kevent: Core files.
Subsystems (12)
A message that is not in the archive (sent off-list or not yet received): [email protected]
cover kevent: Generic event handling mechanism.
Generic event handling mechanism.
Kevent is a generic subsytem which allows to handle event notifications. It supports both level and edge triggered events. It is similar to poll/epoll in some cases, but it is more scalable, it is faster and allows to work with essentially eny kind of events.
Events are provided into kernel through control syscall and can be read back through ring buffer or using usual syscalls.
Kevent update (i.e. readiness switching) happens directly from internals of the appropriate state machine of the underlying subsytem (like network, filesystem, timer or any other).
Homepage:
http://tservice.net.ru/~s0mbre/old/?section=projects&item=kevent
Documentation page:
http://linux-net.osdl.org/index.php/Kevent
Consider for inclusion.
With this release I start 3 days resending timeout - i.e. each third day I will send either new version (if something new was requested and agreed to be implemented) or resending with back counter started from three. When back counter hits zero after three resendings I consider there is no interest in subsystem and I will stop further sending.
I really doubt it is a good way to tell the world about my work, and I bet you all tired from those pathos words, but I really would like to get some feedback, since I want to start to work on network AIO, but sending mails into unfeedbackable 'destination' really does not motivate me for that.
Thanks for understanding and your time.
P.S. If you want to be removed from Cc: list just drop me a mail.
Changes from 'take31' patchset:
* Added aio_sendfile_path() - this syscall allows to asynchronosly transfer file specified by provided pathname to destination socket. Opened file descriptor is returned.
* Added trivial scheduler which selects execution thread. It allows to specify given thread 'by-hands', but since kaio provides '-1' it uses round-robin to get processing thread. In theory it can be bound to scheduler statistics or gamma-ray receiver data.
* Number of bug fixes in kevent based AIO mpage_readpages().
Benchmark of the 100 1Mb files transfer (files are in VFS already) using sync sendfile or this new version shows about 10Mb/sec performance win for aio_sendfile_path().
Changes from 'take30' patchset:
* AIO state machine.
* aio_sendfile() implementation.
* moved kevent_user_get/kevent_user_put into header.
* use *zalloc where needed.
Changes from 'take29' patchset:
* new private userspace notifications - allows to queue any userspace private event and then mark it as ready using kevent_ctl(KEVENT_READY) command
* KEVENT_REQ_READY flag - if set kevent will be marked as ready at enqueue time
* port to 2.6.20-rc2 tree (54abb5fcdae74a811ed440ec6556cabc6b24f404 commit)
* use struct kmem_cache instead of kmem_cache_t
* added notificaion type into search key, this allows to have the same id for different types of notifications
Changes from 'take28' patchset:
* optimized af_unix to use socket notifications
* changed ALWAYS_QUEUE behaviour with poll/select notifications - previously kevent was not queued into poll wait queue when ALWAYS_QUEUE flag is set
* added KEVENT_POLL_POLLRDHUP definition into ukevent.h header
* libevent-1.2 patch (Jamal, your request is completed, so I'm waiting two weeks before starting final countdown :)
All regression tests passed successfully except test_evbuffer(), which is crashed on my amd64 linux 2.6 test machine for all types of notifications, probably it was fixed in libevent-1.2a version, I did not check. Patch and README can be found at project homepage.
Changes from 'take27' patchset:
* made kevent default yes in non embedded case.
* added falgs to callback structures - currently used to check if kevent can be requested from kernelspace only (posix timers) or userspace (all others)
Changes from 'take26' patchset:
* made kevent visible in config only in case of embedded setup.
* added comment about KEVENT_MAX number.
* spell fix.
Changes from 'take25' patchset:
* use timespec as timeout parameter.
* added high-resolution timer to handle absolute timeouts.
* added flags to waiting and initialization syscalls.
* kevent_commit() has new_uidx parameter.
* kevent_wait() has old_uidx parameter, which, if not equal to u->uidx, results in immediate wakeup (usefull for the case when entries
are added asynchronously from kernel (not supported for now)).
* added interface to mark any event as ready.
* event POSIX timers support.
* return -ENOSYS if there is no registered event type.
* provided file descriptor must be checked for fifo type (spotted by Eric Dumazet).
* signal notifications.
* documentation update.
* lighttpd patch updated (the latest benchmarks with lighttpd patch can be found in blog).
Changes from 'take24' patchset:
* new (old (new)) ring buffer implementation with kernel and user indexes.
* added initialization syscall instead of opening /dev/kevent
* kevent_commit() syscall to commit ring buffer entries
* changed KEVENT_REQ_WAKEUP_ONE flag to KEVENT_REQ_WAKEUP_ALL, kevent wakes only first thread always if that flag is not set
* KEVENT_REQ_ALWAYS_QUEUE flag. If set, kevent will be queued into ready queue instead of copying back to userspace when kevent is ready immediately when it is added.
* lighttpd patch (Hail! Although nothing really outstanding compared to epoll)
Changes from 'take23' patchset:
* kevent PIPE notifications
* KEVENT_REQ_LAST_CHECK flag, which allows to perform last check at dequeueing time
* fixed poll/select notifications (were broken due to tree manipulations)
* made Documentation/kevent.txt look nice in 80-col terminal
* fix for copy_to_user() failure report for the first kevent (Andrew Morton)
* minor function renames
Changes from 'take22' patchset:
* new ring buffer implementation in process' memory
* wakeup-one-thread flag
* edge-triggered behaviour
Changes from 'take21' patchset:
* minor cleanups (different return values, removed unneded variables, whitespaces and so on)
* fixed bug in kevent removal in case when kevent being removed
is the same as overflow_kevent (spotted by Eric Dumazet)
Changes from 'take20' patchset:
* new ring buffer implementation
* removed artificial limit on possible number of kevents
Changes from 'take19' patchset:
* use __init instead of __devinit
* removed 'default N' from config for user statistic
* removed kevent_user_fini() since kevent can not be unloaded
* use KERN_INFO for statistic output
Changes from 'take18' patchset:
* use __init instead of __devinit
* removed 'default N' from config for user statistic
* removed kevent_user_fini() since kevent can not be unloaded
* use KERN_INFO for statistic output
Changes from 'take17' patchset:
* Use RB tree instead of hash table.
At least for a web sever, frequency of addition/deletion of new kevent is comparable with number of search access, i.e. most of the time events are added, accesed only couple of times and then removed, so it justifies RB tree usage over AVL tree, since the latter does have much slower deletion time (max O(log(N)) compared to 3 ops),
although faster search time (1.44*O(log(N)) vs. 2*O(log(N))).
So for kevents I use RB tree for now and later, when my AVL tree implementation is ready, it will be possible to compare them.
* Changed readiness check for socket notifications.
With both above changes it is possible to achieve more than 3380 req/second compared to 2200, sometimes 2500 req/second for epoll() for trivial web-server and httperf client on the same hardware.
It is possible that above kevent limit is due to maximum allowed kevents in a time limit, which is 4096 events.
Changes from 'take16' patchset:
* misc cleanups (__read_mostly, const ...)
* created special macro which is used for mmap size (number of pages) calculation
* export kevent_socket_notify(), since it is used in network protocols which can be built as modules (IPv6 for example)
Changes from 'take15' patchset:
* converted kevent_timer to high-resolution timers, this forces timer API update at
http://linux-net.osdl.org/index.php/Kevent
* use struct ukevent* instead of void * in syscalls (documentation has been updated)
* added warning in kevent_add_ukevent() if ring has broken index (for testing)
Changes from 'take14' patchset:
* added kevent_wait()
This syscall waits until either timeout expires or at least one event becomes ready. It also commits that @num events from @start are processed by userspace and thus can be be removed or rearmed (depending on it's flags). It can be used for commit events read by userspace through mmap interface. Example userspace code (evtest.c) can be found on project's homepage.
