Quartus II keeps crashing on Ubuntu 14.10

[Altera_Forum]

Quartus II will start on Ubuntu 14.10. And I've successfully programmed .sof file to Cyclone V. But, if idle for a few minutes, it crashes with the following report: 

   problem details Error:   *** Fatal Error: Segment Violation at 0xffffffff00000000  Module: quartus  Stack Trace:   0x214ed: addbyter + 0xd (curl.so.4)   0x22403: dprintf_formatf + 0xd33 (curl.so.4)   0x22a45: curl_mvsnprintf + 0x25 (curl.so.4)   0x216c2: curl_msnprintf + 0x82 (curl.so.4)   0x12646: Curl_failf + 0x126 (curl.so.4)   0x910f: Curl_resolv_timeout + 0x12f (curl.so.4)     End-trace    Executable: quartus Comment: None  system information Platform: linux64 OS name: Ubuntu 14.10 OS version: 14  quartus ii information Address bits: 64 Version: 14.1.0 Build: 186 Edition: Web Edition     Anyone having similar problems? Any solutions?

[Altera_Forum]

This past weekend I migrated a 14.0 project into 14.1 -- I'm running ubuntu 14.04 (has been very stable with 14.0 tools for me for many months). 

 I had a few crashes at first -- couldn't complete a compile. I've seen this before, no idea why, but the same sort of crashes initially working with an upgrade (I don't have any specifics stored). It seemed to blow chunks at different stages then for some reason the project will start compiling again (I even archived the project and brought it into 14.1 cleanly, I thought). In any case, I successfully compiled on Centos then moved back to Ubuntu with the project and it seemed to be behaving.  Everything seemed to be working fine, but TimeQuest was crashing for me still (both on Centos and Ubuntu after testing). I'm working with the plain cyclone V on the arrow bemicroCV board and the DDR3 pin assignment tcl script doesn't execute successfully on 14.1 for me.  This is the second time I tried to get up and running with 14.1 and I've gone back to 14.0.2 or whatever the last was.   The two issues I had in the end (TimeQuest crashing and the DDR3 pin assignment script failures) happend on Centos6.5/6.6, I decided to bail. If it were a Ubuntu problem I would have maybe stuck it out a little longer and tried to narrow it down.  I'm using the full version, FWIW.  Good luck, Lance

[Altera_Forum]

I'm seeing the same crash on both Ubuntu 12.04.5 and CentOS 7. I tried contacting Altera support, but they only referred me to the supported OS list, which lists RHEL5 and RHEL6 as the only supported Linux distributions. In fairness, I tried installing Quartus 14.1 on CentOS 6.6, and it didn't crash.

[Altera_Forum]

Per Lance's suggestion, I ran Quartus 14.0 all day yesterday on Ubuntu 14.10 -- NO CRASHES. So it does seem the Altera move to Quartus 14.1 is indeed causing the problem on Ubuntu. Thanks for the help Lance. 

 I'm planning to run and monitor stability of 14.0 base for awhile longer, and then eventually try to install Update 2 to 14.0   andersm, have you sent the crash report to Altera? I did. It would be nice to get this fixed.  -- Jeff

[Altera_Forum]

--- Quote Start ---  andersm, have you sent the crash report to Altera? I did. It would be nice to get this fixed.  --- Quote End ---   I did, and received this reply:   --- Quote Start ---  We don’t actually support Ubuntu. http://www.altera.com/download/os-support/oss-index.html I’ll contact the factory and see if they can suggest anything but I don’t think we have any Unbuntu machines for us to try to replicate this problem.   --- Quote End ---  

[Altera_Forum]

Please update the thread if you figure out what is going on. 

 There have been some external tutorials I've found for hacking things to get modelsim to run properly under ubuntu, getting the USB blaster to work, etc, but usually the basic software and other core stuff would run. The clock domain stuff in Qsys would be nice to be able to use...along with any other updates behind the scenes to IP and such.   The other real plus of 14.1 is that it seemed like the built in help links worked on ubuntu for the first time (pulling up a browser and not failing). That stuff worked fine under CentOS/Redhat -- but I can't make myself run Centos as a main OS on my laptops due to hardware support; just too painful. I keep VM's up to date on all the ubuntu boxes with Centos 6.6 if I really need to run stuff there.   Lance

[Altera_Forum]

I received an update from Altera support: 

  --- Quote Start ---  The Curl library shipped with Quartus II is not thread-safe and this causes the error you have seen. You can work around this problem by creating an environment variable to set CURLOPT_NOSIGNAL to 1 with the following command:   For csh-style shells:   setenv CURLOPT_NOSIGNAL 1   For sh-style shells   export CURLOPT_NOSIGNAL=1 If you could let me know if this resolves your problem then I can confirm this solution with the factory.   --- Quote End ---   The suggested workaround did not work for me.

