
folio_size()/page_size() are used by the following popular headers:
- linux/highmem.h
- linux/iov_iter.h
- linux/pagemap.h
Moving them to a separate lean header will allow us to avoid the
dependency on linux/mm.h. Additionally, it allows us to move
folio_next() to a separate header, because it needs folio_nr_pages()
which is also moved to folio_size.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 140 +------------------------------
include/linux/mm/folio_size.h | 150 ++++++++++++++++++++++++++++++++++
2 files changed, 151 insertions(+), 139 deletions(-)
create mode 100644 include/linux/mm/folio_size.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 713cedc03b88..0d291ff61db3 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,6 +2,7 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
+#include <linux/mm/folio_size.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/errno.h>
@@ -1068,38 +1069,6 @@ int vma_is_stack_for_current(struct vm_area_struct *vma);
struct mmu_gather;
struct inode;
-/*
- * compound_order() can be called without holding a reference, which means
- * that niceties like page_folio() don't work. These callers should be
- * prepared to handle wild return values. For example, PG_head may be
- * set before the order is initialised, or this may be a tail page.
- * See compaction.c for some good examples.
- */
-static inline unsigned int compound_order(struct page *page)
-{
- struct folio *folio = (struct folio *)page;
-
- if (!test_bit(PG_head, &folio->flags))
- return 0;
- return folio->_flags_1 & 0xff;
-}
-
-/**
- * folio_order - The allocation order of a folio.
- * @folio: The folio.
- *
- * A folio is composed of 2^order pages. See get_order() for the definition
- * of order.
- *
- * Return: The order of the folio.
- */
-static inline unsigned int folio_order(struct folio *folio)
-{
- if (!folio_test_large(folio))
- return 0;
- return folio->_flags_1 & 0xff;
-}
-
#include <linux/huge_mm.h>
/*
@@ -1306,39 +1275,6 @@ unsigned long nr_free_buffer_pages(void);
void destroy_large_folio(struct folio *folio);
-/* Returns the number of bytes in this potentially compound page. */
-static inline unsigned long page_size(struct page *page)
-{
- return PAGE_SIZE << compound_order(page);
-}
-
-/* Returns the number of bits needed for the number of bytes in a page */
-static inline unsigned int page_shift(struct page *page)
-{
- return PAGE_SHIFT + compound_order(page);
-}
-
-/**
- * thp_order - Order of a transparent huge page.
- * @page: Head page of a transparent huge page.
- */
-static inline unsigned int thp_order(struct page *page)
-{
- VM_BUG_ON_PGFLAGS(PageTail(page), page);
- return compound_order(page);
-}
-
-/**
- * thp_size - Size of a transparent huge page.
- * @page: Head page of a transparent huge page.
- *
- * Return: Number of bytes in this page.
- */
-static inline unsigned long thp_size(struct page *page)
-{
- return PAGE_SIZE << thp_order(page);
-}
-
#ifdef CONFIG_MMU
/*
* Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
@@ -1979,23 +1915,6 @@ static inline void set_page_links(struct page *page, enum zone_type zone,
#endif
}
-/**
- * folio_nr_pages - The number of pages in the folio.
- * @folio: The folio.
- *
- * Return: A positive power of two.
- */
-static inline long folio_nr_pages(struct folio *folio)
-{
- if (!folio_test_large(folio))
- return 1;
-#ifdef CONFIG_64BIT
- return folio->_folio_nr_pages;
-#else
- return 1L << (folio->_flags_1 & 0xff);
-#endif
-}
-
/* Only hugetlbfs can allocate folios larger than MAX_ORDER */
#ifdef CONFIG_ARCH_HAS_GIGANTIC_PAGE
#define MAX_FOLIO_NR_PAGES (1UL << PUD_ORDER)
@@ -2003,33 +1922,6 @@ static inline long folio_nr_pages(struct folio *folio)
#define MAX_FOLIO_NR_PAGES MAX_ORDER_NR_PAGES
#endif
-/*
- * compound_nr() returns the number of pages in this potentially compound
- * page. compound_nr() can be called on a tail page, and is defined to
- * return 1 in that case.
- */
-static inline unsigned long compound_nr(struct page *page)
-{
- struct folio *folio = (struct folio *)page;
-
- if (!test_bit(PG_head, &folio->flags))
- return 1;
-#ifdef CONFIG_64BIT
- return folio->_folio_nr_pages;
-#else
- return 1L << (folio->_flags_1 & 0xff);
-#endif
-}
-
-/**
- * thp_nr_pages - The number of regular pages in this huge page.
- * @page: The head page of a huge page.
- */
-static inline int thp_nr_pages(struct page *page)
-{
- return folio_nr_pages((struct folio *)page);
-}
-
/**
* folio_next - Move to the next physical folio.
* @folio: The folio we're currently operating on.
@@ -2049,36 +1941,6 @@ static inline struct folio *folio_next(struct folio *folio)
return (struct folio *)folio_page(folio, folio_nr_pages(folio));
}
-/**
- * folio_shift - The size of the memory described by this folio.
- * @folio: The folio.
- *
- * A folio represents a number of bytes which is a power-of-two in size.
- * This function tells you which power-of-two the folio is. See also
- * folio_size() and folio_order().
- *
- * Context: The caller should have a reference on the folio to prevent
- * it from being split. It is not necessary for the folio to be locked.
- * Return: The base-2 logarithm of the size of this folio.
- */
-static inline unsigned int folio_shift(struct folio *folio)
-{
- return PAGE_SHIFT + folio_order(folio);
-}
-
-/**
- * folio_size - The number of bytes in a folio.
- * @folio: The folio.
- *
- * Context: The caller should have a reference on the folio to prevent
- * it from being split. It is not necessary for the folio to be locked.
- * Return: The number of bytes in this folio.
- */
-static inline size_t folio_size(struct folio *folio)
-{
- return PAGE_SIZE << folio_order(folio);
-}
-
/**
* folio_estimated_sharers - Estimate the number of sharers of a folio.
* @folio: The folio.
diff --git a/include/linux/mm/folio_size.h b/include/linux/mm/folio_size.h
new file mode 100644
index 000000000000..dd8af39ef572
--- /dev/null
+++ b/include/linux/mm/folio_size.h
@@ -0,0 +1,150 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_FOLIO_SIZE_H
+#define _LINUX_MM_FOLIO_SIZE_H
+
+#include <linux/bitops.h> // for test_bit()
+#include <linux/mmdebug.h> // for VM_BUG_ON_PGFLAGS()
+#include <linux/mm_types.h> // for struct page
+#include <linux/page-flags.h> // for folio_test_large()
+#include <asm/page.h> // for PAGE_SIZE, PAGE_SHIFT
+
+/*
+ * compound_order() can be called without holding a reference, which means
+ * that niceties like page_folio() don't work. These callers should be
+ * prepared to handle wild return values. For example, PG_head may be
+ * set before the order is initialised, or this may be a tail page.
+ * See compaction.c for some good examples.
+ */
+static inline unsigned int compound_order(struct page *page)
+{
+ struct folio *folio = (struct folio *)page;
+
+ if (!test_bit(PG_head, &folio->flags))
+ return 0;
+ return folio->_flags_1 & 0xff;
+}
+
+/**
+ * folio_order - The allocation order of a folio.
+ * @folio: The folio.
+ *
+ * A folio is composed of 2^order pages. See get_order() for the definition
+ * of order.
+ *
+ * Return: The order of the folio.
+ */
+static inline unsigned int folio_order(struct folio *folio)
+{
+ if (!folio_test_large(folio))
+ return 0;
+ return folio->_flags_1 & 0xff;
+}
+
+/* Returns the number of bytes in this potentially compound page. */
+static inline unsigned long page_size(struct page *page)
+{
+ return PAGE_SIZE << compound_order(page);
+}
+
+/* Returns the number of bits needed for the number of bytes in a page */
+static inline unsigned int page_shift(struct page *page)
+{
+ return PAGE_SHIFT + compound_order(page);
+}
+
+/**
+ * thp_order - Order of a transparent huge page.
+ * @page: Head page of a transparent huge page.
+ */
+static inline unsigned int thp_order(struct page *page)
+{
+ VM_BUG_ON_PGFLAGS(PageTail(page), page);
+ return compound_order(page);
+}
+
+/**
+ * thp_size - Size of a transparent huge page.
+ * @page: Head page of a transparent huge page.
+ *
+ * Return: Number of bytes in this page.
+ */
+static inline unsigned long thp_size(struct page *page)
+{
+ return PAGE_SIZE << thp_order(page);
+}
+
+/**
+ * folio_nr_pages - The number of pages in the folio.
