FreeSWITCH

mod_xml_rpc, windows 2003 x86, debug/crash

Details

  • Type: Bug Bug
  • Status: Closed Closed
  • Priority: Minor Minor
  • Resolution: Fixed
  • Affects Version/s: 1.0.6
  • Fix Version/s: None
  • Component/s: mod_xml_rpc
  • Labels:
    None
  • Environment:
    Windows Server 2003 x86
  • Platform:
    Win32/VS
  • CPU Info:
    Hide
    OS Name Microsoft(R) Windows(R) Server 2003, Standard Edition
    Version 5.2.3790 Service Pack 2 Build 3790
    Other OS Description R2
    OS Manufacturer Microsoft Corporation
    System Name PBX
    System Manufacturer Dell Computer Corporation
    System Model PowerEdge 2850
    System Type X86-based PC
    Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz
    Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz
    Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz
    Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz
    BIOS Version/Date Dell Computer Corporation A02, 1/19/2005
    SMBIOS Version 2.3
    Windows Directory C:\WINDOWS
    System Directory C:\WINDOWS\system32
    Boot Device \Device\HarddiskVolume2
    Locale United States
    Hardware Abstraction Layer Version = "5.2.3790.3959 (srv03_sp2_rtm.070216-1710)"
    User Name Not Available
    Time Zone Central Daylight Time
    Total Physical Memory 4,095.08 MB
    Available Physical Memory 2.99 GB
    Total Virtual Memory 5.83 GB
    Available Virtual Memory 4.76 GB
    Page File Space 2.00 GB
    Page File C:\pagefile.sys
    Show
    OS Name Microsoft(R) Windows(R) Server 2003, Standard Edition Version 5.2.3790 Service Pack 2 Build 3790 Other OS Description R2 OS Manufacturer Microsoft Corporation System Name PBX System Manufacturer Dell Computer Corporation System Model PowerEdge 2850 System Type X86-based PC Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz Processor x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3192 Mhz BIOS Version/Date Dell Computer Corporation A02, 1/19/2005 SMBIOS Version 2.3 Windows Directory C:\WINDOWS System Directory C:\WINDOWS\system32 Boot Device \Device\HarddiskVolume2 Locale United States Hardware Abstraction Layer Version = "5.2.3790.3959 (srv03_sp2_rtm.070216-1710)" User Name Not Available Time Zone Central Daylight Time Total Physical Memory 4,095.08 MB Available Physical Memory 2.99 GB Total Virtual Memory 5.83 GB Available Virtual Memory 4.76 GB Page File Space 2.00 GB Page File C:\pagefile.sys
  • FreeSWITCH GIT Revision:
    GIT head 9/1/2010
  • Reproduced with GIT HEAD?:
    yes

Description

When making repeated calls to mod_xml_rpc to check fifo status a debug dialog popup on the freeswitch console. Pressing Debug causes a freeswitch crash, pressing cancel cause freeswitch to lockup but not crash, leaving the popup alone allows freeswitch to continue to operate for some amount of time. The following error shows up in then windows application event log.

Event Type: Error
Event Source: .NET Runtime 2.0 Error Reporting
Event Category: None
Event ID: 1000
Date: 9/29/2010
Time: 9:20:58 PM
User: N/A
Computer: PBX
Description:
Faulting application freeswitch.exe, version 0.0.0.0, stamp 4c86794d, faulting module mod_xml_rpc.dll, version 0.0.0.0, stamp 4c98df77, debug? 0, fault address 0x0000239e.

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
Data:
0000: 41 00 70 00 70 00 6c 00 A.p.p.l.
0008: 69 00 63 00 61 00 74 00 i.c.a.t.
0010: 69 00 6f 00 6e 00 20 00 i.o.n. .
0018: 46 00 61 00 69 00 6c 00 F.a.i.l.
0020: 75 00 72 00 65 00 20 00 u.r.e. .
0028: 20 00 66 00 72 00 65 00 .f.r.e.
0030: 65 00 73 00 77 00 69 00 e.s.w.i.
0038: 74 00 63 00 68 00 2e 00 t.c.h...
0040: 65 00 78 00 65 00 20 00 e.x.e. .
0048: 30 00 2e 00 30 00 2e 00 0...0...
0050: 30 00 2e 00 30 00 20 00 0...0. .
0058: 34 00 63 00 38 00 36 00 4.c.8.6.
0060: 37 00 39 00 34 00 64 00 7.9.4.d.
0068: 20 00 69 00 6e 00 20 00 .i.n. .
0070: 6d 00 6f 00 64 00 5f 00 m.o.d._.
0078: 78 00 6d 00 6c 00 5f 00 x.m.l._.
0080: 72 00 70 00 63 00 2e 00 r.p.c...
0088: 64 00 6c 00 6c 00 20 00 d.l.l. .
0090: 30 00 2e 00 30 00 2e 00 0...0...
0098: 30 00 2e 00 30 00 20 00 0...0. .
00a0: 34 00 63 00 39 00 38 00 4.c.9.8.
00a8: 64 00 66 00 37 00 37 00 d.f.7.7.
00b0: 20 00 66 00 44 00 65 00 .f.D.e.
00b8: 62 00 75 00 67 00 20 00 b.u.g. .
00c0: 30 00 20 00 61 00 74 00 0. .a.t.
00c8: 20 00 6f 00 66 00 66 00 .o.f.f.
00d0: 73 00 65 00 74 00 20 00 s.e.t. .
00d8: 30 00 30 00 30 00 30 00 0.0.0.0.
00e0: 32 00 33 00 39 00 65 00 2.3.9.e.
00e8: 0d 00 0a 00 ....
  1. File
    20111031-xmlrpc.diff
    31/Oct/11 10:28 AM
    2.11 MB
    garmt
  2. File
    Freeswitch.2010.sln
    02/Apr/11 10:38 PM
    295 kB
    Jeff Lenk
  3. File
    FS-2746_20120923.diff
    23/Sep/12 2:06 PM
    2.48 MB
    garmt
  4. Text File
    gitstatus.txt
    21/Apr/11 11:39 AM
    23 kB
    garmt
  5. File
    newrpc.diff
    12/Jan/11 9:48 PM
    311 kB
    Jeff Lenk
  6. File
    newrpc1.diff
    13/Jan/11 11:22 PM
    318 kB
    Jeff Lenk
  7. File
    testxmlrpc.cs
    13/Sep/12 3:02 PM
    5 kB
    garmt
  8. File
    testxmlrpc.cs
    31/Oct/11 10:28 AM
    4 kB
    garmt
  9. File
    testxmlrpc.cs
    01/Apr/11 4:47 PM
    4 kB
    garmt
  11. File
    xmlrpc.diff
    01/Apr/11 4:45 PM
    2.55 MB
    garmt
  12. File
    xmlrpc1.diff
    05/Apr/11 7:25 AM
    1.99 MB
    garmt
  13. File
    xmlrpc2.diff
    06/Apr/11 4:45 AM
    2.00 MB
    garmt
  14. GZip Archive
    xmlrpc4.diff.gz
    22/Apr/11 11:12 AM
    390 kB
    garmt
  15. GZip Archive
    xmlrpc-c.tar.gz
    21/Apr/11 12:20 PM
    804 kB
    garmt

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Anthony Minessale II added a comment -
if you reported it today, and you are using the version from 9/1
the the answer with "reproduced with git head" is no
you MUST update and reproduce with git head from the day you post the bug.
Show
Anthony Minessale II added a comment - if you reported it today, and you are using the version from 9/1 the the answer with "reproduced with git head" is no you MUST update and reproduce with git head from the day you post the bug.
Hide
Permalink
Jeff Lenk added a comment -
you will also have to provide a stacktrace from visual studio and provide more background as to what you are doing to cause this.
Show
Jeff Lenk added a comment - you will also have to provide a stacktrace from visual studio and provide more background as to what you are doing to cause this.
Hide
Permalink
Brian West added a comment -
Downgrading to minor pending more information.
Show
Brian West added a comment - Downgrading to minor pending more information.
Hide
Permalink
Dennis Young added a comment -
OK, here is more information.

