Adding a configure flag to disable GENA notification reordering as even
with an imillisleep(1), this mechanism consumes too much CPU on embedded
devices when there is a burst of notifications.
(cherry picked from commit c73d870f46)
When a device with embedded devices (like IGD) when created and one of
the embedded devices did not have any service, there was a Segmentation
Fault (see SF Tracker [ 2688125 ]).
(cherry picked from commit ab54cb3dc5)
When a lot of notifications were generated by a device in a short
period of time then 100% of the CPU was used to reorder those
notifications by pushing back the thread in the job queue. This
mechanism has been modified so now thread sleep 1 ms before being
pushed back into the job queue.
Removing DEFAULT_SCHED_PARAM parameter and use
sched_get_priority_min(DEFAULT_POLICY) instead.
(cherry picked from commit c33b11d09f)
Devices must respond to M-SEARCH requests for any supported version and the
response should specify the same version as was contained in the search target.
Previously, the device did not answer if the M-SEARCH request did not
contain the same version number than the version number of the device.
(cherry picked from commit 4966423d96)
Add Content-Language header in the response if and only if there is an Accept-Language header in the request.
Manually ported from revision d2238615e3.
This patch adds the WEB_SERVER_CONTENT_LANGUAGE parameter so the user can specify
the language used by the device during Description and Presentation steps of UPnP
through the HTTP CONTENT-LANGUAGE header.
By default, the WEB_SERVER_CONTENT_LANGUAGE is an empty string so no
CONTENT-LANGUAGE is added.
(cherry picked from commit 2fcbe6df52)
This patch allows a user to customize the stack size of the threads used by
pupnp through the new THREAD_STACK_SIZE variable. This is especially useful
on embedded systems with limited memory where the user can set THREAD_STACK_SIZE
to ITHREAD_STACK_MIN.
However, as this modification can have side effects, I set 0 as the default
value, so threads will continue to use the default stack size of the system
(which varies greatly as stated in
https://computing.llnl.gov/tutorials/pthreads/).
(cherry picked from commit 467f9987a1)
IPv6 is currently broken in latest release of branch-1.6.x, so find
a patch attached that correct the issue (small fixes on define,
undef and retVal).
(cherry picked from commit 55d581481f)
Hi,
I've made some modification to the libupnp v1.6.5
I've add a timeout for each TCP connect.
It is very useful when an UPnP device stop working and do not accept
connection for an UPnP action.
Modifications are only located in
upnp/src/genlib/net/http/httpreadwrite.c
For every TCP connection, I set the socket to non-blocking, perform
connect,
check result and wait during a timeout if necessary, then reset the
socket to blocking.
Please see this patch in attached file.
I hope it helps.
Regards,
David Hoeung
Consultant Extia
Orange Labs R&D
----
(cherry picked from commit 67009170d1)
This is a follow up from issue 6 in tracker id 3056713: calling UpnpSetMaxContentLength() by passing '0' disables the content length checking. This is useful for developing some prototype applications that deal with a lot of XML/SOAP data, and for debugging.
The corresponding c file change is already in the pupnp tree. Copy/pasting the relevant block of code here for clarity:
In upnp/src/genlib/net/http/httpreadwrite.c:
if (g_maxContentLength > 0 && parser->content_length > (unsigned int)g_maxContentLength) {
*http_error_code = HTTP_REQ_ENTITY_TOO_LARGE;
line = __LINE__;
ret = UPNP_E_OUTOF_BOUNDS;
goto ExitFunction;
}
This block of code checks only does the bounds check if g_maxContentLength > 0, and it's only place g_maxContentLength is checked.
Attached is a patch against the latest sources.
src/ssdp/ssdp_ctrlpt.c: In function ‘SearchByTarget’:
src/ssdp/ssdp_ctrlpt.c:634: warning: format ‘%s’ expects type ‘char *’, but argument 6 has type ‘int’
(cherry picked from commit 92b241b560)
Sending messages over UDP is broken in some Apple OSes
such as OS X and iOS. This might be broken in other OSes to but didn't
verify.
The fix is to modify the socket lenght argument of sendto to use the correct
sockaddr lenght dependng on whether the socket is IPV4 or IPV6.
Also added some error checks and debugging related to the issue
(cherry picked from commit 2b3ab1799b)
crash. This happens when the file being downloaded exceeds the device
memory - entirely possible when transferring video files.
The programmatic cause is that the logic implemented in the function
http_ReadHttpGet (which UpnpReadHttpGet calls) reads the entire file
into memory. The fix modifies the existing logic to discard data after
it's been read; there's no reason to keep it around since the caller
of UpnpReadHttpGet already has a copy of it.
This issue exists in 1.6.6 as well as the latest sources.
Patch submitted by Chandra (inactiveneurons).
