| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/tags/APR_0_9_3@64459 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64457 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
is not a noop for some mutexes
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64456 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64455 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
those platforms where we can't/shouldn't shove a sem_t * into
an int -- bluntly reported by Joe Orton
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64454 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64453 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
process, and we are using anonymous proc_mutex or global_mutex methods,
so win32 should return APR_SUCCESS for proc_mutex_child_init.
We also introduced the kernel 'object'
folding function for shared memory
section names, so reuse that folding function here to provide reliable
Win2K/XP names (prefixed with \global\) and fold away any slashes or
backslashes from that path.
Based on issues observed by "Andre Schild" <A.Schild@aarboard.ch> with
the httpd mod_ssl implementation for Win32, with input from Andre and
JimJ.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64452 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64451 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
based on much discussion on dev@apr, and Madhu pointed out that they
had their own threads implementation that was abandoned in HPUX 11,
and therefore not worth implementing.
PR: 9457
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64450 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64449 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
Submitted by: Craig Rodrigues <rodrigc@attbi.com>
Reviewed by: Jeff Trawick
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64448 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
Submitted by: Garrett Rooney <rooneg@electricjellyfish.net>
Reviewed by: Justin Erenkrantz
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64447 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
| |
builds, such as mod_perl. Let's only override this tag if a macro
workaround wasn't already defined.
Submitted by: Stas Bekman
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64446 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64445 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64443 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
we will make no presumtion that fd 0..2 are special cases from
apr_file_dup(), and remain uninherited as in the previous release of APR.
Although it's a common Unix construct to;
close(2)
fd = dup(2, x)
it's certainly not portable and shouldn't be encouraged.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64442 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
1) track the target file's existing flags and register the proper
sort of cleanup (a bug in the new design) or
2) revert to the previous behavior and retain the existing cleanup
I've gone with option 2, since Joe Orton has expressed concern with
introducing too many changes in the coming release. However, this
implies that;
apr_file_close(fd1);
apr_file_dup2(fd1, fd2);
is absolutely unsupported. Since it wouldn't work on Win32 in the
first place, I'm not terribly concerned about this limitation.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64441 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64440 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64439 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
into 0.9 and 1.0 categories, and knocking off one issue on Win32 that
was fixed by closing apr_proc_t::hproc under the same conditions as
unix would prune zombie processes.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64438 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64437 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
local variable in apr_file_dup.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64436 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
BTW - that code *definately* violates the style guide; either repeat
the assignments or indent the second line!!!
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64435 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
no cleanup - so enforce APR_FILE_NOCLEANUP to force apr_file_inherit_[un]set
to choke.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64434 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64433 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
and the exact sequence is critical.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64432 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64431 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
by returning APR_EINVAL if the user attempts to toggle the inherit state.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64430 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
register a cleanup whatsoever.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64429 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
into one
of the std[in|out|err] fd's - we should automatically toggle the inherit
bit. If the user calls apr_file_inherit_set() again it will be a noop.
This brings Unix into line with Win32's implementation of dup. If folks
feel we should *only* apply this code to which_dup==2 cases, that's fine
with me; although close(1); dup(n) would generally create a new fd 1 on
unix, that code isn't portable to Win32 and should be strongly discouraged.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64428 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
in apr_file_setaside(), based on the ->flags of the fd being set-aside.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64427 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
called apr_file_close() against the original handle, and then invoke
apr_file_dup2(). However, in that case the apr_file_t has already been
unregistered from the cleanups.
Always register a cleanup within _file_dup(), yet kill any remaining
cleanup for the existing apr_file_t in the apr_file_dup2() situation.
This also prevents double-registration, but in a more robust manner.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64426 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
APR_ENOMEM checks are noops.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64425 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
apart to solve two major bugs.
1) APR attempts to defer creation of objects until the underlying atomic
call actually succeeds - in this case, dup() for the apr_file_dup()
flavor (which_dup == 1).
2) APR cannot trust the LHS point passed in via apr_file_dup(), nowhere
else do we require the user to 'NULL' out the target variable.
This patch always creates a new apr_pcalloc() for apr_file_dup()
ignoring the old value of *new_file.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64424 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
it belongs (this will be evident in bug #2). The patch fixes a fd
leak on Netware where each call to apr_file_dup2 failed to close the
original fd.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64423 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
Also define APR_INT64_T_FMT_LEN on Windows and Netware; Unix already
defines that symbol.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64422 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
propogate flags as passed without enforcing their values - the rest
of APR code will make invalid assumptions.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64421 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64420 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
Add a comment about the order in which cleanups are run. This has been
the case for quite a while, but was never documented.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64419 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
since NetWare doesn't yet have a way to kill a spawned NLM.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64418 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64417 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
Needs *much* more work.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64416 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
(apr_mmap_create): Return errno instead of APR_ENOMEM.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64415 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
as libapr.pdb which is sufficient for 99% of all purposes (see dev@ dialog
on the problems and reasons for this change.)
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64414 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64413 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64412 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
via apr.h, so there is no need to use APR_FLAG_*, and no need to check
for CODESET in configure.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64411 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
Submitted by: Craig Rodrigues <rodrigc@attbi.com>
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64410 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
files are flushed before calling fstat.
[Port of jorton's rev. 1.65 commit of apr/file_io/unix/filestat.c]
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64409 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
apr_private.h in misc/unix/charset.c so that the Unix version of
apr_os_locale_encoding actually has a chance to return something
different from apr_os_default_encoding.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@64408 13f79535-47bb-0310-9956-ffa450edef68
|