* Add test coverage for atomics
Signed-off-by: Elijah Shaw-Rutschman <elijahr@gmail.com>
* Fix compareExchange bugs for non-trivial objects
Bugs fixed:
1. compareExchange would not set the desired value in the success case.
2. compareExchange would not set var expected to the found value in the failure case.
3. withLock would spin forever running the unit tests. try..body..finally prevents this. Not sure why this makes a difference, since an exception wasn’t being raised, but clearing the guard in a finally block seems correct anyways.
Signed-off-by: Elijah Shaw-Rutschman <elijahr@gmail.com>
* net: allow close() to ignore SSL failures due to disconnections
Comes with this PR is also a SIGPIPE handling contraption.
* net: don't do selectSigpipe() on macOS
macOS sockets have SO_NOSIGPIPE set, so an EPIPE doesn't necessary mean
that a SIGPIPE happened.
* net: fix alreadyBlocked logic
* net: WSAESHUTDOWN is also a disconnection error
* asyncnet, net: don't attempt SSL_shutdown if a fatal error occurred
Per TLS standard and SSL_shutdown(3ssl). This should prevent errors
coming from a close() after a bad event (ie. the other end of the pipe
is closed before shutdown can be negotiated).
Ref #9867
* tssl: try sending until an error occur
* tssl: cleanup
* tssl: actually run the test
I forgot to make the test run :P
* tssl: run the test on ARC, maybe then it'll be happy
* tssl: turns off ARC, switch tlsEmulation on for freebsd
* tssl: document why tlsEmulation is employed
* net: move SafeDisconn handling logic to socketError
* Fix#14906 by wrapping outputStream with PipeOutStream
* Fix compile error when ./build_all.sh
* Use PipeOutStream on posix
* Fix compile error when build_all.sh
* Use ptr UncheckedArray
* Replace copyRefObj
* Remove tmp buffer from posPeekData
* Add more tests for outputStream
* Add comments about PipeOutStream.buffer
* Fix bug in posReadLine
* Move implementation of newPipeOutStream to streamwrapper module
* fix#14475; make unittest work with -d:nodejs
* fixup
* fixup
* disable inim, delaunay which failed after unittest.require got fixed
* re-enable tests that have been fixed
* add debug format string
* remove try except
* add changelog
* add docs and more tests
* Update lib/pure/strformat.nim
Co-authored-by: Juan Carlos <juancarlospaco@gmail.com>
* minor
Co-authored-by: Juan Carlos <juancarlospaco@gmail.com>
* json custom serialization; application for strtabs
* serialize using nesting
* make toJson more feature complete
* add since
* Revert "Improve JSON serialisation of strtabs (#14549)"
This reverts commit 7cb4ef26ad.
* better approach via mixin
* toJson, jsonTo
* fix test
* address comments
* move to jsonutils
* doc
* cleanups
* also test for js
* also test for vm
Per SSL_get_error(3):
The current thread's error queue must be empty before the TLS/SSL I/O
operation is attempted, or SSL_get_error() will not work reliably.
There has been records of not clearing the error queue causing weird SSL
errors when there shouldn't be any, see:
https://github.com/openssl/openssl/issues/11889
The error we're looking for is "certificate verify failed". The routine
that reports this will be different between openssl versions, so it
makes no sense to track the routine name as well.
* tfdleak_multiple: introduce stress tester for tfdleak
Imported from #14548 and tweaked for consumption by testament.
This test seems to be really good at bringing out the flakyness of
tfdleadk.
Co-authored-by: Timothee Cour <timothee.cour2@gmail.com>
* tfdleak: increase accuracy of the test on Windows
This commit implements a new testing strategy for Windows:
1. We duplicate the handle that will be tested and enable inheritance.
This duplicate will serve as a reference handle.
2. In addition to checking whether the handle is valid, we also verify
whether the handle is the same as the reference. This gives us
complete certainty on whether the handle in question is inherited
from the parent.
A side effect is that this uses Windows 10+ APIs. But since
this is just for the test, we don't have to be picky about it.
Ideally we would want to do something like this for other POSIX-based
system, but most of them lack a facility to do this, and as of writing
there isn't any false positive for them, so we won't need the additional
checks.
MemFile.fHandle will also no longer be tested, as this handle defaults
to being invalid.
Co-authored-by: Timothee Cour <timothee.cour2@gmail.com>
* use check
* trunner now works with cpp
* cleanup: move compiler/unittest_light => stdtest/unittest_light
* fix tests/readme.md
* remove deadcode references to rodfiles
* fix for windows
* added bit operations based on bit slices, clarified documentation, made non-mutating versions of mask ops
* Added since annotations, some runnable examples
* Added mask()/masked() functions, changed internal workings of mask ops to use new bit* funcs
* Changelog updated for new bitops improvements
* Reorganization, added runnable examples
* Documentation adjustments
* Add incltrl for since annotation
* Fix masked() impl
* Fix mask() return type
* Don't call toUnsigned on already unsigned types
* Remove improper `var T` for flipMasked()
* Fix return types for flipMasked()
* Slight syntactic cleanup for *masked ops
* Added tests for bitslice operations, new mask() operation, non-mutating mask ops
* Fix setmasked() var T issue
* More comprehensive tests
* Fix runnable example for bitsliced()
* Fix runnable example for mask()
* Error -> Defect for defects
The distinction between Error and Defect is subjective,
context-dependent and somewhat arbitrary, so when looking at an
exception, it's hard to guess what it is - this happens often when
looking at a `raises` list _without_ opening the corresponding
definition and digging through layers of inheritance.