When I said I used the 9/1 GIT head, that was production. I did test using the latest head; we just don't want to push that into production yet. Sorry if my answers seem contradictory

Today I had some time to put together a test program that will make 10/4/10 GIT head fail in our test environment.

I compiled 10/4 GIT head in VS 2008 x86 DEBUG. I attached DEBUG to the freeswitch.exe process and ran my test program. It failed after about 4 1/2 minutes with the following error:

"Unhandled exception at 0x0fe21874 (msvcr90d.dll) in FreeSwitch.exe: 0xC0000005: Access violation reading location 0xddddddd8."

The debug pointer is on line 1692 in dbgheap.c

Here is the VS stack trace:
> msvcr90d.dll!CheckBytes(unsigned char * pb=0xddddddd8, unsigned char bCheck='í', unsigned int nSize=3) Line 1692 + 0x7 bytes C++
  msvcr90d.dll!_free_dbg_nolock(void * pUserData=0xdddddddd, int nBlockUse=1) Line 1295 + 0x19 bytes C++
  msvcr90d.dll!_free_dbg(void * pUserData=0xdddddddd, int nBlockUse=1) Line 1258 + 0xd bytes C++
  msvcr90d.dll!free(void * pUserData=0xdddddddd) Line 49 + 0xb bytes C++
  mod_xml_rpc.dll!destroyChannel(void * const userHandle=0x056090d0) Line 891 + 0xf bytes C
  mod_xml_rpc.dll!connDone(_TConn * const connectionP=0x056090d0) Line 61 + 0xe bytes C
  mod_xml_rpc.dll!threadDone(void * const userHandle=0x056090d0) Line 73 + 0x9 bytes C
  mod_xml_rpc.dll!threadRun(void * const arg=0x05587b68) Line 50 + 0x11 bytes C
  msvcr90d.dll!_callthreadstartex() Line 348 + 0xf bytes C
  msvcr90d.dll!_threadstartex(void * ptd=0x02ec0c68) Line 331 C
  kernel32.dll!75911194()
  [Frames below may be incorrect and/or missing, no symbols loaded for kernel32.dll]
  ntdll.dll!7704b495()
  ntdll.dll!7704b468()
Show
Dennis Young added a comment - OK, here is more information. When I said I used the 9/1 GIT head, that was production. I did test using the latest head; we just don't want to push that into production yet. Sorry if my answers seem contradictory Today I had some time to put together a test program that will make 10/4/10 GIT head fail in our test environment. I compiled 10/4 GIT head in VS 2008 x86 DEBUG. I attached DEBUG to the freeswitch.exe process and ran my test program. It failed after about 4 1/2 minutes with the following error: "Unhandled exception at 0x0fe21874 (msvcr90d.dll) in FreeSwitch.exe: 0xC0000005: Access violation reading location 0xddddddd8." The debug pointer is on line 1692 in dbgheap.c Here is the VS stack trace: > msvcr90d.dll!CheckBytes(unsigned char * pb=0xddddddd8, unsigned char bCheck='í', unsigned int nSize=3) Line 1692 + 0x7 bytes C++   msvcr90d.dll!_free_dbg_nolock(void * pUserData=0xdddddddd, int nBlockUse=1) Line 1295 + 0x19 bytes C++   msvcr90d.dll!_free_dbg(void * pUserData=0xdddddddd, int nBlockUse=1) Line 1258 + 0xd bytes C++   msvcr90d.dll!free(void * pUserData=0xdddddddd) Line 49 + 0xb bytes C++   mod_xml_rpc.dll!destroyChannel(void * const userHandle=0x056090d0) Line 891 + 0xf bytes C   mod_xml_rpc.dll!connDone(_TConn * const connectionP=0x056090d0) Line 61 + 0xe bytes C   mod_xml_rpc.dll!threadDone(void * const userHandle=0x056090d0) Line 73 + 0x9 bytes C   mod_xml_rpc.dll!threadRun(void * const arg=0x05587b68) Line 50 + 0x11 bytes C   msvcr90d.dll!_callthreadstartex() Line 348 + 0xf bytes C   msvcr90d.dll!_threadstartex(void * ptd=0x02ec0c68) Line 331 C   kernel32.dll!75911194()   [Frames below may be incorrect and/or missing, no symbols loaded for kernel32.dll]   ntdll.dll!7704b495()   ntdll.dll!7704b468()
Hide
Permalink
Dennis Young added a comment -
Here is the C# test program that starts 5 threads that call http://localhost:8080/webapi/xml_locate?root every 2 seconds. I hope this helps.

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Web;
using System.Net;
using System.Reflection;
using System.Threading;
using System.Configuration;

namespace FreeSwitchBreakRPC
{
    class Program
    {
        private const int NumThreads = 5;
        private static ManualResetEvent[] resetEvents;
        private static DateTime start = DateTime.Now;

        static void Main(string[] args)
        {
            resetEvents = new ManualResetEvent[NumThreads];
            Thread.CurrentThread.Name = "Main";
            SetUseUnsafeHeaderParsing(true);
            for (int i = 0; i < NumThreads; i++)
            {
                resetEvents[i] = new ManualResetEvent(false);
                ThreadPool.QueueUserWorkItem(new WaitCallback(dowork), i);
            }
            Console.WriteLine("Waiting...");
            WaitHandle.WaitAll(resetEvents);
        }

        private static void dowork(object o)
        {
            int index = (int)o;
            Thread.CurrentThread.Name = index.ToString();
            System.Net.WebClient client = new System.Net.WebClient();
            client.Headers.Add("user-agent", "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; .NET CLR 1.0.3705;)");
            // create login credentials
            client.Credentials = new NetworkCredential("freeswitch","works");
            string ret = string.Empty;
            while ((ret = client.DownloadString(@"http://localhost:8080/webapi/xml_locate?root")) != null)
            {
                Console.WriteLine("Thread: {0}, time: {2}, xml size = {1}",Thread.CurrentThread.Name, ret.Length,DateTime.Now - start);
                Thread.Sleep(2000);
            }
            resetEvents[index].Set();
        }

        // no config file.
        public static bool SetUseUnsafeHeaderParsing(bool b)
        {
            Assembly a = Assembly.GetAssembly(typeof(System.Net.Configuration.SettingsSection));
            if (a == null) return false;

            Type t = a.GetType("System.Net.Configuration.SettingsSectionInternal");
            if (t == null) return false;

            object o = t.InvokeMember("Section",
                BindingFlags.Static | BindingFlags.GetProperty | BindingFlags.NonPublic, null, null, new object[] { });
            if (o == null) return false;