* added socket notifications (send/recv/accept)
Changes from 'take13' patchset:
* do not get lock aroung user data check in __kevent_search()
* fail early if there were no registered callbacks for given type of kevent
* trailing whitespace cleanup
Changes from 'take12' patchset:
* remove non-chardev interface for initialization
* use pointer to kevent_mring instead of unsigned longs
* use aligned 64bit type in raw user data (can be used by high-res timer if needed)
* simplified enqueue/dequeue callbacks and kevent initialization
* use nanoseconds for timeout
* put number of milliseconds into timer's return data
* move some definitions into user-visible header
* removed filenames from comments
Changes from 'take11' patchset:
* include missing headers into patchset
* some trivial code cleanups (use goto instead of if/else games and so on)
* some whitespace cleanups
* check for ready_callback() callback before main loop which should save us some ticks
Changes from 'take10' patchset:
* removed non-existent prototypes
* added helper function for kevent_registered_callbacks
* fixed 80 lines comments issues
* added shared between userspace and kernelspace header instead of embedd them in one
* core restructuring to remove forward declarations
* s o m e w h i t e s p a c e c o d y n g s t y l e c l e a n u p
* use vm_insert_page() instead of remap_pfn_range()
Changes from 'take9' patchset:
* fixed ->nopage method
Changes from 'take8' patchset:
* fixed mmap release bug
* use module_init() instead of late_initcall()
* use better structures for timer notifications
Changes from 'take7' patchset:
* new mmap interface (not tested, waiting for other changes to be acked)
- use nopage() method to dynamically substitue pages
- allocate new page for events only when new added kevent requres it
- do not use ugly index dereferencing, use structure instead
- reduced amount of data in the ring (id and flags), maximum 12 pages on x86 per kevent fd
Changes from 'take6' patchset:
* a lot of comments!
* do not use list poisoning for detection of the fact, that entry is in the list
* return number of ready kevents even if copy*user() fails
* strict check for number of kevents in syscall
* use ARRAY_SIZE for array size calculation
* changed superblock magic number
* use SLAB_PANIC instead of direct panic() call
* changed -E* return values
* a lot of small cleanups and indent fixes
Changes from 'take5' patchset:
* removed compilation warnings about unused wariables when lockdep is not turned on
* do not use internal socket structures, use appropriate (exported) wrappers instead
* removed default 1 second timeout
* removed AIO stuff from patchset
Changes from 'take4' patchset:
* use miscdevice instead of chardevice
* comments fixes
Changes from 'take3' patchset:
* removed serializing mutex from kevent_user_wait()
* moved storage list processing to RCU
* removed lockdep screaming - all storage locks are initialized in the same function, so it was
learned
to differentiate between various cases
* remove kevent from storage if is marked as broken after callback
* fixed a typo in mmaped buffer implementation which would end up in wrong index calcualtion
Changes from 'take2' patchset:
* split kevent_finish_user() to locked and unlocked variants
* do not use KEVENT_STAT ifdefs, use inline functions instead
* use array of callbacks of each type instead of each kevent callback initialization
* changed name of ukevent guarding lock
* use only one kevent lock in kevent_user for all hash buckets instead of per-bucket locks
* do not use kevent_user_ctl structure instead provide needed arguments as syscall parameters
* various indent cleanups
* added optimisation, which is aimed to help when a lot of kevents are being copied from
userspace
* mapped buffer (initial) implementation (no userspace yet)
Changes from 'take1' patchset:
- rebased against 2.6.18-git tree
- removed ioctl controlling
- added new syscall kevent_get_events(int fd, unsigned int min_nr, unsigned int max_nr, unsigned int timeout, void __user *buf, unsigned flags)
- use old syscall kevent_ctl for creation/removing, modification and initial kevent initialization
- use mutuxes instead of semaphores
- added file descriptor check and return error if provided descriptor does not match kevent file operations
- various indent fixes
- removed aio_sendfile() declarations.
Thank you.
PATCH 1/10 kevent: Description.
Description.
Documentation/kevent.txt new file +259 -0
@@ -0,0 +1,259 @@ +Description. + +int kevent_init(struct kevent_ring *ring, unsigned int ring_size, + unsigned int flags); + +num - size of the ring buffer in events +ring - pointer to allocated ring buffer +flags - various flags, see KEVENT_FLAGS_* definitions. + +Return value: kevent control file descriptor or negative error value. + + struct kevent_ring + { + unsigned int ring_kidx, ring_over; + struct ukevent event[0]; + } + +ring_kidx - index in the ring buffer where kernel will put new events + when kevent_wait() or kevent_get_events() is called +ring_over - number of overflows of ring_uidx happend from the start. + Overflow counter is used to prevent situation when two threads + are going to free the same events, but one of them was scheduled + away for too long, so ring indexes were wrapped, so when that + thread will be awakened, it will free not those events, which + it suppose to free. + +Example userspace code (ring_buffer.c) can be found on project's homepage. + +Each kevent syscall can be so called cancellation point in glibc, i.e. when +thread has been cancelled in kevent syscall, thread can be safely removed +and no events will be lost, since each syscall (kevent_wait() or +kevent_get_events()) will copy event into special ring buffer, accessible +from other threads or even processes (if shared memory is used). + +When kevent is removed (not dequeued when it is ready, but just removed), +even if it was ready, it is not copied into ring buffer, since if it is +removed, no one cares about it (otherwise user would wait until it becomes +ready and got it through usual way using kevent_get_events() or kevent_wait()) +and thus no need to copy it to the ring buffer. + +------------------------------------------------------------------------------- + + +int kevent_ctl(int fd, unsigned int cmd, unsigned int num, struct ukevent *arg); + +fd - is the file descriptor referring to the kevent queue to manipulate. +It is created by opening "/dev/kevent" char device, which is created with +dynamic minor number and major number assigned for misc devices. + +cmd - is the requested operation. It can be one of the following: + KEVENT_CTL_ADD - add event notification + KEVENT_CTL_REMOVE - remove event notification + KEVENT_CTL_MODIFY - modify existing notification + KEVENT_CTL_READY - mark existing events as ready, if number of events is zero, + it just wakes up parked in syscall thread + +num - number of struct ukevent in the array pointed to by arg +arg - array of struct ukevent + +Return value: + number of events processed or negative error value. + +When called, kevent_ctl will carry out the operation specified in the +cmd parameter. +------------------------------------------------------------------------------- + + int kevent_get_events(int ctl_fd, unsigned int min_nr, unsigned int max_nr, + struct timespec timeout, struct ukevent *buf, unsigned flags); + +ctl_fd - file descriptor referring to the kevent queue +min_nr - minimum number of completed events that kevent_get_events will block + waiting for +max_nr - number of struct ukevent in buf +timeout - time to wait before returning less than min_nr + events. If this is -1, then wait forever. +buf - pointer to an array of struct ukevent. +flags - various flags, see KEVENT_FLAGS_* definitions. + +Return value: + number of events copied or negative error value. + +kevent_get_events will wait timeout milliseconds for at least min_nr completed +events, copying