nim-chronos/chronos/ioselects/ioselectors_kqueue.nim

717 lines
23 KiB
Nim
Raw Normal View History

exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
#
#
# Nim's Runtime Library
# (c) Copyright 2016 Eugene Kabanov
#
# See the file "copying.txt", included in this
# distribution, for details about the copyright.
#
# This module implements BSD kqueue().
{.push raises: [Defect].}
import std/[kqueue, deques, tables]
import stew/base10
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
const
# SIG_IGN and SIG_DFL declared in posix.nim as variables, but we need them
# to be constants and GC-safe.
SIG_DFL = cast[proc(x: cint) {.raises: [], noconv, gcsafe.}](0)
SIG_IGN = cast[proc(x: cint) {.raises: [], noconv, gcsafe.}](1)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
type
SelectorImpl[T] = object
kqFd: cint
fds: Table[int32, SelectorKey[T]]
virtualHoles: Deque[int32]
virtualId: int32
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
Selector*[T] = ref SelectorImpl[T]
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
SelectEventImpl = object
rfd: cint
wfd: cint
SelectEvent* = ptr SelectEventImpl
# SelectEvent is declared as `ptr` to be placed in `shared memory`,
# so you can share one SelectEvent handle between threads.
proc getVirtualId[T](s: Selector[T]): SelectResult[int32] =
if len(s.virtualHoles) > 0:
ok(s.virtualHoles.popLast())
else:
if s.virtualId == low(int32):
err(EMFILE)
else:
dec(s.virtualId)
ok(s.virtualId)
proc isVirtualId(ident: int32): bool =
ident < 0'i32
proc toString(key: int32|cint|SocketHandle|int): string =
let fdi32 = when key is int32: key else: int32(key)
if isVirtualId(fdi32):
if fdi32 == -1:
"InvalidIdent"
else:
"V" & Base10.toString(uint32(-fdi32))
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
else:
Base10.toString(uint32(fdi32))
proc toPointer(data: int32): pointer =
when sizeof(int) == 8:
cast[pointer](uint64(uint32(data)))
else:
cast[pointer](uint32(data))
template addKey[T](s: Selector[T], key: int32, skey: SelectorKey[T]) =
if s.fds.hasKeyOrPut(key, skey):
raiseAssert "Descriptor [" & key.toString() &
"] is already registered in the selector!"
template getKey[T](s: Selector[T], key: int32): SelectorKey[T] =
let
defaultKey = SelectorKey[T](ident: InvalidIdent)
pkey = s.fds.getOrDefault(key, defaultKey)
doAssert(pkey.ident != InvalidIdent, "Descriptor [" & key.toString() &
"] is not registered in the selector!")
pkey
template checkKey[T](s: Selector[T], key: int32): bool =
s.fds.contains(key)
proc freeKey[T](s: Selector[T], key: int32) =
s.fds.del(key)
if isVirtualId(key):
s.virtualHoles.addFirst(key)
template getIdent(event: KEvent): int32 =
doAssert(event.ident <= uint(high(uint32)),
"Invalid event ident value [" & Base10.toString(event.ident) &
"] in the kqueue event object")
cast[int32](uint32(event.ident))
template getUdata(event: KEvent): int32 =
let udata = cast[uint](event.udata)
doAssert(event.ident <= uint(high(uint32)),
"Invalid event udata value [" & Base10.toString(udata) &
"] in the kqueue event object with ident [" &
Base10.toString(event.ident) & "]")
cast[int32](uint32(udata))
proc new*(t: typedesc[Selector], T: typedesc): SelectResult[Selector[T]] =
let kqFd =
block:
let res = handleEintr(kqueue())
if res == -1:
return err(osLastError())
cint(res)
let selector = Selector[T](
kqFd: kqFd,
fds: initTable[int32, SelectorKey[T]](asyncInitialSize),
virtualId: -1'i32, # Should start with -1, because `InvalidIdent` == -1
virtualHoles: initDeque[int32]()
)
ok(selector)
proc close2*[T](s: Selector[T]): SelectResult[void] =
s.fds.clear()
s.virtualHoles.clear()
s.virtualId = -1'i32
if handleEintr(osdefs.close(s.kqFd)) != 0:
err(osLastError())
else:
ok()
proc new*(t: typedesc[SelectEvent]): SelectResult[SelectEvent] =
