| Age | Commit message (Collapse) | Author |
|---|
|
from pjanzen; ok marc
|
|
uninitialized lock ends up in a locked state. This lead to a deadlock if we
called pthread_key_create().
ok marc@, kurt@
|
|
pthread_stackseg_np(). With input and okay marc@
|
|
by replicating part of $$dyncall in the code that sets up a thread's initial
stack frame. Also make sure we actually reserve some space for that initial
stack frame.
ok miod@
|
|
isolate its usage to libpthread only and replace with generic non-static
mutex support in the one place it is needed:
- remove _FD_LOCK/UNLOCK from lseek and ftruncate in libc and make the
functions weak so that libpthread can override with its own new
versions that do the locking.
- remove _thread_fd_lock/unlock() weak functions from libc and adjust
libpthread for the change.
- add generic _thread_mutex_lock/unlock/destroy() weak functions in libc
to support non-static mutexes in libc and add libpthread and librthread
implementations for them. libc can utilize non-static mutexes via the
new _MUTEX_LOCK/UNLOCK/DESTROY() macros. Actually these new macros can
support both static and non-static mutexes but currently only using
them for non-static.
- make opendir/closedir/readdir/readdir_r/seekdir/telldir() thread-safe
for both thread libraries by using a non-static mutex in the struct
_dirdesc (typedef DIR), utilizing it in the *dir functions and remove
remaining and incorrect _FD_LOCK/UNLOCK() use in libc.
- add comments to both thread libraries to indicate libc depends on the
current implementation of static mutex initialization. suggested by
marc@
- major bump libc and libpthread due to function removal, structure
change and weak symbol conversions.
okay marc@, tedu@
|
|
|
|
|
|
|
|
casting when safe or adding ARGSUSED where needed. Reviewed and
improvements from millert@ and marc@. okay marc@
|
|
- use the correct types
- mark _thread_gc() w/ARGSUSED
- fix a 'a cast does not yield an lvalue'
okay marc@
|
|
allignment. okay marc@
|
|
|
|
|
|
calling _get_curthread() again. fixes some lint warnings. okay marc@
|
|
merge _thread_fd_unlock_thread() into _thread_fd_unlock(). okay marc@
|
|
applications that increase nofiles using setrlimit(2). ok marc@
|
|
|
|
|
|
from the parent to all non-blocking. This prevents the child from restoring
the file to be blocking upon close() and causing the parent to deadlock
when it later attempts to utilize the blocking fd. Corrects the netbeans
build failure caused by the jdk forkAndExec() function that replaces the
std fd's before execvp(). okay marc@
|
|
build warnings noticed by david@. okay millert@ marc@
|
|
conditionally compile code using these defines will work.
okay tedu@, marc@, milllert@
|
|
always saved upon context switches, as FPU registers are heavily used for
long long computations (don't ask). Gcc default to -m4.
Credits to drahn@ otto@ and deraadt@ for feedback and help testing.
Upgrade procedure if you don't want to use the damn snapshots:
- build and install new kernel, reboot off it
- build new gcc, do not install it yet
- make includes
- install new gcc
- build and install lib/csu and lib/libc
- make build
|
|
|
|
ok marc@, krw@
|
|
closed so that a threaded child process can still read it without blocking.
However, leaving stdin/out/err non-blocking when closed is bad because it
can be shared with non-threaded apps that can't deal with a non-blocking
file descriptor (i.e. cat). Therefore special case stdin/out/err pipes so
that they are reset to blocking upon a close(). Tested by robert@, jolan@
and myself with multiple OOo builds on mp systems where the problem was
seen more frequently.
|
|
|
|
|
|
need to bail from _thread_kern_poll() when a file descriptor is in
closing state. corrects segfault reported by ckuethe@
|
|
this can potentially block indefinitely. Instead just
protect against fd state transitions and perform
the _thread_sys_shutdown(). Fixes deadlock reported by
Tero Koskinen <tero.koskinen at iki.fi>. okay kettenis@
|
|
NetBSD.
|
|
|
|
When a fd enters the closing state prevent any threads from
polling the fd and reschedule the thread with the closing_fd
flag set. This fixes a class of deadlocks where a thread is
blocked waiting for data (that may never arrive) and a later
thread calls close() or dup2() on the fd. okay brad@
|
|
_thread_fd_table_init()'s return value and blow it away. fixes
firefox plugin problem noticed by jolan@
|
|
stacks; report by Antoine Jacoutot; ok kurt@
|
|
|
|
applications that install their own thread stack red-zones with mprotect
PROT_NONE from entering the malloc pool when the pages are freed.
- round up requested thread stack size to closest page boundary to
simplify the red-zone calculations.
okay brad@
|
|
Adjust design of file descriptor table to eliminate races
with both opening and closing of file descriptor entries
and eliminates one class of deadlocks. One nice side effect
of this change in design should be better performance for
applications that open and close many file descriptors due
to reduced fd_table_lock contention and fd entry reuse.
- Add entry states to manage use of entry and eliminate
some closing races. fd entries are not deallocated upon
close() now.
- Call _thread_fd_table_init with one of five discreet
modes to properly initialize an entry and manage the
state transition to open.
- When closing an entry hold the entry spinlock locked
across the state transition and the _thread_sys_close
call to close another race.
- Introduce a new lock type FD_RDWR_CLOSE that transitions
either a closed entry or an open entry into closing state
and then waits for a RDWR lock so that the lock queue can
unwind normally. All subsequent fd lock attempts for that
entry are rejected with EBADF until the fd is fully closed,
or reopened by dup2(). Once a thread holds the FD_RDWR_LOCK
it is safe to close() it or dup2() on it.
- When a thread creates a new fd there is a window of time
when another thread could attempt to use the fd before the
creating thread has initialized the entry for it. This can
result in improper status_flags for the entry, so record
the entries init mode, detect when this has happened and
correct the status_flags when needed.
reviewed by marc@ & brad@, tested by several, okay brad@
|
|
|
|
File status flags should be shared for dup'ed file descriptors.
However fd_table_entry's should not be shared for dup'ed file
descriptors so they can be independently be closed without
interfering with dup'ed fd's.
- split out file status flags into its own structure
fs_flags to manage sharing of status flags between
dup'ed file descriptors.
- when duplicating a fd, initialize a new fd_table_entry
for the new fd, but share the status flags via status_flags.
- consolidate the code that sets the underlying system fd
to be non-blocking to a new function _thread_fs_flags_init()
- consolidate the code that sets the underlying system
fd back to blocking into a new function _thread_fs_flags_replace()
This change is needed as a prerequisite to the coming race
and deadlock corrections. okay marc@
|
|
instructions once every few runs. Problem noted by Botond Botyanszki.
ok marc@
|
|
to cast away constness whenever we want to pass in a string that is
already const -- we copy the string if we want to use it, so having it
be non-const in the first place does not make any sense.
From tholo@
ok tedu@
|
|
it need to be matched with a call to _thread_machdep_restore_float_state(),
so add missing one in _thread_machdep_init().
- 16-byte align fp frame
- call fwait() before save and and after restoring fp state.
lots of guidance and ok mickey@
|
|
suggested by kjell@; ok otto@ pat@ millert@ jaredy@
|
|
weak functions; ok millert@.
|
|
originally reported by Gergely Kovacs; help from dhartmei@;
ok tedu@ millert@
|
|
prevents userland from causing an overflow of tv_sec. okay marc@
|
|
|
|
Add new sys/_types.h header
Include machine/_types.h or sys/_types.h where applicable
|
|
OK deraadt@
|
|
|