completed struct ukevents to buf and deleting any +KEVENT_REQ_ONESHOT event requests. In nonblocking mode it returns as many +events as possible, but not more than max_nr. In blocking mode it waits until +timeout or if at least min_nr events are ready. + +This function copies event into ring buffer if it was initialized, if ring buffer +is full, KEVENT_RET_COPY_FAILED flag is set in ret_flags field. +------------------------------------------------------------------------------- + + int kevent_wait(int ctl_fd, unsigned int num, unsigned int old_uidx, + struct timespec timeout, unsigned int flags); + +ctl_fd - file descriptor referring to the kevent queue +num - number of processed kevents +old_uidx - the last index user is aware of +timeout - time to wait until there is free space in kevent queue +flags - various flags, see KEVENT_FLAGS_* definitions. + +Return value: + number of events copied into ring buffer or negative error value. + +This syscall waits until either timeout expires or at least one event becomes +ready. It also copies events into special ring buffer. If ring buffer is full, +it waits until there are ready events and then return. +If kevent is one-shot kevent it is removed in this syscall. +If kevent is edge-triggered (KEVENT_REQ_ET flag is set in 'req_flags') it is +requeued in this syscall for performance reasons. +------------------------------------------------------------------------------- + + int kevent_commit(int ctl_fd, unsigned int new_idx, unsigned int over); + +ctl_fd - file descriptor referring to the kevent queue +new_uidx - the last committed kevent +over - overflow count for given $new_idx value + +Return value: + number of committed kevents or negative error value. + +This function commits, i.e. marks as empty, slots in the ring buffer, so +they can be reused when userspace completes that entries processing. + +Overflow counter is used to prevent situation when two threads are going +to free the same events, but one of them was scheduled away for too long, +so ring indexes were wrapped, so when that thread will be awakened, it +will free not those events, which it suppose to free. + +It is possible that returned number of committed events will be smaller than +requested number - it is possible when several threads try to commit the +same events. +------------------------------------------------------------------------------- + +long aio_sendfile(int kevent_fd, int sock_fd, int in_fd, off_t offset, size_t count); + +kevent_fd - file descriptor referring to the kevent queue +sock_fd - destination socket file descriptor +in_fd - source file descriptor +offset - offset from the beginning of the source file +count - number of bytes to transfer + +Async sendfile implementation. +Returned coockie can be used to determine which entry has been returned by +kevent_get_events() - it will be stored in event.ptr. +event.ret_data will contain number of bytes actually transferred. +------------------------------------------------------------------------------- + + +The bulk of the interface is entirely done through the ukevent struct. +It is used to add event requests, modify existing event requests, +specify which event requests to remove, and return completed events. + +struct ukevent contains the following members: + +struct kevent_id id + Id of this request, e.g. socket number, file descriptor and so on +__u32 type + Event type, e.g. KEVENT_SOCK, KEVENT_INODE, KEVENT_TIMER and so on +__u32 event + Event itself, e.g. SOCK_ACCEPT, INODE_CREATED, TIMER_FIRED +__u32 req_flags + Per-event request flags, + + KEVENT_REQ_ONESHOT + event will be removed when it is ready + + KEVENT_REQ_WAKEUP_ALL + Kevent wakes up only first thread interested in given event, + or all threads if this flag is set. + + KEVENT_REQ_ET + Edge Triggered behaviour. It is an optimisation which allows to move + ready and dequeued (i.e. copied to userspace) event to move into set + of interest for given storage (socket, inode and so on) again. It is + very usefull for cases when the same event should be used many times + (like reading from pipe). It is similar to epoll()'s EPOLLET flag. + + KEVENT_REQ_LAST_CHECK + if set allows to perform the last check on kevent (call appropriate + callback) when kevent is marked as ready and has been removed from + ready queue. If it will be confirmed that kevent is ready + (k->callbacks.callback(k) returns true) then kevent will be copied + to userspace, otherwise it will be requeued back to storage. + Second (checking) call is performed with this bit cleared, so callback + can detect when it was called from kevent_storage_ready() - bit is set, + or kevent_dequeue_ready() - bit is cleared. If kevent will be requeued, + bit will be set again. + + KEVENT_REQ_ALWAYS_QUEUE + If this flag is set kevent will be queued into ready queue if it is + ready at enqueue time, otherwise it will be copied back to userspace + and will not be queued into the storage. + + KEVENT_REQ_READY + If this flag is set, kevent will be marked as ready immediately at enqueue + time. + +__u32 ret_flags + Per-event return flags + + KEVENT_RET_BROKEN + Kevent is broken + + KEVENT_RET_DONE + Kevent processing was finished successfully + + KEVENT_RET_COPY_FAILED + Kevent was not copied into ring buffer due to some error conditions. + +__u32 ret_data + Event return data. Event originator fills it with anything it likes + (for example timer notifications put number of milliseconds when timer + has fired +union { __u32 user[2]; void *ptr; } + User's data. It is not used, just copied to/from user. The whole structure + is aligned to 8 bytes already, so the last union is aligned properly. + +------------------------------------------------------------------------------- + +Kevent waiting syscall flags. + +KEVENT_FLAGS_ABSTIME - provided timespec parameter contains absolute time, + for example Aug 27, 2194, or time(NULL) + 10. + +------------------------------------------------------------------------------- + +Usage + +For KEVENT_CTL_ADD, all fields relevant to the event type must be filled +(id, type, event, req_flags). +After kevent_ctl(..., KEVENT_CTL_ADD, ...) returns each struct's ret_flags +should be checked to see if the event is already broken or done. + +For KEVENT_CTL_MODIFY, the id, req_flags, and user and event fields must be +set and an existing kevent request must have matching id and user fields. If +match is found, req_flags and event are replaced with the newly supplied +values and requeueing is started, so modified kevent can be checked and +probably marked as ready immediately. If a match can't be found, the +passed in ukevent's ret_flags has KEVENT_RET_BROKEN set. KEVENT_RET_DONE is +always set. + +For KEVENT_CTL_REMOVE, the id and user fields must be set and an existing +kevent request must have matching id and user fields. If a match is found, +the kevent request is removed. If a match can't be found, the passed in +ukevent's ret_flags has KEVENT_RET_BROKEN set. KEVENT_RET_DONE is always set. + +For kevent_get_events, the entire structure is returned. + +------------------------------------------------------------------------------- + +Usage cases + +kevent_timer +struct ukevent should contain following fields: + type - KEVENT_TIMER + event - KEVENT_TIMER_FIRED + req_flags - KEVENT_REQ_ONESHOT if you want to fire that timer only once + id.raw[0] - number of seconds after commit when this timer shout expire + id.raw[0] - additional to number of seconds number of nanoseconds
PATCH 2/10 kevent: Core files.
Core files.
This patch includes core kevent files:
* userspace controlling
* kernelspace interfaces
* initialization
* notification state machines
Some bits of documentation can be found on project's homepage (and links from there):
http://tservice.net.ru/~s0mbre/old/?section=projects&item=kevent
arch/i386/kernel/syscall_table.S +7 -0
@@ -319,3 +319,10 @@ ENTRY(sys_call_table) .long sys_move_pages .long sys_getcpu .long sys_epoll_pwait + .long sys_kevent_get_events + .long sys_kevent_ctl /* 320 */ + .long sys_kevent_wait + .long sys_kevent_commit + .long sys_kevent_init + .lond sys_aio_sendfile + .lond sys_aio_sendfile_path
arch/x86_64/ia32/ia32entry.S +8 -1
@@ -714,8 +714,15 @@ ia32_sys_call_table: .quad compat_sys_get_robust_list .quad sys_splice .quad sys_sync_file_range - .quad sys_tee + .quad sys_tee /* 315 */ .quad compat_sys_vmsplice .quad compat_sys_move_pages .quad sys_getcpu + .quad sys_kevent_get_events + .quad sys_kevent_ctl /* 320 */ + .quad sys_kevent_wait + .quad sys_kevent_commit + .quad sys_kevent_init + .quad sys_aio_sendfile + .quad sys_aio_sendfile_path ia32_syscall_end:
include/asm-i386/unistd.h +8 -1
@@ -325,10 +325,17 @@#define __NR_move_pages 317#define __NR_getcpu 318#define __NR_epoll_pwait 319 +#define __NR_kevent_get_events 320 +#define __NR_kevent_ctl 321 +#define __NR_kevent_wait 322 +#define __NR_kevent_commit 323 +#define __NR_kevent_init 324 +#define __NR_aio_sendfile 325 +#define __NR_aio_sendfile_path 326#ifdef __KERNEL__ -#define NR_syscalls 320 +#define NR_syscalls 327#define __ARCH_WANT_IPC_PARSE_VERSION#define __ARCH_WANT_OLD_READDIR
include/asm-x86_64/unistd.h +16 -2
@@ -619,8 +619,22 @@ __SYSCALL(__NR_sync_file_range, sys_sync_file_range) __SYSCALL(__NR_vmsplice, sys_vmsplice)#define __NR_move_pages 279 __SYSCALL(__NR_move_pages, sys_move_pages) - -#define __NR_syscall_max __NR_move_pages +#define __NR_kevent_get_events 280 +__SYSCALL(__NR_kevent_get_events, sys_kevent_get_events) +#define __NR_kevent_ctl 281 +__SYSCALL(__NR_kevent_ctl, sys_kevent_ctl) +#define __NR_kevent_wait 282 +__SYSCALL(__NR_kevent_wait, sys_kevent_wait) +#define __NR_kevent_commit 283 +__SYSCALL(__NR_kevent_commit, sys_kevent_commit) +#define __NR_kevent_init 284 +__SYSCALL(__NR_kevent_init, sys_kevent_init) +#define __NR_aio_sendfile 285 +__SYSCALL(__NR_aio_sendfile, sys_aio_sendfile) +#define __NR_aio_sendfile_path 286 +__SYSCALL(__NR_aio_sendfile_path, sys_aio_sendfile_path) + +#define __NR_syscall_max __NR_aio_sendfile_path#ifndef __NO_STUBS#define __ARCH_WANT_OLD_READDIR
include/linux/kevent.h new file +268 -0
268 lines not on this page · raw message
include/linux/kevent_storage.h new file +11 -0
@@ -0,0 +1,11 @@ +#ifndef __KEVENT_STORAGE_H +#define __KEVENT_STORAGE_H + +struct kevent_storage +{ +void *origin;/* Originator's pointer, e.g. struct sock or struct file. Can be NULL. */ +struct list_head list;/* List of queued kevents. */ +spinlock_t lock;/* Protects users queue. */ +}; + +#endif /* __KEVENT_STORAGE_H */
include/linux/syscalls.h +13 -0
@@ -54,6 +54,8 @@ struct compat_stat;struct compat_timeval ;struct robust_list_head ;struct getcpu_cache ; +struct ukevent ; +struct kevent_ring ;#include <linux/types.h> #include <linux/aio_abi.h>
@@ -603,6 +605,17 @@ asmlinkage long sys_set_robust_list(struct robust_list_head __user *head, size_t len);asmlinkage long sys_getcpu (unsigned __user *cpu,unsigned __user *node,struct getcpu_cache __user *cache); +asmlinkage long sys_kevent_get_events (int ctl_fd,unsigned int min,unsigned int max, +struct timespec timeout,struct ukevent __user *buf,unsigned flags); +asmlinkage long sys_kevent_ctl (int ctl_fd,unsigned int cmd,unsigned int num,struct ukevent __user *buf); +asmlinkage long sys_kevent_wait (int ctl_fd,unsigned int num,unsigned int old_uidx, +struct timespec timeout,unsigned int flags); +asmlinkage long sys_kevent_commit (int ctl_fd,unsigned int new_uidx,unsigned int over); +asmlinkage long sys_kevent_init (int ctl_fd,struct kevent_ring __user *ring,unsigned int num,unsigned int flags); +int kernel_execve (const char *filename,char *const argv[],char *const envp[]); +asmlinkage long sys_aio_sendfile (int kevent_fd,int sock_fd,int in_fd,off_t offset,size_t count); +asmlinkage long sys_aio_sendfile_path (int kevent_fd,int sock_fd,char __user *filename,off_t offset,size_t count); +#endif
include/linux/ukevent.h new file +191 -0
191 lines not on this page · raw message
init/Kconfig +2 -0
@@ -219,6 +219,8 @@ config AUDITSYSCALL such as SELinux. To use audit's filesystem watch feature, please ensure that INOTIFY is configured. +source "kernel/kevent/Kconfig" + config IKCONFIG tristate "Kernel .config support" ---help---
kernel/Makefile +1 -0
@@ -46,6 +46,7 @@ obj-$(CONFIG_DETECT_SOFTLOCKUP) += softlockup.o obj-$(CONFIG_GENERIC_HARDIRQS) += irq/ obj-$(CONFIG_SECCOMP) += seccomp.o obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o +obj-$(CONFIG_KEVENT) += kevent/ obj-$(CONFIG_RELAY) += relay.o obj-$(CONFIG_UTS_NS) += utsname.o obj-$(CONFIG_TASK_DELAY_ACCT) += delayacct.o
kernel/kevent/Kconfig new file +82 -0
82 lines not on this page · raw message
kernel/kevent/Makefile new file +8 -0
@@ -0,0 +1,8 @@ +obj-y := kevent.o kevent_user.o +obj-$(CONFIG_KEVENT_TIMER) += kevent_timer.o +obj-$(CONFIG_KEVENT_POLL) += kevent_poll.o +obj-$(CONFIG_KEVENT_SOCKET) += kevent_socket.o +obj-$(CONFIG_KEVENT_PIPE) += kevent_pipe.o +obj-$(CONFIG_KEVENT_SIGNAL) += kevent_signal.o +obj-$(CONFIG_KEVENT_UNOTIFY) += kevent_unotify.o +obj-$(CONFIG_KEVENT_AIO) += kevent_aio.o
kernel/kevent/kevent.c new file +267 -0
267 lines not on this page · raw message
kernel/kevent/kevent_user.c new file +1360 -0
1,360 lines not on this page · raw message
kernel/sys_ni.c +9 -0
@@ -123,6 +123,15 @@ cond_syscall(ppc_rtas);cond_syscall (sys_spu_run);cond_syscall (sys_spu_create); +cond_syscall (sys_kevent_get_events); +cond_syscall (sys_kevent_ctl); +cond_syscall (sys_kevent_wait); +cond_syscall (sys_kevent_commit); +cond_syscall (sys_kevent_init); + +cond_syscall (sys_aio_sendfile); +cond_syscall (sys_aio_sendfile_path); +/* mmu depending weak syscall entries */ cond_syscall (sys_mprotect);cond_syscall (sys_msync);
PATCH 3/10 kevent: poll/select() notifications.
poll/select() notifications.
This patch includes generic poll/select notifications. kevent_poll works simialr to epoll and has the same issues (callback is invoked not from internal state machine of the caller, but through process awake, a lot of allocations and so on).
fs/file_table.c +3 -0
@@ -20,6 +20,7 @@#include <linux/cdev.h> #include <linux/fsnotify.h> #include <linux/sysctl.h> +#include <linux/kevent.h> #include <linux/percpu_counter.h> #include <asm/atomic.h>
@@ -119,6 +120,7 @@ struct file *get_empty_filp(void) f->f_uid = tsk->fsuid; f->f_gid = tsk->fsgid;eventpoll_init_file (f); +kevent_init_file (f);/* f->f_version: 0 */ return f;
@@ -164,6 +166,7 @@ void fastcall __fput(struct file *file)* in the file cleanup chain. */ eventpoll_release (file); +kevent_cleanup_file (file);locks_remove_flock (file);if (file->f_op && file->f_op->release)
include/linux/fs.h +10 -0
@@ -280,6 +280,7 @@ extern int dir_notify_enable;#include <linux/init.h> #include <linux/pid.h> #include <linux/mutex.h> +#include <linux/kevent_storage.h> #include <asm/atomic.h> #include <asm/semaphore.h>
@@ -408,6 +409,8 @@ struct address_space_operations {int (*readpages)(struct file *filp,struct address_space *mapping,struct list_head *pages,unsigned nr_pages); +int (*aio_readpages)(struct file *filp,struct address_space *mapping, +struct list_head *pages,unsigned nr_pages,void *priv);/* * ext3 requires that a successful prepare_write() call be followed
@@ -578,6 +581,10 @@ struct inode {struct mutex inotify_mutex;/* protects the watches list */ #endif +#if definedCONFIG_KEVENT_SOCKET || definedCONFIG_KEVENT_PIPE +struct kevent_storage st; +#endif +unsigned long i_state;unsigned long dirtied_when;/* jiffies of first dirtying */
@@ -737,6 +744,9 @@ struct file {struct list_head f_ep_links;spinlock_t f_ep_lock;#endif /* #ifdef CONFIG_EPOLL */ +#ifdef CONFIG_KEVENT_POLL +struct kevent_storage st; +#endif struct address_space *f_mapping; };extern spinlock_t files_lock;
kernel/kevent/kevent_poll.c new file +234 -0
234 lines not on this page · raw message
PATCH 4/10 kevent: Socket notifications.
Socket notifications.
This patch includes socket send/recv/accept notifications. Using trivial web server based on kevent and this features instead of epoll it's performance increased more than noticebly. More details about various benchmarks and server itself (evserver_kevent.c) can be found on project's homepage.
fs/inode.c +7 -0
@@ -21,6 +21,7 @@#include <linux/cdev.h> #include <linux/bootmem.h> #include <linux/inotify.h> +#include <linux/kevent.h> #include <linux/mount.h> /*
@@ -164,12 +165,18 @@ static struct inode *alloc_inode(struct super_block *sb) } inode->i_private =NULL ; inode->i_mapping = mapping; +#if definedCONFIG_KEVENT_SOCKET || definedCONFIG_KEVENT_PIPE +kevent_storage_init (inode, &inode->st); +#endif }return inode; }void destroy_inode (struct inode *inode) { +#if definedCONFIG_KEVENT_SOCKET || definedCONFIG_KEVENT_PIPE +kevent_storage_fini (&inode->st); +#endif BUG_ON (inode_has_buffers (inode));security_inode_free (inode);if (inode->i_sb->s_op->destroy_inode)
include/net/sock.h +17 -15
@@ -49,6 +49,7 @@#include <linux/skbuff.h> /* struct sk_buff */ #include <linux/mm.h> #include <linux/security.h> +#include <linux/kevent.h> #include <linux/filter.h>
@@ -451,6 +452,21 @@ static inline int sk_stream_memory_free(struct sock *sk)extern void sk_stream_rfree (struct sk_buff *skb); +struct socket_alloc { +struct socket socket; +struct inode vfs_inode; +}; + +static inline struct socket *SOCKET_I (struct inode *inode) +{ + return &container_of(inode, struct socket_alloc, vfs_inode)->socket; +} + +static inline struct inode *SOCK_INODE (struct socket *socket) +{ + return &container_of(socket, struct socket_alloc, socket)->vfs_inode; +} +static inline void sk_stream_set_owner_r (struct sk_buff *skb,struct sock *sk) { skb->sk = sk;
@@ -478,6 +494,7 @@ static inline void sk_add_backlog(struct sock *sk, struct sk_buff *skb) sk->sk_backlog.tail = skb; } skb->next =NULL ; +kevent_socket_notify (sk,KEVENT_SOCKET_RECV ); }#define sk_wait_event (__sk, __timeo, __condition) \
@@ -679,21 +696,6 @@ static inline struct kiocb *siocb_to_kiocb(struct sock_iocb *si)return si->kiocb; } -struct socket_alloc { -struct socket socket; -struct inode vfs_inode; -}; - -static inline struct socket *SOCKET_I (struct inode *inode) -{ - return &container_of(inode, struct socket_alloc, vfs_inode)->socket; -} - -static inline struct inode *SOCK_INODE (struct socket *socket) -{ - return &container_of(socket, struct socket_alloc, socket)->vfs_inode; -} -extern void __sk_stream_mem_reclaim (struct sock *sk);extern int sk_stream_mem_schedule (struct sock *sk,int size,int kind);
include/net/tcp.h +1 -0
@@ -864,6 +864,7 @@ static inline int tcp_prequeue(struct sock *sk, struct sk_buff *skb) tp->ucopy.memory =0 ; }else if (skb_queue_len (&tp->ucopy.prequeue) ==1 ) {wake_up_interruptible (sk->sk_sleep); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV |KEVENT_SOCKET_SEND );if (!inet_csk_ack_scheduled (sk)) inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, (3 * TCP_RTO_MIN) / 4,
kernel/kevent/kevent_socket.c new file +144 -0
144 lines not on this page · raw message
net/core/sock.c +9 -1
@@ -1393,6 +1393,7 @@ static void sock_def_wakeup(struct sock *sk)if (sk->sk_sleep &&waitqueue_active (sk->sk_sleep))wake_up_interruptible_all (sk->sk_sleep);read_unlock (&sk->sk_callback_lock); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV |KEVENT_SOCKET_SEND ); }static void sock_def_error_report (struct sock *sk)
@@ -1402,6 +1403,7 @@ static void sock_def_error_report(struct sock *sk)wake_up_interruptible (sk->sk_sleep);sk_wake_async (sk,0 ,POLL_ERR );read_unlock (&sk->sk_callback_lock); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV |KEVENT_SOCKET_SEND ); }static void sock_def_readable (struct sock *sk,int len)
@@ -1411,6 +1413,7 @@ static void sock_def_readable(struct sock *sk, int len)wake_up_interruptible (sk->sk_sleep);sk_wake_async (sk,1 ,POLL_IN );read_unlock (&sk->sk_callback_lock); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV |KEVENT_SOCKET_SEND ); }static void sock_def_write_space (struct sock *sk)
@@ -1430,6 +1433,7 @@ static void sock_def_write_space(struct sock *sk) }read_unlock (&sk->sk_callback_lock); +kevent_socket_notify (sk,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV ); }static void sock_def_destruct (struct sock *sk)
@@ -1480,6 +1484,8 @@ void sock_init_data(struct socket *sock, struct sock *sk) sk->sk_state =TCP_CLOSE ; sk->sk_socket = sock; +kevent_sk_reinit (sk); +sock_set_flag (sk,SOCK_ZAPPED );if (sock)
@@ -1546,8 +1552,10 @@ void fastcall release_sock(struct sock *sk)if (sk->sk_backlog.tail)__release_sock (sk); sk->sk_lock.owner =NULL ; -if (waitqueue_active (&sk->sk_lock.wq)) +if (waitqueue_active (&sk->sk_lock.wq)) {wake_up (&sk->sk_lock.wq); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV |KEVENT_SOCKET_SEND ); + }spin_unlock_bh (&sk->sk_lock.slock); }EXPORT_SYMBOL (release_sock);
net/core/stream.c +1 -0
@@ -36,6 +36,7 @@ void sk_stream_write_space(struct sock *sk)wake_up_interruptible (sk->sk_sleep);if (sock->fasync_list && !(sk->sk_shutdown &SEND_SHUTDOWN ))sock_wake_async (sock,2 ,POLL_OUT ); +kevent_socket_notify (sk,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV ); } }
net/ipv4/tcp_input.c +1 -0
@@ -3127,6 +3127,7 @@ static void tcp_ofo_queue(struct sock *sk)__skb_unlink (skb, &tp->out_of_order_queue);__skb_queue_tail (&sk->sk_receive_queue, skb); +kevent_socket_notify (sk,KEVENT_SOCKET_RECV ); tp->rcv_nxt =TCP_SKB_CB (skb)->end_seq;if (skb->h.th->fin)tcp_fin (skb, sk, skb->h.th);
net/ipv4/tcp_ipv4.c +2 -0
@@ -61,6 +61,7 @@#include <linux/jhash.h> #include <linux/init.h> #include <linux/times.h> +#include <linux/kevent.h> #include <net/icmp.h> #include <net/inet_hashtables.h>
@@ -1391,6 +1392,7 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)reqsk_free (req); }else {inet_csk_reqsk_queue_hash_add (sk, req,TCP_TIMEOUT_INIT ); +kevent_socket_notify (sk,KEVENT_SOCKET_ACCEPT ); }return 0 ;
net/socket.c +3 -0
@@ -84,6 +84,7 @@#include <linux/kmod.h> #include <linux/audit.h> #include <linux/wireless.h> +#include <linux/kevent.h> #include <asm/uaccess.h> #include <asm/unistd.h>
@@ -496,6 +497,8 @@ static struct socket *sock_alloc(void) inode->i_uid = current->fsuid; inode->i_gid = current->fsgid; +kevent_socket_reinit (sock); +get_cpu_var (sockets_in_use)++;put_cpu_var (sockets_in_use);return sock;
net/unix/af_unix.c +21 -1
@@ -1563,8 +1563,10 @@ static int unix_dgram_recvmsg(struct kiocb *iocb, struct socket *sock,struct scm_cookie tmp_scm;struct sock *sk = sock->sk;struct unix_sock *u =unix_sk (sk); +struct sock *other;int noblock = flags &MSG_DONTWAIT ;struct sk_buff *skb; +int err; err = -EOPNOTSUPP ;
@@ -1580,6 +1582,12 @@ static int unix_dgram_recvmsg(struct kiocb *iocb, struct socket *sock, goto out_unlock;wake_up_interruptible (&u->peer_wait); + other =unix_peer_get (sk); +if (other) { +kevent_socket_notify (other,KEVENT_SOCKET_SEND ); +sock_put (other); + }else +kevent_socket_notify (sk,KEVENT_SOCKET_RECV );if (msg->msg_name)unix_copy_addr (msg, skb->sk);
@@ -1674,7 +1682,7 @@ static int unix_stream_recvmsg(struct kiocb *iocb, struct socket *sock, {struct sock_iocb *siocb =kiocb_to_siocb (iocb);struct scm_cookie tmp_scm; -struct sock *sk = sock->sk; +struct sock *sk = sock->sk, *other;struct unix_sock *u =unix_sk (sk);struct sockaddr_un *sunaddr=msg->msg_name;int copied =0 ;
@@ -1803,6 +1811,14 @@ static int unix_stream_recvmsg(struct kiocb *iocb, struct socket *sock, } }while (size); + other =unix_peer_get (sk); +if (other) { +kevent_socket_notify (other,KEVENT_SOCKET_SEND ); +sock_put (other); + } +if (sk->sk_shutdown &RCV_SHUTDOWN || !other) +kevent_socket_notify (sk,KEVENT_SOCKET_RECV ); +mutex_unlock (&u->readlock);scm_recv (sock, msg, siocb->scm, flags); out:
@@ -1824,6 +1840,10 @@ static int unix_shutdown(struct socket *sock, int mode)sock_hold (other);unix_state_wunlock (sk); sk->sk_state_change (sk); +kevent_socket_notify (sk,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV ); +if (other) +kevent_socket_notify (other,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV ); +kevent_socket_notify (sk,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV );if (other && (sk->sk_type ==SOCK_STREAM || sk->sk_type ==SOCK_SEQPACKET )) {
PATCH 5/10 kevent: Timer notifications.
Timer notifications.
Timer notifications can be used for fine grained per-process time management, since interval timers are very inconvenient to use, and they are limited.
This subsystem uses high-resolution timers.
id.raw[0] is used as number of seconds
id.raw[1] is used as number of nanoseconds
kernel/kevent/kevent_timer.c new file +114 -0
@@ -0,0 +1,114 @@ +/* +* 2006 Copyright (c) Evgeniy Polyakov <[email protected]> +* All rights reserved. +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +*/ + +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/list.h> +#include <linux/slab.h> +#include <linux/spinlock.h> +#include <linux/hrtimer.h> +#include <linux/jiffies.h> +#include <linux/kevent.h> + +struct kevent_timer +{ +struct hrtimer ktimer; +struct kevent_storage ktimer_storage; +struct kevent *ktimer_event; +}; + +static int kevent_timer_func (struct hrtimer *timer) +{ + struct kevent_timer *t = container_of(timer, struct kevent_timer, ktimer); +struct kevent *k = t->ktimer_event; + +kevent_storage_ready (&t->ktimer_storage,NULL ,KEVENT_MASK_ALL ); +hrtimer_forward (timer, timer->base->softirq_time, +ktime_set (k->event.id.raw[0 ], k->event.id.raw[1 ])); +return HRTIMER_RESTART ; +} + +static struct lock_class_key kevent_timer_key; + +static int kevent_timer_enqueue (struct kevent *k) +{ +int err; +struct kevent_timer *t; + + t =kmalloc (sizeof (struct kevent_timer ),GFP_KERNEL ); +if (!t) +return -ENOMEM ; + +hrtimer_init (&t->ktimer,CLOCK_MONOTONIC ,HRTIMER_REL ); + t->ktimer.expires =ktime_set (k->event.id.raw[0 ], k->event.id.raw[1 ]); + t->ktimer.function = kevent_timer_func; + t->ktimer_event = k; + + err =kevent_storage_init (&t->ktimer, &t->ktimer_storage); +if (err) + goto err_out_free; +lockdep_set_class (&t->ktimer_storage.lock, &kevent_timer_key); + + err =kevent_storage_enqueue (&t->ktimer_storage, k); +if (err) + goto err_out_st_fini; + +hrtimer_start (&t->ktimer, t->ktimer.expires,HRTIMER_REL ); + +return 0 ; + +err_out_st_fini: +kevent_storage_fini (&t->ktimer_storage); +err_out_free: +kfree (t); + +return err; +} + +static int kevent_timer_dequeue (struct kevent *k) +{ +struct kevent_storage *st = k->st; + struct kevent_timer *t = container_of(st, struct kevent_timer, ktimer_storage); + +hrtimer_cancel (&t->ktimer); +kevent_storage_dequeue (st, k); +kfree (t); + +return 0 ; +} + +static int kevent_timer_callback (struct kevent *k) +{ + k->event.ret_data[0 ] =jiffies_to_msecs (jiffies); +return 1 ; +} + +static int __init kevent_init_timer (void ) +{ +struct kevent_callbacks tc = { + .callback = &kevent_timer_callback, + .enqueue = &kevent_timer_enqueue, + .dequeue = &kevent_timer_dequeue, + .flags =0 , + }; + +return kevent_add_callbacks (&tc,KEVENT_TIMER ); +} +module_init (kevent_init_timer); +
PATCH 6/10 kevent: Pipe notifications.
Pipe notifications.
fs/pipe.c +6 -0
@@ -16,6 +16,7 @@#include <linux/uio.h> #include <linux/highmem.h> #include <linux/pagemap.h> +#include <linux/kevent.h> #include <asm/uaccess.h> #include <asm/ioctls.h>
@@ -313,6 +314,7 @@ redo:break ; }if (do_wakeup) { +kevent_pipe_notify (inode,KEVENT_SOCKET_SEND );wake_up_interruptible_sync (&pipe->wait);kill_fasync (&pipe->fasync_writers,SIGIO ,POLL_OUT ); }
@@ -322,6 +324,7 @@ redo:/* Signal writers asynchronously that there is more room. */ if (do_wakeup) { +kevent_pipe_notify (inode,KEVENT_SOCKET_SEND );wake_up_interruptible (&pipe->wait);kill_fasync (&pipe->fasync_writers,SIGIO ,POLL_OUT ); }
@@ -484,6 +487,7 @@ redo2:break ; }if (do_wakeup) { +kevent_pipe_notify (inode,KEVENT_SOCKET_RECV );wake_up_interruptible_sync (&pipe->wait);kill_fasync (&pipe->fasync_readers,SIGIO ,POLL_IN ); do_wakeup =0 ;
@@ -495,6 +499,7 @@ redo2: out:mutex_unlock (&inode->i_mutex);if (do_wakeup) { +kevent_pipe_notify (inode,KEVENT_SOCKET_RECV );wake_up_interruptible (&pipe->wait);kill_fasync (&pipe->fasync_readers,SIGIO ,POLL_IN ); }
@@ -590,6 +595,7 @@ pipe_release(struct inode *inode, int decr, int decw)free_pipe_info (inode); }else {wake_up_interruptible (&pipe->wait); +kevent_pipe_notify (inode,KEVENT_SOCKET_SEND |KEVENT_SOCKET_RECV );kill_fasync (&pipe->fasync_readers,SIGIO ,POLL_IN );kill_fasync (&pipe->fasync_writers,SIGIO ,POLL_OUT ); }
kernel/kevent/kevent_pipe.c new file +123 -0
123 lines not on this page · raw message
PATCH 7/10 kevent: Signal notifications.
Signal notifications.
This type of notifications allows to deliver signals through kevent queue. One can find example application signal.c on project homepage.
If KEVENT_SIGNAL_NOMASK bit is set in raw_u64 id then signal will be delivered only through queue, otherwise both delivery types are used - old through update of mask of pending signals and through queue.
If signal is delivered only through kevent queue mask of pending signals is not updated at all, which is equal to putting signal into blocked mask, but with delivery of that signal through kevent queue.
include/linux/sched.h +5 -0
@@ -82,6 +82,7 @@ struct sched_param {#include <linux/resource.h> #include <linux/timer.h> #include <linux/hrtimer.h> +#include <linux/kevent_storage.h> #include <linux/task_io_accounting.h> #include <asm/processor.h>
@@ -1048,6 +1049,10 @@ struct task_struct {#ifdef CONFIG_TASK_DELAY_ACCT struct task_delay_info *delays;#endif +#ifdef CONFIG_KEVENT_SIGNAL +struct kevent_storage st; +u32 kevent_signals; +#endif #ifdef CONFIG_FAULT_INJECTION int make_it_fail;#endif
kernel/fork.c +8 -0
@@ -49,6 +49,7 @@#include <linux/delayacct.h> #include <linux/taskstats_kern.h> #include <linux/random.h> +#include <linux/kevent.h> #include <asm/pgtable.h> #include <asm/pgalloc.h>
@@ -118,6 +119,9 @@ void __put_task_struct(struct task_struct *tsk)WARN_ON (atomic_read (&tsk->usage));WARN_ON (tsk == current); +#ifdef CONFIG_KEVENT_SIGNAL +kevent_storage_fini (&tsk->st); +#endif security_task_free (tsk);free_uid (tsk->user);put_group_info (tsk->group_info);
@@ -1126,6 +1130,10 @@ static struct task_struct *copy_process(unsigned long clone_flags,if (retval) goto bad_fork_cleanup_namespaces; +#ifdef CONFIG_KEVENT_SIGNAL +kevent_storage_init (p, &p->st); +#endif + p->set_child_tid = (clone_flags &CLONE_CHILD_SETTID ) ? child_tidptr :NULL ;/* * Clear TID on mm_release()?
kernel/kevent/kevent_signal.c new file +94 -0
94 lines not on this page · raw message
kernel/signal.c +26 -0
@@ -23,6 +23,7 @@#include <linux/ptrace.h> #include <linux/signal.h> #include <linux/capability.h> +#include <linux/kevent.h> #include <linux/freezer.h> #include <linux/pid_namespace.h> #include <linux/nsproxy.h>
@@ -714,6 +715,9 @@ static int send_signal(int sig, struct siginfo *info, struct task_struct *t, {struct sigqueue * q =NULL ;int ret =0 ; + +if (kevent_signal_notify (t, sig)) +return 1 ;/* * fast-pathed signals for kernel-internal things like SIGSTOP
@@ -793,6 +797,17 @@ specific_send_sig_info(int sig, struct siginfo *info, struct task_struct *t) ret =send_signal (sig, info, t, &t->pending);if (!ret && !sigismember (&t->blocked, sig))signal_wake_up (t, sig ==SIGKILL ); +#ifdef CONFIG_KEVENT_SIGNAL +/* +* Kevent allows to deliver signals through kevent queue, +* it is possible to setup kevent to not deliver +* signal through the usual way, in that case send_signal() +* returns 1 and signal is delivered only through kevent queue. +* We simulate successfull delivery notification through this hack: +*/ +if (ret ==1 ) + ret =0 ; +#endif out:return ret; }
@@ -982,6 +997,17 @@ __group_send_sig_info(int sig, struct siginfo *info, struct task_struct *p)* to avoid several races. */ ret =send_signal (sig, info, p, &p->signal->shared_pending); +#ifdef CONFIG_KEVENT_SIGNAL +/* +* Kevent allows to deliver signals through kevent queue, +* it is possible to setup kevent to not deliver +* signal through the usual way, in that case send_signal() +* returns 1 and signal is delivered only through kevent queue. +* We simulate successfull delivery notification through this hack: +*/ +if (ret ==1 ) + ret =0 ; +#endif if (unlikely (ret))return ret;
PATCH 8/10 kevent: Kevent posix timer notifications.
Kevent posix timer notifications.
Simple extensions to POSIX timers which allows
to deliver notification of the timer expiration
through kevent queue.
Example application posix_timer.c can be found
in archive on project homepage.
include/asm-generic/siginfo.h +3 -0
@@ -235,6 +235,7 @@ typedef struct siginfo {#define SIGEV_NONE 1/* other notification: meaningless */ #define SIGEV_THREAD 2/* deliver via thread creation */ #define SIGEV_THREAD_ID 4/* deliver to thread */ +#define SIGEV_KEVENT 8/* deliver through kevent queue */ /* * This works because the alignment is ok on all current architectures
@@ -260,6 +261,8 @@ typedef struct sigevent {void (*_function)(sigval_t );void *_attribute;/* really pthread_attr_t */ } _sigev_thread; + +int kevent_fd; } _sigev_un; } sigevent_t;
include/linux/posix-timers.h +4 -0
@@ -4,6 +4,7 @@#include <linux/spinlock.h> #include <linux/list.h> #include <linux/sched.h> +#include <linux/kevent_storage.h> union cpu_time_count {cputime_t cpu;
@@ -49,6 +50,9 @@ struct k_itimer {sigval_t it_sigev_value;/* value word of sigevent struct */ struct task_struct *it_process;/* process to send signal to */ struct sigqueue *sigq;/* signal queue entry. */ +#ifdef CONFIG_KEVENT_TIMER +struct kevent_storage st; +#endif union { struct { struct hrtimer timer;
kernel/posix-timers.c +171 -48
264 lines not on this page · raw message
PATCH 9/10 kevent: Private userspace notifications.
Private userspace notifications.
Allows to register notifications of any private userspace events over kevent. Events can be marked as readt using kevent_ctl(KEVENT_READY) command.
kernel/kevent/kevent_unotify.c new file +62 -0
@@ -0,0 +1,62 @@ +/* +* 2006 Copyright (c) Evgeniy Polyakov <[email protected]> +* All rights reserved. +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +*/ + +#include <linux/kernel.h> +#include <linux/kevent.h> + +static int kevent_unotify_callback (struct kevent *k) +{ +return 1 ; +} + +int kevent_unotify_enqueue (struct kevent *k) +{ +int err; + + err =kevent_storage_enqueue (&k->user->st, k); +if (err) + goto err_out_exit; + +if (k->event.req_flags &KEVENT_REQ_ALWAYS_QUEUE ) +kevent_requeue (k); + +return 0 ; + +err_out_exit: +return err; +} + +int kevent_unotify_dequeue (struct kevent *k) +{ +kevent_storage_dequeue (k->st, k); +return 0 ; +} + +static int __init kevent_init_unotify (void ) +{ +struct kevent_callbacks sc = { + .callback = &kevent_unotify_callback, + .enqueue = &kevent_unotify_enqueue, + .dequeue = &kevent_unotify_dequeue, + .flags =0 , + }; + +return kevent_add_callbacks (&sc,KEVENT_UNOTIFY ); +} +module_init (kevent_init_unotify);
PATCH 10/10 kevent: Kevent based AIO (aio_sendfile()).
Kevent based AIO (aio_sendfile()).
aio_sendfile() contains of two major parts: AIO state machine and page processing code. The former is just a small subsystem, which allows to queue callback for theirs invocation in process' context on behalf of pool of kernel threads. It allows to queue caches of callbacks to the local thread or to any other specified. Each cache of callbacks is processed until there are callbacks in it, callbacks can requeue themselfs into the same cache.
Real work is being done in page processing code - code which populates pages into VFS cache and then sends pages to the destination socket via ->sendpage(). Unlike previous aio_sendfile() implementation, new one does not require low-level filesystem specific callbacks (->get_block()) at all, instead I extended struct address_space_operations to contain new member called ->aio_readpages(), which is exactly the same as ->readpage() (read: mpage_readpages()) except different BIO allocation and sumbission routines. I changed mpage_readpages() to provide mpage_alloc() and mpage_bio_submit() to the new function called __mpage_readpages(), which is exactly old mpage_readpages() with provided callback invocation instead of usage for old functions. mpage_readpages_aio() provides kevent specific callbacks, which calls old functions, but with different destructor callbacks, which are essentially the same, except that they reschedule AIO processing.
Benchmarks of simple sendfile vs. aio_sendfile did not show noticeble win of any approach, but I want to notice, that my receiving side is not that best in my case (I managed to create a test where usual senfile() stuck until signal received, the same hapend with aio_sendfile() too, but the latter can be buggy).
It would be good to use it in lighttpd (I will create a patch after some feedback received and approach will be considered as good.
AIO state machine is a base for network AIO (which becomes quite trivial), but I will not start implementation until roadback of kevent as a whole and AIO implementation become more clear.
fs/bio.c +1 -1
@@ -120,7 +120,7 @@ void bio_free(struct bio *bio, struct bio_set *bio_set)/* * default destructor for a bio allocated with bio_alloc_bioset() */ -static void bio_fs_destructor (struct bio *bio) +void bio_fs_destructor (struct bio *bio) {bio_free (bio, fs_bio_set); }
fs/ext3/inode.c +8 -0
@@ -1650,6 +1650,13 @@ ext3_readpages(struct file *file, struct address_space *mapping,return mpage_readpages (mapping, pages, nr_pages, ext3_get_block); } +static int +ext3_readpages_aio (struct file *file,struct address_space *mapping, +struct list_head *pages,unsigned nr_pages,void *priv) +{ +return mpage_readpages_aio (mapping, pages, nr_pages, ext3_get_block, priv); +} +static void ext3_invalidatepage (struct page *page,unsigned long offset) {journal_t *journal =EXT3_JOURNAL (page->mapping->host);
@@ -1768,6 +1775,7 @@ static int ext3_journalled_set_page_dirty(struct page *page) }static const struct address_space_operations ext3_ordered_aops = { + .aio_readpages = ext3_readpages_aio, .readpage = ext3_readpage, .readpages = ext3_readpages, .writepage = ext3_ordered_writepage,
fs/mpage.c +76 -41
188 lines not on this page · raw message
include/linux/mpage.h +2 -0
@@ -16,6 +16,8 @@ typedef int (writepage_t)(struct page *page, struct writeback_control *wbc);int mpage_readpages (struct address_space *mapping,struct list_head *pages,unsigned nr_pages,get_block_t get_block); +int mpage_readpages_aio (struct address_space *mapping,struct list_head *pages, +unsigned nr_pages,get_block_t get_block,void *priv);int mpage_readpage (struct page *page,get_block_t get_block);int mpage_writepages (struct address_space *mapping,struct writeback_control *wbc,get_block_t get_block);
kernel/kevent/kevent_aio.c new file +881 -0
881 lines not on this page · raw message
Re: [take32 0/10] kevent: Generic event handling mechanism.
In reply to Evgeniy PolyakovQuoted 19 lines · Evgeniy Polyakov
Generic event handling mechanism.
Kevent is a generic subsytem which allows to handle event notifications. It supports both level and edge triggered events. It is similar to poll/epoll in some cases, but it is more scalable, it is faster and allows to work with essentially eny kind of events.
Events are provided into kernel through control syscall and can be read back through ring buffer or using usual syscalls.
Kevent update (i.e. readiness switching) happens directly from internals of the appropriate state machine of the underlying subsytem (like network, filesystem, timer or any other).Homepage:
http://tservice.net.ru/~s0mbre/old/?section=projects&item=keventDocumentation page:
http://linux-net.osdl.org/index.php/Kevent
Quoted 8 lines
Consider for inclusion.
With this release I start 3 days resending timeout - i.e. each third day I will send either new version (if something new was requested and agreed to be implemented) or resending with back counter started from three. When back counter hits zero after three resendings I consider there is no interest in subsystem and I will stop further sending.
I really doubt it is a good way to tell the world about my work, and I bet you all tired from those pathos words, but I really would like to get some feedback, since I want to start to work on network AIO, but sending mails into unfeedbackable 'destination' really does not motivate me for that.
Thanks for understanding and your time.
Once the rate of change slows, Andrew should IMO definitely pick this up.
If you wanted to make this process automatic, create a git branch that Andrew and others can pull.
I like the direction so far, and think it should be in -mm for wider testing and review.
Jeff
Re: [take32 0/10] kevent: Generic event handling mechanism.
Quoted 1 line · Jeff Garzik
Once the rate of change slows, Andrew should IMO definitely pick this up.
There are _tons_ of ideas to implement with kevent - so if we want, rate will not slow down. As you can see, from take26 I only send new
features: signals, posix timers, AIO, userspace notifications, various flags and the like. I test it on my machines (recently one them died, so only amd64 right now (running kernel) and i386 compile-only) and some bug-fixes withoout any additioanl feature requests (almost, Ingo asked for AIO before New Year), but broader testing is welcome indeed.
Quoted 2 lines
If you wanted to make this process automatic, create a git branch that Andrew and others can pull.
Exported git tree would be good, but I do not have enough disk space on web-site, and do you really want to read comments written in bad english with russian transliterated indecent words?
Quoted 2 lines
I like the direction so far, and think it should be in -mm for wider testing and review.
It was there, but Andrew dropped it somewhere about take25 :)
Quoted 1 line
Jeff
Signature · 1 line
Evgeniy Polyakov
Re: [take32 0/10] kevent: Generic event handling mechanism.
Quoted 11 lines · Evgeniy Polyakov
Quoted 1 line · Jeff Garzik
Once the rate of change slows, Andrew should IMO definitely pick this up.
There are _tons_ of ideas to implement with kevent - so if we want, rate will not slow down. As you can see, from take26 I only send new
features: signals, posix timers, AIO, userspace notifications, various flags and the like. I test it on my machines (recently one them died, so only amd64 right now (running kernel) and i386 compile-only) and some bug-fixes withoout any additioanl feature requests (almost, Ingo asked for AIO before New Year), but broader testing is welcome indeed.
If the rate doesn't slow (if only artificially), people are discouraged from reviewing, because it becomes a moving target.
Quoted 3 lines
Quoted 1 line
If you wanted to make this process automatic, create a git branch that Andrew and others can pull.
Exported git tree would be good, but I do not have enough disk space on
Request an account on http://www.foo-projects.org/ which supports git. The Intel guys use it to send me e1000/ixgb changes, for example.
Quoted 2 lines
web-site, and do you really want to read comments written in bad english with russian transliterated indecent words?
The only thing exported to -mm is the code changes, as a patch. git merely automates the process, so that Andrew doesn't have to spend time [that he doesn't have] tracking a project with a high rate of change.
Quoted 3 lines
Quoted 1 line
I like the direction so far, and think it should be in -mm for wider testing and review.
It was there, but Andrew dropped it somewhere about take25 :)
Probably because it was a moving target with a high rate of change, requiring time that Andrew did not have just to keep in sync and fix build conflicts with other -mm patches.
Jeff
Re: [take32 0/10] kevent: Generic event handling mechanism.
Quoted 5 lines · Jeff Garzik
Quoted 1 line
It was there, but Andrew dropped it somewhere about take25 :)
Probably because it was a moving target with a high rate of change, requiring time that Andrew did not have just to keep in sync and fix build conflicts with other -mm patches.
Ok, I understood.
Let's freeze kevent for a while and allow things to settle down, but...
How can I detect that it is in use, since I get no feedback? I suppose that if Andrew will not pick it up, I will resend/push/pull the latest set, but without new features implemented (or as addon), only bugfixes.
Quoted 2 lines
Jeff
Signature · 1 line
Evgeniy Polyakov
Kevent bonus: epoll implementaion over kevent.
In reply to Evgeniy PolyakovAs a usage scenario, compile-tested only.
Replace fs/eventpoll.c with this code and see,
how your kernel crashes. Or works.
:)
#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/kevent.h>
#include <linux/ukevent.h>
#include <linux/syscalls.h>
#include <asm/poll.h>
#include <asm/uaccess.h>
#include <linux/eventpoll.h>
asmlinkage long sys_epoll_create(int size)
{
return sys_kevent_init(NULL, 0, 0);
}#define EP_PRIVATE_BITS (EPOLLONESHOT | EPOLLET)
asmlinkage long sys_epoll_ctl(int epfd, int op, int fd, struct epoll_event __user *event)
{
struct epoll_event epds;
struct ukevent uk;
int cmd, err;
switch (op) {
case EPOLL_CTL_DEL:
cmd = KEVENT_CTL_REMOVE;
break;
case EPOLL_CTL_ADD:
cmd = KEVENT_CTL_ADD;
break;
case EPOLL_CTL_MOD:
cmd = KEVENT_CTL_MODIFY;
break;
default:
err = -EINVAL;
goto err_out_exit;
}
if (copy_from_user(&epds, event, sizeof(struct epoll_event))) {
err = -EFAULT;
goto err_out_exit;
}
memset(&uk, 0, sizeof(struct ukevent));
uk.id.raw[0] = fd;
memcpy(uk.user, &epds.data, sizeof(u64));
uk.type = KEVENT_POLL;
uk.event = (epds.events & ~EP_PRIVATE_BITS)| POLLERR | POLLHUP;
uk.req_flags = KEVENT_REQ_ALWAYS_QUEUE | KEVENT_REQ_LAST_CHECK;
if (epds.events & EPOLLONESHOT)
uk.req_flags |= KEVENT_REQ_ONESHOT;
if (epds.events & EPOLLET)
uk.req_flags |= KEVENT_REQ_ET;
return sys_kevent_ctl(epfd, cmd, 1, &uk);err_out_exit:
return err; }
asmlinkage long sys_epoll_wait(int epfd, struct epoll_event __user *events,
int maxevents, int timeout)
{
struct ukevent uk[3];
struct timespec ts;
long num, i;
int err;
if (!access_ok(VERIFY_WRITE, events, maxevents * sizeof(struct epoll_event))) {
err = -EFAULT;
goto err_out_exit;
}
memset(uk, 0, sizeof(uk));
ts.tv_sec = timeout/1000;
ts.tv_nsec = (timeout - ts.tv_sec*1000)*1000;
num = sys_kevent_get_events(epfd, 1, min(3, maxevents), ts, uk, 0);
if (num <= 0)
return num;
for (i=0; i<num; ++i) {
__u64 data;
memcpy(&data, &uk[i].user, sizeof(u64));
if (__put_user(uk[i].ret_data[0], &events[i].events) ||
__put_user(data, &events[i].data))
return -EFAULT;
}
return num;err_out_exit:
return err; }
Signature · 1 line
Evgeniy Polyakov