From c01a5d89e5c8abe638107be2a4ea9e4c7fcdd7f6 Mon Sep 17 00:00:00 2001 From: Wang Long Date: Fri, 2 Aug 2024 17:19:01 +0800 Subject: percpu-rwsem: remove the unused parameter 'read' In the function percpu_rwsem_release, the parameter `read` is unused, so remove it. Signed-off-by: Wang Long Link: https://lore.kernel.org/r/20240802091901.2546797-1-w@laoqinren.net Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- include/linux/fs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index fb0426f349fc..c0286d479c9d 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1683,7 +1683,7 @@ static inline bool __sb_start_write_trylock(struct super_block *sb, int level) #define __sb_writers_acquired(sb, lev) \ percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) #define __sb_writers_release(sb, lev) \ - percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) + percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], _THIS_IP_) /** * __sb_write_started - check if sb freeze level is held -- cgit v1.2.3 From 8447d848e1dc6d42d1dcd00f133d7715fc732c47 Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sat, 10 Aug 2024 08:47:53 +0200 Subject: vfs: only read fops once in fops_get/put In do_dentry_open() the usage is: f->f_op = fops_get(inode->i_fop); In generated asm the compiler emits 2 reads from inode->i_fop instead of just one. This popped up due to false-sharing where loads from that offset end up bouncing a cacheline during parallel open. While this is going to be fixed, the spurious load does not need to be there. This makes do_dentry_open() go down from 1177 to 1154 bytes. fops_put() is patched to maintain some consistency. No functional changes. Reviewed-by: Josef Bacik Signed-off-by: Mateusz Guzik Link: https://lore.kernel.org/r/20240810064753.1211441-1-mjguzik@gmail.com Signed-off-by: Christian Brauner --- include/linux/fs.h | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index c0286d479c9d..696c620f90e0 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2553,10 +2553,17 @@ struct super_block *sget(struct file_system_type *type, struct super_block *sget_dev(struct fs_context *fc, dev_t dev); /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ -#define fops_get(fops) \ - (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) -#define fops_put(fops) \ - do { if (fops) module_put((fops)->owner); } while(0) +#define fops_get(fops) ({ \ + const struct file_operations *_fops = (fops); \ + (((_fops) && try_module_get((_fops)->owner) ? (_fops) : NULL)); \ +}) + +#define fops_put(fops) ({ \ + const struct file_operations *_fops = (fops); \ + if (_fops) \ + module_put((_fops)->owner); \ +}) + /* * This one is to be used *ONLY* from ->open() instances. * fops must be non-NULL, pinned down *and* module dependencies -- cgit v1.2.3 From 641bb4394f405cba498b100b44541ffc0aed5be1 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 9 Aug 2024 12:38:56 +0200 Subject: fs: move FMODE_UNSIGNED_OFFSET to fop_flags This is another flag that is statically set and doesn't need to use up an FMODE_* bit. Move it to ->fop_flags and free up another FMODE_* bit. (1) mem_open() used from proc_mem_operations (2) adi_open() used from adi_fops (3) drm_open_helper(): (3.1) accel_open() used from DRM_ACCEL_FOPS (3.2) drm_open() used from (3.2.1) amdgpu_driver_kms_fops (3.2.2) psb_gem_fops (3.2.3) i915_driver_fops (3.2.4) nouveau_driver_fops (3.2.5) panthor_drm_driver_fops (3.2.6) radeon_driver_kms_fops (3.2.7) tegra_drm_fops (3.2.8) vmwgfx_driver_fops (3.2.9) xe_driver_fops (3.2.10) DRM_GEM_FOPS (3.2.11) DEFINE_DRM_GEM_DMA_FOPS (4) struct memdev sets fmode flags based on type of device opened. For devices using struct mem_fops unsigned offset is used. Mark all these file operations as FOP_UNSIGNED_OFFSET and add asserts into the open helper to ensure that the flag is always set. Link: https://lore.kernel.org/r/20240809-work-fop_unsigned-v1-1-658e054d893e@kernel.org Reviewed-by: Jeff Layton Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- drivers/char/adi.c | 8 +------- drivers/char/mem.c | 3 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 1 + drivers/gpu/drm/drm_file.c | 3 ++- drivers/gpu/drm/gma500/psb_drv.c | 1 + drivers/gpu/drm/i915/i915_driver.c | 1 + drivers/gpu/drm/nouveau/nouveau_drm.c | 1 + drivers/gpu/drm/radeon/radeon_drv.c | 1 + drivers/gpu/drm/tegra/drm.c | 1 + drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 1 + drivers/gpu/drm/xe/xe_device.c | 1 + fs/proc/base.c | 10 ++++------ fs/read_write.c | 2 +- include/drm/drm_accel.h | 3 ++- include/drm/drm_gem.h | 3 ++- include/drm/drm_gem_dma_helper.h | 1 + include/linux/fs.h | 5 +++-- mm/mmap.c | 2 +- 18 files changed, 27 insertions(+), 21 deletions(-) (limited to 'include/linux/fs.h') diff --git a/drivers/char/adi.c b/drivers/char/adi.c index 751d7cc0da1b..1c76c8758f0f 100644 --- a/drivers/char/adi.c +++ b/drivers/char/adi.c @@ -14,12 +14,6 @@ #define MAX_BUF_SZ PAGE_SIZE -static int adi_open(struct inode *inode, struct file *file) -{ - file->f_mode |= FMODE_UNSIGNED_OFFSET; - return 0; -} - static int read_mcd_tag(unsigned long addr) { long err; @@ -206,9 +200,9 @@ static loff_t adi_llseek(struct file *file, loff_t offset, int whence) static const struct file_operations adi_fops = { .owner = THIS_MODULE, .llseek = adi_llseek, - .open = adi_open, .read = adi_read, .write = adi_write, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static struct miscdevice adi_miscdev = { diff --git a/drivers/char/mem.c b/drivers/char/mem.c index 7c359cc406d5..169eed162a7f 100644 --- a/drivers/char/mem.c +++ b/drivers/char/mem.c @@ -643,6 +643,7 @@ static const struct file_operations __maybe_unused mem_fops = { .get_unmapped_area = get_unmapped_area_mem, .mmap_capabilities = memory_mmap_capabilities, #endif + .fop_flags = FOP_UNSIGNED_OFFSET, }; static const struct file_operations null_fops = { @@ -693,7 +694,7 @@ static const struct memdev { umode_t mode; } devlist[] = { #ifdef CONFIG_DEVMEM - [DEVMEM_MINOR] = { "mem", &mem_fops, FMODE_UNSIGNED_OFFSET, 0 }, + [DEVMEM_MINOR] = { "mem", &mem_fops, 0, 0 }, #endif [3] = { "null", &null_fops, FMODE_NOWAIT, 0666 }, #ifdef CONFIG_DEVPORT diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 094498a0964b..d7ef8cbecf6c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2908,6 +2908,7 @@ static const struct file_operations amdgpu_driver_kms_fops = { #ifdef CONFIG_PROC_FS .show_fdinfo = drm_show_fdinfo, #endif + .fop_flags = FOP_UNSIGNED_OFFSET, }; int amdgpu_file_to_fpriv(struct file *filp, struct amdgpu_fpriv **fpriv) diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c index 714e42b05108..f8de3cba1a08 100644 --- a/drivers/gpu/drm/drm_file.c +++ b/drivers/gpu/drm/drm_file.c @@ -318,6 +318,8 @@ int drm_open_helper(struct file *filp, struct drm_minor *minor) if (dev->switch_power_state != DRM_SWITCH_POWER_ON && dev->switch_power_state != DRM_SWITCH_POWER_DYNAMIC_OFF) return -EINVAL; + if (WARN_ON_ONCE(!