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
var fds: array[2, cint]
when declared(pipe2):
if osdefs.pipe2(fds, osdefs.O_NONBLOCK or osdefs.O_CLOEXEC) == -1:
return err(osLastError())
var res = cast[SelectEvent](allocShared0(sizeof(SelectEventImpl)))
res.rfd = fds[0]
res.wfd = fds[1]
ok(res)
else:
if osdefs.pipe(fds) == -1:
return err(osLastError())
let res1 = setDescriptorFlags(fds[0], true, true)
if res1.isErr():
discard closeFd(fds[0])
discard closeFd(fds[1])
return err(res1.error())
let res2 = setDescriptorFlags(fds[1], true, true)
if res2.isErr():
discard closeFd(fds[0])
discard closeFd(fds[1])
return err(res2.error())
var res = cast[SelectEvent](allocShared0(sizeof(SelectEventImpl)))
res.rfd = fds[0]
res.wfd = fds[1]
ok(res)
proc trigger2*(event: SelectEvent): SelectResult[void] =
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
var data: uint64 = 1
let res = handleEintr(osdefs.write(event.wfd, addr data, sizeof(uint64)))
if res == -1:
err(osLastError())
elif res != sizeof(uint64):
err(oserrno.EINVAL)
else:
ok()
proc close2*(ev: SelectEvent): SelectResult[void] =
let
rfd = ev.rfd
wfd = ev.wfd
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
deallocShared(cast[pointer](ev))
if closeFd(rfd) != 0:
let errorCode = osLastError()
discard closeFd(wfd)
err(errorCode)
else:
if closeFd(wfd) != 0:
err(osLastError())
else:
ok()
template modifyKQueue(changes: var openArray[KEvent], index: int, nident: uint,
nfilter: cshort, nflags: cushort, nfflags: cuint,
ndata: int, nudata: pointer) =
changes[index] = KEvent(ident: nident, filter: nfilter, flags: nflags,
fflags: nfflags, data: ndata, udata: nudata)
proc registerHandle2*[T](s: Selector[T], fd: cint, events: set[Event],
data: T): SelectResult[void] =
let selectorKey = SelectorKey[T](ident: fd, events: events,
param: 0, data: data)
s.addKey(fd, selectorKey)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if events != {}:
var
changes: array[2, KEvent]
k = 0
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if Event.Read in events:
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_READ, EV_ADD, 0, 0, nil)
inc(k)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if Event.Write in events:
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_WRITE, EV_ADD, 0, 0, nil)
inc(k)
if k > 0:
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(k), nil,
0, nil)) == -1:
s.freeKey(fd)
return err(osLastError())
ok()
proc updateHandle2*[T](s: Selector[T], fd: cint,
events: set[Event]): SelectResult[void] =
let EventsMask = {Event.Timer, Event.Signal, Event.Process, Event.Vnode,
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
Event.User, Event.Oneshot, Event.Error}
s.fds.withValue(int32(fd), pkey):
doAssert(pkey[].events * EventsMask == {},
"Descriptor [" & fd.toString() & "] could not be updated!")
if pkey.events != events:
var
changes: array[4, KEvent]
k = 0
if (Event.Read in pkey[].events) and (Event.Read notin events):
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_READ, EV_DELETE,
0, 0, nil)
inc(k)
if (Event.Write in pkey[].events) and (Event.Write notin events):
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_WRITE, EV_DELETE,
0, 0, nil)
inc(k)
if (Event.Read notin pkey[].events) and (Event.Read in events):
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_READ, EV_ADD,
0, 0, nil)
inc(k)
if (Event.Write notin pkey[].events) and (Event.Write in events):
changes.modifyKQueue(k, uint(uint32(fd)), EVFILT_WRITE, EV_ADD,
0, 0, nil)
inc(k)
if k > 0:
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(k), nil,
0, nil)) == -1:
return err(osLastError())
pkey[].events = events
do:
raiseAssert "Descriptor [" & fd.toString() &
"] is not registered in the selector!"
