Can't compile console.c: gcc gets signal 11

Can't compile console.c: gcc gets signal 11

Post by Tod Hag » Mon, 08 May 1995 04:00:00



I'm running into a problem compiling a new Linux kernel. I was
able to compile a slimmer kernel previously, but now that I've
configured for networking the compilation dies on console.c
with this error message:

gcc: internal compiler error: program cc1 got fatal signal 11

I'm compiling linux 1.2.0 (Infomagic Slackware 2.2) using gcc
version 2.6.3.

I'm running a 486-66 with 20MB. The error is repeatable (at
least 6 times now).

Any assistance will be appreciated. I need the new kernel, and
this error is preventing me from building it.

Thanks.
--


Campton, NH

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Andries Brouw » Mon, 08 May 1995 04:00:00


: I'm running into a problem compiling a new Linux kernel. I was
: able to compile a slimmer kernel previously, but now that I've
: configured for networking the compilation dies on console.c
: with this error message:

: gcc: internal compiler error: program cc1 got fatal signal 11

: I'm compiling linux 1.2.0 (Infomagic Slackware 2.2) using gcc
: version 2.6.3.

: I'm running a 486-66 with 20MB. The error is repeatable (at
: least 6 times now).

Nothing is wrong with the source - it compiles fine with gcc 2.6.3.
So, probably you have some bad memory. Try replacing.
[console.c is larger than average, so errors more often show up there]

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Timothy Murp » Tue, 09 May 1995 04:00:00



>: gcc: internal compiler error: program cc1 got fatal signal 11
>Nothing is wrong with the source - it compiles fine with gcc 2.6.3.
>So, probably you have some bad memory. Try replacing.

Several people have suggested this,
but I find it a bit hard to swallow.
Surely all these people can't have bad memory ?
How come it never turns up running MS-Windows ?

I may be mistaken, but I think the problem occurs less frequently with me
since I installed the new bdflush/update.

Also when it does occur, I find if I repeat the command,
it usually does not recur the second time.

Finally, it seems to occur much less frequently (if at all)
once gcc is re-compiled.

--
Timothy Murphy  

tel: +353-1-2842366
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Xiaoguang Zha » Tue, 09 May 1995 04:00:00



: >: gcc: internal compiler error: program cc1 got fatal signal 11

: >Nothing is wrong with the source - it compiles fine with gcc 2.6.3.
: >So, probably you have some bad memory. Try replacing.

: Several people have suggested this,
: but I find it a bit hard to swallow.
: Surely all these people can't have bad memory ?
: How come it never turns up running MS-Windows ?

MS-Windows never fully utilizes the computer, therefore
marginal hardware are ok for MS windows. There are more poor
quality memory chips on the market than you realize.

: I may be mistaken, but I think the problem occurs less frequently with me
: since I installed the new bdflush/update.

: Also when it does occur, I find if I repeat the command,
: it usually does not recur the second time.

: Finally, it seems to occur much less frequently (if at all)
: once gcc is re-compiled.

: --
: Timothy Murphy  

: tel: +353-1-2842366
: s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Andries Brouw » Tue, 09 May 1995 04:00:00



:: : gcc: internal compiler error: program cc1 got fatal signal 11

:: Nothing is wrong with the source - it compiles fine with gcc 2.6.3.
:: So, probably you have some bad memory. Try replacing.

: Several people have suggested this,
: but I find it a bit hard to swallow.
: Surely all these people can't have bad memory ?

Why not? Flaky memory is a rather common phenomenon.
At the factory the memory is tested, and a lot of rubbish is thrown out,
but not everything that passes one test is perfect.
(My experience so far has been that I had to exchange about 1 in every 8
memory chips bought.)

: Also when it does occur, I find if I repeat the command,
: it usually does not recur the second time.

Explanation 1: flaky memory does not fail all the time.
Explanation 2: when you run a command a second time it does
not necessarily use the same parts of memory as it did the first time.

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Tod Hag » Tue, 09 May 1995 04:00:00


I got the compilation to complete by running another task in the background
so that the compilation would end up in a different area in memory. I
created a small perl program that did an infinite loop; the perl interpreter
is big enough to displace the compilation.

So the bad memory theory looks good. Gcc is a poor diagnostic -- although it
does indicate that the memory is bad, it doesn't indicate which SIMM. Is
there a good memory diagnostic on the net?