            FieldInfo f = t.GetField("useUnsafeHeaderParsing", BindingFlags.NonPublic | BindingFlags.Instance);
            if (f == null) return false;

            f.SetValue(o, b);
            return true;
        }
    }
}
Show
Dennis Young added a comment - Here is the C# test program that starts 5 threads that call http://localhost:8080/webapi/xml_locate?root every 2 seconds. I hope this helps. using System; using System.Collections.Generic; using System.Linq; using System.Text; using System.Web; using System.Net; using System.Reflection; using System.Threading; using System.Configuration; namespace FreeSwitchBreakRPC {     class Program     {         private const int NumThreads = 5;         private static ManualResetEvent[] resetEvents;         private static DateTime start = DateTime.Now;         static void Main(string[] args)         {             resetEvents = new ManualResetEvent[NumThreads];             Thread.CurrentThread.Name = "Main";             SetUseUnsafeHeaderParsing(true);             for (int i = 0; i < NumThreads; i++)             {                 resetEvents[i] = new ManualResetEvent(false);                 ThreadPool.QueueUserWorkItem(new WaitCallback(dowork), i);             }             Console.WriteLine("Waiting...");             WaitHandle.WaitAll(resetEvents);         }         private static void dowork(object o)         {             int index = (int)o;             Thread.CurrentThread.Name = index.ToString();             System.Net.WebClient client = new System.Net.WebClient();             client.Headers.Add("user-agent", "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; .NET CLR 1.0.3705;)");             // create login credentials             client.Credentials = new NetworkCredential("freeswitch","works");             string ret = string.Empty;             while ((ret = client.DownloadString(@" http://localhost:8080/webapi/xml_locate?root ")) != null)             {                 Console.WriteLine("Thread: {0}, time: {2}, xml size = {1}",Thread.CurrentThread.Name, ret.Length,DateTime.Now - start);                 Thread.Sleep(2000);             }             resetEvents[index].Set();         }         // no config file.         public static bool SetUseUnsafeHeaderParsing(bool b)         {             Assembly a = Assembly.GetAssembly(typeof(System.Net.Configuration.SettingsSection));             if (a == null) return false;             Type t = a.GetType("System.Net.Configuration.SettingsSectionInternal");             if (t == null) return false;             object o = t.InvokeMember("Section",                 BindingFlags.Static | BindingFlags.GetProperty | BindingFlags.NonPublic, null, null, new object[] { });             if (o == null) return false;             FieldInfo f = t.GetField("useUnsafeHeaderParsing", BindingFlags.NonPublic | BindingFlags.Instance);             if (f == null) return false;             f.SetValue(o, b);             return true;         }     } }
Hide
Permalink
Jeff Lenk added a comment -
Maybe someone else knows or has experience with this code under windows but it seems to me that a new version of xml-rpc might be needed see-
http://xmlrpc-c.sourceforge.net/windows.php

They do not advise anything before 1.13 (I think the current fs code is based on somewhere around 1.06.27) because alot of changes went into making that work under windows.
Show
Jeff Lenk added a comment - Maybe someone else knows or has experience with this code under windows but it seems to me that a new version of xml-rpc might be needed see- http://xmlrpc-c.sourceforge.net/windows.php They do not advise anything before 1.13 (I think the current fs code is based on somewhere around 1.06.27) because alot of changes went into making that work under windows.
Hide
Permalink
garmt added a comment -

with current git head of 5-jan-11 version.h says
#define XMLRPC_C_VERSION "Xmlrpc-c 1.14.99"
which is supposed to be stable version, although a version 1.16 is recommended.

I tried to reproduce the crash. After 20 minutes it did not crash on my win7 32bit release version, but I think (i'm no expert) freeswitch is leaking memory. I see the private bytes increasing (using sysinternals process explorer ).
It took 20 minutes or so to see CRIT errors being logged at the freeswitch console (probably memory related) and I stopped the test.

I tried to reproduce it using 10 threads, requesting data every 500 ms.
On my simple laptop I had a load of about 50% cpu (25 % for freeswitch).
I also had one softphone registered playing music on hold (9664).
Show
garmt added a comment - with current git head of 5-jan-11 version.h says #define XMLRPC_C_VERSION "Xmlrpc-c 1.14.99" which is supposed to be stable version, although a version 1.16 is recommended. I tried to reproduce the crash. After 20 minutes it did not crash on my win7 32bit release version, but I think (i'm no expert) freeswitch is leaking memory. I see the private bytes increasing (using sysinternals process explorer ). It took 20 minutes or so to see CRIT errors being logged at the freeswitch console (probably memory related) and I stopped the test. I tried to reproduce it using 10 threads, requesting data every 500 ms. On my simple laptop I had a load of about 50% cpu (25 % for freeswitch). I also had one softphone registered playing music on hold (9664).
Hide
Permalink
Jeff Lenk added a comment -
Ok thanks - I may have been wrong on my version estimation, I was not able to really figure out the version other than the file you mentioned.
upgrading the version of xmlrpc is going to be a large task especially if it only benefits windows. if there were needed changes for other platforms that would make it easier to swallow.

garmt, are you willing to look at the upgraded version? possibly submitting a preliminary patch.
Show
Jeff Lenk added a comment - Ok thanks - I may have been wrong on my version estimation, I was not able to really figure out the version other than the file you mentioned. upgrading the version of xmlrpc is going to be a large task especially if it only benefits windows. if there were needed changes for other platforms that would make it easier to swallow. garmt, are you willing to look at the upgraded version? possibly submitting a preliminary patch.
Hide
Permalink
garmt added a comment -
Jeff, your fix of Jan 10th (Commit:719416f66b1757d33364b6ffb37119e29596a5ae * mem leak) seems to solve the problem.
I tested today's git with Dennis' script and the memory leak is gone. There must have been a (much smaller) leak in linux as well. Contrary to the simpleVasprintf function used in windows, the linux "vasprintf" will not allocate a 4k buffer to contain the timestamp. So indeed it explains why stkn could not reproduce this with his linux.

I managed to build the xmlrpc-c projects (advanced 1.25.01 2011-01-06 and stable 1.16.32 2011-12-30) on windows. I was about to start integrating them with fs when I saw your fix.

So, I will not continue for the moment to upgrade to the latest and greatest xmlrpc-c (also keeping in mind anthm's warnings about the quality of the abyss server code and the changes that needed to be made to get things running and stable).

Thanks a lot, Jeff.

I think you can close this "issue"
Show
garmt added a comment - Jeff, your fix of Jan 10th (Commit:719416f66b1757d33364b6ffb37119e29596a5ae * mem leak) seems to solve the problem. I tested today's git with Dennis' script and the memory leak is gone. There must have been a (much smaller) leak in linux as well. Contrary to the simpleVasprintf function used in windows, the linux "vasprintf" will not allocate a 4k buffer to contain the timestamp. So indeed it explains why stkn could not reproduce this with his linux. I managed to build the xmlrpc-c projects (advanced 1.25.01 2011-01-06 and stable 1.16.32 2011-12-30) on windows. I was about to start integrating them with fs when I saw your fix. So, I will not continue for the moment to upgrade to the latest and greatest xmlrpc-c (also keeping in mind anthm's warnings about the quality of the abyss server code and the changes that needed to be made to get things running and stable). Thanks a lot, Jeff. I think you can close this "issue"
Hide
Permalink
Jeff Lenk added a comment -
garmt,

Please check commit 6f103acd796ebcd7c0001a8d58568a3d77d2b3e6 also.

It fixes the problem with the module not unloading when stuck reading from socket. This change only effect the windows build.

So you dont have any exceptions running that test program any more? Even when more that 10 threads with a 100msec delay?
Show
Jeff Lenk added a comment - garmt, Please check commit 6f103acd796ebcd7c0001a8d58568a3d77d2b3e6 also. It fixes the problem with the module not unloading when stuck reading from socket. This change only effect the windows build. So you dont have any exceptions running that test program any more? Even when more that 10 threads with a 100msec delay?
Hide
Permalink
garmt added a comment -
I tested a bit more, but have not experienced problems with unloading the module, so I can't tell if your update to socket_win.c makes a difference.

