Re: Xine Build - Segmentation faults

[Declan Moriarty]

Segmentation Faults are a sign of dud software. In one of the generations of 
cpu in the past (The My useless cpu is better than your useless cpu stage) 
somebody (Intel) went to 16 bit data bus and registers, but 20 bit address 
bus. How do you address the last four bits? Enter a crude paging system with  
four bit registers, which allowed pcs only to address 64K of upper memory at 
a time. This, I think, is Real Mode in the 386+ and reality for anything 
below, and a huge compatability nightmare. It also explains why Dos is 
comfortable in 16 bits, and windows has such difficulty with it. This 
segmentation thing is hidden in there always.

The result is in this limited mode, that if you try to grab something off the 
wrong page, you get a segmentation fault. But linux requires a 386+ because 
it never wants to use this stupid way of going on in the first place (GOOD 
WORK, LINUS!). When you see Segmentation Fault, that cpu is in real mode, 
the application is has gone off the rails and switched the cpu to real mode 
which I don't think linux ever uses. So there's either a dud compiler, dud 
support files, dud disk, dud rpm, or dud something. Time for the M$ solution 
(delete  reinstall).  Lastly, don't take all of this as gospel; I'm an 
electronic hardware type, with no authority on software issues. 

-- 
Regards,


Declan Moriarty




Applied Researches - Ireland's Foremost Electronic Hardware Genius

A Slightly Serious(TM) Company

Experience is like a comb, 
that Life gives you - AFTER all your hair has fallen out!

On Sunday 09 December 2001 17:32, you wrote:
 On Sun, 9 Dec 2001 21:48:14 +1000, Keith Antoine wrote:
 Why the hell use srpms and have compile probs, when you can use the
  generic rpms ??

 Well,

 The rpms are built on a SuSE box and all I get from them is a
 segmentation fault.  But it would seem that this is common as I get the
 same fault after I got the thing built from tarballs.

 Is there a dump log that saves data from segmentation fault errors?

 stayler

 ___
 Linux-users mailing list
 Archives, Digests, etc at http://linux.nf/mailman/listinfo/linux-users

___
Linux-users mailing list
Archives, Digests, etc at http://linux.nf/mailman/listinfo/linux-users

Re: Xine Build - Segmentation faults

[Kurt Wall]

Declan Moriarty wrote:

[...]

% 
% The result is in this limited mode, that if you try to grab something off the 
% wrong page, you get a segmentation fault. But linux requires a 386+ because 
% it never wants to use this stupid way of going on in the first place (GOOD 
% WORK, LINUS!). When you see Segmentation Fault, that cpu is in real mode, 
% the application is has gone off the rails and switched the cpu to real mode 
% which I don't think linux ever uses. So there's either a dud compiler, dud 
% support files, dud disk, dud rpm, or dud something. Time for the M$ solution 
% (delete  reinstall).  Lastly, don't take all of this as gospel; I'm an 
% electronic hardware type, with no authority on software issues. 

Segmentation faults occur when there is an illegal memory access of
any type. In software, it most often occurs when when referencing a
undefined pointer or accessing unallocated memory (outside of an
array, for example). In Linux, applications cannot drop the CPU
into real mode -- the kernel controls this. Moreover, the kernel
*does* use real mode, but only at boot time, and it jumpts to
protected mode fairly early. 

Kurt
-- 
Ben, why didn't you tell me?
-- Luke Skywalker
___
Linux-users mailing list
Archives, Digests, etc at http://linux.nf/mailman/listinfo/linux-users