+ * @folio: The folio.
+ *
+ * Return: A positive power of two.
+ */
+static inline long folio_nr_pages(struct folio *folio)
+{
+ if (!folio_test_large(folio))
+ return 1;
+#ifdef CONFIG_64BIT
+ return folio->_folio_nr_pages;
+#else
+ return 1L << (folio->_flags_1 & 0xff);
+#endif
+}
+
+/*
+ * compound_nr() returns the number of pages in this potentially compound
+ * page. compound_nr() can be called on a tail page, and is defined to
+ * return 1 in that case.
+ */
+static inline unsigned long compound_nr(struct page *page)
+{
+ struct folio *folio = (struct folio *)page;
+
+ if (!test_bit(PG_head, &folio->flags))
+ return 1;
+#ifdef CONFIG_64BIT
+ return folio->_folio_nr_pages;
+#else
+ return 1L << (folio->_flags_1 & 0xff);
+#endif
+}
+
+/**
+ * thp_nr_pages - The number of regular pages in this huge page.
+ * @page: The head page of a huge page.
+ */
+static inline int thp_nr_pages(struct page *page)
+{
+ return folio_nr_pages((struct folio *)page);
+}
+
+/**
+ * folio_shift - The size of the memory described by this folio.
+ * @folio: The folio.
+ *
+ * A folio represents a number of bytes which is a power-of-two in size.
+ * This function tells you which power-of-two the folio is. See also
+ * folio_size() and folio_order().
+ *
+ * Context: The caller should have a reference on the folio to prevent
+ * it from being split. It is not necessary for the folio to be locked.
+ * Return: The base-2 logarithm of the size of this folio.
+ */
+static inline unsigned int folio_shift(struct folio *folio)
+{
+ return PAGE_SHIFT + folio_order(folio);
+}
+
+/**
+ * folio_size - The number of bytes in a folio.
+ * @folio: The folio.
+ *
+ * Context: The caller should have a reference on the folio to prevent
+ * it from being split. It is not necessary for the folio to be locked.
+ * Return: The number of bytes in this folio.
+ */
+static inline size_t folio_size(struct folio *folio)
+{
+ return PAGE_SIZE << folio_order(folio);
+}
+
+#endif /* _LINUX_MM_FOLIO_SIZE_H_H */
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
put_devmap_managed_page() is called by put_page(). Moving it to a
separate header allows us to move put_page() to a separate lean header
as well.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 25 +--------------------
include/linux/mm/devmap_managed.h | 37 +++++++++++++++++++++++++++++++
2 files changed, 38 insertions(+), 24 deletions(-)
create mode 100644 include/linux/mm/devmap_managed.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index e6c2df977abc..61f1312a626e 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,6 +2,7 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
+#include <linux/mm/devmap_managed.h>
#include <linux/mm/folio_next.h>
#include <linux/mm/folio_size.h>
#include <linux/mm/page_address.h>
@@ -1357,30 +1358,6 @@ vm_fault_t finish_fault(struct vm_fault *vmf);
* back into memory.
*/
-#if defined(CONFIG_ZONE_DEVICE) && defined(CONFIG_FS_DAX)
-DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
-
-bool __put_devmap_managed_page_refs(struct page *page, int refs);
-static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
-{
- if (!static_branch_unlikely(&devmap_managed_key))
- return false;
- if (!is_zone_device_page(page))
- return false;
- return __put_devmap_managed_page_refs(page, refs);
-}
-#else /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
-static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
-{
- return false;
-}
-#endif /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
-
-static inline bool put_devmap_managed_page(struct page *page)
-{
- return put_devmap_managed_page_refs(page, 1);
-}
-
/* 127: arbitrary random number, small enough to assemble well */
#define folio_ref_zero_or_close_to_overflow(folio) \
((unsigned int) folio_ref_count(folio) + 127u <= 127u)
diff --git a/include/linux/mm/devmap_managed.h b/include/linux/mm/devmap_managed.h
new file mode 100644
index 000000000000..0773529d80b2
--- /dev/null
+++ b/include/linux/mm/devmap_managed.h
@@ -0,0 +1,37 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_DEVMAP_MANAGED_H
+#define _LINUX_MM_DEVMAP_MANAGED_H
+
+#include <linux/types.h> // for bool
+
+struct page;
+
+#if defined(CONFIG_ZONE_DEVICE) && defined(CONFIG_FS_DAX)
+
+#include <linux/jump_label.h> // for DECLARE_STATIC_KEY_FALSE(), static_branch_unlikely()
+#include <linux/mmzone.h> // is_zone_device_page()
+
+DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
+
+bool __put_devmap_managed_page_refs(struct page *page, int refs);
+static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
+{
+ if (!static_branch_unlikely(&devmap_managed_key))
+ return false;
+ if (!is_zone_device_page(page))
+ return false;
+ return __put_devmap_managed_page_refs(page, refs);
+}
+#else /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
+static inline bool put_devmap_managed_page_refs(struct page *page, int refs)
+{
+ return false;
+}
+#endif /* CONFIG_ZONE_DEVICE && CONFIG_FS_DAX */
+
+static inline bool put_devmap_managed_page(struct page *page)
+{
+ return put_devmap_managed_page_refs(page, 1);
+}
+
+#endif /* _LINUX_MM_DEVMAP_MANAGED_H */
--
2.39.2
This is needed to eliminate linux/skbuff.h's dependency on linux/mm.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 28 +--------------------------
include/linux/mm/folio_zone.h | 36 +++++++++++++++++++++++++++++++++++
2 files changed, 37 insertions(+), 27 deletions(-)
create mode 100644 include/linux/mm/folio_zone.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index e8a914e7bebd..b3f83fb26aca 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -5,6 +5,7 @@
#include <linux/mm/folio_next.h>
#include <linux/mm/folio_size.h>
#include <linux/mm/folio_usage.h>
+#include <linux/mm/folio_zone.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/errno.h>
@@ -1370,33 +1371,6 @@ static inline bool is_nommu_shared_mapping(vm_flags_t flags)
}
#endif
-/*
- * The identification function is mainly used by the buddy allocator for
- * determining if two pages could be buddies. We are not really identifying
- * the zone since we could be using the section number id if we do not have
- * node id available in page flags.
- * We only guarantee that it will return the same value for two combinable
- * pages in a zone.
- */
-static inline int page_zone_id(struct page *page)
-{
- return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
-}
-
-#ifdef NODE_NOT_IN_PAGE_FLAGS
-int page_to_nid(const struct page *page);
-#else
-static inline int page_to_nid(const struct page *page)
-{
- return (PF_POISONED_CHECK(page)->flags >> NODES_PGSHIFT) & NODES_MASK;
-}
-#endif
-
-static inline int folio_nid(const struct folio *folio)
-{
- return page_to_nid(&folio->page);
-}
-
#ifdef CONFIG_NUMA_BALANCING
/* page access time bits needs to hold at least 4 seconds */
#define PAGE_ACCESS_TIME_MIN_BITS 12
diff --git a/include/linux/mm/folio_zone.h b/include/linux/mm/folio_zone.h
new file mode 100644
index 000000000000..572fe37068f1
--- /dev/null
+++ b/include/linux/mm/folio_zone.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_FOLIO_ZONE_H
+#define _LINUX_MM_FOLIO_ZONE_H
+
+#include <linux/mm_types.h> // for struct page, struct folio
+#include <linux/mmzone.h> // for ZONEID_*, NODES_*
+#include <linux/page-flags.h> // for PF_POISONED_CHECK()
+
+/*
+ * The identification function is mainly used by the buddy allocator for
+ * determining if two pages could be buddies. We are not really identifying
+ * the zone since we could be using the section number id if we do not have
+ * node id available in page flags.
+ * We only guarantee that it will return the same value for two combinable
+ * pages in a zone.
+ */
+static inline int page_zone_id(struct page *page)
+{
+ return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
+}
+
+#ifdef NODE_NOT_IN_PAGE_FLAGS
+int page_to_nid(const struct page *page);
+#else
+static inline int page_to_nid(const struct page *page)
+{
+ return (PF_POISONED_CHECK(page)->flags >> NODES_PGSHIFT) & NODES_MASK;
+}
+#endif
+
+static inline int folio_nid(const struct folio *folio)
+{
+ return page_to_nid(&folio->page);
+}
+
+#endif /* _LINUX_MM_FOLIO_ZONE_H */
--
2.39.2
This is needed to eliminate linux/skbuff.h's dependency on linux/mm.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 45 +-----------------------------
include/linux/mm/pfmemalloc.h | 52 +++++++++++++++++++++++++++++++++++
2 files changed, 53 insertions(+), 44 deletions(-)
create mode 100644 include/linux/mm/pfmemalloc.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index b3f83fb26aca..07262ae43c5d 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -8,6 +8,7 @@
#include <linux/mm/folio_zone.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
+#include <linux/mm/pfmemalloc.h>
#include <linux/errno.h>
#include <linux/mmdebug.h>
#include <linux/gfp.h>
@@ -1763,50 +1764,6 @@ static inline pgoff_t page_index(struct page *page)
return page->index;
}
-/*
- * Return true only if the page has been allocated with
- * ALLOC_NO_WATERMARKS and the low watermark was not
- * met implying that the system is under some pressure.
- */
-static inline bool page_is_pfmemalloc(const struct page *page)
-{
- /*
- * lru.next has bit 1 set if the page is allocated from the
- * pfmemalloc reserves. Callers may simply overwrite it if
- * they do not need to preserve that information.
- */
- return (uintptr_t)page->lru.next & BIT(1);
-}
-
-/*
- * Return true only if the folio has been allocated with
- * ALLOC_NO_WATERMARKS and the low watermark was not
- * met implying that the system is under some pressure.
- */
-static inline bool folio_is_pfmemalloc(const struct folio *folio)
-{
- /*
- * lru.next has bit 1 set if the page is allocated from the
- * pfmemalloc reserves. Callers may simply overwrite it if
- * they do not need to preserve that information.
- */
- return (uintptr_t)folio->lru.next & BIT(1);
-}
-
-/*
- * Only to be called by the page allocator on a freshly allocated
- * page.
- */
-static inline void set_page_pfmemalloc(struct page *page)
-{
- page->lru.next = (void *)BIT(1);
-}
-
-static inline void clear_page_pfmemalloc(struct page *page)
-{
- page->lru.next = NULL;
-}
-
/*
* Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
*/
diff --git a/include/linux/mm/pfmemalloc.h b/include/linux/mm/pfmemalloc.h
new file mode 100644
index 000000000000..345b215a3566
--- /dev/null
+++ b/include/linux/mm/pfmemalloc.h
@@ -0,0 +1,52 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_PFMEMALLOC_H
+#define _LINUX_MM_PFMEMALLOC_H
+
+#include <linux/bits.h> // for BIT()
+#include <linux/mm_types.h> // for struct page
+
+/*
+ * Return true only if the page has been allocated with
+ * ALLOC_NO_WATERMARKS and the low watermark was not
+ * met implying that the system is under some pressure.
+ */
+static inline bool page_is_pfmemalloc(const struct page *page)
+{
+ /*
+ * lru.next has bit 1 set if the page is allocated from the
+ * pfmemalloc reserves. Callers may simply overwrite it if
+ * they do not need to preserve that information.
+ */
+ return (uintptr_t)page->lru.next & BIT(1);
+}
+
+/*
+ * Return true only if the folio has been allocated with
+ * ALLOC_NO_WATERMARKS and the low watermark was not
+ * met implying that the system is under some pressure.
+ */
+static inline bool folio_is_pfmemalloc(const struct folio *folio)
+{
+ /*
+ * lru.next has bit 1 set if the page is allocated from the
+ * pfmemalloc reserves. Callers may simply overwrite it if
+ * they do not need to preserve that information.
+ */
+ return (uintptr_t)folio->lru.next & BIT(1);
+}
+
+/*
+ * Only to be called by the page allocator on a freshly allocated
+ * page.
+ */
+static inline void set_page_pfmemalloc(struct page *page)
+{
+ page->lru.next = (void *)BIT(1);
+}
+
+static inline void clear_page_pfmemalloc(struct page *page)
+{
+ page->lru.next = NULL;
+}
+
+#endif /* _LINUX_MM_PFMEMALLOC_H */
--
2.39.2
This is needed to eliminate linux/dma-mapping.h's dependency on
linux/mm.h.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 25 +------------------------
include/linux/mm/vmalloc_addr.h | 33 +++++++++++++++++++++++++++++++++
2 files changed, 34 insertions(+), 24 deletions(-)
create mode 100644 include/linux/mm/vmalloc_addr.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 07262ae43c5d..80fc7df2856a 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -9,6 +9,7 @@
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/mm/pfmemalloc.h>
+#include <linux/mm/vmalloc_addr.h>
#include <linux/errno.h>
#include <linux/mmdebug.h>
#include <linux/gfp.h>
@@ -1086,30 +1087,6 @@ enum {
int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
unsigned long desc);
-/* Support for virtually mapped pages */
-struct page *vmalloc_to_page(const void *addr);
-unsigned long vmalloc_to_pfn(const void *addr);
-
-/*
- * Determine if an address is within the vmalloc range
- *
- * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
- * is no special casing required.
- */
-#ifdef CONFIG_MMU
-extern bool is_vmalloc_addr(const void *x);
-extern int is_vmalloc_or_module_addr(const void *x);
-#else
-static inline bool is_vmalloc_addr(const void *x)
-{
- return false;
-}
-static inline int is_vmalloc_or_module_addr(const void *x)
-{
- return 0;
-}
-#endif
-
/*
* How many times the entire folio is mapped as a single unit (eg by a
* PMD or PUD entry). This is probably not what you want, except for
diff --git a/include/linux/mm/vmalloc_addr.h b/include/linux/mm/vmalloc_addr.h
new file mode 100644
index 000000000000..86ad2dc94960
--- /dev/null
+++ b/include/linux/mm/vmalloc_addr.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_VMALLOC_ADDR_H
+#define _LINUX_MM_VMALLOC_ADDR_H
+
+#include <linux/types.h> // for bool
+
+struct page;
+
+/* Support for virtually mapped pages */
+struct page *vmalloc_to_page(const void *addr);
+unsigned long vmalloc_to_pfn(const void *addr);
+
+/*
+ * Determine if an address is within the vmalloc range
+ *
+ * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
+ * is no special casing required.
+ */
+#ifdef CONFIG_MMU
+extern bool is_vmalloc_addr(const void *x);
+extern int is_vmalloc_or_module_addr(const void *x);
+#else
+static inline bool is_vmalloc_addr(const void *x)
+{
+ return false;
+}
+static inline int is_vmalloc_or_module_addr(const void *x)
+{
+ return 0;
+}
+#endif
+
+#endif /* _LINUX_MM_VMALLOC_ADDR_H */
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
This new header contains wrappers for the low-level functions from
page_ref.h. By having those higher-level functions in a separate
header, we can avoid their additional dependencies in the page_ref.h.
Having these in a separate header will allow eliminating the
dependency on linux/mm.h from these headers:
- linux/skbuff.h
- linux/swap.h
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/mm.h | 172 +------------------------------
include/linux/mm/folio_usage.h | 182 +++++++++++++++++++++++++++++++++
2 files changed, 183 insertions(+), 171 deletions(-)
create mode 100644 include/linux/mm/folio_usage.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 61f1312a626e..e8a914e7bebd 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,9 +2,9 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
-#include <linux/mm/devmap_managed.h>
#include <linux/mm/folio_next.h>
#include <linux/mm/folio_size.h>
+#include <linux/mm/folio_usage.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/errno.h>
@@ -1073,51 +1073,6 @@ struct inode;
#include <linux/huge_mm.h>
-/*
- * Methods to modify the page usage count.
- *
- * What counts for a page usage:
- * - cache mapping (page->mapping)
- * - private data (page->private)
- * - page mapped in a task's page tables, each mapping
- * is counted separately
- *
- * Also, many kernel routines increase the page count before a critical
- * routine so they can be sure the page doesn't go away from under them.
- */
-
-/*
- * Drop a ref, return true if the refcount fell to zero (the page has no users)
- */
-static inline int put_page_testzero(struct page *page)
-{
- VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
- return page_ref_dec_and_test(page);
-}
-
-static inline int folio_put_testzero(struct folio *folio)
-{
- return put_page_testzero(&folio->page);
-}
-
-/*
- * Try to grab a ref unless the page has a refcount of zero, return false if
- * that is the case.
- * This can be called when MMU is off so it must not access
- * any of the virtual mappings.
- */
-static inline bool get_page_unless_zero(struct page *page)
-{
- return page_ref_add_unless(page, 1, 0);
-}
-
-static inline struct folio *folio_get_nontail_page(struct page *page)
-{
- if (unlikely(!get_page_unless_zero(page)))
- return NULL;
- return (struct folio *)page;
-}
-
extern int page_is_ram(unsigned long pfn);
enum {
@@ -1266,8 +1221,6 @@ static inline struct folio *virt_to_folio(const void *x)
return page_folio(page);
}
-void __folio_put(struct folio *folio);
-
void put_pages_list(struct list_head *pages);
void split_page(struct page *page, unsigned int order);
@@ -1358,129 +1311,6 @@ vm_fault_t finish_fault(struct vm_fault *vmf);
* back into memory.
*/
-/* 127: arbitrary random number, small enough to assemble well */
-#define folio_ref_zero_or_close_to_overflow(folio) \
- ((unsigned int) folio_ref_count(folio) + 127u <= 127u)
-
-/**
- * folio_get - Increment the reference count on a folio.
- * @folio: The folio.
- *
- * Context: May be called in any context, as long as you know that
- * you have a refcount on the folio. If you do not already have one,
- * folio_try_get() may be the right interface for you to use.
- */
-static inline void folio_get(struct folio *folio)
-{
- VM_BUG_ON_FOLIO(folio_ref_zero_or_close_to_overflow(folio), folio);
- folio_ref_inc(folio);
-}
-
-static inline void get_page(struct page *page)
-{
- folio_get(page_folio(page));
-}
-
-static inline __must_check bool try_get_page(struct page *page)
-{
- page = compound_head(page);
- if (WARN_ON_ONCE(page_ref_count(page) <= 0))
- return false;
- page_ref_inc(page);
- return true;
-}
-
-/**
- * folio_put - Decrement the reference count on a folio.
- * @folio: The folio.
- *
- * If the folio's reference count reaches zero, the memory will be
- * released back to the page allocator and may be used by another
- * allocation immediately. Do not access the memory or the struct folio
- * after calling folio_put() unless you can be sure that it wasn't the
- * last reference.
- *
- * Context: May be called in process or interrupt context, but not in NMI
- * context. May be called while holding a spinlock.
- */
-static inline void folio_put(struct folio *folio)
-{
- if (folio_put_testzero(folio))
- __folio_put(folio);
-}
-
-/**
- * folio_put_refs - Reduce the reference count on a folio.
- * @folio: The folio.
- * @refs: The amount to subtract from the folio's reference count.
- *
- * If the folio's reference count reaches zero, the memory will be
- * released back to the page allocator and may be used by another
- * allocation immediately. Do not access the memory or the struct folio
- * after calling folio_put_refs() unless you can be sure that these weren't
- * the last references.
- *
- * Context: May be called in process or interrupt context, but not in NMI
- * context. May be called while holding a spinlock.
- */
-static inline void folio_put_refs(struct folio *folio, int refs)
-{
- if (folio_ref_sub_and_test(folio, refs))
- __folio_put(folio);
-}
-
-void folios_put_refs(struct folio_batch *folios, unsigned int *refs);
-
-/*
- * union release_pages_arg - an array of pages or folios
- *
- * release_pages() releases a simple array of multiple pages, and
- * accepts various different forms of said page array: either
- * a regular old boring array of pages, an array of folios, or
- * an array of encoded page pointers.
- *
- * The transparent union syntax for this kind of "any of these
- * argument types" is all kinds of ugly, so look away.
- */
-typedef union {
- struct page **pages;
- struct folio **folios;
- struct encoded_page **encoded_pages;
-} release_pages_arg __attribute__ ((__transparent_union__));
-
-void release_pages(release_pages_arg, int nr);
-
-/**
- * folios_put - Decrement the reference count on an array of folios.
- * @folios: The folios.
- *
- * Like folio_put(), but for a batch of folios. This is more efficient
- * than writing the loop yourself as it will optimise the locks which need
- * to be taken if the folios are freed. The folios batch is returned
- * empty and ready to be reused for another batch; there is no need to
- * reinitialise it.
- *
- * Context: May be called in process or interrupt context, but not in NMI
- * context. May be called while holding a spinlock.
- */
-static inline void folios_put(struct folio_batch *folios)
-{
- folios_put_refs(folios, NULL);
-}
-
-static inline void put_page(struct page *page)
-{
- struct folio *folio = page_folio(page);
-
- /*
- * For some devmap managed pages we need to catch refcount transition
- * from 2 to 1:
- */
- if (put_devmap_managed_page(&folio->page))
- return;
- folio_put(folio);
-}
-
/*
* GUP_PIN_COUNTING_BIAS, and the associated functions that use it, overload
* the page's refcount so that two separate items are tracked: the original page
diff --git a/include/linux/mm/folio_usage.h b/include/linux/mm/folio_usage.h
new file mode 100644
index 000000000000..4a7e9cd74909
--- /dev/null
+++ b/include/linux/mm/folio_usage.h
@@ -0,0 +1,182 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_FOLIO_USAGE_H
+#define _LINUX_MM_FOLIO_USAGE_H
+
+#include <linux/mm/devmap_managed.h> // for put_devmap_managed_page()
+#include <linux/mmdebug.h> // for VM_BUG_ON_PAGE()
+#include <linux/mm_types.h> // for struct folio
+#include <linux/page_ref.h>
+
+struct folio_batch;
+
+/*
+ * Methods to modify the page usage count.
+ *
+ * What counts for a page usage:
+ * - cache mapping (page->mapping)
+ * - private data (page->private)
+ * - page mapped in a task's page tables, each mapping
+ * is counted separately
+ *
+ * Also, many kernel routines increase the page count before a critical
+ * routine so they can be sure the page doesn't go away from under them.
+ */
+
+/*
+ * Drop a ref, return true if the refcount fell to zero (the page has no users)
+ */
+static inline int put_page_testzero(struct page *page)
+{
+ VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
+ return page_ref_dec_and_test(page);
+}
+
+static inline int folio_put_testzero(struct folio *folio)
+{
+ return put_page_testzero(&folio->page);
+}
+
+/*
+ * Try to grab a ref unless the page has a refcount of zero, return false if
+ * that is the case.
+ * This can be called when MMU is off so it must not access
+ * any of the virtual mappings.
+ */
+static inline bool get_page_unless_zero(struct page *page)
+{
+ return page_ref_add_unless(page, 1, 0);
+}
+
+static inline struct folio *folio_get_nontail_page(struct page *page)
+{
+ if (unlikely(!get_page_unless_zero(page)))
+ return NULL;
+ return (struct folio *)page;
+}
+
+void __folio_put(struct folio *folio);
+
+/* 127: arbitrary random number, small enough to assemble well */
+#define folio_ref_zero_or_close_to_overflow(folio) \
+ ((unsigned int) folio_ref_count(folio) + 127u <= 127u)
+
+/**
+ * folio_get - Increment the reference count on a folio.
+ * @folio: The folio.
+ *
+ * Context: May be called in any context, as long as you know that
+ * you have a refcount on the folio. If you do not already have one,
+ * folio_try_get() may be the right interface for you to use.
+ */
+static inline void folio_get(struct folio *folio)
+{
+ VM_BUG_ON_FOLIO(folio_ref_zero_or_close_to_overflow(folio), folio);
+ folio_ref_inc(folio);
+}
+
+static inline void get_page(struct page *page)
+{
+ folio_get(page_folio(page));
+}
+
+static inline __must_check bool try_get_page(struct page *page)
+{
+ page = compound_head(page);
+ if (WARN_ON_ONCE(page_ref_count(page) <= 0))
+ return false;
+ page_ref_inc(page);
+ return true;
+}
+
+/**
+ * folio_put - Decrement the reference count on a folio.
+ * @folio: The folio.
+ *
+ * If the folio's reference count reaches zero, the memory will be
+ * released back to the page allocator and may be used by another
+ * allocation immediately. Do not access the memory or the struct folio
+ * after calling folio_put() unless you can be sure that it wasn't the
+ * last reference.
+ *
+ * Context: May be called in process or interrupt context, but not in NMI
+ * context. May be called while holding a spinlock.
+ */
+static inline void folio_put(struct folio *folio)
+{
+ if (folio_put_testzero(folio))
+ __folio_put(folio);
+}
+
+/**
+ * folio_put_refs - Reduce the reference count on a folio.
+ * @folio: The folio.
+ * @refs: The amount to subtract from the folio's reference count.
+ *
+ * If the folio's reference count reaches zero, the memory will be
+ * released back to the page allocator and may be used by another
+ * allocation immediately. Do not access the memory or the struct folio
+ * after calling folio_put_refs() unless you can be sure that these weren't
+ * the last references.
+ *
+ * Context: May be called in process or interrupt context, but not in NMI
+ * context. May be called while holding a spinlock.
+ */
+static inline void folio_put_refs(struct folio *folio, int refs)
+{
+ if (folio_ref_sub_and_test(folio, refs))
+ __folio_put(folio);
+}
+
+void folios_put_refs(struct folio_batch *folios, unsigned int *refs);
+
+/*
+ * union release_pages_arg - an array of pages or folios
+ *
+ * release_pages() releases a simple array of multiple pages, and
+ * accepts various different forms of said page array: either
+ * a regular old boring array of pages, an array of folios, or
+ * an array of encoded page pointers.
+ *
+ * The transparent union syntax for this kind of "any of these
+ * argument types" is all kinds of ugly, so look away.
+ */
+typedef union {
+ struct page **pages;
+ struct folio **folios;
+ struct encoded_page **encoded_pages;
+} release_pages_arg __attribute__ ((__transparent_union__));
+
+void release_pages(release_pages_arg, int nr);
+
+/**
+ * folios_put - Decrement the reference count on an array of folios.
+ * @folios: The folios.
+ *
+ * Like folio_put(), but for a batch of folios. This is more efficient
+ * than writing the loop yourself as it will optimise the locks which need
+ * to be taken if the folios are freed. The folios batch is returned
+ * empty and ready to be reused for another batch; there is no need to
+ * reinitialise it.
+ *
+ * Context: May be called in process or interrupt context, but not in NMI
+ * context. May be called while holding a spinlock.
+ */
+static inline void folios_put(struct folio_batch *folios)
+{
+ folios_put_refs(folios, NULL);
+}
+
+static inline void put_page(struct page *page)
+{
+ struct folio *folio = page_folio(page);
+
+ /*
+ * For some devmap managed pages we need to catch refcount transition
+ * from 2 to 1:
+ */
+ if (put_devmap_managed_page(&folio->page))
+ return;
+ folio_put(folio);
+}
+
+#endif /* _LINUX_MM_FOLIO_USAGE_H */
--
2.39.2
This variable is used by lots of arch/*/include/asm/ headers, but
these do not (and should not) include the huge linux/mm.h header.
Let's move this "extern" variable to a separate header and include
this one in arch/.
Signed-off-by: Max Kellermann <[email protected]>
---
arch/arm/include/asm/memory.h | 4 ++++
arch/arm/include/asm/pgtable.h | 2 ++
arch/csky/include/asm/page.h | 1 +
arch/hexagon/include/asm/mem-layout.h | 4 ++++
arch/m68k/include/asm/page_mm.h | 1 +
arch/m68k/include/asm/pgtable_mm.h | 1 +
arch/parisc/include/asm/floppy.h | 1 +
arch/powerpc/include/asm/book3s/32/pgtable.h | 4 ++++
arch/powerpc/include/asm/nohash/32/pgtable.h | 1 +
arch/powerpc/include/asm/page.h | 1 +
arch/x86/include/asm/floppy.h | 1 +
arch/x86/include/asm/pgtable_32_areas.h | 4 ++++
drivers/mtd/nand/onenand/onenand_samsung.c | 1 +
include/linux/mm.h | 2 +-
include/linux/mm/high_memory.h | 7 +++++++
15 files changed, 34 insertions(+), 1 deletion(-)
create mode 100644 include/linux/mm/high_memory.h
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h
index ef2aa79ece5a..a67afb213e2e 100644
--- a/arch/arm/include/asm/memory.h
+++ b/arch/arm/include/asm/memory.h
@@ -387,6 +387,10 @@ static inline unsigned long __virt_to_idmap(unsigned long x)
*/
#define ARCH_PFN_OFFSET PHYS_PFN_OFFSET
+#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
+#endif
+
#define virt_to_page(kaddr) pfn_to_page(virt_to_pfn(kaddr))
#define virt_addr_valid(kaddr) (((unsigned long)(kaddr) >= PAGE_OFFSET && (unsigned long)(kaddr) < (unsigned long)high_memory) \
&& pfn_valid(virt_to_pfn(kaddr)))
diff --git a/arch/arm/include/asm/pgtable.h b/arch/arm/include/asm/pgtable.h
index be91e376df79..eb80f6a65619 100644
--- a/arch/arm/include/asm/pgtable.h
+++ b/arch/arm/include/asm/pgtable.h
@@ -11,6 +11,8 @@
#include <asm/proc-fns.h>
#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
+
/*
* ZERO_PAGE is a global shared page that is always zero: used
* for zero-mapped memory areas etc..
diff --git a/arch/csky/include/asm/page.h b/arch/csky/include/asm/page.h
index 0ca6c408c07f..17ccc91c3cd6 100644
--- a/arch/csky/include/asm/page.h
+++ b/arch/csky/include/asm/page.h
@@ -32,6 +32,7 @@
#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
#include <linux/pfn.h>
#define virt_addr_valid(kaddr) ((void *)(kaddr) >= (void *)PAGE_OFFSET && \
diff --git a/arch/hexagon/include/asm/mem-layout.h b/arch/hexagon/include/asm/mem-layout.h
index e2f99413fe56..a09116c50043 100644
--- a/arch/hexagon/include/asm/mem-layout.h
+++ b/arch/hexagon/include/asm/mem-layout.h
@@ -10,6 +10,10 @@
#include <linux/const.h>
+#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
+#endif
+
/*
* Have to do this for ginormous numbers, else they get printed as
* negative numbers, which the linker no likey when you try to
diff --git a/arch/m68k/include/asm/page_mm.h b/arch/m68k/include/asm/page_mm.h
index e0ae4d5fc985..f958655d1931 100644
--- a/arch/m68k/include/asm/page_mm.h
+++ b/arch/m68k/include/asm/page_mm.h
@@ -5,6 +5,7 @@
#ifndef __ASSEMBLY__
#include <linux/compiler.h>
+#include <linux/mm/high_memory.h>
#include <asm/module.h>
/*
diff --git a/arch/m68k/include/asm/pgtable_mm.h b/arch/m68k/include/asm/pgtable_mm.h
index dbdf1c2b2f66..ec593656bdc5 100644
--- a/arch/m68k/include/asm/pgtable_mm.h
+++ b/arch/m68k/include/asm/pgtable_mm.h
@@ -13,6 +13,7 @@
#ifndef __ASSEMBLY__
#include <asm/processor.h>
+#include <linux/mm/high_memory.h>
#include <linux/sched.h>
#include <linux/threads.h>
diff --git a/arch/parisc/include/asm/floppy.h b/arch/parisc/include/asm/floppy.h
index b318a7df52f6..98ed37c5dc59 100644
--- a/arch/parisc/include/asm/floppy.h
+++ b/arch/parisc/include/asm/floppy.h
@@ -8,6 +8,7 @@
#ifndef __ASM_PARISC_FLOPPY_H
#define __ASM_PARISC_FLOPPY_H
+#include <linux/mm/high_memory.h>
#include <linux/vmalloc.h>
diff --git a/arch/powerpc/include/asm/book3s/32/pgtable.h b/arch/powerpc/include/asm/book3s/32/pgtable.h
index 52971ee30717..03e0a32e1c2c 100644
--- a/arch/powerpc/include/asm/book3s/32/pgtable.h
+++ b/arch/powerpc/include/asm/book3s/32/pgtable.h
@@ -4,6 +4,10 @@
#include <asm-generic/pgtable-nopmd.h>
+#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
+#endif
+
/*
* The "classic" 32-bit implementation of the PowerPC MMU uses a hash
* table containing PTEs, together with a set of 16 segment registers,
diff --git a/arch/powerpc/include/asm/nohash/32/pgtable.h b/arch/powerpc/include/asm/nohash/32/pgtable.h
index 9164a9e41b02..8a9f5b546e4a 100644
--- a/arch/powerpc/include/asm/nohash/32/pgtable.h
+++ b/arch/powerpc/include/asm/nohash/32/pgtable.h
@@ -5,6 +5,7 @@
#include <asm-generic/pgtable-nopmd.h>
#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
#include <linux/sched.h>
#include <linux/threads.h>
#include <asm/mmu.h> /* For sub-arch specific PPC_PIN_SIZE */
diff --git a/arch/powerpc/include/asm/page.h b/arch/powerpc/include/asm/page.h
index e411e5a70ea3..9ae87fc2c648 100644
--- a/arch/powerpc/include/asm/page.h
+++ b/arch/powerpc/include/asm/page.h
@@ -7,6 +7,7 @@
*/
#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/bug.h>
diff --git a/arch/x86/include/asm/floppy.h b/arch/x86/include/asm/floppy.h
index 6ec3fc969ad5..7756e984f146 100644
--- a/arch/x86/include/asm/floppy.h
+++ b/arch/x86/include/asm/floppy.h
@@ -10,6 +10,7 @@
#ifndef _ASM_X86_FLOPPY_H
#define _ASM_X86_FLOPPY_H
+#include <linux/mm/high_memory.h>
#include <linux/vmalloc.h>
/*
diff --git a/arch/x86/include/asm/pgtable_32_areas.h b/arch/x86/include/asm/pgtable_32_areas.h
