fasmarm bug, "error: undefined symbol " [fixed]

Index > Compiler Internals > fasmarm bug, "error: undefined symbol " [fixed]

Author

Thread

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20298
Location: In your JS exploiting you and your system

revolution 16 Apr 2008, 04:21

With version 1.67.26, I have seen with the wrong label name printed with the error message "error: undefined symbol 'somevariable'".

So far I have been unable to make minimal code to reproduce the problem.

Basically, the bug appears when there are multiple missing labels in separate lines of code. fasm displays the first line of code and a different missing symbol.

The following example doesn't yield to a simple few lines of code to reproduce the problem Sad

It seems to only occur in the context of the larger program. I can't post the whole code here, so this will have to serve as a preliminary report until I can narrow down the problem some more.

Code:

...
mov r0,not_a_symbol
mov r1,also_not_a_symbol
...

Code:

flat assembler for ARM  version 1.67.26  (160000 kilobytes memory)
test_gpio.asm [7173]:
mov r0,not_a_symbol
error: undefined symbol 'also_not_a_symbol'.

Last edited by revolution on 13 Jul 2008, 06:38; edited 1 time in total

16 Apr 2008, 04:21

edfed

Joined: 20 Feb 2006
Posts: 4330
Location: Now

edfed 16 Apr 2008, 08:36

this problem was still there in fasm 1.67.25

16 Apr 2008, 08:36

vid
Verbosity in development

Joined: 05 Sep 2003
Posts: 7105
Location: Slovakia

vid 16 Apr 2008, 08:41

i think i'll start new policy for tomasz to easier notice bugreports when he comes here: highlight bugreport thread

16 Apr 2008, 08:41

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20298
Location: In your JS exploiting you and your system

revolution 16 Apr 2008, 09:00

vid wrote:

i think i'll start new policy for tomasz to easier notice bugreports when he comes here: highlight bugreport thread

Good idea. Can I assume they are only sticky's until confirmed fixed?

16 Apr 2008, 09:00

vid
Verbosity in development

Joined: 05 Sep 2003
Posts: 7105
Location: Slovakia

vid 16 Apr 2008, 09:47

I hope Tomasz will realize that Wink

16 Apr 2008, 09:47

edfed

Joined: 20 Feb 2006
Posts: 4330
Location: Now

edfed 16 Apr 2008, 09:51

maybe he is a little bored to code FASM since 10 years now.
FASM is also a bit complete since 1.64, it just need a possibility to extend the opcodes/mnemonics lists with a configuration file or an include...

16 Apr 2008, 09:51

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20298
Location: In your JS exploiting you and your system

revolution 16 Apr 2008, 10:03

edfed wrote:

maybe he is a little bored to code FASM since 10 years now.
FASM is also a bit complete since 1.64, it just need a possibility to extend the opcodes/mnemonics lists with a configuration file or an include...

There has been such a thing around for a long time, it is called a table assembler. They are usually not very flexible and mostly you have to make lots of sacrifices in terms of performance and code generation.

16 Apr 2008, 10:03

AlexP

Joined: 14 Nov 2007
Posts: 561
Location: Out the window. Yes, that one.

AlexP 16 Apr 2008, 22:37

I think there should be a plug-in for the tables used by the assembler. Wouldn't that help? Then whatever instructions or different types of processors came out it'd work

16 Apr 2008, 22:37

Tomasz Grysztar

Joined: 16 Jun 2003
Posts: 8349
Location: Kraków, Poland

Tomasz Grysztar 09 Jul 2008, 08:38

Thank you, vid, for making those sticky threads for me. Things are now much simpler for me (as I begin to work on the new release - yeah, finally).

09 Jul 2008, 08:38

Tomasz Grysztar

Joined: 16 Jun 2003
Posts: 8349
Location: Kraków, Poland

Tomasz Grysztar 09 Jul 2008, 16:29

revolution: have you got any sources that would reproduce the problem for me?

09 Jul 2008, 16:29

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20298
Location: In your JS exploiting you and your system

revolution 09 Jul 2008, 16:39

Tomasz Grysztar wrote:

revolution: have you got any sources that would reproduce the problem for me?

Oh dear, I didn't keep any non-working versions. But even if I did keep them there are three major problems. 1: It involved multiple files (40+), 2: The total size is more than 6MB, and 3: I would not be at liberty to post the entire sources.

I found it extremely difficult to make a small version of the problem. Each time I tried, fasm would correctly show the offending line (i.e. it did not produce the problem).

So, how about this: if/when I spot this problem again then instead of me trying to isolate a small code snippet I will go straight to Olly and see if I can solve it that way.

09 Jul 2008, 16:39

revolution
When all else fails, read the source

Joined: 24 Aug 2004
Posts: 20298
Location: In your JS exploiting you and your system

revolution 13 Jul 2008, 06:37

I found the problem. In the ARM version I was not correctly saving/restoring the label_info variable. Sorry to put this as a fasm bug, it was a fasmarm bug not dealing with the new error information properly, I'm not sure how I got confused with the versions. Anyhow, I will upload a new version of fasmarm combined with the new 1.67.27 (displays as 1.67.28!) version soon.

13 Jul 2008, 06:37

< Last Thread | Next Thread >

Forum Rules:

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum