| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@63117 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@63079 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
memory for metadata
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@63078 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62985 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62947 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
Thanks Jin.
Submitted by: Jin Hong <jinh@cnet.com>
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62912 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
| |
code.
PR: 9710
Submitted by: Per Ekman [pek@pdc.kth.se]
Reviewed by: Jeff Trawick, who expanded Per's fix to a couple of
more places (so all breakage is mine :) )
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62911 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62875 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
creating process' pool (cleans up after apr_shm_create) and an attaching
process' pool (cleans up after apr_shm_attach).
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62848 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Nearly, the apr.hw patch needs some review. In short, several quite
standard ansi headers (e.g. time.h) aren't present in the WinCT port,
but these changes require corresponding changes to apr.h.in.
I changed Mladen's #define'd symbol names for the macro blocks to
IF_WIN_OS_IS_UNICODE and ELSE_WIN_OS_IS_ANSI to make the code a bit more
readable, and drop the global apr_os_level from each macro invocation.
Also, I changed the scope of his APR_HAS_ANSI_FS to local scope, since
it has no application in the public headers.
Mladen's patch helps NT as well, allowing the /D WINNT flag to define
NT-only compilations. With WINNT defined, all UNICODE/ANSI os version
tests drop out entirely.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62837 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62836 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
obscure is already obvious to OS2/BeOS folks, feel free to round out.
I suspect it's simply the pointer to the phys mem, filling out the
internals of apr_shm_t, but I'm not certain.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62831 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
take the handle and do interesting things, such as passing it (after
duping it) into the child.
Unix implementation is simply the address of the shm region, AFAICT,
but Aaron is reviewing.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62830 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
temporary but gets the server running again.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62826 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
| |
creating either anon or named memory now.
Of course, reattaching a detached processes anon handle requires some
further efforts. An apr_shm_os_get/put should help.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62822 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Name-based shmem now works on all tested platforms (linux, solaris)
and on all shmem flavors.
- All critical FIXMEs are now fixed.
- Much more robust error checking.
- Properly implemented attach/detach and destroy routines.
- Fixed a couple bugs having to do with the segment length.
- Temp files/mmaped files are now removed after use.
- mmaped files, temp files, and segments are now created exclusively,
meaning that the create function will fail if the file already exists
(this has important security implications).
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62816 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
| |
Changed the return error from apr_shm_attach if filename == NULL. Normally
in APR input parameters are not explicitly checked, but since in apr_shm
a NULL filename has a special meaning (anonymous memory) it will now
return APR_EINVAL in that case.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62814 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62800 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
memory. When a segment has been marked for deletion, other processes
are not allowed to attach to that segment.
This also fixes the problem with the name-based portion of the testshm.c
test.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62798 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62796 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
so if we crash, they are removed
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62792 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62777 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
- change the p -> pool and add a pool accessor
- add the attach/detach code
- adjust the name we give an area so it's more descriptive
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62762 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62758 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
follow.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62754 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62753 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62747 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62745 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62740 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62739 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
memory API, which was unusable on Win32 and other platforms that do
not have inherited shared memory segments since it only supported
anonymous memory. We now support both anonymous and name-based
shared memory.
Anonymous shmem tested to work on the following platforms:
Linux 2.2, Linux 2.4, Solaris 8, FreeBSD 5.0-CURRENT
Name-based shmem is _not_ fully functional on UNIX, but this API
replaces all of the preexisting apr_shmem_t functionality.
Stubs were provided for Beos and OS/2, and as much relevant code
as possible was preserved, but no guarantees of correctness.
Reviewed by: Justin Erenkrantz, Will Rowe
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62735 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
physical length after apr_file_trunc] and change the api a bit
for Aaron's new apr_shm.h
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62733 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62730 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62729 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
| |
into the libapr/apr.dsp files until that header is committed.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62728 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
The reason for this, is that it is very possible to want a temp
file that isn't deleted when the file is closed. It also makes sense
to have the flags in the apr_file_t if possible.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62716 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62708 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62707 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62706 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
and get rid of apr_lock_create_np(); apr_lock_create() has a new
parameter for specifying the lock mechanism (or APR_LOCK_DEFAULT to let
APR choose)
(same for apr_proc_mutex_create_np() and apr_proc_mutex_create())
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62684 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
Allocate some extra space for heap structures so that a block of the same
size as the reqsize can be allocated later with apr_shm_malloc().
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62681 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62264 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
to specify that the segment will go away when the last
user detaches
This fixes a problem with Apache leaving behind a shared memory
segments at termination.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62256 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
| |
shared memory code for unix
there are still some unexpected uses of APR_EGENERAL and a general
lack of cleanup of previous steps when something goes wrong
(not always a reasonable thing to do with shared memory)
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62254 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
that shared memory segment).
apr_shm_malloc will either return valid memory or NULL, so this is a
useful check. When using libc's malloc(), that isn't necessarily the
case. However, this patch requires the caller to check for a NULL
return which they probably don't do anyway, so the segfault gets
moved out of APR and into the caller. That's good enough for now...
PR: Graham's posts to dev@apr
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62250 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62214 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
| |
and SHM_W (e.g. OS/390)
Submitted by: Jeff Trawick
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62203 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
| |
this definitely clears up warnings on Tru64 and should get APR building
again on HP-UX, where this is an error
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62165 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
| |
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62161 13f79535-47bb-0310-9956-ffa450edef68
|
|
|
|
|
|
|
|
|
|
| |
an apr_file_t to apr_file_open. Also, apr_os_file_get returns
a status value, not the OS file descriptor.
This gets Apache running again on Linux.
git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@62160 13f79535-47bb-0310-9956-ffa450edef68
|