However, there seems to be a significantly different behaviour between FS versions: release/debug and also VS2008/VS2010. (I tested only 32bit versions)
Using VS2010, I can easily reproduce the crash of Dennis Young. The release version crashes quite quickly, the debug version takes a bit more time (with more than 20 threads and 100 ms delay still within 15 sec)

With VS2008 it takes more time, but FS will crash.

The memory leak you solved is another problem, probably not related to the crash Dennis Young reported. I initially used VS2008/win32/release versions, and just happened to have more problems with the memory leak, before the "Dennis" crash as reported occured.
Show
garmt added a comment - I tested a bit more, but have not experienced problems with unloading the module, so I can't tell if your update to socket_win.c makes a difference. However, there seems to be a significantly different behaviour between FS versions: release/debug and also VS2008/VS2010. (I tested only 32bit versions) Using VS2010, I can easily reproduce the crash of Dennis Young. The release version crashes quite quickly, the debug version takes a bit more time (with more than 20 threads and 100 ms delay still within 15 sec) With VS2008 it takes more time, but FS will crash. The memory leak you solved is another problem, probably not related to the crash Dennis Young reported. I initially used VS2008/win32/release versions, and just happened to have more problems with the memory leak, before the "Dennis" crash as reported occured.
Hide
Permalink
Jeff Lenk added a comment -
This patch fixes the crash problems under windows, this patch is 1.16.23 from the stable branch of xml-rpc.

Unfortunately this patch changes alot that I havent had time to verify that there are no regressions. It does seem to work fine under windows.

If someone could check this patch under linux that would be very helpful.
Show
Jeff Lenk added a comment - This patch fixes the crash problems under windows, this patch is 1.16.23 from the stable branch of xml-rpc. Unfortunately this patch changes alot that I havent had time to verify that there are no regressions. It does seem to work fine under windows. If someone could check this patch under linux that would be very helpful.
Hide
Permalink
garmt added a comment -
It's a lot more robust now, a small mem leak remains.

We should not forget to update "version.h"
There are a significant number of changes (mostly initialization of variables) that may or may not be important.
The unix specific code was significantly changed by FS people. So I don't think we should use this patch for unix as it is now.

I can and will update and test more next week.

Thanks
Show
garmt added a comment - It's a lot more robust now, a small mem leak remains. We should not forget to update "version.h" There are a significant number of changes (mostly initialization of variables) that may or may not be important. The unix specific code was significantly changed by FS people. So I don't think we should use this patch for unix as it is now. I can and will update and test more next week. Thanks
Hide
Permalink
Jeff Lenk added a comment -
This version of the patch fixes the leak newrpc1.diff

really only two changes search for jlenk
Show
Jeff Lenk added a comment - This version of the patch fixes the leak newrpc1.diff really only two changes search for jlenk
Hide
Permalink
garmt added a comment -
You spent some serious time debugging this. As far as my testing goes, you have resolved all the memory leaks.
Makes one wonder if anyone actually seriously uses abyss as a web server.

Maybe I can dust off my unix skills and do some updates and testing on linux so that we can get this into trunk.
Hope to have some time by the end of next week. Thanks a lot.
Show
garmt added a comment - You spent some serious time debugging this. As far as my testing goes, you have resolved all the memory leaks. Makes one wonder if anyone actually seriously uses abyss as a web server. Maybe I can dust off my unix skills and do some updates and testing on linux so that we can get this into trunk. Hope to have some time by the end of next week. Thanks a lot.
Hide
Permalink
garmt added a comment -
A long week (since mid januari) but I finally managed to make a version that compiles and works on Centos.
The version of xml_rpc I used is 1.26.0 (advanced) which is currently the same as 1.25.xx (stable), (stable was recently updated in the svn trunk of xmlrpc-c)
I found and hopefully resolved some more memory leaks (when unloading the module)
A fair amount of changes to both source code and the build environments and I may have made some mistakes.
Tested with a (slightly) modified version of Dennis' test program.

Note: I did not update the VS2008 and express version of the solution and project files (as I did not install 2008 on my dev machine yet)
Show
garmt added a comment - A long week (since mid januari) but I finally managed to make a version that compiles and works on Centos. The version of xml_rpc I used is 1.26.0 (advanced) which is currently the same as 1.25.xx (stable), (stable was recently updated in the svn trunk of xmlrpc-c) I found and hopefully resolved some more memory leaks (when unloading the module) A fair amount of changes to both source code and the build environments and I may have made some mistakes. Tested with a (slightly) modified version of Dennis' test program. Note: I did not update the VS2008 and express version of the solution and project files (as I did not install 2008 on my dev machine yet)
Hide
Permalink
Anthony Minessale II added a comment -
Did you compare all the bugs we fixed in it to see if they had also been fixed by them?
Especially the ones that were randomly closing fd numbers all over the machine.... =D
Show
Anthony Minessale II added a comment - Did you compare all the bugs we fixed in it to see if they had also been fixed by them? Especially the ones that were randomly closing fd numbers all over the machine.... =D
Hide
Permalink
Jeff Lenk added a comment -
I intentionally took a conservative approach to my patch by keeping it as small as possible which should make it easier to verify that there are no regressions.

While I appreciate that you put a lot of effort into the newer version it will be more difficult to merge because of the massive changes that occurred.

I would like to get newrpc1.diff committed first and then in the future we can upgrade to the newer version(1.26). I was unable to test these changes(newrpc1.diff) in linux so I was unable to verify whether these changes broke the build.
Show
Jeff Lenk added a comment - I intentionally took a conservative approach to my patch by keeping it as small as possible which should make it easier to verify that there are no regressions. While I appreciate that you put a lot of effort into the newer version it will be more difficult to merge because of the massive changes that occurred. I would like to get newrpc1.diff committed first and then in the future we can upgrade to the newer version(1.26). I was unable to test these changes(newrpc1.diff) in linux so I was unable to verify whether these changes broke the build.
Hide
Permalink
garmt added a comment -
It's a bit by accident I went all the way to the latest version of xmlrpc-c
I had prepared everything to update to the latest stable xmlrp-c version (1.16.xx) with all the patches from Jeff and some more of my own.
I have been running that in a rather small win32 based production enviroment, and it was very stable (no memory problems, running for a 3 months now, about 200 outbound calls a day). I indeed spent quite some time to upgrade the linux (centos) version, is was mostly a journey into understanding how the build systems work.

However, somewhat by incident I svn updated the stable version of xmlrpc-c in the freeswitch libs folder and after some confusion it turned out to be the same as the latest advanced version, since it was recently updated. Because I had done most of the digging to find out how things worked, I made the decision to continue, meanwhile making sure all changes from Jeff were incorporated.
I carefully windiffed alle source and header files. Overall, the changes are less dramatic then appears at first sight when looking at all the files that are involved. My diff should work on a freshly check-ed-out git system, both on windows and centos. It should be not so hard to test it. (probably a ./bootstrap.sh is required after applying the patch).

I'm not sure that newrpc1.diff is a less conservative approach... The abyss code is not brilliant, but it seems to do its job, as far as my testing goes (stress and some (limited) functionality), both on linux and windows.
I would recommend going "all the way", - it makes things easier for me -. Maybe Seven can do some more non-windows based testing before you commit this to the tree? I'm happy to closely monitor any issue that pops up with this code (at least in the next weeks).
As I do not have a comprehensive test suite for this, I cannot guarantee if everything works as previously advertized, but my limited testing seems to be fine.

anthm: I bases my changes on Jeff's diff. Would it help to attach a list of the files that changed rather than the whole diff?
 
