[mythtv-users] Compilation segmentation faults

Ray Olszewski ray at comarre.com
Tue Jul 22 09:19:18 EDT 2003


At 10:27 AM 7/22/2003 -0400, Lev Pertsov wrote:
>Here is the story:
>I installed RedHat 9 and did the update which updated my kernel as well. 
>Then I got the code from MythTV cvs and tried to compile it but 
>compilation fails every time with segmentation fault in different places.

Do you really mean the compiler (g++) itself segfaults while compiling? If 
so, you clearly have a system level problem, not a Myth problem. (A Myth 
source problem could cause compilation to fail -- but with an error message 
refereicing a content proble, not a segfault.) Or do you mean that the 
source compiles, but when you run it, it segfaults? In that case, who knows 
where the problem is. You'll have to provide the usual details.

You say you "updated my kernel as well". Did you compile it locally or just 
install a newer stock kernel? Kenel compiles are another good test for 
system-level problems.

>I also got IVTV stuff from their CVS and that compiles without a problem.
>Does anybody have any idea why that might be happening? What should I do?

I haven't compiled ivtv, but I'd guess it is a considerably smaller compile 
than Myth (modules usually are small). So either of the hardware issues I 
mention below could affect the two differently.

>Different problem (secondary):
>I then got MythTv via apt-get install and I can view live TV without 
>problems. However after about half an hour of watching MythTV froze and I 
>had to do hard reboot. Any idea?

Too many.

1. Bad hardware. Fits both the earlier problem and this one. A bad memory 
location high in RAM could cause both problems (it would also manifest 
itself during large ftp transfers, for example, or anything that caused 
buffer/cache use of memory to grow appreciably). So could intermitant 
problems with disk access. Or a power supply that delivered too little 
maximum power for CPU-intensive activities. Or a heat problem. Your 
hardware report is too selective to permit more than speculation here.

2. Swap problem. I've seen this only on one system, but I did see it there 
very clearly: if the system did not have swap available, it crashed in an 
astounding variety of ways. If it was available, the system worked fine, 
and, with rare exceptions, never actually used the swap space. (This one 
was a nightmare to track down, BTW.)

The rest of these explain the Myth problem in isolation but do not explain 
a compilation segfault as well, so they are less likely, if I've understood 
your first problem correctly.

3. Problem with the ringbuffer, like running out of  disk space. (In a 
thread yesterday, I asked Isaac what would happen if Myth tried to allocate 
buffer space but the drive was full. He said he didn't know ... not 
surprising, since experienced users are unlikely to run their systems that 
close to the edge. But newcomers might.) If RH still defaults to dividing 
the drive into a gazillion separate partitions, the way it once did, you 
might hit this limit without realizing it, depending on where your 
ringbuffer is located.

4. Unknown problem with IVTV.

5. Hard drive does not properly support DMA. This would likely manifest 
itself sooner than a half hour, but maybe not with "live" TV.

6. Your system somehow imposes a 2 GB limit on individual files, and the 
ringbuffer hit it. I thought newer Linux filesystems did not have this 
problem, but I don't know your setup well enough to know that yours does 
not have it.

I could suggest more, but now we are getting into real longshots. Oh, one 
thing ... how "frozen" is the system, really? I assume you mean that the 
local console is completely unresponsive, even to CRTL-ALT-F* and 
CTRL-ALT-DEL, but can you ssh in? ping successfully? Does the RESET button 
work? Or do you have to do a power-cycle reboot (people can use "hard" 
reboot to refer to either of these things, I've found).





More information about the mythtv-users mailing list