uv__fs_buf_iter currently sets req->bufs to NULL after it is done, but
if the operation fails with EINTR then it will be retried, at which
point it expects the bufs to not be NULL, causing a seg fault as in
https://github.com/nodejs/node/issues/4291.
uv__fs_buf_iter should not set req->bufs to NULL if the operation
fails with EINTR.
Also, when it sets req->bufs to NULL, it should set req->nbufs to 0 as
well, so we don't have the messy situation of a positive nbufs with no
actual bufs.
PR-URL: https://github.com/libuv/libuv/pull/661
Reviewed-By: Fedor Indutny <fedor@indutny.com>
Trying to acquire a rwlock that the current thread already owns may fail
(not "shall fail") with EDEADLK. Libuv considers that a fatal error and
rightly so in my opinion; it normally means the program contains a logic
error.
The test had indeed logic errors. This commit splits off the UV_EBUSY
return code testing into a separate thread, eliminating the potential
for self-deadlock.
Fixes: https://github.com/libuv/libuv/issues/544
PR-URL: https://github.com/libuv/libuv/pull/649
Reviewed-By: Jeremy Whitlock <jwhitlock@apache.org>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
The following changeset
442b8a5a84 "unix: use QUEUE_MOVE when iterating over lists"
introduced a new assert failure:
`queue_foreach_delete` failed: exit code 6
Output from process `queue_foreach_delete`:
run-tests: src/unix/linux-inotify.c:244: uv_fs_event_stop: Assertion `w != ((void *)0)' failed.
Simplest test case for this:
1. create and start two uv_fs_event_t for the same path;
2. in the callback for the first one, call uv_close() on it;
3. assert/segfault while accessing the second uv_fs_event_t from uv__inotify_read().
PR-URL: https://github.com/libuv/libuv/pull/621
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
The test adds all types of handles, that were known to be affected by
QUEUE_REMOVE()-within-QUEUE_FOREACH() bug, to a loop.
It then calls uv_close() to trigger QUEUE_REMOVE()-within-QUEUE_FOREACH() case
and checks whether a particular QUEUE is corrupted or not.
Restrict the test to Linux only for now as it fails on other platforms
for various reasons.
PR-URL: https://github.com/libuv/libuv/pull/621
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
On Windows the pipe implementation could read more from a stream than
was available and it would create an assertion failure. This change
will make it so we read the minimum of the available data or the length
of the data.
To test this, I took the existing ipc_send_recv_tcp test and updated it
to do two writes and two read on each side of the pipe since that was the
reproduction recipe used by the reporter. This approach reproduced the
issue on Windows and the committed fix resolved the issue.
Fixes: https://github.com/libuv/libuv/issues/505
PR-URL: https://github.com/libuv/libuv/pull/549
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Previous behavior on Windows was to set the path to NULL for removed
and renamed fs events. This was because the path provided by
ReadDirectoryChangesW might (in rare cases) be an 8.3 short name which
could then no longer be converted to a long name after the path had
been removed or renamed. This meant that the user had to detect which
path was actually deleted or renamed and required the user to rescan
the entire watched subtree, taking several seconds or more for large
subtrees.
However, ReadDirectoryChangesW is publicly documented to emit 8.3
short names if the original handle for the changed path was opened
using an 8.3 short name, and libuv may already emit 8.3 short names for
other events if the path cannot be explicitly resolved to a long name.
This commit fixes the path for removed and renamed fs events, and does
not set the path to NULL, even if the path might be an 8.3 short name.
This makes it possible for the user to do a partial scan of the
subtree, restricting the scan to paths which match the long form or 8.3
short name (even if some of these are false positive matches). This
means that deletes and renames can now be detected accurately on
Windows within a few milliseconds.
Fixes: https://github.com/libuv/libuv/issues/634
Refs: https://github.com/libuv/libuv/pull/199
PR-URL: https://github.com/libuv/libuv/pull/639
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Equivalent to realpath(3), returns the full resolved absolute path of a
file or directory.
PR-URL: https://github.com/libuv/libuv/pull/531
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
'/dev/random' causes 'isatty()' to return 1 on AIX[1]. This causes
an issue where 'dev/tty' is opened (hard coded in
src/unix/tty.c:uv_tty_init). Tried to replace the hardcoded value with
what is returned by 'ttyname()' but on AIX all that is returned is
'/dev/'.
This seems to be a bug with the AIX kernel.