Thanks.
--


Campton, NH

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Onno Ro » Wed, 10 May 1995 04:00:00




>>I got the compilation to complete by running another task in the background
>>so that the compilation would end up in a different area in memory. I
>>created a small perl program that did an infinite loop; the perl interpreter
>>is big enough to displace the compilation.
>>So the bad memory theory looks good. Gcc is a poor diagnostic -- although it
>>does indicate that the memory is bad, it doesn't indicate which SIMM. Is
>>there a good memory diagnostic on the net?
>But did you try just running the program again
>without making any changes ?
>I find that usually it runs on the second or third try.
>In fact I have always got through compilations by repetition in this way.
>I remain unconvinced of the bad memory theory.
>Is the standard start-up memory test really defective ?
>What is it doing if not testing the memory ?
>--
>Timothy Murphy  

>tel: +353-1-2842366
>s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

There is another theory about signal 11 problems (gcc). I tried
running a dx4/100 ay 2x50Mhz  I got also problems with compiling the linux
kernel. NOT the first files but when disk activity became more and more.

The solution  for me was I didn't change the ext. cache waitstaites to
a cpu running at 50 Mhz. So maybe there are problems with the cache chips.
So try to add waitstates to the ext. cache maybe it helps.

I don't know a program that can test a proper working memory&cache system.
The only I know at this moment is compiling a linux kernel, and if this
works without signal errors it should be ok. Msdos/ Windows don't detect
problems but somtimes hang the system.

Onno.

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Timothy Murp » Wed, 10 May 1995 04:00:00



>I got the compilation to complete by running another task in the background
>so that the compilation would end up in a different area in memory. I
>created a small perl program that did an infinite loop; the perl interpreter
>is big enough to displace the compilation.
>So the bad memory theory looks good. Gcc is a poor diagnostic -- although it
>does indicate that the memory is bad, it doesn't indicate which SIMM. Is
>there a good memory diagnostic on the net?

But did you try just running the program again
without making any changes ?
I find that usually it runs on the second or third try.
In fact I have always got through compilations by repetition in this way.

I remain unconvinced of the bad memory theory.

Is the standard start-up memory test really defective ?
What is it doing if not testing the memory ?

--
Timothy Murphy  

tel: +353-1-2842366
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Thomas Reineck » Wed, 10 May 1995 04:00:00





>: >: gcc: internal compiler error: program cc1 got fatal signal 11
>: >Nothing is wrong with the source - it compiles fine with gcc 2.6.3.
>: >So, probably you have some bad memory. Try replacing.
>: Several people have suggested this,
>: but I find it a bit hard to swallow.
>: Surely all these people can't have bad memory ?
>: How come it never turns up running MS-Windows ?

I had the very same prob: cc1 got fatal signal 11 with new kernel releases 1.2.x
After I was told here that this might caused by bad memory timing I checked
my BIOS Setup (during cold start booting) and changed the entry for
RAM access from FAST to NORMAL -- now I can compile for hours without fatals.

I never had this prob before running Linux, WINDOOFS, OS/WARP.  So I guess that
the RAM access times vary by thermal influence from full CPU/RAM access loads
during compiling huge code segments.

Quote:>: I may be mistaken, but I think the problem occurs less frequently with me
>: since I installed the new bdflush/update.
>: Also when it does occur, I find if I repeat the command,
>: it usually does not recur the second time.
>: Finally, it seems to occur much less frequently (if at all)
>: once gcc is re-compiled.

If you can't modify your RAM timings you should give your machine a beak
while comipling:  nice it down!  Better compiling successfully and slow than
restart compiling after aech fatal signal you get.

Tom

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Tod Hag » Thu, 11 May 1995 04:00:00



Quote:>But did you try just running the program again
>without making any changes ?

Yes, roughly six times over a period of 24 hours.

Quote:>I find that usually it runs on the second or third try.
>In fact I have always got through compilations by repetition in this way.

>I remain unconvinced of the bad memory theory.

>Is the standard start-up memory test really defective ?
>What is it doing if not testing the memory ?

It's testing that the memory is THERE, not that it works.

I've not yet found a free memory diagnostic. The only ones I'm aware of are
commerical and fairly costly.
--


Campton, NH

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Justin Maxwe » Sat, 13 May 1995 04:00:00




>I remain unconvinced of the bad memory theory.
>Is the standard start-up memory test really defective ?
>What is it doing if not testing the memory ?

Counting it...

Seriously, I've seen alot of memory related problems, and only once
has the boot-up check picked it up.  The nature of RAM is that it's
very difficult to check properly.  Unless you write gazoodles of bit
combinations to bazillions of permutations of neiboughing bytes,
you'll never get a 100% certain that the RAM is reliable.

eg: two bits stuck together, either in one byte, or nearby bytes...
You'll never pick it unless your pattern check writes different data
to those particular bits...

Anyway, I've now seen about a dozen people say it's memory corruption,
and nobody say it isn't so....

However, when I asked the question, what about SWAP memory, the answer
was, yes, it could be that (ie: the disk channel)...

