Tickets :: [#14550] Decoding certain TNEF files results in memory exhaustion

6.0.0-beta1

7/29/25

Summary	Decoding certain TNEF files results in memory exhaustion
Queue	Horde Framework Packages
Type	Bug
State	Not A Bug
Priority	1. Low
Owners	mrubinsk (at) horde (dot) org
Requester	michael.klein (at) hypaerion (dot) net
Created	12/23/2016 (3140 days ago)
Due
Updated	02/21/2017 (3080 days ago)
Assigned	02/20/2017 (3081 days ago)
Resolved	02/21/2017 (3080 days ago)
Github Issue Link
Github Pull Request
Milestone
Patch	No

02/21/2017 12:19:11 AM	Michael Rubinsky	Summary ⇒ Decoding certain TNEF files results in memory exhaustion

02/21/2017 12:18:32 AM	Michael Rubinsky	Comment #18 State ⇒ Not A Bug Priority ⇒ 1. Low	Reply to this comment
Well, I've narrowed it down to some bug in the "built in" implementation of some of the algorithms. Trying to fix this in the PEAR Math_BigInteger library is becoming a huge undertaking. For now, try installing either the bcmath or the gmp php extension. I've tested both locally and either one will prevent this issue. If I manually update to version 2.x of BigInteger (directly from the phpseclib project, and not from PEAR), then the problem is also solved. So, going to mark this as not a bug, since this is an issue with the implementation in the version of Math_BigInteger currently available via PEAR.

02/20/2017 11:37:29 PM	Michael Rubinsky	Comment #17 State ⇒ Feedback	Reply to this comment
Well, I've narrowed it down to some bug in the "built in" implementation of some of the algorithms. Trying to fix this in the PEAR Math_BigInteger library is becoming a huge undertaking. For now, try installing either the bcmath or the gmp php extension. I've tested both locally and either one will prevent this issue.

02/20/2017 10:21:36 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #16	Reply to this comment
I have just recieved another mail causing this behaviour: The file you sent me triggers an out of memory when attempting to open it even in IMP. ... though oddly, only on one of my machines. I have the same in IMP aswell as active sync.

02/20/2017 09:45:27 PM	Michael Rubinsky	Comment #15	Reply to this comment
I have just recieved another mail causing this behaviour: The file you sent me triggers an out of memory when attempting to open it even in IMP. ... though oddly, only on one of my machines.

02/20/2017 09:44:08 PM	Michael Rubinsky	Comment #14 State ⇒ Assigned	Reply to this comment
I have just recieved another mail causing this behaviour: The file you sent me triggers an out of memory when attempting to open it even in IMP.

02/20/2017 01:10:03 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #13	Reply to this comment
I have just recieved another mail causing this behaviour: A fatal error has occurred Allowed memory size of 268435456 bytes exhausted (tried to allocate 301989888 bytes) 1. Horde_ErrorHandler::catchFatalError() Details The full error message is logged in Horde's log file, and is shown below only to administrators. Non-administrative users will not see error details. ErrorException Object ( [message:protected] => Allowed memory size of 268435456 bytes exhausted (tried to allocate 301989888 bytes) [string:Exception:private] => [code:protected] => 0 [file:protected] => /usr/share/php/Math/BigInteger.php [line:protected] => 1090 [trace:Exception:private] => Array ( [0] => Array ( [function] => catchFatalError [class] => Horde_ErrorHandler [type] => :: [args] => Array ( ) ) ) [previous:Exception:private] => [severity:protected] => 1 [logged] => 1 ) [Show Quoted Text - 40 lines][Hide Quoted Text] I think i nailed it down... Looks like a whole bunch of mails which ran throu a yahoo group some very long time ago were the rootcause.. i send you one example via mail. The sample email is sync'd with no issues to all of my test clients - including Outlook. No display issues and no out of memory issues. Additionaly, the email you sent me did not contain any TNEF or DAT attachments, which is the ONLY way the code where the memory exhaustion occurred would have been called. I have removed the mail from the trash.. Still the error occurres. I have pulled a new report under the same address as before. As this mail was related to a calendar entry, do you think it might be related to the entry in the calendar? Yes, meeting requests are sometimes sent as TNEF or DAT file attachments from Outlook. I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/23/2017 09:23:52 PM	Michael Rubinsky	Comment #12	Reply to this comment
I think i nailed it down... Looks like a whole bunch of mails which ran throu a yahoo group some very long time ago were the rootcause.. i send you one example via mail. The sample email is sync'd with no issues to all of my test clients - including Outlook. No display issues and no out of memory issues. Additionaly, the email you sent me did not contain any TNEF or DAT attachments, which is the ONLY way the code where the memory exhaustion occurred would have been called. [Show Quoted Text - 28 lines][Hide Quoted Text] I have removed the mail from the trash.. Still the error occurres. I have pulled a new report under the same address as before. As this mail was related to a calendar entry, do you think it might be related to the entry in the calendar? Yes, meeting requests are sometimes sent as TNEF or DAT file attachments from Outlook. I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/22/2017 09:24:38 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #11	Reply to this comment
I think i nailed it down... Looks like a whole bunch of mails which ran throu a yahoo group some very long time ago were the rootcause.. i send you one example via mail. [Show Quoted Text - 27 lines][Hide Quoted Text] I have removed the mail from the trash.. Still the error occurres. I have pulled a new report under the same address as before. As this mail was related to a calendar entry, do you think it might be related to the entry in the calendar? Yes, meeting requests are sometimes sent as TNEF or DAT file attachments from Outlook. I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/20/2017 03:59:37 AM	Michael Rubinsky	Comment #10	Reply to this comment
I have removed the mail from the trash.. Still the error occurres. I have pulled a new report under the same address as before. As this mail was related to a calendar entry, do you think it might be related to the entry in the calendar? Yes, meeting requests are sometimes sent as TNEF or DAT file attachments from Outlook. [Show Quoted Text - 18 lines][Hide Quoted Text] I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/19/2017 09:54:28 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #9	Reply to this comment
I have removed the mail from the trash.. Still the error occurres. I have pulled a new report under the same address as before. As this mail was related to a calendar entry, do you think it might be related to the entry in the calendar? [Show Quoted Text - 17 lines][Hide Quoted Text] I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/19/2017 01:23:12 PM	Michael Rubinsky	State ⇒ No Feedback