[1]http://goo.gl/wGa7mf
PR-URL: https://github.com/libuv/libuv/pull/624
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
~~~~
test/test-get-loadavg.c:30:14: warning: comparison of array 'avg' not
equal to a null pointer is always true [-Wtautological-pointer-compare]
test/test-getnameinfo.c:76:18: warning: comparison of array 'req.host'
not equal to a null pointer is always true
[-Wtautological-pointer-compare]
test/test-getnameinfo.c:77:18: warning: comparison of array
'req.service' not equal to a null pointer is always true
[-Wtautological-pointer-compare]
~~~~
PR-URL: https://github.com/libuv/libuv/pull/622
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
By default, FreeBSD has net.inet6.ip6.v6only set to 1 (i.e., no dual stack
support). A new function can_ipv6_ipv4_dual is added to check its value and
decide whether `udp_dual_stack` test must be skipped or not.
PR-URL: https://github.com/libuv/libuv/pull/614
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
uv_(get|set)_process_title is not implemented for AIX. See unix/aix.c
PR-URL: https://github.com/libuv/libuv/pull/616
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
When duplicating the socket handle being sent the target process
id is defaulted to the current process id. This enables uv_write2
to be used for thread-clustering in addition to process-clustering on
multi-threaded programming languages.
The ipc tests are updated to run in two modes. In the _inproc mode
the echo work is done on a second thread instead of in a second
process.
An internal function int uv_current_pid() is added to the windows
specific code which caches the value of GetCurrentProcessId(). This
means uv_write2 does not call GetCurrentProcessId() every inprocess
send.
Refs: https://github.com/joyent/libuv/issues/926
PR-URL: https://github.com/libuv/libuv/pull/540
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
In FreeBSD 10.2 the test sometimes times out because the "touch file"
timer fires before the "watch file" event has been registered in the
kqueue. Increasing the timeout value seems to fix the issue.
PR-URL: https://github.com/libuv/libuv/pull/581
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Valgrind blocks the setrlimit() system call. Print a warning and skip
the test, don't consider it a fatal error.
PR-URL: https://github.com/libuv/libuv/pull/565
Reviewed-By: Fedor Indutny <fedor.indutny@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
uv_os_fd_t is HANDLE on Windows, but closesocket needs a SOCKET.
PR-URL: https://github.com/libuv/libuv/pull/534
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
In FreeBSD if connect is called and the queue of pending connections(backlog)
is full, the call is automatically rejected with ECONNRESET. In the tests, 100
connections were tried and the backlog of both servers is 12, so the error.
To fix ipc_listen_before_write and ipc_listen_after_write, the backlog of the
server has been increased to 128.
PR-URL: https://github.com/libuv/libuv/pull/504
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
fs_access, fs_async_dir and fs_symlink_dir could fail when the
filesystem where the tests are run does not fully support getting the
file type via getdents. Having getdents support does not imply that the
underlying filesystem fully supports what is necessary for these tests
to have passed prior to this commit. So the failing tests were updated
to check for UV_DIRENT_FILE as it did previously but it will also check for
UV_DIRENT_UNKNOWN as well to handle the filesystems that do not fully support
getting the file type via getdents. For OS X and Windows, which are
known to fully support getting the file type, the tests work as they did
before and will not check for UV_DIRENT_UNKNOWN. We could/should update the
preprocessor directive that chooses "rigorous" or "lax" checks
accordingly when we learn of new environments that should always do the
original "rigorous" checks.
Fixes: https://github.com/libuv/libuv/issues/501
PR-URL: https://github.com/libuv/libuv/pull/508
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
It's an error to do so, so just ignore it. The test would cause an
invalid memory access if the fix is undone.
PR-URL: https://github.com/libuv/libuv/pull/488
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
On Windows we create multiple pipe handles (system handles) which are
attached to pending accept requests. Each of these will take turns in
replacing the reference in handle->handle, so make sure we allow for
that **only** for pipe servers.
PR-URL: https://github.com/libuv/libuv/pull/488
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
fs_event_watch_dir and fs_event_watch_dir_recursive could fail randomly
due to the way in which the tests were written. Originally timers were
used to create, remove and recreate the test files but this could lead
to a race condition if the timeout used to delete the test files ran
before all file creation fs events were handled. On top of that, the
file removal timer scheduled another timer to recreate the test files
and that timer's timeout could also lead to the same condition.
The refactoring removed timers for the removal/recreation of the test
files and instead the fs event callback was updated to have the
necessary logic to drive the test file removal. We no longer recreate
the test files since it appears to be unnecessary.