I'm running a Pentium-75 with a NCR 810 PCI SCSI interface...

Justin

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Versus Versa » Sun, 14 May 1995 04:00:00




: >I remain unconvinced of the bad memory theory.

: >Is the standard start-up memory test really defective ?
: >What is it doing if not testing the memory ?

: Counting it...

: Seriously, I've seen alot of memory related problems, and only once
: has the boot-up check picked it up.  The nature of RAM is that it's
: very difficult to check properly.  Unless you write gazoodles of bit
: combinations to bazillions of permutations of neiboughing bytes,
: you'll never get a 100% certain that the RAM is reliable.

: eg: two bits stuck together, either in one byte, or nearby bytes...
: You'll never pick it unless your pattern check writes different data
: to those particular bits...

: Anyway, I've now seen about a dozen people say it's memory corruption,
: and nobody say it isn't so....

: However, when I asked the question, what about SWAP memory, the answer
: was, yes, it could be that (ie: the disk channel)...

: I'm running a Pentium-75 with a NCR 810 PCI SCSI interface...

: Justin

    I have just brought a 16M RAM and exchanged my original 2x8M RAM with
    my friend's 16M to make my machine 32M. Now, I have the exact problem
    as stated. Could somebody please confirm that it is really bad memory
    problem before I go for a refund?

    Many thanx in advance, Terry

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Mike You » Sun, 14 May 1995 04:00:00






>: >I remain unconvinced of the bad memory theory.

>: >Is the standard start-up memory test really defective ?
>: >What is it doing if not testing the memory ?

>: Counting it...

>: Seriously, I've seen alot of memory related problems, and only once
>: has the boot-up check picked it up.  The nature of RAM is that it's
>: very difficult to check properly.  Unless you write gazoodles of bit
>: combinations to bazillions of permutations of neiboughing bytes,
>: you'll never get a 100% certain that the RAM is reliable.

>: eg: two bits stuck together, either in one byte, or nearby bytes...
>: You'll never pick it unless your pattern check writes different data
>: to those particular bits...

>: Anyway, I've now seen about a dozen people say it's memory corruption,
>: and nobody say it isn't so....

>: However, when I asked the question, what about SWAP memory, the answer
>: was, yes, it could be that (ie: the disk channel)...

>: I'm running a Pentium-75 with a NCR 810 PCI SCSI interface...

>: Justin

>    I have just brought a 16M RAM and exchanged my original 2x8M RAM with
>    my friend's 16M to make my machine 32M. Now, I have the exact problem
>    as stated. Could somebody please confirm that it is really bad memory
>    problem before I go for a refund?

>    Many thanx in advance, Terry

----------
Check your memory if you like; I'm curious why only gcc has a problem...

Follow up with your gcc version. If you're using gcc i2.6.3, check the
README's for optimization flag options.

Mike.

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Timothy Murp » Sun, 14 May 1995 04:00:00



>Anyway, I've now seen about a dozen people say it's memory corruption,
>and nobody say it isn't so....

I'm not a complete atheist on this matter,
but there are three things about the memory theory
that I don't understand.

Firstly, it seems mainly Pentium/PCI users who suffer from this.
But does PCI treat memory any differently to VLB ?
(This is not a rhetorical question; I'm quite ignorant of hardware.)

Secondly, it seemed to me to tend to occur at the same place,
even when there were other things going on.
I would have thought the chances of hitting the same spot in memory
were pretty small.

Thirdly, I found reducing the CPU speed had no effect
(I have a LO/HI switch on my machine.)

--
Timothy Murphy  

tel: +353-1-2842366
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

 
 
 

Can't compile console.c: gcc gets signal 11

Post by Justin Maxwe » Tue, 16 May 1995 04:00:00




>>Anyway, I've now seen about a dozen people say it's memory corruption,
>>and nobody say it isn't so....
>I'm not a complete atheist on this matter,
>but there are three things about the memory theory
>that I don't understand.

I have to confess to being a little doubtful too, to be honest.

Quote:>Firstly, it seems mainly Pentium/PCI users who suffer from this.

Perhaps time for a hardware audit of those having problems?

Quote:>But does PCI treat memory any differently to VLB ?

I believe it is quite alot different.

Quote:>Secondly, it seemed to me to tend to occur at the same place,
>even when there were other things going on.
>I would have thought the chances of hitting the same spot in memory
>were pretty small.

To me it happens all over the place, if I'm lucky it'll get a third of
the way through the kernel before going belly up.

Quote:>Thirdly, I found reducing the CPU speed had no effect
>(I have a LO/HI switch on my machine.)

And I've tried disabling all cache and adding wait states and
replacing RAM...

Justin