(filp->f_op->fop_flags & FOP_UNSIGNED_OFFSET))) + return -EINVAL; drm_dbg_core(dev, "comm=\"%s\", pid=%d, minor=%d\n", current->comm, task_pid_nr(current), minor->index); @@ -335,7 +337,6 @@ int drm_open_helper(struct file *filp, struct drm_minor *minor) } filp->private_data = priv; - filp->f_mode |= FMODE_UNSIGNED_OFFSET; priv->filp = filp; mutex_lock(&dev->filelist_mutex); diff --git a/drivers/gpu/drm/gma500/psb_drv.c b/drivers/gpu/drm/gma500/psb_drv.c index 8b64f61ffaf9..d67c2b3ad901 100644 --- a/drivers/gpu/drm/gma500/psb_drv.c +++ b/drivers/gpu/drm/gma500/psb_drv.c @@ -498,6 +498,7 @@ static const struct file_operations psb_gem_fops = { .mmap = drm_gem_mmap, .poll = drm_poll, .read = drm_read, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static const struct drm_driver driver = { diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c index fb8e9c2fcea5..cf276299bccb 100644 --- a/drivers/gpu/drm/i915/i915_driver.c +++ b/drivers/gpu/drm/i915/i915_driver.c @@ -1671,6 +1671,7 @@ static const struct file_operations i915_driver_fops = { #ifdef CONFIG_PROC_FS .show_fdinfo = drm_show_fdinfo, #endif + .fop_flags = FOP_UNSIGNED_OFFSET, }; static int diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index a58c31089613..e243b42f8582 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -1274,6 +1274,7 @@ nouveau_driver_fops = { .compat_ioctl = nouveau_compat_ioctl, #endif .llseek = noop_llseek, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static struct drm_driver diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 7bf08164140e..ac49779ed03d 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -520,6 +520,7 @@ static const struct file_operations radeon_driver_kms_fops = { #ifdef CONFIG_COMPAT .compat_ioctl = radeon_kms_compat_ioctl, #endif + .fop_flags = FOP_UNSIGNED_OFFSET, }; static const struct drm_ioctl_desc radeon_ioctls_kms[] = { diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c index 03d1c76aec2d..108c26a33edb 100644 --- a/drivers/gpu/drm/tegra/drm.c +++ b/drivers/gpu/drm/tegra/drm.c @@ -801,6 +801,7 @@ static const struct file_operations tegra_drm_fops = { .read = drm_read, .compat_ioctl = drm_compat_ioctl, .llseek = noop_llseek, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static int tegra_drm_context_cleanup(int id, void *p, void *data) diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 50ad3105c16e..2825dd3149ed 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1609,6 +1609,7 @@ static const struct file_operations vmwgfx_driver_fops = { .compat_ioctl = vmw_compat_ioctl, #endif .llseek = noop_llseek, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static const struct drm_driver driver = { diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c index f2f1d8ddb221..2e5537d6608d 100644 --- a/drivers/gpu/drm/xe/xe_device.c +++ b/drivers/gpu/drm/xe/xe_device.c @@ -238,6 +238,7 @@ static const struct file_operations xe_driver_fops = { #ifdef CONFIG_PROC_FS .show_fdinfo = drm_show_fdinfo, #endif + .fop_flags = FOP_UNSIGNED_OFFSET, }; static struct drm_driver driver = { diff --git a/fs/proc/base.c b/fs/proc/base.c index 72a1acd03675..1409d1003101 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -827,12 +827,9 @@ static int __mem_open(struct inode *inode, struct file *file, unsigned int mode) static int mem_open(struct inode *inode, struct file *file) { - int ret = __mem_open(inode, file, PTRACE_MODE_ATTACH); - - /* OK to pass negative loff_t, we can catch out-of-range */ - file->f_mode |= FMODE_UNSIGNED_OFFSET; - - return ret; + if (WARN_ON_ONCE(!(file->f_op->fop_flags & FOP_UNSIGNED_OFFSET))) + return -EINVAL; + return __mem_open(inode, file, PTRACE_MODE_ATTACH); } static ssize_t mem_rw(struct file *file, char __user *buf, @@ -932,6 +929,7 @@ static const struct file_operations proc_mem_operations = { .write = mem_write, .open = mem_open, .release = mem_release, + .fop_flags = FOP_UNSIGNED_OFFSET, }; static int environ_open(struct inode *inode, struct file *file) diff --git a/fs/read_write.c b/fs/read_write.c index 90e283b31ca1..89d4af0e3b93 100644 --- a/fs/read_write.c +++ b/fs/read_write.c @@ -36,7 +36,7 @@ EXPORT_SYMBOL(generic_ro_fops); static inline bool unsigned_offsets(struct file *file) { - return file->f_mode & FMODE_UNSIGNED_OFFSET; + return file->f_op->fop_flags & FOP_UNSIGNED_OFFSET; } /** diff --git a/include/drm/drm_accel.h b/include/drm/drm_accel.h index f4d3784b1dce..41c78b7d712c 100644 --- a/include/drm/drm_accel.h +++ b/include/drm/drm_accel.h @@ -28,7 +28,8 @@ .poll = drm_poll,\ .read = drm_read,\ .llseek = noop_llseek, \ - .mmap = drm_gem_mmap + .mmap = drm_gem_mmap, \ + .fop_flags = FOP_UNSIGNED_OFFSET /** * DEFINE_DRM_ACCEL_FOPS() - macro to generate file operations for accelerators drivers diff --git a/include/drm/drm_gem.h b/include/drm/drm_gem.h index bae4865b2101..d8b86df2ec0d 100644 --- a/include/drm/drm_gem.h +++ b/include/drm/drm_gem.h @@ -447,7 +447,8 @@ struct drm_gem_object { .poll = drm_poll,\ .read = drm_read,\ .llseek = noop_llseek,\ - .mmap = drm_gem_mmap + .mmap = drm_gem_mmap, \ + .fop_flags = FOP_UNSIGNED_OFFSET /** * DEFINE_DRM_GEM_FOPS() - macro to generate file operations for GEM drivers diff --git a/include/drm/drm_gem_dma_helper.h b/include/drm/drm_gem_dma_helper.h index a827bde494f6..f2678e7ecb98 100644 --- a/include/drm/drm_gem_dma_helper.h +++ b/include/drm/drm_gem_dma_helper.h @@ -267,6 +267,7 @@ unsigned long drm_gem_dma_get_unmapped_area(struct file *filp, .read = drm_read,\ .llseek = noop_llseek,\ .mmap = drm_gem_mmap,\ + .fop_flags = FOP_UNSIGNED_OFFSET, \ DRM_GEM_DMA_UNMAPPED_AREA_FOPS \ } diff --git a/include/linux/fs.h b/include/linux/fs.h index 696c620f90e0..0145bda465ff 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -146,8 +146,7 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, /* Expect random access pattern */ #define FMODE_RANDOM ((__force fmode_t)(1 << 12)) -/* File is huge (eg. /dev/mem): treat loff_t as unsigned */ -#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)(1 << 13)) +/* FMODE_* bit 13 */ /* File is opened with O_PATH; almost nothing can be done with it */ #define FMODE_PATH ((__force fmode_t)(1 << 14)) @@ -2073,6 +2072,8 @@ struct file_operations { #define FOP_DIO_PARALLEL_WRITE ((__force fop_flags_t)(1 << 3)) /* Contains huge pages */ #define FOP_HUGE_PAGES ((__force fop_flags_t)(1 << 4)) +/* Treat loff_t as unsigned (e.g., /dev/mem) */ +#define FOP_UNSIGNED_OFFSET ((__force fop_flags_t)(1 << 5)) /* Wrap a directory iterator that needs exclusive inode access */ int wrap_directory_iterator(struct file *, struct dir_context *, diff --git a/mm/mmap.c b/mm/mmap.c index d0dfc85b209b..6ddb278a5ee8 100644 --- a/mm/mmap.c +++ b/mm/mmap.c @@ -1229,7 +1229,7 @@ static inline u64 file_mmap_size_max(struct file *file, struct inode *inode) return MAX_LFS_FILESIZE; /* Special "we do even unsigned file positions" case */ - if (file->f_mode & FMODE_UNSIGNED_OFFSET) + if (file->f_op->fop_flags & FOP_UNSIGNED_OFFSET) return 0; /* Yes, random drivers might want more. But I'm tired of buggy drivers */ -- cgit v1.2.3 From 1c48d441468c425e878d81c5c3e7ff8a9a594cc0 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 16 Aug 2024 16:35:52 +0200 Subject: inode: remove __I_DIO_WAKEUP Afaict, we can just rely on inode->i_dio_count for waiting instead of this