With the help of a little consistency in naming, it's at least possible
to start disentangling the two error types and the standard lib can set
a good example here.
* base `parseEnum` on a case statement, fixes#14030
* apply simplifactions / clean up, remove `norm` node, use strVal
* export `normalize` in json.nim
* cmp using nimIdentNormalize, error at CT if ambiguous enum found
`nimIdentNormalize` provided by @cooldome.
We track all names of the branches we have created so far and error if
a duplicate is found.
Dummy change to make github react...
* fix docstring of `nimIdentNormalize`
* make `typ` arg `typedesc`, add lineinfo, call norm. only once
* * relativePath(foo) now works
* relativePath(rel, abs) and relativePath(abs, rel) now work (fixes#13222)
* relativePath, absolutePath, getCurrentDir now available in more targets (eg: vm, nodejs etc)
* fix bug: isAbsolutePath now works with -d:js; add tests
* workaround https://github.com/nim-lang/Nim/issues/13469
* remove `relativePath(path)` overload for now
* add back changelog after rebase
* Working on OpenBSD CI
* Condense steps into 2 steps to make output easier to follow.
* Move up one directory after csources build.
* Remove FreeBSD build manifest and add OpenBSD test ignores for coroutines and hot code reloading.
* If runCI fails, run the test results script.
* Add email trigger for build failure
* Remove .git from repository URL
* Disable SFML test on OpenBSD
* Disable tgetaddrinfo on OpenBSD as only UDP and TCP protocols are supported.
* Remove getFilePermissions as it causes CI test failures with NimScript.
* Set clang as cc in nim.cfg and use gmake to build csources.
* Add getCurrentDir to nimscript.
* Remove duplicate getCurrentDir and check for not weirdTarget.
* Add CI badge for OpenBSD.
* Disable tests which allocate lots of memory for OpenBSD.
* Use `CORO_BACKEND_SETJMP` on OpenBSD rather than ucontext.
* Simplify building of koch
* Disable t8657 on OpenBSD. See issue #13760.
* Fix#12142 - tarray_of_channels fails on OpenBSD
* Disable thhtpclient_ssl and tosprocterminate on OpenBSD. These tests can be enabled at a later date after fixing them.
* Install libffi.
* Set path to libc for openbsd.
* Disable tevalffi for now.
* Remove tevalffi.nim.
* Use ncpuonline sysctl rather than ncpu.
* Disable tacceptcloserace and tasynchttpserver on OpenBSD.
* Enable tacceptcloserace and tasynchttpserver.
* Fix#13775 as suggested by @alaviss - use /bin/cat on OpenBSD rather than /bin/sh.
* Enable test on OpenBSD.
* Disable tflowvar on OpenBSD.
* io: make file descriptors non-inheritable by default
This prevents file descriptors/handles leakage to child processes
that might cause issues like running out of file descriptors, or potential
security issues like leaking a file descriptor to a restricted file.
While this breaks backward compatibility, I'm rather certain that not
many programs (if any) actually make use of this implementation detail.
A new API `setInheritable` is provided for the few that actually want to
use this functionality.
* io: disable inheritance at file creation time for supported platforms
Some platforms provide extension to fopen-family of functions to allow
for disabling descriptor inheritance atomically during File creation.
This guards against possible leaks when a child process is spawned
before we managed to disable the file descriptor inheritance
(ie. in a multi-threaded program).
* net, nativesockets: make sockets non inheritable by default
With this commit, sockets will no longer leak to child processes when
you don't want it to. Should solves a lot of "address in use" that might
occur when your server has just restarted.
All APIs that create sockets in these modules now expose a `inheritable`
flag that allow users to toggle inheritance for the resulting sockets.
An implementation of `setInheritance()` is also provided for SocketHandle.
While atomically disabling inheritance at creation time is supported on
Windows, it's only implemented by native winsock2, which is too much for
now. This support can be implemented in a future patch.
* posix: add F_DUPFD_CLOEXEC
This command duplicates file descriptor with close-on-exec flag set.
Defined in POSIX.1-2008.
* ioselectors_kqueue: don't leak file descriptors
File descriptors internally used by ioselectors on BSD/OSX are now
shielded from leakage.
* posix: add O_CLOEXEC
This flag allows file descriptors to be open() with close-on-exec flag
set atomically.
This flag is specified in POSIX.1-2008
* tfdleak: test for selectors leakage
Also simplified the test by using handle-type agnostic APIs to test for
validity.
* ioselectors_epoll: mark all fd created close-on-exec
File descriptors from ioselectors should no longer leaks on Linux.
* tfdleak: don't check for selector leakage on Windows
The getFd proc for ioselectors_select returns a hardcoded -1
* io: add NoInheritFlag at compile time
* io: add support for ioctl-based close-on-exec
This allows for the flag to be set/unset in one syscall. While the
performance gains might be negliable, we have one less failure point
to deal with.
* tfdleak: add a test for setInheritable
* stdlib: add nimInheritHandles to restore old behaviors
* memfiles: make file handle not inheritable by default for posix
* io: setInheritable now operates on OS file handle
On Windows, the native handle is the only thing that's inheritable, thus
we can assume that users of this function will already have the handle
available to them. This also allows users to pass down file descriptors
from memfiles on Windows with ease, should that be desired.
With this, nativesockets.setInheritable can be made much simpler.
* changelog: clarify
* nativesockets: document setInheritable return value
* posix_utils: atomically disable fd inheritance for mkstemp