Show
garmt added a comment - It's a bit by accident I went all the way to the latest version of xmlrpc-c I had prepared everything to update to the latest stable xmlrp-c version (1.16.xx) with all the patches from Jeff and some more of my own. I have been running that in a rather small win32 based production enviroment, and it was very stable (no memory problems, running for a 3 months now, about 200 outbound calls a day). I indeed spent quite some time to upgrade the linux (centos) version, is was mostly a journey into understanding how the build systems work. However, somewhat by incident I svn updated the stable version of xmlrpc-c in the freeswitch libs folder and after some confusion it turned out to be the same as the latest advanced version, since it was recently updated. Because I had done most of the digging to find out how things worked, I made the decision to continue, meanwhile making sure all changes from Jeff were incorporated. I carefully windiffed alle source and header files. Overall, the changes are less dramatic then appears at first sight when looking at all the files that are involved. My diff should work on a freshly check-ed-out git system, both on windows and centos. It should be not so hard to test it. (probably a ./bootstrap.sh is required after applying the patch). I'm not sure that newrpc1.diff is a less conservative approach... The abyss code is not brilliant, but it seems to do its job, as far as my testing goes (stress and some (limited) functionality), both on linux and windows. I would recommend going "all the way", - it makes things easier for me -. Maybe Seven can do some more non-windows based testing before you commit this to the tree? I'm happy to closely monitor any issue that pops up with this code (at least in the next weeks). As I do not have a comprehensive test suite for this, I cannot guarantee if everything works as previously advertized, but my limited testing seems to be fine. anthm: I bases my changes on Jeff's diff. Would it help to attach a list of the files that changed rather than the whole diff?  
Hide
Permalink
Anthony Minessale II added a comment -
I'm all for the idea of updating and making it more stable.

I just was curious if you looked on the revision history of patches we did over time and confirmed its still fixed. (there are only a handful and the ones regarding sockets or uninit mem are the serious ones)

We went from a stable release ofxmlrpc-c to the random svn checkout in 1.0.1 because the one we had in 1.0.0 didn't work at all in windows.
Since then I found some pretty horrible issues with uninit socket ints being passed to close which would randomly close files in the system such as sqlite connections or other important OS sockets not even in the FS process.

I can try running the patch on our conference server that uses it pretty heavily. http://conference.freeswitch.org
Show
Anthony Minessale II added a comment - I'm all for the idea of updating and making it more stable. I just was curious if you looked on the revision history of patches we did over time and confirmed its still fixed. (there are only a handful and the ones regarding sockets or uninit mem are the serious ones) We went from a stable release ofxmlrpc-c to the random svn checkout in 1.0.1 because the one we had in 1.0.0 didn't work at all in windows. Since then I found some pretty horrible issues with uninit socket ints being passed to close which would randomly close files in the system such as sqlite connections or other important OS sockets not even in the FS process. I can try running the patch on our conference server that uses it pretty heavily. http://conference.freeswitch.org
Hide
Permalink
Jeff Lenk added a comment -
I'm ok with this either way but the windows project files and the solution need some changes made because there are some which are not correct in garmts latest path.

I will look over those tonight.
Show
Jeff Lenk added a comment - I'm ok with this either way but the windows project files and the solution need some changes made because there are some which are not correct in garmts latest path. I will look over those tonight.
Hide
Permalink
Jeff Lenk added a comment -
Garmt would you take these files and merge them into your patch? The solution and the zip file of the Windows folder from libs\xmlrpc-c\Windows.

I have not had a chance to see how these changes effect 2008 yet.
Show
Jeff Lenk added a comment - Garmt would you take these files and merge them into your patch? The solution and the zip file of the Windows folder from libs\xmlrpc-c\Windows. I have not had a chance to see how these changes effect 2008 yet.
Hide
Permalink
garmt added a comment -
Ok, I checked all file histories and put back the relevant changes of earlier modifications. Not sure most of these were still needed, because there were some changes upstream, but most of them looked sane to me, so I put them back in (most notably the sane_close(), and some initializations).

I also included the changes by Jeff (don't know why things disappeared, must have removed too much when including the gennmtab project
I tested this with VS2010 win32 and x64 and centos (using valgrind and the test program) and did not see any complaints about uninitialized code / conditional jumps / memory leaks (at least with respect to mod_xml_rpc and it's xmlrpc-c libraries)

I made an initial attempt at updating the VS2008 sln file (so that is work in progress)

Not sure what to do exactly with the header files that are generated/copied. I commented out some header files from libs/.svnignore, and removed them from git. At some point they may have to be removed by a make clean.

Did any of you use/test the previous diff?
Show
garmt added a comment - Ok, I checked all file histories and put back the relevant changes of earlier modifications. Not sure most of these were still needed, because there were some changes upstream, but most of them looked sane to me, so I put them back in (most notably the sane_close(), and some initializations). I also included the changes by Jeff (don't know why things disappeared, must have removed too much when including the gennmtab project I tested this with VS2010 win32 and x64 and centos (using valgrind and the test program) and did not see any complaints about uninitialized code / conditional jumps / memory leaks (at least with respect to mod_xml_rpc and it's xmlrpc-c libraries) I made an initial attempt at updating the VS2008 sln file (so that is work in progress) Not sure what to do exactly with the header files that are generated/copied. I commented out some header files from libs/.svnignore, and removed them from git. At some point they may have to be removed by a make clean. Did any of you use/test the previous diff?
Hide
Permalink
Jeff Lenk added a comment -
I tested the build on windows vs2010 but have not had a chance to do much else. I will finish the vs2008 changes when this gets merged.
Show
Jeff Lenk added a comment - I tested the build on windows vs2010 but have not had a chance to do much else. I will finish the vs2008 changes when this gets merged.
Hide
Permalink
garmt added a comment -
includes VS2008 sln and proj files

also contains a few modifications to .gitignore files

I there a way to transfer the changes of my local repository (the files I added and removed to/from git) to my centos freeswitch installation?
Show
garmt added a comment - includes VS2008 sln and proj files also contains a few modifications to .gitignore files I there a way to transfer the changes of my local repository (the files I added and removed to/from git) to my centos freeswitch installation?
Hide
Permalink
garmt added a comment -
almost clean tar file
which
Show
garmt added a comment - almost clean tar file which
Hide
Permalink
garmt added a comment -
the common.mk was missing from xmlrpc3.diff file leading to various problems on centos.
I just tested the attached xmlrpc4.diff file on centos x64 and VS2010

on centos:
from freeswitch root directory:

cd libs/xmlrpc-c
rm -r *
git checkout -f .
cd ..\..
patch -p1 <xmlrpc4.diff
make

I ran this version with valgrind and the small stress test from earlier comments to this issue, and saw no memory leaks (or other strange things) related to xml_rpc
Show
garmt added a comment - the common.mk was missing from xmlrpc3.diff file leading to various problems on centos. I just tested the attached xmlrpc4.diff file on centos x64 and VS2010 on centos: from freeswitch root directory: cd libs/xmlrpc-c rm -r * git checkout -f . cd ..\.. patch -p1 <xmlrpc4.diff make I ran this version with valgrind and the small stress test from earlier comments to this issue, and saw no memory leaks (or other strange things) related to xml_rpc
Hide
Permalink
garmt added a comment -
hehe,
common.mk is missing from xmlrpc2.diff
Show
garmt added a comment - hehe, common.mk is missing from xmlrpc2.diff
Hide
Permalink
Jeff Lenk added a comment -
Is the step to rm -r * needed because that will cause endless problems for upgrades with this? Or is it only needed for getting a clean patch imported?
Show
Jeff Lenk added a comment - Is the step to rm -r * needed because that will cause endless problems for upgrades with this? Or is it only needed for getting a clean patch imported?
Hide
Permalink
Anthony Minessale II added a comment -
I also had to:

cd libs/xmlrpc-c
CFLAGS="-fPIC" make
cd ../..
make mod_xml_rpc-install

so we have some build sorting to do but it's live at http://conference.freeswitch.org
Show
Anthony Minessale II added a comment - I also had to: cd libs/xmlrpc-c CFLAGS="-fPIC" make cd ../.. make mod_xml_rpc-install so we have some build sorting to do but it's live at http://conference.freeswitch.org
Hide
Permalink
garmt added a comment -
jlenk: the rm -r is there for now to be sure things work.
Although while doing this stuff I learned a lot about the GNU buildtools, I'm not an expert (hopefully never will be).