awkward indirection through __I_DIO_WAKEUP. This survives LTP dio and xfstests dio tests. Link: https://lore.kernel.org/r/20240816-vfs-misc-dio-v1-1-80fe21a2c710@kernel.org Reviewed-by: Josef Bacik Reviewed-by: Jeff Layton Signed-off-by: Christian Brauner --- fs/inode.c | 23 +++++++++++------------ fs/netfs/locking.c | 22 +++++----------------- include/linux/fs.h | 8 +++----- 3 files changed, 19 insertions(+), 34 deletions(-) (limited to 'include/linux/fs.h') diff --git a/fs/inode.c b/fs/inode.c index de0969d6009c..248a131a02c3 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -2500,18 +2500,11 @@ EXPORT_SYMBOL(inode_owner_or_capable); /* * Direct i/o helper functions */ -static void __inode_dio_wait(struct inode *inode) +bool inode_dio_finished(const struct inode *inode) { - wait_queue_head_t *wq = bit_waitqueue(&inode->i_state, __I_DIO_WAKEUP); - DEFINE_WAIT_BIT(q, &inode->i_state, __I_DIO_WAKEUP); - - do { - prepare_to_wait(wq, &q.wq_entry, TASK_UNINTERRUPTIBLE); - if (atomic_read(&inode->i_dio_count)) - schedule(); - } while (atomic_read(&inode->i_dio_count)); - finish_wait(wq, &q.wq_entry); + return atomic_read(&inode->i_dio_count) == 0; } +EXPORT_SYMBOL(inode_dio_finished); /** * inode_dio_wait - wait for outstanding DIO requests to finish @@ -2525,11 +2518,17 @@ static void __inode_dio_wait(struct inode *inode) */ void inode_dio_wait(struct inode *inode) { - if (atomic_read(&inode->i_dio_count)) - __inode_dio_wait(inode); + wait_var_event(&inode->i_dio_count, inode_dio_finished(inode)); } EXPORT_SYMBOL(inode_dio_wait); +void inode_dio_wait_interruptible(struct inode *inode) +{ + wait_var_event_interruptible(&inode->i_dio_count, + inode_dio_finished(inode)); +} +EXPORT_SYMBOL(inode_dio_wait_interruptible); + /* * inode_set_flags - atomically set some inode flags * diff --git a/fs/netfs/locking.c b/fs/netfs/locking.c index 75dc52a49b3a..21eab56ee2f9 100644 --- a/fs/netfs/locking.c +++ b/fs/netfs/locking.c @@ -19,25 +19,13 @@ * Must be called under a lock that serializes taking new references * to i_dio_count, usually by inode->i_mutex. */ -static int inode_dio_wait_interruptible(struct inode *inode) +static int netfs_inode_dio_wait_interruptible(struct inode *inode) { - if (!atomic_read(&inode->i_dio_count)) + if (inode_dio_finished(inode)) return 0; - wait_queue_head_t *wq = bit_waitqueue(&inode->i_state, __I_DIO_WAKEUP); - DEFINE_WAIT_BIT(q, &inode->i_state, __I_DIO_WAKEUP); - - for (;;) { - prepare_to_wait(wq, &q.wq_entry, TASK_INTERRUPTIBLE); - if (!atomic_read(&inode->i_dio_count)) - break; - if (signal_pending(current)) - break; - schedule(); - } - finish_wait(wq, &q.wq_entry); - - return atomic_read(&inode->i_dio_count) ? -ERESTARTSYS : 0; + inode_dio_wait_interruptible(inode); + return !inode_dio_finished(inode) ? -ERESTARTSYS : 0; } /* Call with exclusively locked inode->i_rwsem */ @@ -46,7 +34,7 @@ static int netfs_block_o_direct(struct netfs_inode *ictx) if (!test_bit(NETFS_ICTX_ODIRECT, &ictx->flags)) return 0; clear_bit(NETFS_ICTX_ODIRECT, &ictx->flags); - return inode_dio_wait_interruptible(&ictx->inode); + return netfs_inode_dio_wait_interruptible(&ictx->inode); } /** diff --git a/include/linux/fs.h b/include/linux/fs.h index 0145bda465ff..8f8d274929d7 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2373,8 +2373,6 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, * * I_REFERENCED Marks the inode as recently references on the LRU list. * - * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). - * * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to * synchronize