ok()
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc registerTimer*[T](s: Selector[T], timeout: int, oneshot: bool,
data: T): SelectResult[cint] =
let
fdi32 = ? s.getVirtualId()
events = if oneshot: {Event.Timer, Event.Oneshot} else: {Event.Timer}
flags: cushort = if oneshot: EV_ONESHOT or EV_ADD else: EV_ADD
selectorKey = SelectorKey[T](ident: fdi32, events: events, param: timeout,
data: data)
var changes: array[1, KEvent]
s.addKey(fdi32, selectorKey)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
# EVFILT_TIMER on Open/Net(BSD) has granularity of only milliseconds,
# but MacOS and FreeBSD allow use `0` as `fflags` to use milliseconds
# too
changes.modifyKQueue(0, uint(uint32(fdi32)), EVFILT_TIMER, flags, 0,
cint(timeout), nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil, 0, nil)) == -1:
s.freeKey(fdi32)
return err(osLastError())
ok(cint(fdi32))
proc blockSignal(signal: int): SelectResult[void] =
var omask, nmask: Sigset
if sigemptyset(nmask) < 0:
return err(osLastError())
if sigemptyset(omask) < 0:
return err(osLastError())
if sigaddset(nmask, cint(signal)) < 0:
return err(osLastError())
? blockSignals(nmask, omask)
ok()
proc unblockSignal(signal: int): SelectResult[void] =
var omask, nmask: Sigset
if sigemptyset(nmask) < 0:
return err(osLastError())
if sigemptyset(omask) < 0:
return err(osLastError())
if sigaddset(nmask, cint(signal)) < 0:
return err(osLastError())
? unblockSignals(nmask, omask)
ok()
template checkSignal(signal: int) =
doAssert((signal >= 0) and (signal <= int(high(int32))),
"Invalid signal value [" & $signal & "]")
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc registerSignal*[T](s: Selector[T], signal: int,
data: T): SelectResult[cint] =
checkSignal(signal)
let
fdi32 = ? s.getVirtualId()
events = {Event.Signal}
selectorKey = SelectorKey[T](ident: fdi32, events: events,
param: signal, data: data)
var changes: array[1, KEvent]
s.addKey(fdi32, selectorKey)
let res = blockSignal(signal)
if res.isErr():
s.freeKey(fdi32)
return err(res.error())
# To be compatible with linux semantic we need to "eat" signals
signal(cint(signal), SIG_IGN)
changes.modifyKQueue(0, uint(signal), EVFILT_SIGNAL, EV_ADD, 0, 0,
fdi32.toPointer())
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil, 0, nil)) == -1:
let errorCode = osLastError()
s.freeKey(fdi32)
discard unblockSignal(signal)
return err(errorCode)
ok(cint(fdi32))
template checkPid(pid: int) =
when sizeof(int) == 8:
doAssert(pid >= 0 and pid <= int(high(uint32)),
"Invalid process idientified (pid) value")
else:
doAssert(pid >= 0 and pid <= high(int32),
"Invalid process idientified (pid) value")
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc registerProcess*[T](s: Selector[T], pid: int,
data: T): SelectResult[cint] =
checkPid(pid)
let
fdi32 = ? s.getVirtualId()
events = {Event.Process, Event.Oneshot}
flags: cushort = EV_ONESHOT or EV_ADD
selectorKey = SelectorKey[T](ident: fdi32, events: events,
param: pid, data: data)
var changes: array[1, KEvent]
s.addKey(fdi32, selectorKey)
changes.modifyKQueue(0, uint(uint32(pid)), EVFILT_PROC, flags, NOTE_EXIT,
0, fdi32.toPointer())
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil, 0, nil)) == -1:
s.freeKey(fdi32)
return err(osLastError())
ok(cint(fdi32))
proc registerEvent2*[T](s: Selector[T], ev: SelectEvent,
data: T): SelectResult[cint] =
doAssert(not(isNil(ev)))
let
selectorKey = SelectorKey[T](ident: ev.rfd, events: {Event.User},
param: 0, data: data)
var changes: array[1, KEvent]
s.addKey(ev.rfd, selectorKey)
changes.modifyKQueue(0, uint(uint32(ev.rfd)), EVFILT_READ, EV_ADD, 0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil, 0, nil)) == -1:
s.freeKey(ev.rfd)
return err(osLastError())
ok(ev.rfd)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
template processVnodeEvents(events: set[Event]): cuint =
var rfflags = cuint(0)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if events == {Event.VnodeWrite, Event.VnodeDelete, Event.VnodeExtend,
Event.VnodeAttrib, Event.VnodeLink, Event.VnodeRename,
Event.VnodeRevoke}:
rfflags = NOTE_DELETE or NOTE_WRITE or NOTE_EXTEND or NOTE_ATTRIB or
NOTE_LINK or NOTE_RENAME or NOTE_REVOKE
else:
if Event.VnodeDelete in events: rfflags = rfflags or NOTE_DELETE
if Event.VnodeWrite in events: rfflags = rfflags or NOTE_WRITE
if Event.VnodeExtend in events: rfflags = rfflags or NOTE_EXTEND
if Event.VnodeAttrib in events: rfflags = rfflags or NOTE_ATTRIB
if Event.VnodeLink in events: rfflags = rfflags or NOTE_LINK
if Event.VnodeRename in events: rfflags = rfflags or NOTE_RENAME
if Event.VnodeRevoke in events: rfflags = rfflags or NOTE_REVOKE
rfflags
proc registerVnode2*[T](s: Selector[T], fd: cint, events: set[Event],
data: T): SelectResult[cint] =
let
events = {Event.Vnode} + events
fflags = processVnodeEvents(events)
selectorKey = SelectorKey[T](ident: fd, events: events,
param: 0, data: data)
var changes: array[1, KEvent]
s.addKey(fd, selectorKey)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
changes.modifyKQueue(0, uint(uint32(fd)), EVFILT_VNODE, EV_ADD or EV_CLEAR,
fflags, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil, 0, nil)) == -1:
s.freeKey(fd)
return err(osLastError())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
ok(fd)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc unregister2*[T](s: Selector[T], fd: cint): SelectResult[void] =
let
fdi32 = int32(fd)
pkey = s.getKey(fdi32)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
var changes: array[2, KEvent]
var k = 0
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if pkey.events != {}:
if pkey.events * {Event.Read, Event.Write} != {}:
if Event.Read in pkey.events:
changes.modifyKQueue(k, uint(uint32(fdi32)), EVFILT_READ, EV_DELETE,
0, 0, nil)
inc(k)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
if Event.Write in pkey.events:
changes.modifyKQueue(k, uint(uint32(fdi32)), EVFILT_WRITE, EV_DELETE,
0, 0, nil)
inc(k)
if k > 0:
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(k), nil,
0, nil)) == -1:
return err(osLastError())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
elif Event.Timer in pkey.events:
if Event.Finished notin pkey.events:
changes.modifyKQueue(0, uint(uint32(fdi32)), EVFILT_TIMER, EV_DELETE,
0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil,
0, nil)) == -1:
return err(osLastError())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
elif Event.Signal in pkey.events:
let sig = cint(pkey.param)
osdefs.signal(sig, SIG_DFL)
changes.modifyKQueue(0, uint(uint32(pkey.param)), EVFILT_SIGNAL,
EV_DELETE, 0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil,
0, nil)) == -1:
discard unblockSignal(sig)
return err(osLastError())
? unblockSignal(sig)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
elif Event.Process in pkey.events:
if Event.Finished notin pkey.events:
changes.modifyKQueue(0, uint(uint32(pkey.param)), EVFILT_PROC,
EV_DELETE, 0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil,
0, nil)) == -1:
return err(osLastError())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
elif Event.Vnode in pkey.events:
changes.modifyKQueue(0, uint(uint32(fdi32)), EVFILT_VNODE, EV_DELETE,
0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil,
0, nil)) == -1:
return err(osLastError())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
elif Event.User in pkey.events:
changes.modifyKQueue(0, uint(uint32(fdi32)), EVFILT_READ, EV_DELETE,
0, 0, nil)
if handleEintr(kevent(s.kqFd, addr(changes[0]), cint(1), nil,
0, nil)) == -1:
return err(osLastError())
s.freeKey(fdi32)
ok()
proc unregister2*[T](s: Selector[T], event: SelectEvent): SelectResult[void] =
s.unregister2(event.rfd)
proc prepareKey[T](s: Selector[T], event: KEvent): Opt[ReadyKey] =
let fdi32 = event.getIdent()
var rkey = ReadyKey(fd: fdi32, events: {})
var pkey =
case event.filter:
of EVFILT_READ, EVFILT_WRITE, EVFILT_TIMER, EVFILT_VNODE:
s.getKey(fdi32)
of EVFILT_SIGNAL, EVFILT_PROC:
let virtualFd = event.getUdata()
s.getKey(virtualFd)
else:
raiseAssert "Unsupported kqueue filter [" & $event.filter & "] reported!"
case event.filter
of EVFILT_READ:
if (event.flags and EV_EOF) != 0:
rkey.events.incl(Event.Error)
rkey.errorCode = oserrno.ECONNRESET
if Event.User in pkey.events:
var data: uint64 = 0
if handleEintr(osdefs.read(cint(event.ident), addr data,
sizeof(uint64))) != sizeof(uint64):
let errorCode = osLastError()
if errorCode == oserrno.EAGAIN:
# Someone already consumed event data
return Opt.none(ReadyKey)
else:
rkey.events.incl(Event.Error)
rkey.errorCode = errorCode
rkey.events.incl(Event.User)
else:
rkey.events.incl(Event.Read)
of EVFILT_WRITE:
if (event.flags and EV_EOF) != 0:
rkey.events.incl(Event.Error)
rkey.errorCode = oserrno.ECONNRESET
rkey.events.incl(Event.Write)
of EVFILT_TIMER:
rkey.events.incl(Event.Timer)
if Event.Oneshot in pkey.events:
# we are marking key with `Finished` event, to avoid double decrease.
pkey.events.incl(Event.Finished)
rkey.events.incl({Event.Oneshot, Event.Finished})
s.fds[fdi32] = pkey
of EVFILT_VNODE:
rkey.events.incl(Event.Vnode)
if (event.fflags and NOTE_DELETE) != 0: rkey.events.incl(Event.VnodeDelete)
if (event.fflags and NOTE_WRITE) != 0: rkey.events.incl(Event.VnodeWrite)
if (event.fflags and NOTE_EXTEND) != 0: rkey.events.incl(Event.VnodeExtend)
if (event.fflags and NOTE_ATTRIB) != 0: rkey.events.incl(Event.VnodeAttrib)
if (event.fflags and NOTE_LINK) != 0: rkey.events.incl(Event.VnodeLink)
if (event.fflags and NOTE_RENAME) != 0: rkey.events.incl(Event.VnodeRename)
if (event.fflags and NOTE_REVOKE) != 0: rkey.events.incl(Event.VnodeRevoke)
of EVFILT_SIGNAL:
rkey.events.incl(Event.Signal)
rkey.fd = pkey.ident
of EVFILT_PROC:
rkey.events.incl({Event.Process, Event.Oneshot, Event.Finished})
rkey.fd = pkey.ident
pkey.events.incl(Event.Finished)
s.fds[int32(pkey.ident)] = pkey
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
else:
raiseAssert "Unsupported kqueue filter [" & $event.filter & "] reported!"
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
ok(rkey)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc selectInto2*[T](s: Selector[T], timeout: int,
readyKeys: var openArray[ReadyKey]
): SelectResult[int] =
var
tv: Timespec
queueEvents: array[asyncEventsCount, KEvent]
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
verifySelectParams(timeout, -1, high(int))
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
let
ptrTimeout =
if timeout != -1:
if timeout >= 1000:
tv.tv_sec = Time(timeout div 1_000)
tv.tv_nsec = (timeout %% 1_000) * 1_000_000
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
else:
tv.tv_sec = Time(0)
tv.tv_nsec = timeout * 1_000_000
addr tv
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
else:
nil
maxEventsCount = cint(min(asyncEventsCount, len(readyKeys)))
eventsCount =
block:
var res = 0
while true:
res = kevent(s.kqFd, nil, cint(0), addr(queueEvents[0]),
maxEventsCount, ptrTimeout)
if res < 0:
let errorCode = osLastError()
if errorCode == oserrno.EINTR:
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
continue
return err(errorCode)
else:
break
res
var k = 0
for i in 0 ..< eventsCount:
let rkey = s.prepareKey(queueEvents[i]).valueOr: continue
readyKeys[k] = rkey
inc(k)
ok(k)
proc select2*[T](s: Selector[T],
timeout: int): Result[seq[ReadyKey], OSErrorCode] =
var res = newSeq[ReadyKey](asyncEventsCount)
let count = ? selectInto2(s, timeout, res)
res.setLen(count)
ok(res)
proc newSelector*[T](): owned(Selector[T]) {.
raises: [Defect, IOSelectorsException].} =
let res = Selector.new(T)
if res.isErr():
raiseIOSelectorsError(res.error())
res.get()
proc newSelectEvent*(): SelectEvent {.
raises: [Defect, IOSelectorsException].} =
let res = SelectEvent.new()
if res.isErr():
raiseIOSelectorsError(res.error())
res.get()
proc trigger*(ev: SelectEvent) {.
raises: [Defect, IOSelectorsException].} =
let res = ev.trigger2()
if res.isErr():
raiseIOSelectorsError(res.error())
proc close*(ev: SelectEvent) {.
raises: [Defect, IOSelectorsException].} =
let res = ev.close2()
if res.isErr():
raiseIOSelectorsError(res.error())
proc registerHandle*[T](s: Selector[T], fd: cint | SocketHandle,
events: set[Event], data: T) {.
raises: [Defect, IOSelectorsException].} =
let res = registerHandle2(s, cint(fd), events, data)
if res.isErr():
raiseIOSelectorsError(res.error())
proc updateHandle*[T](s: Selector[T], fd: cint | SocketHandle,
events: set[Event]) {.
raises: [Defect, IOSelectorsException].} =
let res = updateHandle2(s, cint(fd), events)
if res.isErr():
raiseIOSelectorsError(res.error())
proc registerEvent*[T](s: Selector[T], ev: SelectEvent, data: T) {.
raises: [Defect, IOSelectorsException].} =
let res = registerEvent2(s, ev, data)
if res.isErr():
raiseIOSelectorsError(res.error())
proc registerVnode*[T](s: Selector[T], fd: cint, events: set[Event], data: T) {.
raises: [Defect, IOSelectorsException].} =
let res = registerVnode2(s, fd, events, data)
if res.isErr():
raiseIOSelectorsError(res.error())
proc unregister*[T](s: Selector[T], event: SelectEvent) {.
raises: [Defect, IOSelectorsException].} =
let res = unregister2(s, event)
if res.isErr():
raiseIOSelectorsError(res.error())
proc unregister*[T](s: Selector[T], fd: cint|SocketHandle) {.
raises: [Defect, IOSelectorsException].} =
let res = unregister2(s, fd)
if res.isErr():
raiseIOSelectorsError(res.error())
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
proc selectInto*[T](s: Selector[T], timeout: int,
results: var openArray[ReadyKey]): int {.
raises: [Defect, IOSelectorsException].} =
let res = selectInto2(s, timeout, results)
if res.isErr():
raiseIOSelectorsError(res.error())
res.get()
proc select*[T](s: Selector[T], timeout: int): seq[ReadyKey] {.
raises: [Defect, IOSelectorsException].} =
let res = select2(s, timeout)
if res.isErr():
raiseIOSelectorsError(res.error())
res.get()
proc close*[T](s: Selector[T]) {.raises: [Defect, IOSelectorsException].} =
let res = s.close2()
if res.isErr():
raiseIOSelectorsError(res.error())
proc contains*[T](s: Selector[T], fd: SocketHandle|cint): bool {.inline.} =
s.checkKey(int32(fd))
proc setData*[T](s: Selector[T], fd: SocketHandle|cint, data: T): bool =
s.fds.withValue(int32(fd), skey):
skey[].data = data
return true
do:
return false
template withData*[T](s: Selector[T], fd: SocketHandle|cint, value,
body: untyped) =
s.fds.withValue(int32(fd), skey):
var value = addr(skey[].data)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
body
template withData*[T](s: Selector[T], fd: SocketHandle|cint, value, body1,
body2: untyped) =
s.fds.withValue(int32(fd), skey):
var value = addr(skey[].data)
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
body1
do:
exception tracking (#166) * exception tracking This PR adds minimal exception tracking to chronos, moving the goalpost one step further. In particular, it becomes invalid to raise exceptions from `callSoon` callbacks: this is critical for writing correct error handling because there's no reasonable way that a user of chronos can possibly _reason_ about exceptions coming out of there: the event loop will be in an indeterminite state when the loop is executing an _random_ callback. As expected, there are several issues in the error handling of chronos: in particular, it will end up in an inconsistent internal state whenever the selector loop operations fail, because the internal state update functions are not written in an exception-safe way. This PR turns this into a Defect, which probably is not the optimal way of handling things - expect more work to be done here. Some API have no way of reporting back errors to callers - for example, when something fails in the accept loop, there's not much it can do, and no way to report it back to the user of the API - this has been fixed with the new accept flow - the old one should be deprecated. Finally, there is information loss in the API: in composite operations like `poll` and `waitFor` there's no way to differentiate internal errors from user-level errors originating from callbacks. * store `CatchableError` in future * annotate proc's with correct raises information * `selectors2` to avoid non-CatchableError IOSelectorsException * `$` should never raise * remove unnecessary gcsafe annotations * fix exceptions leaking out of timer waits * fix some imports * functions must signal raising the union of all exceptions across all platforms to enable cross-platform code * switch to unittest2 * add `selectors2` which supercedes the std library version and fixes several exception handling issues in there * fixes * docs, platform-independent eh specifiers for some functions * add feature flag for strict exception mode also bump version to 3.0.0 - _most_ existing code should be compatible with this version of exception handling but some things might need fixing - callbacks, existing raises specifications etc. * fix AsyncCheck for non-void T
2021-03-24 09:08:33 +00:00
body2
proc getFd*[T](s: Selector[T]): cint = s.kqFd