index b6355416a15a..b339137b4f4e 100644
--- a/arch/x86/include/asm/pgtable_32_areas.h
+++ b/arch/x86/include/asm/pgtable_32_areas.h
@@ -3,6 +3,10 @@
#include <asm/cpu_entry_area.h>
+#ifndef __ASSEMBLY__
+#include <linux/mm/high_memory.h>
+#endif
+
/*
* Just any arbitrary offset to the start of the vmalloc VM area: the
* current 8MB value just means that there will be a 8MB "hole" after the
diff --git a/drivers/mtd/nand/onenand/onenand_samsung.c b/drivers/mtd/nand/onenand/onenand_samsung.c
index fd6890a03d55..7c3cc270386d 100644
--- a/drivers/mtd/nand/onenand/onenand_samsung.c
+++ b/drivers/mtd/nand/onenand/onenand_samsung.c
@@ -15,6 +15,7 @@
#include <linux/platform_device.h>
#include <linux/sched.h>
#include <linux/slab.h>
+#include <linux/mm/high_memory.h>
#include <linux/mtd/mtd.h>
#include <linux/mtd/onenand.h>
#include <linux/mtd/partitions.h>
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 80fc7df2856a..1d7b4c781995 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -6,6 +6,7 @@
#include <linux/mm/folio_size.h>
#include <linux/mm/folio_usage.h>
#include <linux/mm/folio_zone.h>
+#include <linux/mm/high_memory.h>
#include <linux/mm/page_address.h>
#include <linux/mm/page_section.h>
#include <linux/mm/pfmemalloc.h>
@@ -82,7 +83,6 @@ static inline void totalram_pages_add(long count)
atomic_long_add(count, &_totalram_pages);
}
-extern void * high_memory;
extern int page_cluster;
extern const int page_cluster_max;
diff --git a/include/linux/mm/high_memory.h b/include/linux/mm/high_memory.h
new file mode 100644
index 000000000000..e504aafd0d48
--- /dev/null
+++ b/include/linux/mm/high_memory.h
@@ -0,0 +1,7 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_HIGH_MEMORY_H
+#define _LINUX_MM_HIGH_MEMORY_H
+
+extern void * high_memory;
+
+#endif /* _LINUX_MM_HIGH_MEMORY_H */
--
2.39.2
Replace <linux/mm.h> with the smaller pieces that were just splitted
out. This affects a few headers that are included by many, e.g. bio.h
and highmem.h, which now no longer depend on the fat <linux/mm.h>
header.
For this, several missing includes need to be added because they are
no longer indirectly included, e.g. <linux/fs.h>.
Signed-off-by: Max Kellermann <[email protected]>
---
include/linux/bio.h | 2 ++
include/linux/dma-mapping.h | 1 +
include/linux/highmem-internal.h | 2 ++
include/linux/highmem.h | 4 +++-
include/linux/scatterlist.h | 2 +-
include/linux/skbuff.h | 4 ++++
lib/scatterlist.c | 1 +
7 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/include/linux/bio.h b/include/linux/bio.h
index 875d792bffff..e2f24d7235d3 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -5,7 +5,9 @@
#ifndef __LINUX_BIO_H
#define __LINUX_BIO_H
+#include <linux/fs.h> // for struct kiocb, IOCB_NOWAIT
#include <linux/mempool.h>
+#include <linux/mm/folio_next.h>
/* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
#include <linux/blk_types.h>
#include <linux/uio.h>
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 4a658de44ee9..37d5591039c7 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -11,6 +11,7 @@
#include <linux/scatterlist.h>
#include <linux/bug.h>
#include <linux/mem_encrypt.h>
+#include <linux/mm/vmalloc_addr.h>
/**
* List of possible attributes associated with a DMA mapping. The semantics
diff --git a/include/linux/highmem-internal.h b/include/linux/highmem-internal.h
index a3028e400a9c..a479e6b7c54c 100644
--- a/include/linux/highmem-internal.h
+++ b/include/linux/highmem-internal.h
@@ -2,6 +2,8 @@
#ifndef _LINUX_HIGHMEM_INTERNAL_H
#define _LINUX_HIGHMEM_INTERNAL_H
+#include <linux/mm/page_address.h>
+
/*
* Outside of CONFIG_HIGHMEM to support X86 32bit iomap_atomic() cruft.
*/
diff --git a/include/linux/highmem.h b/include/linux/highmem.h
index 00341b56d291..10a165a913ad 100644
--- a/include/linux/highmem.h
+++ b/include/linux/highmem.h
@@ -7,7 +7,9 @@
#include <linux/bug.h>
#include <linux/cacheflush.h>
#include <linux/kmsan.h>
-#include <linux/mm.h>
+#include <linux/mm/folio_size.h> // for page_size()
+#include <linux/mm/folio_usage.h> // for folio_put()
+#include <linux/mm/page_address.h>
#include <linux/uaccess.h>
#include <linux/hardirq.h>
diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h
index 0516e64dc03e..9221bba82b23 100644
--- a/include/linux/scatterlist.h
+++ b/include/linux/scatterlist.h
@@ -5,7 +5,7 @@
#include <linux/string.h>
#include <linux/types.h>
#include <linux/bug.h>
-#include <linux/mm.h>
+#include <linux/mm/page_address.h>
#include <asm/io.h>
#ifdef CONFIG_UML
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 3023bc2be6a1..7b61eac3baa4 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -16,6 +16,10 @@
#include <linux/bug.h>
#include <linux/bvec.h>
#include <linux/cache.h>
+#include <linux/gfp.h> // for alloc_pages_node()
+#include <linux/mm/folio_usage.h> // for get_page(), put_page()
+#include <linux/mm/folio_zone.h> // for page_to_nid()
+#include <linux/mm/pfmemalloc.h> // for page_is_pfmemalloc()
#include <linux/rbtree.h>
#include <linux/socket.h>
#include <linux/refcount.h>
diff --git a/lib/scatterlist.c b/lib/scatterlist.c
index 68b45c82c37a..03599396f15d 100644
--- a/lib/scatterlist.c
+++ b/lib/scatterlist.c
@@ -9,6 +9,7 @@
#include <linux/scatterlist.h>
#include <linux/highmem.h>
#include <linux/kmemleak.h>
+#include <linux/mm.h>
#include <linux/bvec.h>
#include <linux/uio.h>
--
2.39.2
After eliminating includes of linux/mm.h, the following build failure
occurred:
./include/drm/drm_file.h:443:45: warning: ‘struct poll_table_struct’ declared inside parameter list will not be visible outside of this definition or declaration
443 | __poll_t drm_poll(struct file *filp, struct poll_table_struct *wait);
| ^~~~~~~~~~~~~~~~~
In file included from drivers/gpu/drm/imagination/pvr_gem.h:12,
from drivers/gpu/drm/imagination/pvr_fw.h:9:
./include/drm/drm_gem.h:447:27: error: initialization of ‘__poll_t (*)(struct file *, struct poll_table_struct *)’ {aka ‘unsigned int (*)(struct file *, struct poll_table_struct *)’} from incompatible pointer type ‘__poll_t (*)(struct file *, struct poll_table_struct *)’ {aka ‘unsigned int (*)(struct file *, struct poll_table_struct *)’} [-Werror=incompatible-pointer-types]
447 | .poll = drm_poll,\
| ^~~~~~~~
The compiler is confused, and that can be fixed easily by
forward-declaring the struct expicitly.
Signed-off-by: Max Kellermann <[email protected]>
---
include/drm/drm_file.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/drm/drm_file.h b/include/drm/drm_file.h
index ab230d3af138..f24ade9f766f 100644
--- a/include/drm/drm_file.h
+++ b/include/drm/drm_file.h
@@ -44,6 +44,7 @@ struct drm_device;
struct drm_printer;
struct device;
struct file;
+struct poll_table_struct;
/*
* FIXME: Not sure we want to have drm_minor here in the end, but to avoid
--
2.39.2
Prepare to reduce dependencies on linux/mm.h.
page_kasan_tag() is used by the page_to_virt() macro in ARM64 (but
asm/memory.h does not include linux/mm.h). Thus, in order to be able to use
anything that calls page_to_virt(), linux/mm.h needs to be included.
This would prevent us from moving page_address() to a separate header,
because it calls lowmem_page_address() which in turn calls
page_to_virt(). To prepare for this move, we move page_kasan_tag()
out of linux/mm.h into a separate lean header.