(cherry picked from commit 4657e57766)
use connect() are broken. More specifically, connect() in these methods
is returning with an EINVAL. The programatic cause is that the address_len
argument passed to connect() is different in IPV4 vs IPV6 (as described in:
http://www.opengroup.org/onlinepubs/009695399/functions/connect.html).
The current code always uses the IPV6 size. The fix modifies each use of
connect() to use the correct size based on the address family being used.
Patch submitted by Chandra (inactiveneurons).
(cherry picked from commit 21660334e4)
at the end of the notification ("\r\n") in notify_send_and_recv() in
upnp/src/gena/gena_device.c.
Patch by Fabrice Fontaine.
(cherry picked from commit ebc941f265)
* InitHandleList() has never been implemented, I guess no one has ever
called it, so remove it.
* GetFreeHandle() and FreeHandle() are now static as they should.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@567 119443c7-1b9e-41f8-b6fc-b9c35fce742c
* New internal buffer added to store global/ula IPV6 address.
* Macros to test whether an IPV6 address is global or ula.
* UpnpGetServerUlaGuaIp6Address(): added interface.
* IN6_IS_ADDR_GLOBAL, IN6_IS_ADDR_ULA: new macros.
* gIF_IPV6_ULA_GUA: new buffer.
* UpnpRegisterRootDevice3(): Change to the test of already registered
devices for IPV6.
* UpnpGetIfInfo(): gua/ula issues.
Patch submitted by Ronan Menard.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@565 119443c7-1b9e-41f8-b6fc-b9c35fce742c
Hello,
When my patch for tracker ID 3007407 was accepted, the definition of the
serviceList string was changed from
#define SERVICELIST_STR "serviceList"
to
static const char *SERVICELIST_STR = "serviceList";
During internal code review of the final patch, it was pointed out that
sizeof(SERVICELIST_STR) == 4 since SERVICELIST_STR is now declared as
a pointer instead of an array.
If you wish to use a variable instead of a define, I suggest the
following instead:
static const char SERVICELIST_STR[] = "serviceList";
Thanks,
Chuck Thomason
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@556 119443c7-1b9e-41f8-b6fc-b9c35fce742c
Submitted: Chuck Thomason ( cyt4 ) - 2010-05-26 15:07:39 UTC
When the UPnP server is started, one alive message is broadcast for each
service in each device. It appears that libupnp's implementation of the
alive message generation does not correctly navigate the XML description
document when locating the services. This can result in the wrong UDN
being used in the alive message sent for a service.
In my specific case (see attached XML), the root EchoSTB device contains
no services, but its embedded MediaServer device contains 2 services.
When the existing libupnp code traverses the EchoSTB device in the XML,
it searches the global list of serviceLists within the document instead
of searching for a serviceList that is its direct child node. The
ContentDirectory and ConnectionManager services are then announced with
the UDN of EchoSTB1 (the root device) instead of with the UDN of
MediaServer, which is actually their parent device.
I discovered this behavior using libupnp-1.6.6. I have generated a patch
against branch-1.6.x that corrects the XML navigation such that all
services are traversed from their parent device, which results in the
correct UDN being sent in the alive message for each service. I built
from branch-1.6.x without this patch, tested, and confirmed that the
issue still exists as I observed it in libupnp-1.6.6. I then built
from branch-1.6.x with this patch, tested, and confirmed that the
issue was resolved.
Thanks,
Chuck Thomason
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@552 119443c7-1b9e-41f8-b6fc-b9c35fce742c
[svn] SF Bug Tracker [ 2995758 ] libupnp 1.6.6, wrong bind when reuseaddr is 1.
Submitted: viallard anthony ( homer242 )
When trying to use reuseaddr option in miniserver/miniserver.c, there
isn't a affectation of the port chosen (serverAddr.sin_port isn't
receive listen_port variable value).
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@549 119443c7-1b9e-41f8-b6fc-b9c35fce742c
Thanks for the load of updates, I'm still assimilating them ! Could I make
a suggestion though? The addition of printNodes(IXML_Node) to upnpdebug a
dds a new dependency on ixml.h for anything using upnpdebug.h. I'm making
quite a bit of use of upnpdebug in porting things to version 1.8.0, and I'd
prefer it if printNodes could be added to ixmldebug.h instead. I'm attach
ing a patch, what do you think ?
Nick
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@532 119443c7-1b9e-41f8-b6fc-b9c35fce742c
- int ithread_initialize_library(void);
- int ithread_cleanup_library(void);
- int ithread_initialize_thread(void);
- int ithread_cleanup_thread(void);
* SF Bug Tracker [ 2876374 ] Access Violation when compiling with Visual Studio 2008
Submitted: Stulle ( stulleamgym ) - 2009-10-10 19:05
Hi,
I am one of the devs of the MorphXT project and I use this lib in some
other of my projects, too. When I tried to upgrade the lib earlier for one
of my projects I had to realise that something did not work at first and
while most of the things were reasonably ease to be fixed. Now, the last
thing I encountered was not so easy to fix and I am uncertain if my fix is
any good so I'll just post it here and wait for some comments.
The problem was that I got an Access Violation when calling "UpnpInit". It
would call "ithread_rwlock_init(&GlobalHndRWLock, NULL)" which eventually
led to calling "pthread_cond_init" and I got the error notice at
"EnterCriticalSection (&ptw32_cond_list_lock);". It appeared that
"ptw32_cond_list_lock" was NULL. Now, I found two ways to fix this. Firstly
moving the whole block after at least one of the "ThreadPoolInit" calls
will fix the issue. Secondly, you could add:
#ifdef WIN32
#ifdef PTW32_STATIC_LIB
// to get the following working we need this... is it a good patch or
not... I do not know!
pthread_win32_process_attach_np();
#endif
#endif
right before "ithread_rwlock_init(&GlobalHndRWLock, NULL)".
Just so you know, I am using libupnp 1.6.6 and libpthreads 2.8.0 and both
are linked static into the binaries. I am currently using Visual Studio
2008 for development with Windows being the target OS. Any comment at your
end?
Regards, Stulle
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@527 119443c7-1b9e-41f8-b6fc-b9c35fce742c
* SF Bug Tracker [ 2392166 ] ithread_detach not called for finished worker thread
Submitted: Ulrik ( ulsv_enea ) - 2008-12-05 08:24
Valgrind reports a memory leak due to that the function ithread_detach is
not called for finished worker threads in ThreadPool.c.
==21137== 2,176 bytes in 8 blocks are possibly lost in loss record 5 of 5
==21137== at 0x4C20F3F: calloc (vg_replace_malloc.c:279)
==21137== by 0x4010F58: _dl_allocate_tls (in /lib/ld-2.6.1.so)
==21137== by 0x544BA92: pthread_create@@GLIBC_2.2.5 (in
/lib/libpthread-2.6.1.so)
==21137== by 0x5F94592: CreateWorker (ThreadPool.c:639)
==21137== by 0x5F95079: ThreadPoolInit (ThreadPool.c:784)
I'm using libupnp 1.6.6
For more info on pthread_detach, see:
http://gelorakan.wordpress.com/2007/11/26/pthead_create-valgrind-memory-lea
k-solved/
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@521 119443c7-1b9e-41f8-b6fc-b9c35fce742c
* SF Bug Tracker [ 2392304 ] Memory leak in SSDP AdvertiseAndReply
Submitted: Ulrik ( ulsv_enea ) - 2008-12-05 08:24
Valgrind reports a memory leak function in AdvertiseAndReply
(ssdp/ssdp_server.c) in libupnp 1.6.6
There are continue statements in many places in AdvertiseAndReply. In some
of those error handling cases the variable nodelist is not free'ed before
continuing to the next iteration. The next iteration will take care of
free'ing the nodelist from the previous iteration in most cases, but not
when breaking out of the for loop after the last element.
I belive this memory leak can be solved by makeing sure that the rows
ixmlNodeList_free( nodeList );
nodeList = NULL;
are always executed, also in the beginning of the last iteration when we
found out that there are not more elements.
==29110== at 0x4C21C16: malloc (vg_replace_malloc.c:149)
==29110== by 0x5D8DE0E: ixmlNodeList_addToNodeList (nodeList.c:106)
==29110== by 0x5D8B7E2: ixmlNode_getElementsByTagNameRecursive
(node.c:1438)
==29110== by 0x5D8E587: ixmlElement_getElementsByTagName
(element.c:491)
==29110== by 0x5B6C0F1: AdvertiseAndReply (ssdp_server.c:201)
==29110== by 0x5B7AB74: UpnpSendAdvertisement (upnpapi.c:1495)
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@519 119443c7-1b9e-41f8-b6fc-b9c35fce742c
...blah... with...
Submitted: Nick Leverton ( leveret ) - 2010-03-07 05:18
Full error:
/usr/bin/install: will not overwrite just-created
`/tmp/buildd/libupnp-1.6.6/debian/tmp/usr/share/doc/libupnp3-dev/examples/s
ample_util.c' with `common/sample_util.c'
This seems to be from Automake 1.11 which doesn't like having duplicate
files in a Makefile.am. Patch attached, kindly provided by Stefan Potyra
for Debian (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=543068)
This fix will be needed for both 1.6.x and 1.8.x branches.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@510 119443c7-1b9e-41f8-b6fc-b9c35fce742c
SF Patch Tracker [ 2836704 ] Search for nested serviceList (not
stopping at the first lis
Submitted By: zephyrus ( zephyrus00jp )
Internet Gateway Device description contains nested serviceList (rootdevice
-> servicelist, subdevice
and subdevice has the lower-level serviceList, etc..)
Unfrotunately, the sample code sample_util.c used by tv_device sample,
etc.
has a code that looks for only the first top-level serviceList.
This results in the failure to read all the services of an IGD xml
description.
Attached patch modifies this behavior and looks for the service by
visiting all the serviceList in xml document in turn.
With the modified patch (ad additional modification), I could
simulate an IGD device and created a modified control program for that.
Patch against 1.6.6
TIA.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@503 119443c7-1b9e-41f8-b6fc-b9c35fce742c
code list
Submitted By: Nick Leverton ( leveret )
ErrorMessage[] in upnptools.c has got a bit out of sync, the attached
patch (generated from grep 'define UPNP_E_') should bring it up to date.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@495 119443c7-1b9e-41f8-b6fc-b9c35fce742c
(void) argument list.
Submitted By: zephyrus ( zephyrus00jp )
In a publicly installed headers, a few functions are declared without any
arguments at all, a la "()".
When I used gcc's -Wimplict and -Wstrict-prototypes to check for the
mismatch of
function prototype declarations and their usage in my own program,
some headers from libupnp-1.6.6 produced warnings.
They are not strictly bugs, but pretty much annoying. This is 2009, and
almost all the important compilers
understand ISO-C.
So the offending functions are declared as "(void") to show that they have
no arguments at all.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@493 119443c7-1b9e-41f8-b6fc-b9c35fce742c
too-short renewal interval
Submitted By: Nick Leverton (leveret)
Auto-renewals send an invalid SID due to a missing UpnpString_get_String
call. They also send a renewal interval of 0 instead of copying it from
the original subscription.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@492 119443c7-1b9e-41f8-b6fc-b9c35fce742c
Seems like SF's tracker won't let me add a patch to someone else's issue ?!
This refers to https://sourceforge.net/support/tracker.php?aid=2724578
The calls to inet_ntoa are in getlocalhostname(), which is called from
UpnpInit when it is returning the bound IP address.
UpnpInit/getlocalhostname hasn't been updated to IPv6, I presume this is
deliberate so that it doesn't start returning IPv6 addresses and
overwriting the caller's IPv4-sized allocation.
The attached patch just updates getlocalhostname to use inet_ntop instead
of inet_ntoa, and also documents the fact that UpnpInit is IPv4 only whilst
UpnpInnit2 is both IPv4 and IPv6.
A fuller solution might be to change UpnpInit to use some variant on
UpnpGetIfInfo. UpnpInit could still be left as IPv4 only if desired -
perhaps UpnpGetIfInfo could take an option for the desired address family.
getlocalhostname and its own copy of the interface scanning code would then
be redundant. I don't have IPv6 capability here though so I'm reluctant to
change the IPv6 code, as I have no way to test it.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@491 119443c7-1b9e-41f8-b6fc-b9c35fce742c
add devices.
each time a device been added, UpnpInit() is called, on exit, UpnpFinish()
is called, but the memories allocated by ThreadPoolInit() may lost because
there's no code to call ThreadPoolShutdown() to release the memories. And
inet_ntoa() is not thread safe, so in my patch, I substitute inet_ntoa()
with inet_ntop().
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@489 119443c7-1b9e-41f8-b6fc-b9c35fce742c
object API
As per email to pupnp-devel, this is the patch to add the _strget_
accessors for string-like objects in the interface.
Will add a further patch shortly to udpate the sample programs.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@488 119443c7-1b9e-41f8-b6fc-b9c35fce742c
SOAPACTION and User-Agent headers.
There is something going wrong in soap_ctrlpt.c at line 931 (based on
version 1.6.6 release).
The http_Makemessage call looks as follows:
if (http_MakeMessage(
&request, 1, 1,
"Q" "sbc" "N" "s" "s" "Ucc" "sss",
SOAPMETHOD_POST, path.buf, path.length,
"HOST: ", host.buf, host.length,
content_length,
ContentTypeHeader,
"SOAPACTION:
\"urn:schemas-upnp-org:control-1-0#QueryStateVariable\"",
xml_start, var_name, xml_end ) != 0 ) {
return UPNP_E_OUTOF_MEMORY;
}
This will result in the SOAPACTION header to be immediately followed by the
User-Agent header, while a cr-lf should separate the two. I propose to fix
this by changing the second "s" to "sc" to force the addition of a cr-lf
after the SOAPACTION. This looks consistent to the other Makemessage calls.
git-svn-id: https://pupnp.svn.sourceforge.net/svnroot/pupnp/trunk@484 119443c7-1b9e-41f8-b6fc-b9c35fce742c