[Altera_Forum]

--- Quote Start ---  I received an update from Altera support:  The suggested workaround did not work for me.  --- Quote End ---    In the altera/14.1/quartus/linux64 folder, rename libcurl.so.4 to libcurl.so.4.bak. Then, restart Quartus. This will cause the runtime library loader to fall back to the system's own libcurl, which is hopefully compatible. This worked on RHEL 7, no crashes for a couple of hours when it used to crash after every compilation.

[Altera_Forum]

--- Quote Start ---  In the altera/14.1/quartus/linux64 folder, rename libcurl.so.4 to libcurl.so.4.bak. Then, restart Quartus. This will cause the runtime library loader to fall back to the system's own libcurl, which is hopefully compatible. This worked on RHEL 7, no crashes for a couple of hours when it used to crash after every compilation.  --- Quote End ---    Thanks! That seems to have worked for me. Been running over an hour now and no crashes. That is much longer than the minute or two I would get before.  I should also note that I haven't just done the fix Pinhedd described. All three of these things I have done to make it work, but not necessarily needing numbers 1 and 2. I could try reverting the change made in number 1, and rebooting to revert number 2. It would be real nice if Altera would add proper support for Debian based linux distros, such as Ubuntu. Ubuntu is so much more useful and easier to use than RHEL based. I don't want to spend time tinkering to get things working - I want to get down to work.    

1. Make certain 127.0.0.1 is set to resolve to localhost.localdomain localhost: mikeon84@mikes-laptop:~$ cat /etc/hosts | head -1 127.0.0.1 localhost.localdomain localhost 
2. Set env variable export CURLOPT_NOSIGNAL=1 
3. Most likely the real fix, noted by Pinhedd. There are many libraries in that folder, which many are already included in the OS - so are probably not necessary? mikeon84@mikes-laptop:~/$ sudo mv /altera/14.1/quartus/linux64/libcurl.so.4 /altera/14.1/quartus/linux64/libcurl.so.4.bak password for mikeon84: mikeon84@mikes-laptop:~/$  

   Thanks again, Pinhedd. I was afraid I was going to have to install Windows in a VM or something.

[Altera_Forum]

--- Quote Start ---  Thanks! That seems to have worked for me. Been running over an hour now and no crashes. That is much longer than the minute or two I would get before.  I should also note that I haven't just done the fix Pinhedd described. All three of these things I have done to make it work, but not necessarily needing numbers 1 and 2. I could try reverting the change made in number 1, and rebooting to revert number 2. It would be real nice if Altera would add proper support for Debian based linux distros, such as Ubuntu. Ubuntu is so much more useful and easier to use than RHEL based. I don't want to spend time tinkering to get things working - I want to get down to work.    

1. Make certain 127.0.0.1 is set to resolve to localhost.localdomain localhost: mikeon84@mikes-laptop:~$ cat /etc/hosts | head -1 127.0.0.1 localhost.localdomain localhost 
2. Set env variable export CURLOPT_NOSIGNAL=1 
3. Most likely the real fix, noted by Pinhedd. There are many libraries in that folder, which many are already included in the OS - so are probably not necessary? mikeon84@mikes-laptop:~/$ sudo mv /altera/14.1/quartus/linux64/libcurl.so.4 /altera/14.1/quartus/linux64/libcurl.so.4.bak password for mikeon84: mikeon84@mikes-laptop:~/$  

   Thanks again, Pinhedd. I was afraid I was going to have to install Windows in a VM or something.  --- Quote End ---    I'd followed these Pinhedd steps and them works, you don't need to set a new environment variable just rename the libcurl.so.4 to libcurl.so.4.bak. Why? Because in Fedora Quartus don't need to load this lib, just in CentOS or RedHat versions! Thanks guys, this solution solve my problem. I hope that be useful for you!

[Altera_Forum]

yes, this set of solutions seems to be working for me as well. 

I think I am interpreting the results a little differently than before (I misinterpreted some warning messages that were labelled as errors as being more severe, and I had some other things project related that were not due to 14.1 that were fixed since the last time I tried it).  I only did a couple compiles today but saw no crashes.  Lance

[Altera_Forum]

It ended up crashing in spectacular fashion a couple of times (windows vanished) last evening during a signaltap session, so I'm giving up on 14.1 again for a while.  

 I would seek support but since Ubuntu isn't supported and I don't feel like taking the time to test this on Centos (some of the previous crashes were happening on Centos 6.5/6.6) right now, it probably will not provide any useful result. These published tweaks certainly helped get further than I had gotten before though.  One thing that still bugged me is that wile I think it was only "warnings falsely classified as errors", the result of the DDR pin assignment script execution is a little discouraging. It throws a lot of red text up on the console.  To anyone who has had success on Ubuntu, have you exercised the tools pretty thoroughly and things still seem stable or are you just doing compiles and aren't deep in yet?  Lance

[Altera_Forum]

>> To anyone who has had success on Ubuntu, have you exercised the tools pretty thoroughly and things still seem stable or are you just doing compiles and aren't deep in yet? 

 Since downgrading back to Quartus 14.0.2, as I reported in# 4 above, I have not gone back to Quartus 14.1, so I have not tried any of the reported fixes. So, sorry I can't help at the moment. I can say that Quartus 14.0.2 on Ubuntu has not crashed once, and I'm working a Cyclone V SoC FPGA, so I've been using all the Quartus tools: HDL coding, Qsys, TimeQuest, ModelSim, Pin Planner, Programmer. Also, SoC EDS 14.0.2 for embedded Linux App development has been stable under Ubuntu.

[Altera_Forum]

That is good to hear. I had done some work with the SoC EDS stuff last year but nothing lately; otherwise everything I've done with 14.02 has been fairly reliable. I'm on Ubuntu 14.04LTS and probably staying there for the time being. I have kept CentOS VM's running because I could never get the CV SocKit GSRD yocto stuff to build properly under ubuntu 14.04 (or 13.10?); I had success under 12.04. I didn't investigate it too deeply. I had higher expectations for enterprise linux (unrealistic perhaps) and the laptop hardware support was not good for me; Ubuntu on my main machines has been a pretty good working OS.

[Altera_Forum]

Hello guys. I have same problem with Quartus II in Ubuntu 14.04LTS. The error is 

Inconsistency detected by ld.so: dl-close.c: 762: _dl_close: Assertion `map->l_init_called' failed! Program is crashed in 5-10minutes or when I compillng project. The patch http://www.altera.com/support/kdb/solutions/rd01272015_37.html didn`t help me. Can somebody help me, please? Thanks.

[Altera_Forum]

Hi, 

 I've been running quartus 14.1 on Debian for some time now, and aftere having these crashes every 5 minutes i did som digging as well and found that they were curl related. So I unplugged the network - and crashes stopped immediately.  Now, I run quartus inside a VMbox VM, and disconnecting the network interface of the VM does not affect the network capabilities of the host nor any shared directories between host and VM, so I find this a workable solution.  So there appear to be two simple solutions:  1) remove libcurl.so.4 (more elegant, thanks Pinhedd)  2) unplug the network cable.  regards, Theo     --- Quote Start ---  Hello guys. I have same problem with Quartus II in Ubuntu 14.04LTS. The error is Inconsistency detected by ld.so: dl-close.c: 762: _dl_close: Assertion `map->l_init_called' failed! Program is crashed in 5-10minutes or when I compillng project. The patch http://www.altera.com/support/kdb/solutions/rd01272015_37.html didn`t help me. Can somebody help me, please? Thanks.  --- Quote End ---  

[Altera_Forum]

There is also this thread, that has a link to a patch (which I havent yet tried). 

 http://stackoverflow.com/questions/28538645/quartusii-14-1-0-debian-linux-crash

[Altera_Forum]

For what it is worth, quartus II v15.0 seems the smoothest out of the gate for me on Ubuntu 14.04 LTS. 

 I gave up on 14.1 a couple of times due to random crashes. 14.0.2 was quite stable for me but it had a couple quirks. 1) needed hacks to get modelsim to run... not painful, but annoying. 2) the integrated help wasn't working at all. I lived with that...  15.0 (downloaded today) has modelsim running out of the box with no mods, help launches a browser properly, it seems "zippy", and without changing any settings the compile time is down maybe 10% on my cyclone V project. There may be issues of course, but this is all pretty promising.  -Lance

[Altera_Forum]

--- Quote Start ---  For what it is worth, quartus II v15.0 seems the smoothest out of the gate for me on Ubuntu 14.04 LTS.  -Lance  --- Quote End ---    Hi Lance:   Can you check something in with your 15.0 build. I'm getting a hard hang when I try to generate a PLL for the Max 10 under Ubuntu 14.04 LTS. (Mega wizard starts, but hangs as soon as I try to change in input PLL frequency)  I've tried all the fixes listed in this thread, but still get the ld.so error message. But that happens shortly after I start quartus. Quartus seems to run OK, but the megawizard hangs. (At least for ALTPLL)  Thanks  Pete

[Altera_Forum]

Pete, Funny you should mention it... I attended one of the max10 workshops a couple weeks ago and kept running into that. 

 It is a GUI bug. Right after you open the dialog, stretch the window out a little (drag a corner) and make it so scrolling is not necessary.  For me, once I adopted that "use pattern", I avoided the crash. It slowed me down in the workshop something crazy though because of the way it would just hang.   Hopefully I'm remembering this correctly.  Let me know if that solves it -- sorry I didn't see your post until just now.   -Lance

[Altera_Forum]

Thanks Lance. 

 When I attemped move the window then expand it the first time, it still hung, but I used the maximize button in the alt-pll window, and it worked as well as the windows version.. (Still quirky, in that when you enter numbers, you can not delete the last digit. so you need to first type in the new number then erase the old number)  Email notifications seem to be delayed from Alteraforums lately. So I just saw your message.  Pete