A side effect of this patch is that the <linux/kasan.h> include line
is moved inside the "#ifdef CONFIG_KASAN..." block, i.e. it is not
included at all if KASAN is disabled.
Signed-off-by: Max Kellermann <[email protected]>
---
MAINTAINERS | 1 +
include/linux/mm.h | 56 +-------------------------
include/linux/mm/page_kasan_tag.h | 66 +++++++++++++++++++++++++++++++
3 files changed, 68 insertions(+), 55 deletions(-)
create mode 100644 include/linux/mm/page_kasan_tag.h
diff --git a/MAINTAINERS b/MAINTAINERS
index b43102ca365d..f2d32e2da433 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -14218,6 +14218,7 @@ F: include/linux/mempolicy.h
F: include/linux/mempool.h
F: include/linux/memremap.h
F: include/linux/mm.h
+F: include/linux/mm/*.h
F: include/linux/mm_*.h
F: include/linux/mmzone.h
F: include/linux/mmu_notifier.h
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 0436b919f1c7..e140ee45f49c 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2,6 +2,7 @@
#ifndef _LINUX_MM_H
#define _LINUX_MM_H
+#include <linux/mm/page_kasan_tag.h>
#include <linux/errno.h>
#include <linux/mmdebug.h>
#include <linux/gfp.h>
@@ -27,7 +28,6 @@
#include <linux/sizes.h>
#include <linux/sched.h>
#include <linux/pgtable.h>
-#include <linux/kasan.h>
#include <linux/memremap.h>
#include <linux/slab.h>
@@ -1818,60 +1818,6 @@ static inline void vma_set_access_pid_bit(struct vm_area_struct *vma)
}
#endif /* CONFIG_NUMA_BALANCING */
-#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
-
-/*
- * KASAN per-page tags are stored xor'ed with 0xff. This allows to avoid
- * setting tags for all pages to native kernel tag value 0xff, as the default
- * value 0x00 maps to 0xff.
- */
-
-static inline u8 page_kasan_tag(const struct page *page)
-{
- u8 tag = KASAN_TAG_KERNEL;
-
- if (kasan_enabled()) {
- tag = (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK;
- tag ^= 0xff;
- }
-
- return tag;
-}
-
-static inline void page_kasan_tag_set(struct page *page, u8 tag)
-{
- unsigned long old_flags, flags;
-
- if (!kasan_enabled())
- return;
-
- tag ^= 0xff;
- old_flags = READ_ONCE(page->flags);
- do {
- flags = old_flags;
- flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT);
- flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
- } while (unlikely(!try_cmpxchg(&page->flags, &old_flags, flags)));
-}
-
-static inline void page_kasan_tag_reset(struct page *page)
-{
- if (kasan_enabled())
- page_kasan_tag_set(page, KASAN_TAG_KERNEL);
-}
-
-#else /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
-
-static inline u8 page_kasan_tag(const struct page *page)
-{
- return 0xff;
-}
-
-static inline void page_kasan_tag_set(struct page *page, u8 tag) { }
-static inline void page_kasan_tag_reset(struct page *page) { }
-
-#endif /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
-
static inline struct zone *page_zone(const struct page *page)
{
return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
diff --git a/include/linux/mm/page_kasan_tag.h b/include/linux/mm/page_kasan_tag.h
new file mode 100644
index 000000000000..1210c62170a3
--- /dev/null
+++ b/include/linux/mm/page_kasan_tag.h
@@ -0,0 +1,66 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_MM_PAGE_KASAN_TAG_H
+#define _LINUX_MM_PAGE_KASAN_TAG_H
+
+#include <linux/types.h>
+
+struct page;
+
+#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
+
+#include <linux/kasan.h>
+#include <linux/mm_types.h> // for struct page
+
+/*
+ * KASAN per-page tags are stored xor'ed with 0xff. This allows to avoid
+ * setting tags for all pages to native kernel tag value 0xff, as the default
+ * value 0x00 maps to 0xff.
+ */
+
+static inline u8 page_kasan_tag(const struct page *page)
+{
+ u8 tag = KASAN_TAG_KERNEL;
+
+ if (kasan_enabled()) {
+ tag = (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK;
+ tag ^= 0xff;
+ }
+
+ return tag;
+}
+
+static inline void page_kasan_tag_set(struct page *page, u8 tag)
+{
+ unsigned long old_flags, flags;
+
+ if (!kasan_enabled())
+ return;
+
+ tag ^= 0xff;
+ old_flags = READ_ONCE(page->flags);
+ do {
+ flags = old_flags;
+ flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT);
+ flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
+ } while (unlikely(!try_cmpxchg(&page->flags, &old_flags, flags)));
+}
+
+static inline void page_kasan_tag_reset(struct page *page)
+{
+ if (kasan_enabled())
+ page_kasan_tag_set(page, KASAN_TAG_KERNEL);
+}
+
+#else /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
+
+static inline u8 page_kasan_tag(const struct page *page)
+{
+ return 0xff;
+}
+
+static inline void page_kasan_tag_set(struct page *page, u8 tag) { }
+static inline void page_kasan_tag_reset(struct page *page) { }
+
+#endif /* CONFIG_KASAN_SW_TAGS || CONFIG_KASAN_HW_TAGS */
+
+#endif /* _LINUX_MM_PAGE_KASAN_TAG_H */
--
2.39.2
On 12.03.24 10:41, Max Kellermann wrote:
> This patch set aims to clean up the linux/mm.h header and reduce
> dependencies on it by moving parts out.
>
> The goal was to eliminate dependencies on linux/mm.h from other
> popular headers such as highmem.h and dma-mapping.h, and I started by
> checking which symbols were really used and moved those declarations
> to separate slim headers.
>
[...]
> include/linux/mm.h | 583 +-----------------
> include/linux/mm/devmap_managed.h | 37 ++
> include/linux/mm/folio_next.h | 27 +
Isn't that a bit excessive? Do we really need that many folio headers?
--
Cheers,
David / dhildenb
On Tue, Mar 12, 2024 at 5:10 PM David Hildenbrand <[email protected]> wrote:
> > include/linux/mm.h | 583 +-----------------
> > include/linux/mm/devmap_managed.h | 37 ++
> > include/linux/mm/folio_next.h | 27 +
>
> Isn't that a bit excessive? Do we really need that many folio headers?
It would technically be possible to have fewer headers by merging
several of those headers back into one, but then the dependencies will
be heavier, and eventually we'd be back at the large "mm.h" mess that
I would like to get rid of.
My patch set constitutes the balance of "not too many headers" vs
"small set of unnecessary dependencies for each including source file"
according to my taste.
Max
On 12.03.24 17:27, Max Kellermann wrote:
> On Tue, Mar 12, 2024 at 5:10 PM David Hildenbrand <[email protected]> wrote:
>>> include/linux/mm.h | 583 +-----------------
>>> include/linux/mm/devmap_managed.h | 37 ++
>>> include/linux/mm/folio_next.h | 27 +
>>
>> Isn't that a bit excessive? Do we really need that many folio headers?
>
> It would technically be possible to have fewer headers by merging
> several of those headers back into one, but then the dependencies will
> be heavier, and eventually we'd be back at the large "mm.h" mess that
> I would like to get rid of.
Just curious: why? Usually build time, do you have some numbers?
> My patch set constitutes the balance of "not too many headers" vs
> "small set of unnecessary dependencies for each including source file"
> according to my taste.
I'm not against splitting out stuff. But one function per header is a
bit excessive IMHO. Ideally, we'd have some MM guideline that we'll be
able to follow in the future. So we don't need "personal taste".
For example, if I were to write a folio_prev(), should I put it in
include/linux/mm/folio_prev.h ? Likely we'd put it into folio_next.h,
but then the name doesn't make any sense.
The point I am trying to make: if there was a single folio_ops.h, it
would be clearer where it would go.
Just my 2 cents, seeing one-function-per-file headers.
--
Cheers,
David / dhildenb
On Tue, Mar 12, 2024 at 5:33 PM David Hildenbrand <[email protected]> wrote:
> Just curious: why? Usually build time, do you have some numbers?
(This has been discussed so often and I thought having smaller header
dependencies was already established as general consensus among kernel
devs, and everybody agreed that mm.h and kernel.h are a big mess that
needs cleanup.)
Why: Correctness, less namespace pollution, and the least important
aspect is reduced build times. However, the gains by this patch set
are very small; each optimization gains very little.
Some time ago, I posted a much larger patch set with a few numbers:
https://lore.kernel.org/lkml/[email protected]/
- the speedup was measurable, but not amazing, because even after that
patch set, everybody still includes everything, and much more cleanup
is needed to make a bigger difference. Once the big knots like
kernel.h and mm.h are broken up, we will have better results. And as I
said: build times are nice, but the lesser advantage.