competing switching instances and to tell * wb stat updates to grab the i_pages lock. See @@ -2409,8 +2407,6 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, #define __I_SYNC 7 #define I_SYNC (1 << __I_SYNC) #define I_REFERENCED (1 << 8) -#define __I_DIO_WAKEUP 9 -#define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) #define I_LINKABLE (1 << 10) #define I_DIRTY_TIME (1 << 11) #define I_WB_SWITCH (1 << 13) @@ -3227,7 +3223,9 @@ static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, } #endif +bool inode_dio_finished(const struct inode *inode); void inode_dio_wait(struct inode *inode); +void inode_dio_wait_interruptible(struct inode *inode); /** * inode_dio_begin - signal start of a direct I/O requests @@ -3251,7 +3249,7 @@ static inline void inode_dio_begin(struct inode *inode) static inline void inode_dio_end(struct inode *inode) { if (atomic_dec_and_test(&inode->i_dio_count)) - wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); + wake_up_var(&inode->i_dio_count); } extern void inode_set_flags(struct inode *inode, unsigned int flags, -- cgit v1.2.3 From 71ff58ce3428b2471efae5b00594e892b285c97c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 22 Aug 2024 11:30:58 +0200 Subject: fs: s/__u32/u32/ for s_fsnotify_mask The underscore variants are for uapi whereas the non-underscore variants are for in-kernel consumers. Link: https://lore.kernel.org/r/20240822-anwerben-nutzung-1cd6c82a565f@brauner Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- include/linux/fs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index 8f8d274929d7..cda1f2545ea0 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1266,7 +1266,7 @@ struct super_block { time64_t s_time_min; time64_t s_time_max; #ifdef CONFIG_FSNOTIFY - __u32 s_fsnotify_mask; + u32 s_fsnotify_mask; struct fsnotify_sb_info *s_fsnotify_info; #endif -- cgit v1.2.3 From da18ecbf0fb6fb73e6f9273e7aa15610f148ce17 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 23 Aug 2024 14:47:35 +0200 Subject: fs: add i_state helpers The i_state member is an unsigned long so that it can be used with the wait bit infrastructure which expects unsigned long. This wastes 4 bytes which we're unlikely to ever use. Switch to using the var event wait mechanism using the address of the bit. Thanks to Linus for the address idea. Link: https://lore.kernel.org/r/20240823-work-i_state-v3-1-5cd5fd207a57@kernel.org Reviewed-by: Josef Bacik Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- fs/inode.c | 11 +++++++++++ include/linux/fs.h | 15 +++++++++++++++ 2 files changed, 26 insertions(+) (limited to 'include/linux/fs.h') diff --git a/fs/inode.c b/fs/inode.c index ba1645a09603..1b0f52ebae27 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -472,6 +472,17 @@ static void __inode_add_lru(struct inode *inode, bool rotate) inode->i_state |= I_REFERENCED; } +struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe, + struct inode *inode, u32 bit) +{ + void *bit_address; + + bit_address = inode_state_wait_address(inode, bit); + init_wait_var_entry(wqe, bit_address, 0); + return __var_waitqueue(bit_address); +} +EXPORT_SYMBOL(inode_bit_waitqueue); + /* * Add inode to LRU if needed (inode is unused and clean). * diff --git a/include/linux/fs.h b/include/linux/fs.h index cda1f2545ea0..d35de348d2ec 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -744,6 +744,21 @@ struct inode { void *i_private; /* fs or device private pointer */ } __randomize_layout; +/* + * Get bit address from inode->i_state to use with wait_var_event() + * infrastructre. + */ +#define inode_state_wait_address(inode, bit) ((char *)&(inode)->i_state + (bit)) + +struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe, + struct inode *inode, u32 bit); + +static inline void inode_wake_up_bit(struct inode *inode, u32 bit) +{ + /* Caller is responsible for correct memory barriers. */ + wake_up_var(inode_state_wait_address(inode, bit)); +} + struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); static inline unsigned int i_blocksize(const struct inode *node) -- cgit v1.2.3 From 2ed634c96ed1d6e4ee0497be176f9980866e81cb Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 23 Aug 2024 14:47:36 +0200 Subject: fs: reorder i_state bits so that we can use the first bits to derive unique addresses from i_state. Link: https://lore.kernel.org/r/20240823-work-i_state-v3-2-5cd5fd207a57@kernel.org Reviewed-by: Josef Bacik Reviewed-by: Jeff Layton Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- include/linux/fs.h | 38 +++++++++++++++++++++----------------- 1 file changed, 21 insertions(+), 17 deletions(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index d35de348d2ec..a868c9823c10 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -2410,28 +2410,32 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, * i_count. * * Q: What is the difference between I_WILL_FREE and I_FREEING? + * + * __I_{SYNC,NEW,LRU_ISOLATING} are used to derive unique addresses to wait + * upon. There's one free address left. */ -#define I_DIRTY_SYNC (1 << 0) -#define I_DIRTY_DATASYNC (1 << 1) -#define I_DIRTY_PAGES (1 << 2) -#define __I_NEW 3 +#define __I_NEW 0 #define I_NEW (1 << __I_NEW) -#define I_WILL_FREE (1 << 4) -#define I_FREEING (1 << 5) -#define I_CLEAR (1 << 6) -#define __I_SYNC 7 +#define __I_SYNC 1 #define I_SYNC (1 << __I_SYNC) -#define I_REFERENCED (1 << 8) +#define __I_LRU_ISOLATING 2 +#define I_LRU_ISOLATING (1 << __I_LRU_ISOLATING) + +#define I_DIRTY_SYNC (1 << 3) +#define I_DIRTY_DATASYNC (1 << 4) +#define I_DIRTY_PAGES (1 << 5) +#define I_WILL_FREE (1 << 6) +#define I_FREEING (1 << 7) +#define I_CLEAR (1 << 8) +#define I_REFERENCED (1 << 9) #define I_LINKABLE (1 << 10) #define I_DIRTY_TIME (1 << 11) -#define I_WB_SWITCH (1 << 13) -#define I_OVL_INUSE (1 << 14) -#define I_CREATING (1 << 15) -#define I_DONTCACHE (1 << 16) -#define I_SYNC_QUEUED (1 << 17) -#define I_PINNING_NETFS_WB (1 << 18) -#define __I_LRU_ISOLATING 19 -#define I_LRU_ISOLATING (1 << __I_LRU_ISOLATING) +#define I_WB_SWITCH (1 << 12) +#define I_OVL_INUSE (1 << 13) +#define I_CREATING (1 << 14) +#define I_DONTCACHE (1 << 15) +#define I_SYNC_QUEUED (1 << 16) +#define I_PINNING_NETFS_WB (1 << 17) #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) -- cgit v1.2.3 From 2b111edbe0a9c441605be5cfb73001dc98ec686f Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 23 Aug 2024 14:47:40 +0200 Subject: inode: make i_state a u32 Now that we use the wait var event mechanism make i_state a u32 and free up 4 bytes. This means we currently have two 4 byte holes in struct inode which we can pack. Link: https://lore.kernel.org/r/20240823-work-i_state-v3-6-5cd5fd207a57@kernel.org Reviewed-by: Josef Bacik Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- include/linux/fs.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index a868c9823c10..a1ef8c65d828 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -681,7 +681,8 @@ struct inode { #endif /* Misc */ - unsigned long i_state; + u32 i_state; + /* 32-bit hole */ struct rw_semaphore i_rwsem; unsigned long dirtied_when; /* jiffies of first dirtying */ -- cgit v1.2.3