Fixes: https://github.com/libuv/libuv/issues/30
PR-URL: https://github.com/libuv/libuv/pull/480
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Join the watchdog thread unconditionally on exit. Fixes the following
harmless but noisy memory leak:
576 bytes in 1 blocks are possibly lost in loss record 1 of 2
at 0x4C2A9C7: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
by 0x40121B4: _dl_allocate_tls (in /usr/lib64/ld-2.21.so)
by 0x5AEA045: pthread_create@@GLIBC_2.2.5 (in /usr/lib64/libpthread-2.21.so)
by 0x450D3E: process_wait (runner-unix.c:212)
by 0x4067F1: run_test (runner.c:284)
by 0x405EC3: maybe_run_test (run-tests.c:180)
by 0x4058AD: main (run-tests.c:57)
PR-URL: https://github.com/libuv/libuv/pull/479
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
This allows writing and reading any amount of buffers,
regardless of what IOV_MAX may be defined as.
It also moves the IOV_MAX test from stream to core.
This is based on the excellent work of @bwijen in #269.
Refs: https://github.com/libuv/libuv/pull/269
PR-URL: https://github.com/libuv/libuv/pull/448
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
file_prefix_in_subdir is only used in Windows and OSX. Fixes the
following warning, which shows up with Clang:
test/test-fs-event.c:40:19: warning: unused variable
'file_prefix_in_subdir' [-Wunused-const-variable]
static const char file_prefix_in_subdir[] = "subdir";
PR-URL: https://github.com/libuv/libuv/pull/450
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Ifdef out the tests that rely on internal symbols when making a shared build.
PR-URL: https://github.com/libuv/libuv/pull/444
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Do not hard-fail network tests when libuv is built on
an isolated host/container.
Signed-off-by: Luca Bruno <lucab@debian.org>
PR-URL: https://github.com/libuv/libuv/pull/437
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Add new UV__POLLRDHUP event to be emitted when EPOLLRDHUP(in Linux) or
EV_EOF(in BSD / OSX) is detected and only if UV_READABLE is set.
When a read returns ECONNRESET after a UV__POLLRDHUP event, emit EOF instead
of the error.
Add tcp-squelch-connreset test. Not to be run on Windows as it returns
ECONNRESET error.
Fixes in test-poll and test-tcp-open so they pass after these changes.
PR-URL: https://github.com/libuv/libuv/pull/403
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Some 32 bits architectures, linux/mips and linux/mipsel in particular,
suffer from address space fragmentation when spawning many threads
with the default 8 MB stack size. The watchdog threads don't need
much stack space, all they do is sleep until the monitored process
exits, so lower it to 256 kB.
Fixes: https://github.com/libuv/libuv/issues/408
PR-URL: https://github.com/libuv/libuv/pull/429
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
UV_HANDLE_CONNECTED was defined, but never used anywhere - outside this if
condition inside uv__stdio_create. So this test can't be true.
UV_HANDLE_CONNECTION was meant.
A test was also added verifying the behaviour.
PR-URL: https://github.com/libuv/libuv/pull/404
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Make the test runner code that waits for other processes to terminate
retry the select() system call on EINTR and take elapsed time into
account.
A small side effect of this change is that millisecond timeout values
are now honored.
PR-URL: https://github.com/libuv/libuv/pull/410
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Introduce two new APIs:
int uv_tcp_init_ex(uv_loop_t*, uv_tcp_t* handle, int flags)
int uv_udp_init_ex(uv_loop_t*, uv_udp_t* handle, int flags)
The lower 8 bits of the flags field are used for the socket domain.
AF_INET, AF_INET6 and AF_UNSPEC are supported. If AF_UNSPEC is specified
the socket is created lazily, just like uv_{tcp,udp}_init.
Some Windows notes:
getsockname fails with WSAEINVAL if the socket is not bound. This could
potentially be improved by detecting the socket family and filling
the sockaddr_in/6 struct manually.
bind returns WSAEFAULT if we try to bind a socket to the wrong family.
Unix returns EINVAL.
PR-URL: https://github.com/libuv/libuv/pull/400
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Prior to MSVC 2015, the standard C99 `inline` keyword was missing,
added a compiler version check and disabled the inline replacement for
MSVC >= 2015 in test/task.h.
Refs: https://github.com/libuv/libuv/pull/341
PR-URL: https://github.com/libuv/libuv/pull/369
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
The implementation will leave the family set to `AF_UNSPEC` if a
netmask is not present, but the test driver would always print the
uninitialized buffer as an `AF_INET4` address. It will now print
"none" if there is no netmask (e.g., for loopback interfaces).
PR-URL: https://github.com/libuv/libuv/pull/373
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Prior to MSVC 2015, there was no standard compliant snprintf(), added a
compiler version check and disabled the snprintf() replacement for MSVC >=
2015 in test/task.h.
PR-URL: https://github.com/libuv/libuv/pull/341
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