Anyway, this first patch set was so extremely large that nobody was
able to review it. So this patch contains just the parts that deal
with mm.h; later, when this patch set is merged, I can continue with
other headers. (I already have a branch that splits kernel.h and I'll
submit it eventually, after the mm.h cleanup.)
> I'm not against splitting out stuff. But one function per header is a
> bit excessive IMHO.
One function per header is certainly not my goal and I agree it's
excessive; but folio_next() in its own header made sense, just in this
special case, because it allowed removing the mm.h dependency from
bio.h. Removing this dependency was the goal, and folio_next.h was the
solution for this particular problem.
> Ideally, we'd have some MM guideline that we'll be
> able to follow in the future. So we don't need "personal taste".
Agree. But lacking such a guideline, all I can do is make suggestions
and submit patches for review, trying to follow what seemed to be
consensus in previous cleanups and what had previously been merged.
> For example, if I were to write a folio_prev(), should I put it in
> include/linux/mm/folio_prev.h ? Likely we'd put it into folio_next.h,
> but then the name doesn't make any sense.
True. But since no folio_prev() function exists, the only name that
made sense for this header was folio_next.h. If folio_prev() gets
added, I'd say put it in that header, but rename it to, let's say,
"folio_iterator.h". But right now, with just this one function (and
nothing else like it that could be added here), I decided to suggest
naming it "folio_next.h". If you think "folio_iterator.h" or something
else is better, I'll rename and resubmit; names don't matter so much
to me; the general idea of cleaning up the headers is what's
important.
> The point I am trying to make: if there was a single folio_ops.h, it
> would be clearer where it would go.
Maybe, but IMO this wouldn't be a good place to add folio_next(),
because folio_next() doesn't "operate" on a folio, but on a folio
pointer (or "iterator"). Again, names don't matter to me -
ideas/concepts matter. I'm only explaining why I decided to submit my
patches that way, but I'll change them any way you people want.
Max
On 12.03.24 19:11, Max Kellermann wrote:
> On Tue, Mar 12, 2024 at 5:33 PM David Hildenbrand <[email protected]> wrote:
>> Just curious: why? Usually build time, do you have some numbers?
>
> (This has been discussed so often and I thought having smaller header
> dependencies was already established as general consensus among kernel
> devs, and everybody agreed that mm.h and kernel.h are a big mess that
> needs cleanup.)
It's always a good idea to include at least some sentences about
history+motivation in the cover letter.
>
> Why: Correctness, less namespace pollution, and the least important
> aspect is reduced build times. However, the gains by this patch set
> are very small; each optimization gains very little.
>
> Some time ago, I posted a much larger patch set with a few numbers:
> https://lore.kernel.org/lkml/[email protected]/
> - the speedup was measurable, but not amazing, because even after that
> patch set, everybody still includes everything, and much more cleanup
> is needed to make a bigger difference. Once the big knots like
> kernel.h and mm.h are broken up, we will have better results. And as I
> said: build times are nice, but the lesser advantage.
Okay, so "Fast kernel headers:" is a bit misleading :P
I'm asking these stupid questions because I remember some header splitup
(from Christoph?) that really did make a significant build-time difference.
Reading all that, I would have written something like this in the cover
letter:
"There is an agreement that we want to split up some of the big kernel
headers, such as kernel.h and mm.h. While not delivering immediate
build-time gains, it's one step into the desired direction. Besides
build-time, smaller headers promise less namespace pollution and
correctness."
(although I don't immediately understand what correctness means in this
context)
>
> Anyway, this first patch set was so extremely large that nobody was
> able to review it. So this patch contains just the parts that deal
> with mm.h; later, when this patch set is merged, I can continue with
> other headers. (I already have a branch that splits kernel.h and I'll
> submit it eventually, after the mm.h cleanup.)
I'd have continued with
"Some of these changes were part of a previous, bigger patchset [1]. I'm
now sending out smaller, reviewable pieces. Splitting up mm.h is the
first step."
[1]
https://lore.kernel.org/lkml/[email protected]/
>
>> I'm not against splitting out stuff. But one function per header is a
>> bit excessive IMHO.
>
> One function per header is certainly not my goal and I agree it's
> excessive; but folio_next() in its own header made sense, just in this
> special case, because it allowed removing the mm.h dependency from
> bio.h. Removing this dependency was the goal, and folio_next.h was the
> solution for this particular problem.
Okay, I see. I still wonder if merging some of the new folio headers in
"folio.h" wouldn't have achieved something close to that. Sure, they
bring in some other dependencies, but hopefully not mm.h
Anyhow, what you are describing here would also have been reasonable to
describe somewhere in few words. IOW, which approach did you chose to
decide when a new header was reasonable.
>
>> Ideally, we'd have some MM guideline that we'll be
>> able to follow in the future. So we don't need "personal taste".
>
> Agree. But lacking such a guideline, all I can do is make suggestions
> and submit patches for review, trying to follow what seemed to be
> consensus in previous cleanups and what had previously been merged.
Possibly it makes sense to have some rough guideline. The problem I see
is that once your stuff is merged, it will start getting messed up again
over time. But maybe that's just the way it works.
>
>> For example, if I were to write a folio_prev(), should I put it in
>> include/linux/mm/folio_prev.h ? Likely we'd put it into folio_next.h,
>> but then the name doesn't make any sense.
>
> True. But since no folio_prev() function exists, the only name that
> made sense for this header was folio_next.h. If folio_prev() gets
> added, I'd say put it in that header, but rename it to, let's say,
> "folio_iterator.h". But right now, with just this one function (and
> nothing else like it that could be added here), I decided to suggest
> naming it "folio_next.h". If you think "folio_iterator.h" or something
> else is better, I'll rename and resubmit; names don't matter so much
> to me; the general idea of cleaning up the headers is what's
> important.
>
>> The point I am trying to make: if there was a single folio_ops.h, it
>> would be clearer where it would go.
>
> Maybe, but IMO this wouldn't be a good place to add folio_next(),
> because folio_next() doesn't "operate" on a folio, but on a folio
> pointer (or "iterator"). Again, names don't matter to me -
> ideas/concepts matter. I'm only explaining why I decided to submit my
> patches that way, but I'll change them any way you people want.
Thanks for the details.
--
Cheers,
David / dhildenb
On Tue, Mar 12, 2024 at 8:26 PM David Hildenbrand <[email protected]> wrote:
> It's always a good idea to include at least some sentences about
> history+motivation in the cover letter.
Agree, thanks for your review and for the suggestions, that was very
useful. Once somebody points it out, it's suddenly obvious what was
missing in my explanations for others to understand my motivation.
I'll add some more text to the cover letter.
> Okay, so "Fast kernel headers:" is a bit misleading :P
Yeah. It was the name chosen by Ingo Molnar for his header cleanup
project, and some of his work was merged several years ago, so I
thought this phrase would be familiar and would bring back memories
from these discussions.
> (although I don't immediately understand what correctness means in this
> context)
Some sources don't include all the headers they need; but by chance,
this happens to work because deep down the include tree, somebody
includes kernel.h, mm.h or fs.h and that pulls in everything else;
thus incorrect includes are never a visible problem. By reducing
includes to the bare minimum, many of those problems are revealed (by
producing compiler errors that then have to be fixed). I very much
like to write code in a way that the compiler points out some kinds of
mistakes - I rather have a build-time error than a silent runtime
breakage.
Incorrect includes are not just a theoretical problem; imagine all the
asm-generic headers that provide fallback definitions for macros that
are not defined on some architectures (e.g. asm-generic/cacheflush.h).
If, on some sources, the arch-specific header happens to be not
included or defines the macro in the wrong header (I've seen it all),
you suddenly randomly get the correct arch macro or the generic
fallback macro, leading to inconsistencies, hidden ABI breakages that
may lead to severe problems that will be hard to debug. One can
imagine many more ways that code can break silently at runtime due to
incorrect code (that usually works most of the time, but only by
chance). Better be explicit about what you really need to include, and
don't rely on other headers to indirectly include what you need.
Include what you need even if somebody else down the tree had already
included it, because that somebody else may change his mind at some
point, and then your source shouldn't break.
> The problem I see
> is that once your stuff is merged, it will start getting messed up again
> over time. But maybe that's just the way it works.
It is. This is like cleaning your house: the following day, somebody
visits you and brings nice presents (feature patches) but has dirty
shoes, and then you have to clean the house again... that's the
infinite cycle of software development, as we all know it.
Max