Index
> DOS > [obsolete / fixed] FASM 1.67.25 for DOS is buggyGoto page 1, 2 Next |
| Author |
|
|
DOS386 15 Jan 2008, 06:12
Sahrian wrote: Did you test the last version of FASM for DOS? When the assembly file has errors FASM crashes, but not constantly. I used the last FASM for DOS under various DPMI hosts with the same bad results. Details please. DOS = ??? DPMI host = ??? FASM version = ??? Crash symptoms = ??? Dangerous stuff (drivers,TSR's) installed ??? _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
15 Jan 2008, 06:12 |
|
|
Sahrian 15 Jan 2008, 07:00
DOS386 wrote:
Read more carefully . I said "the last FASM for DOS". From what I know it is version 1.67.25. The bug has nothing to do with the installed TSR or drivers, and it is evident for anyone who would try to feed FASM under a ring 3 DPMI host with a source code full of errors. The bug manifestation is not constant, but it is easy reproducible. Test and see for yourself. |
|||
|
15 Jan 2008, 07:00 |
|
|
DOS386 15 Jan 2008, 07:18
Sahrian wrote: Read more carefully . I said "the last FASM for DOS". YES, but it's not safe. Quote: it is evident for anyone who would try to feed FASM under a ring 3 DPMI host with a source code full of errors. The bug manifestation is not constant, but it is easy reproducible. Test and see for yourself. YES, there IS A CRITICAL BUG in FASM DOS 1.67.25. Notes (you could have posted  ) :
- No need for a Ring3 DPMI host, Unreal crashes as well - Version with IDE does NOT have this bug (that's why I didn't find it myself, I use almost the IDE only, because the Unreal stuff doesn't work with a DPMI host without EMM386  )
- No need for a big source full of errors (see below) - Bug symptoms: random: no problem, PageFault, "invalid definition provided" - The bug is NEW in 1.67.25, no bug in 1.67.24 Code: format binary as "COM" use16 org $0100 mov ah,256 ; BUG _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
|
15 Jan 2008, 07:18 |
|
|
Mac2004 17 Jan 2008, 18:34
Maybe these problems relate to problems I've encountered. I used FASM for dos as well.
http://board.flatassembler.net/topic.php?t=7902 regards, Mac2004 |
|||
|
17 Jan 2008, 18:34 |
|
|
Tomasz Grysztar 19 Jan 2008, 17:44
I'm still not able to reproduce this bug myself.
|
|||
|
19 Jan 2008, 17:44 |
|
|
Mac2004 19 Jan 2008, 20:04
Quote: I'm still not able to reproduce this bug myself. Maybe the problem appears only on my computer. Something is not working and it causes Fasm to crash. regards, Mac2004 |
|||
|
19 Jan 2008, 20:04 |
|
|
Sahrian 19 Jan 2008, 21:05
Tomasz Grysztar wrote: I'm still not able to reproduce this bug myself. Did you try it on multiple systems? I tested FASM v. 1.67.25 for DOS on multiple systems and configurations with the same bad results: when the source code has errors, FASM randomly crashes. The bug is easily reproducible as the crashes are frequent enough. The problem doesn't appear using the previous DOS version of FASM. |
|||
|
19 Jan 2008, 21:05 |
|
|
edfed 19 Jan 2008, 21:12
can you report error messages? just to know where is the problem?
i obtain the same bug. it's a general protection fault located at 00AF:000A for dosbox on win98 but the bug don't appear if fasm is called from window$ it appear only in command line mode. command.com |
|||
|
19 Jan 2008, 21:12 |
|
|
Sahrian 19 Jan 2008, 21:35
edfed wrote: can you report error messages? just to know where is the problem? DOS386 has described the bug almost exhaustively. edfed wrote: i obtain the same bug. Yes, in Win98 it crashes with this GPF, but the problem is that it crashes even in plain DOS, both using "unreal mode" or under a DPMI server. |
|||
|
19 Jan 2008, 21:35 |
|
|
DOS386 20 Jan 2008, 01:44
Tomasz wrote:
> I'm still not able to reproduce this bug myself. 