From what I understand and experience on my environment, a make clean in mod_xm_rpc should work.
There is an interaction between files that are generared doing the configure phase, or during make or files that should or should not be under version control by git. I'm not sure what the exact preference is. I thought I fixed everything, but apparently my fs on centos and anthm/stkn's versions are different.

anthm: cool that it is live, hopefully it will stay that way ;-)

about the build process:
As far as I know, there is no need to "make" anything in the xmlrpc-c directory.
mod_xml_rpc does not depend on any library from xmlrpc-c (I might be wrong here).
Maybe it is because I tested the build in a devel-bootstrapped environment?
Or maybe I missed a dependency because a xmlrpc-c library is installed already



Show
garmt added a comment - jlenk: the rm -r is there for now to be sure things work. Although while doing this stuff I learned a lot about the GNU buildtools, I'm not an expert (hopefully never will be). From what I understand and experience on my environment, a make clean in mod_xm_rpc should work. There is an interaction between files that are generared doing the configure phase, or during make or files that should or should not be under version control by git. I'm not sure what the exact preference is. I thought I fixed everything, but apparently my fs on centos and anthm/stkn's versions are different. anthm: cool that it is live, hopefully it will stay that way ;-) about the build process: As far as I know, there is no need to "make" anything in the xmlrpc-c directory. mod_xml_rpc does not depend on any library from xmlrpc-c (I might be wrong here). Maybe it is because I tested the build in a devel-bootstrapped environment? Or maybe I missed a dependency because a xmlrpc-c library is installed already
Hide
Permalink
garmt added a comment -
anthm: is it still live? Should I sort out the build issues?
Show
garmt added a comment - anthm: is it still live? Should I sort out the build issues?
Hide
Permalink
Anthony Minessale II added a comment -
its working fine, I have MikeJ tasked to work out the build issues.
Show
Anthony Minessale II added a comment - its working fine, I have MikeJ tasked to work out the build issues.
Hide
Permalink
garmt added a comment -
Good to hear.
However, I recently experienced problems on my windows based platform
(especially when xmlrpc needs to send data larger than about 8k bytes and particularly when using ie9 as a client).
I prepared some patches for this, but I need to do more testing and also check if the same problems occur on linux. I will post an updated patch in two or three days.
Show
garmt added a comment - Good to hear. However, I recently experienced problems on my windows based platform (especially when xmlrpc needs to send data larger than about 8k bytes and particularly when using ie9 as a client). I prepared some patches for this, but I need to do more testing and also check if the same problems occur on linux. I will post an updated patch in two or three days.
Hide
Permalink
Anthony Minessale II added a comment -
it's day 5, any luck?
Show
Anthony Minessale II added a comment - it's day 5, any luck?
Hide
Permalink
garmt added a comment -
Sorry, should have let you know sooner that I indeed had no luck.

I'm still working on it, patches will arrive, but I cannot say when, probably a few days from now (ah, I'm again making a stupid promise...)
It's a lot of work to build/configure and test on both windows and centos...

I made quite a few bug fixes and code improvements. It took a while to realize (when testing) that ie9 was misbehaving too.
And it also took time to realize that I cannot bypass this. ie9 has problems rendering plain/text files with lots of CRLF.

It's not a blocking issue, - probably noone else will experience it, I had bad luck with my test scenario - however I guess we should look for an alternative to the xmlrpc-c abyss server, as it's not a brilliant piece of code and hard to maintain (It may be hard to sell all our changes upstream to the xmlrpc-c project, anyway)
Show
garmt added a comment - Sorry, should have let you know sooner that I indeed had no luck. I'm still working on it, patches will arrive, but I cannot say when, probably a few days from now (ah, I'm again making a stupid promise...) It's a lot of work to build/configure and test on both windows and centos... I made quite a few bug fixes and code improvements. It took a while to realize (when testing) that ie9 was misbehaving too. And it also took time to realize that I cannot bypass this. ie9 has problems rendering plain/text files with lots of CRLF. It's not a blocking issue, - probably noone else will experience it, I had bad luck with my test scenario - however I guess we should look for an alternative to the xmlrpc-c abyss server, as it's not a brilliant piece of code and hard to maintain (It may be hard to sell all our changes upstream to the xmlrpc-c project, anyway)
Hide
Permalink
Anthony Minessale II added a comment -
So should we try to merge this now or wait for more patches?
Show
Anthony Minessale II added a comment - So should we try to merge this now or wait for more patches?
Hide
Permalink
Anthony Minessale II added a comment -
build master
Show
Anthony Minessale II added a comment - build master
Hide
Permalink
garmt added a comment -
I made a lot (and indeed also significant) changes and I'm still busy testing.
I gave a copy to ledr today to help me test.

Still need to fix a linux build issue; when done I will attach the latest and greatest here.

Some of the changes are not really in mod_xml_rpc or in the xmlrpc-c libraries, e.g. several typo's in comments (relevant for doxygen) and issues with mod_commands (not really consistent ERR reporting, proper http and html generation).
) and some fixes to windows compiler warnings. I should probably open separate jira's for those, but it takes time to take them apart, and some do depend on changes posted in this issue.

To conclude, I think it is better to wait for more patches.



Show
garmt added a comment - I made a lot (and indeed also significant) changes and I'm still busy testing. I gave a copy to ledr today to help me test. Still need to fix a linux build issue; when done I will attach the latest and greatest here. Some of the changes are not really in mod_xml_rpc or in the xmlrpc-c libraries, e.g. several typo's in comments (relevant for doxygen) and issues with mod_commands (not really consistent ERR reporting, proper http and html generation). ) and some fixes to windows compiler warnings. I should probably open separate jira's for those, but it takes time to take them apart, and some do depend on changes posted in this issue. To conclude, I think it is better to wait for more patches.
Hide
Permalink
Jeff Lenk added a comment -
Garmt,

Are you ready to merge this? There have been many changes here which would be of benefit. The patch for xmlprc should be committed and if you have other fixes you should open another Jira for those.

Lets try to get this committed.
Show
Jeff Lenk added a comment - Garmt, Are you ready to merge this? There have been many changes here which would be of benefit. The patch for xmlprc should be committed and if you have other fixes you should open another Jira for those. Lets try to get this committed.
Hide
Permalink
garmt added a comment -
find attached my latest patch file
I built and tested this with a fresh git checkout and pull on both windows vs2010 and centos.

Unfortunately, due to a HD crash (actually at Cluecon) I lost my VS2008 installation and did not restore it, so it may or may not build on VS2008 without changes. While at it, I unified the +OK / -ERR texts in mod_commands.
Ultimately I guess we'd be better off having a result code and some extended result code text, but that will break things.

FreeSWITCH Version 1.0.head (git-50328a6 2011-10-30 09-54-17 +0000)

I realize that there are a lot of changes, I did a lot of (functional, and some stress) testing, but I do not know all use cases.

On centos: a "make clean" in mod_xml_rpc may help.

I'm attaching a modified test c# that may help in (stress) testing.

my comments in from mod_xml_rpc.c -- XML RPC:

/*
 * embedded webserver for FS
 * exposes fs api to web (and ajax, javascript, ...)
 * supports GET/POST requests (as defined below)
 * and similar XMLRPC format /RPC2 freeswitch.api and management.api
 *
 * usage:
 * (1) http:/host:port/[txt|web|xml]api/fsapicommand[?arg[ arg]*][ &key=value[+&key=value]*]
 * e.g. http:/host:port/api/show?calls &refresh=5+&weather=nice
 * (2) http:/host:port/filepath - serves files from conf/htdocs
 *
 * NB:
 * ad (1) - key/value pairs are propagated as event headers
 * - if &key=value is "&refresh=xx" - special feature: automatic refresh after xx sec (triggered by browser)
 * note that refresh works only
 * IF response content-type: text/html (i.e. "webapi" or "api")
 * AND fs api command created an event header HTTP-REFRESH before the first write_stream
 * NOTE if "api", fs api command has to overwrite content-type to be text/html i.s.o. text/plain (default)
 *
 * - if api format is "api" mod_xml_rpc will automatically assume plain/text (so webunaware fs commands are rendered apropriately)
 * - xmlapi doesn't seem to be used, however if a fs api command renders xml, you can set the format type to xml
 * txtapi-text/plain or webapi-text/html surround xml with <pre> xml <pre/>
 * - typically fs api command arguments are encoded with UrlPathEncode (spaces -> %20), and k/v pairs are urlencoded (space -> +)
 *
 * ad (2) ms ie may show extra empty lines when serving large txt files as ie has problems rendering content-type "plain/text"
 
 
Show
garmt added a comment - find attached my latest patch file I built and tested this with a fresh git checkout and pull on both windows vs2010 and centos. Unfortunately, due to a HD crash (actually at Cluecon) I lost my VS2008 installation and did not restore it, so it may or may not build on VS2008 without changes. While at it, I unified the +OK / -ERR texts in mod_commands. Ultimately I guess we'd be better off having a result code and some extended result code text, but that will break things. FreeSWITCH Version 1.0.head (git-50328a6 2011-10-30 09-54-17 +0000) I realize that there are a lot of changes, I did a lot of (functional, and some stress) testing, but I do not know all use cases. On centos: a "make clean" in mod_xml_rpc may help. I'm attaching a modified test c# that may help in (stress) testing. my comments in from mod_xml_rpc.c -- XML RPC: /*  * embedded webserver for FS  * exposes fs api to web (and ajax, javascript, ...)  * supports GET/POST requests (as defined below)  * and similar XMLRPC format /RPC2 freeswitch.api and management.api  *  * usage:  * (1) http:/host:port/[txt|web|xml]api/fsapicommand[?arg[ arg]*][ &key=value[+&key=value]*]  * e.g. http:/host:port/api/show?calls &refresh=5+&weather=nice  * (2) http:/host:port/filepath - serves files from conf/htdocs  *  * NB:  * ad (1) - key/value pairs are propagated as event headers  * - if &key=value is "&refresh=xx" - special feature: automatic refresh after xx sec (triggered by browser)  * note that refresh works only  * IF response content-type: text/html (i.e. "webapi" or "api")  * AND fs api command created an event header HTTP-REFRESH before the first write_stream  * NOTE if "api", fs api command has to overwrite content-type to be text/html i.s.o. text/plain (default)  *  * - if api format is "api" mod_xml_rpc will automatically assume plain/text (so webunaware fs commands are rendered apropriately)  * - xmlapi doesn't seem to be used, however if a fs api command renders xml, you can set the format type to xml  * txtapi-text/plain or webapi-text/html surround xml with <pre> xml <pre/>  * - typically fs api command arguments are encoded with UrlPathEncode (spaces -> %20), and k/v pairs are urlencoded (space -> +)  *  * ad (2) ms ie may show extra empty lines when serving large txt files as ie has problems rendering content-type "plain/text"    
Hide
Permalink
Anthony Minessale II added a comment -
sounds cool, can you help round up some testers?
Show
Anthony Minessale II added a comment - sounds cool, can you help round up some testers?
Hide
Permalink
Marc Olivier Chouinard added a comment -
garmt, you got an updated patch... There is lot of stuff in there to review... Upload your latest and I'll try to review the code (linux only)
Show
Marc Olivier Chouinard added a comment - garmt, you got an updated patch... There is lot of stuff in there to review... Upload your latest and I'll try to review the code (linux only)
Hide
Permalink
Ken Rice added a comment -
Any update on this issue?
Show
Ken Rice added a comment - Any update on this issue?
Hide
Permalink
Mike Jerris added a comment -
jeff, can you take a look at this?
Show
Mike Jerris added a comment - jeff, can you take a look at this?
Hide
Permalink
Jeff Lenk added a comment -
yep sure will. Garmt do you still have a current patch (maybe is still current?) I would like to review your latest again. Thanks
Show
Jeff Lenk added a comment - yep sure will. Garmt do you still have a current patch (maybe is still current?) I would like to review your latest again. Thanks
Hide
Permalink
garmt added a comment -
I started working on a new patch today, not sure when I will finish, hopefully in a few days
Show
garmt added a comment - I started working on a new patch today, not sure when I will finish, hopefully in a few days
Hide
Permalink
garmt added a comment -
attached a new patch, should build with vs2010, vs2008 and on centos6.3
To test one can use the updated testxmlrpc.cs, this one uses more sockets.
Show
garmt added a comment - attached a new patch, should build with vs2010, vs2008 and on centos6.3 To test one can use the updated testxmlrpc.cs, this one uses more sockets.
Hide
Permalink
Jeff Lenk added a comment -
garmt that's not a valid diff file -> FS-2746_20120913.diff

Please check it.

Thanks
Show
Jeff Lenk added a comment - garmt that's not a valid diff file -> FS-2746 _20120913.diff Please check it. Thanks
Hide
Permalink
garmt added a comment -
Something must have gone wrong uploading the attachment, not sure why, here's another one: FS-2746_20120923.diff
Show
garmt added a comment - Something must have gone wrong uploading the attachment, not sure why, here's another one: FS-2746 _20120923.diff
Hide
Permalink
Jeff Lenk added a comment -
was that patch generated against git head? It seems to have lots of unrelated stuff in it.
Show
Jeff Lenk added a comment - was that patch generated against git head? It seems to have lots of unrelated stuff in it.
Hide
Permalink
garmt added a comment -
You are right. Git and me, we don't always agree...
Somehow all the changes since the previous time I "git pull"-ed were incorporated.
I pulled again, and now it seems to be better.

I must say that there are a lot of changes, some of them do not immediately seem connected to xml_rpc, like streamlining result codes and lower/upper case in mod_commands, and cleaning up gitignore.
Show
garmt added a comment - You are right. Git and me, we don't always agree... Somehow all the changes since the previous time I "git pull"-ed were incorporated. I pulled again, and now it seems to be better. I must say that there are a lot of changes, some of them do not immediately seem connected to xml_rpc, like streamlining result codes and lower/upper case in mod_commands, and cleaning up gitignore.
Hide
Permalink
Jeff Lenk added a comment -
This looks pretty good so far. If anybody else is watching and could apply the patch and post the results that would be great.
Show
Jeff Lenk added a comment - This looks pretty good so far. If anybody else is watching and could apply the patch and post the results that would be great.
Hide
Permalink
Jeff Lenk added a comment -
garmt can you reapply that to a fresh git head and let me know what steps you did. It seems that some files are missing for autoconf.
Show
Jeff Lenk added a comment - garmt can you reapply that to a fresh git head and let me know what steps you did. It seems that some files are missing for autoconf.
Hide
Permalink
Anthony Minessale II added a comment -
I pulled this into the FS main conf server and it has run fine for a few weeks.
This is good to go pending it builds right everywhere.
Show
Anthony Minessale II added a comment - I pulled this into the FS main conf server and it has run fine for a few weeks. This is good to go pending it builds right everywhere.
Hide
Permalink
Jeff Lenk added a comment -
tony what steps did you have to do to make it build on the conf server?
Show
Jeff Lenk added a comment - tony what steps did you have to do to make it build on the conf server?
Hide
Permalink
Anthony Minessale II added a comment -
I think I just applied the patch.
My tree was already configured with the previous configure though.