01/08/2017 05:25:08 PM	Michael Rubinsky	Comment #8	Reply to this comment
I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] You should be looking at the sync log, not the horde log and should see a LOT more than this. Can i send you the log file so you can have a look? I really can not find anything :-( You can send me the sync log, yes but I will also probably need an email that triggers this. As I said, it's most likely an email with a TNEF or *.DAT file attachment.

01/08/2017 04:38:31 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #7	Reply to this comment
I have setup a new sync with a new device. But the only thing i see until the HORDE[5838]: Allowed memory size of 134217728 bytes exhausted (tried to allocate 150994944 bytes) [pid 5838 on line 1090 of "/usr/share/php/Math/BigInteger.php"] Can i send you the log file so you can have a look? I really can not find anything :-( [Show Quoted Text - 34 lines][Hide Quoted Text] Do you have any hint where i can find the message maybe in the log file? Unfortunately i do not now which mail is causing this. If you know which device it is at least, you can enable a per-device activesync debug log. It would be the last message attempted to be sent. If you don't know the device, you can instead try the consolidated debug log and hope we can figure it out from there. Looking at where the process dies, it is probably an email with some type of TNEF attachment. Please provide a sample email that triggers this error. It looks like it is not dying but still trying to do the sync. Well, when the memory is exhausted, the running process dies and the sync must restart. The only thing i can find is this error message: Previous request processing for synckey {58608cf6-006c-4922-b379-4ed95e72af70}123 failed to be accepted by the client, removing previous state and trying again. Correct. This indicates a previous attempt failed since the client is sending an older synckey. The Client is Outlook 2016

01/04/2017 01:49:00 PM	Michael Rubinsky	Comment #6	Reply to this comment
[Show Quoted Text - 19 lines][Hide Quoted Text] Do you have any hint where i can find the message maybe in the log file? Unfortunately i do not now which mail is causing this. If you know which device it is at least, you can enable a per-device activesync debug log. It would be the last message attempted to be sent. If you don't know the device, you can instead try the consolidated debug log and hope we can figure it out from there. Looking at where the process dies, it is probably an email with some type of TNEF attachment. Please provide a sample email that triggers this error. It looks like it is not dying but still trying to do the sync. Well, when the memory is exhausted, the running process dies and the sync must restart. The only thing i can find is this error message: Previous request processing for synckey {58608cf6-006c-4922-b379-4ed95e72af70}123 failed to be accepted by the client, removing previous state and trying again. Correct. This indicates a previous attempt failed since the client is sending an older synckey. The Client is Outlook 2016

01/04/2017 09:27:23 AM	michael (dot) klein (at) hypaerion (dot) net	Comment #5	Reply to this comment
[Show Quoted Text - 17 lines][Hide Quoted Text] Do you have any hint where i can find the message maybe in the log file? Unfortunately i do not now which mail is causing this. If you know which device it is at least, you can enable a per-device activesync debug log. It would be the last message attempted to be sent. If you don't know the device, you can instead try the consolidated debug log and hope we can figure it out from there. Looking at where the process dies, it is probably an email with some type of TNEF attachment. Please provide a sample email that triggers this error. It looks like it is not dying but still trying to do the sync. The only thing i can find is this error message: Previous request processing for synckey {58608cf6-006c-4922-b379-4ed95e72af70}123 failed to be accepted by the client, removing previous state and trying again. The Client is Outlook 2016

01/02/2017 08:31:22 PM	Michael Rubinsky	Comment #4	Reply to this comment
Do you have any hint where i can find the message maybe in the log file? Unfortunately i do not now which mail is causing this. If you know which device it is at least, you can enable a per-device activesync debug log. It would be the last message attempted to be sent. If you don't know the device, you can instead try the consolidated debug log and hope we can figure it out from there. Looking at where the process dies, it is probably an email with some type of TNEF attachment. Please provide a sample email that triggers this error.

01/02/2017 08:26:35 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #3	Reply to this comment
Do you have any hint where i can find the message maybe in the log file? Unfortunately i do not now which mail is causing this. Please provide a sample email that triggers this error.

01/02/2017 04:36:37 PM	Michael Rubinsky	Comment #2 Assigned to Michael Rubinsky State ⇒ Feedback	Reply to this comment
Please provide a sample email that triggers this error.

12/23/2016 08:21:51 PM	michael (dot) klein (at) hypaerion (dot) net	Comment #1 Priority ⇒ 2. Medium State ⇒ Unconfirmed Patch ⇒ No Milestone ⇒ Summary ⇒ Active sync via outlook resulting in memory exhausted Type ⇒ Bug Queue ⇒ Horde Framework Packages	Reply to this comment
When connecting via outlook (Active sync ) to Horde. Apache is posting a memory exhaust error HORDE[22137]: Allowed memory size of 1073741824 bytes exhausted (tried to allocate 1207959552 bytes) [pid 22137 on line 1090 of "/usr/share/php/Math/BigInteger.php"]