I already had a crash report also but didn't save it ... F**K
Crash reports might be rare but at least bad whining is very well reproductable ... Also, I suspect low memory corruption here
PS: I named FASMD/FASMW the 2 versions without IDE, and using FASMDD/FASMWW for the IDE
PPSS: Move to Compiler internals ? PPPSSS: Please look also at the S&R bug of FASMD (and seems FASMW as well) occasionally 
PPPPSSSS: Tested in DOS only, I don't care about DOS stuff on NiceTruncation/MoreErrors  |
|||
|
20 Jan 2008, 01:44 |
|
|
Sahrian 20 Jan 2008, 09:16
DOS386 wrote: Crash reports might be rare but at least bad whining is very well reproductable ... Use for testing a bigger source code full of errors and you will see that crashes are not so rare. Under DPMI it crashes with a PF, running in "unreal mode" it hangs computer. This horrible bug is newly introduced. I switched to an older version of the FASM for DOS, which works flawlessly. The last version is unusable and better would be removed from the download page. |
|||
|
20 Jan 2008, 09:16 |
|
|
System86 22 Jan 2008, 01:09
I have both 1.67.25 and 1.67.24 on my system (for testing the newer version), in different directories. After running 1.67.25 on a source file, the memory report of the older version changes, and sometimes the older version crashes with an invalid opcode exception (not always). Therefore, it causes memory corruption that lasts even after the program closes. The main problem with the newer version on my system is an "invalid opcode" error. This critical bug is new in the new version, and until it's fixed the older version should be put up on the download page.
Last edited by System86 on 22 Jan 2008, 01:20; edited 2 times in total |
|||
|
22 Jan 2008, 01:09 |
|
|
DOS386 22 Jan 2008, 01:18
Quote: ... Confirm all _________________ Bug Nr.: 12345 Title: Hello World program compiles to 100 KB !!! Status: Closed: NOT a Bug |
|||
|
22 Jan 2008, 01:18 |
|
|
edfed 22 Jan 2008, 02:23
is it possible that version 2 is needed?
hem hem... if it's the case, you need to make a more modular structure of the compiler. there are some insludes that are more than 1000 lines, how to manage this efficiently? alone! |
|||
|
22 Jan 2008, 02:23 |
|
|
System86 22 Jan 2008, 02:30
Quote:
This is probably not a complex bugfix, as the previous version and the IDE don't have this problem. Privalov should just change the DOS version to incorporate the bugfix and probably leave a message for users to update their version to incorporate the bugfix. This defiantly can't wait until version 2, this is an urgent problem. |
|||
|
22 Jan 2008, 02:30 |
|
|
System86 22 Jan 2008, 02:34
I tired comparing the files fasm.asm, system.inc, and modes.inc in the files in both the new and old version with fc, and they were identical! This isn't a bug in the dos directory, it must be somewhere else.
|
|||
|
22 Jan 2008, 02:34 |
|
|
edfed 22 Jan 2008, 02:34
so, after bug fix, version 2?
|
|||
|
22 Jan 2008, 02:34 |
|
|
System86 22 Jan 2008, 21:35
Only the errors.inc, version.inc, preproce.inc, and formats.inc have any changes (only looked at files in the main directory, dos directory has no changes). The bug must be in one of these files (unless the changes made a previously hidden bug elsewhere visible).
|
|||
|
22 Jan 2008, 21:35 |
|
|
rugxulo 25 Jan 2008, 03:50
Yeah, both 1.67.24 and 1.67.25 seem to have problems with "invalid definition" even when none are defined! Tested in pure DOS (not Windows), so maybe it's something to do with that. I dunno, been too busy / preoccupied with bagatela to investigate further (sorry).
|
|||
|
25 Jan 2008, 03:50 |
|
| Goto page 1, 2 Next < Last Thread | Next Thread > |
Forum Rules:
|