I just applied the patch after it was already built once.
Show
Anthony Minessale II added a comment - I think I just applied the patch. My tree was already configured with the previous configure though. I just applied the patch after it was already built once.
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/master
Commit: 5e9537d http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=5e9537d
Updated By: jeff@jefflenk.com

Comment:
FS-2746 updated in tree patch to head


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/master Commit: 5e9537d http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=5e9537d Updated By: jeff@jefflenk.com Comment: FS-2746 updated in tree patch to head FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Anthony Minessale II added a comment -
tried from totally clean tree and it dies on bootstrap

configure.in:27: installing `build/config/config.guess'
configure.in:27: installing `build/config/config.sub'
configure.in:1056: required file `libs/xmlrpc-c/srcdir.mk.in' not found

Show
Anthony Minessale II added a comment - tried from totally clean tree and it dies on bootstrap configure.in:27: installing `build/config/config.guess' configure.in:27: installing `build/config/config.sub' configure.in:1056: required file `libs/xmlrpc-c/srcdir.mk.in' not found
Hide
Permalink
Anthony Minessale II added a comment -
I think you need to git clean -fdx && git reset --hard then apply the patch to realize the problems
Show
Anthony Minessale II added a comment - I think you need to git clean -fdx && git reset --hard then apply the patch to realize the problems
Hide
Permalink
Jeff Lenk added a comment -
I think I messed the patch up - let me rebuild it again
Show
Jeff Lenk added a comment - I think I messed the patch up - let me rebuild it again
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/master
Commit: 37ecad9 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=37ecad9
Updated By: jeff@jefflenk.com

Comment:
FS-2746 updated in tree patch to head 2nd try


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/master Commit: 37ecad9 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=37ecad9 Updated By: jeff@jefflenk.com Comment: FS-2746 updated in tree patch to head 2nd try FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Jeff Lenk added a comment -
the last patch applies cleanly to a CentOS 6.2 system and builds from a fresh clone. Getting pretty close here. What else needs to be done before commit?
Show
Jeff Lenk added a comment - the last patch applies cleanly to a CentOS 6.2 system and builds from a fresh clone. Getting pretty close here. What else needs to be done before commit?
Hide
Permalink
Anthony Minessale II added a comment -
i guess we could verify debian or ubuntu but really if it works its ready to go.
Show
Anthony Minessale II added a comment - i guess we could verify debian or ubuntu but really if it works its ready to go.
Hide
Permalink
Jeff Lenk added a comment -
Verified on Ubunu 12.04 x64 commit imminent
Show
Jeff Lenk added a comment - Verified on Ubunu 12.04 x64 commit imminent
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/master
Commit: 6b6c83a http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=6b6c83a
Closed By: jeff@jefflenk.com

Comment:
FS-2746 --resolve large xmlrpc update thanks garmt


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/master Commit: 6b6c83a http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=6b6c83a Closed By: jeff@jefflenk.com Comment: FS-2746 --resolve large xmlrpc update thanks garmt FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/master
Commit: e3c3f02 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=e3c3f02
Updated By: jeff@jefflenk.com

Comment:
FS-2746 vs express corrections


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/master Commit: e3c3f02 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=e3c3f02 Updated By: jeff@jefflenk.com Comment: FS-2746 vs express corrections FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/video-media-bug
Commit: 5e9537d http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=5e9537d
Updated By: dujinfang@gmail.com

Comment:
FS-2746 updated in tree patch to head


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/video-media-bug Commit: 5e9537d http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=5e9537d Updated By: dujinfang@gmail.com Comment: FS-2746 updated in tree patch to head FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/video-media-bug
Commit: 37ecad9 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=37ecad9
Updated By: dujinfang@gmail.com

Comment:
FS-2746 updated in tree patch to head 2nd try


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/video-media-bug Commit: 37ecad9 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=37ecad9 Updated By: dujinfang@gmail.com Comment: FS-2746 updated in tree patch to head 2nd try FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/video-media-bug
Commit: 6b6c83a http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=6b6c83a
Closed By: dujinfang@gmail.com

Comment:
FS-2746 --resolve large xmlrpc update thanks garmt


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/video-media-bug Commit: 6b6c83a http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=6b6c83a Closed By: dujinfang@gmail.com Comment: FS-2746 --resolve large xmlrpc update thanks garmt FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/video-media-bug
Commit: e3c3f02 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=e3c3f02
Updated By: dujinfang@gmail.com

Comment:
FS-2746 vs express corrections


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/video-media-bug Commit: e3c3f02 http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=e3c3f02 Updated By: dujinfang@gmail.com Comment: FS-2746 vs express corrections FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Git added a comment -
Repository: freeswitch
Branch: refs/heads/master
Commit: fa9a0ed http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=fa9a0ed
Updated By: anthm@freeswitch.org

Comment:
FS-2746 found this assert in a BT on the conf box


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Git added a comment - Repository: freeswitch Branch: refs/heads/master Commit: fa9a0ed http://fisheye.freeswitch.org/changelog/freeswitch.git/?cs=fa9a0ed Updated By: anthm@freeswitch.org Comment: FS-2746 found this assert in a BT on the conf box FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com
Hide
Permalink
Auto Admin added a comment -
Due to a long period of inactivity (13 or more days), this issue is due to be automatically close within 24 hours.
If this issue is not actually resolved, please reopen it and add appropriate comments to help developers fix the issue.

Thanks,
  Jira Admin
Show
Auto Admin added a comment - Due to a long period of inactivity (13 or more days), this issue is due to be automatically close within 24 hours. If this issue is not actually resolved, please reopen it and add appropriate comments to help developers fix the issue. Thanks,   Jira Admin
Hide
Permalink
Auto Admin added a comment -
Since there has been no change to the status of this issue, it is being closed for inactivity

Thanks,
  Jira Admin


FreeSWITCH Support Contracts and Consulting Services available!

Contact us:
Email: consulting@freeswitch.org
Web: http://www.freeswitch.org
Phone: +1-918-420-9266
Tollfree: +1-877-742-2583
Fax: +1-918-420-9267
iNum: +883 5100 1420 9266


Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony!
http://www.cluecon.com

Show
Auto Admin added a comment - Since there has been no change to the status of this issue, it is being closed for inactivity Thanks,   Jira Admin FreeSWITCH Support Contracts and Consulting Services available! Contact us: Email: consulting@freeswitch.org Web: http://www.freeswitch.org Phone: +1-918-420-9266 Tollfree: +1-877-742-2583 Fax: +1-918-420-9267 iNum: +883 5100 1420 9266 Come To ClueCon in August to learn more about FreeSWITCH and Internet Telephony! http://www.cluecon.com

People

  • Assignee:
    Jeff Lenk
    Reporter:
    Dennis Young
Vote (0)
Watch (6)

Dates

  • Created:
    Updated:
    Resolved: