Re: [suse-kde] PROBLEM - gcc 3.2x segfault with 8.1/AthlonXP/VIAKT266A
On Tuesday 17 December 2002 04:11 pm, Matt Gibson scribed these words:
On Tuesday 17 December 2002 19:46, PL O'Smith wrote:
Andrei, Is this from a fresh install of SuSE 8.1 or upgrade from a previous version? Have you checked your ram? Is the gcc the only thing that is causing you these problems?
I'll second one of those questions, at least. If gcc is causing _random_ segmentation faults when compiling something big, it's almost always a RAM or motherboard problem. I had this problem compiling my kernel once, and it turned out to be a faulty motherboard. A friend had the problem, replaced his memory, and the problem disappeared for ever.
Compiling something as big and complex as KDE or a kernel is an ideal way to stress-test a PC -- sometimes it's only that kind of intensive processing, swapping, memory usage, etc. that can show them up, and the machine behaves fine under normal load.
First thing I'd do, Andrei, if I were you, would be to swap the memory for some from another working machine to see if it helps.
M
------------------------ Thanks Matt, I think you meant for this to go back to the list, so I thought I would pass it along with some additions. I think Andrei has changed out the ram at some point. Don't know if he has tested it or not after changing with memtest? The fact that WinXP runs ok on the machine is not much of a test of the hardware or indication that the hardware is ok as Andrei seems to think. If the install is an upgrade over a previous version of SuSE I am suspecting that there might be several gcc 2.95 programs still hanging around. If the install is new though, then my thoughts run to bad hardware or bad install/updates since then. If with a fresh install you have no problems, but after your updates you do, it's pretty likely something has gotten out of sync. Patrick --- KMail v1.4.3 --- SuSE Linux Pro v8.1 --- Registered Linux User #225206
On Tuesday 17 December 2002 21:51, PL O'Smith wrote:
I think you meant for this to go back to the list, so I thought I would pass it along with some additions. I think Andrei has changed out the
Oops! Absolutely right, of course.
changing with memtest? The fact that WinXP runs ok on the machine is not much of a test of the hardware or indication that the hardware is ok as Andrei seems to think.
I agree. The machine I had with the dodgy motherboard was running NT 4.0 dual boot with Linux, and only compiling the kernel under Linux showed up the problem. I saw maybe two bluescreens under NT for the couple of months I had that motherboard, which was no more than average for NT4 *grin*. The person who changed the motherboard in the end was a friend of mine who was in the PC retail business at the time; he used to use gcc as a stress test for new machines.
If the install is an upgrade over a previous version of SuSE I am suspecting that there might be several gcc 2.95 programs still hanging around. If the install is new though, then my thoughts run to bad hardware or bad install/updates since then. If with a fresh install you have no problems, but after your updates you do, it's pretty likely something has gotten out of sync.
Yah. As I said before, I'm definitely thinking hardware myself, simply because these are random crashes -- if gcc is broken in terms of installation, I'd think it would normally die at the same place every time. Cheers, Matt -- "It's the small gaps between the rain that count, and learning how to live amongst them." -- Jeff Noon
Matt Gibson wrote:
On Tuesday 17 December 2002 21:51, PL O'Smith wrote:
Yah. As I said before, I'm definitely thinking hardware myself, simply because these are random crashes -- if gcc is broken in terms of installation, I'd think it would normally die at the same place every time.
Cheers,
Matt
I would agree with you, except that this machine crashed randomly under SuSE 8.0 all the time but runs rock stable under 8.1. Sometimes it stopped crashing for a week or so, then it would start again. Reinstalling *everything* did not help either. -- opinions personal, facts suspect. http://home.arcor.de/36bit/samba.html
participants (3)
-
Andrew Williams -
Matt Gibson -
PL O'Smith