I found some problem with the align directive.

Following discription found in docs:






I write cdr boot layout with FASM and have to fill some sectors, sectorsize 2048 bytes (800h). When I try to fill the sectors with following function align does not fill sector complete. I don't know why. Maybe this is a bug.

My own macro function is:



The following workaround code fills sector correct.



Here I use labels begadr (begin of code) end endaddr (end of code) for calculation.
fillsector2 works fine, fillsector has a functional bug.

Just for discussion, maybe somebody has an idea whats wrong.

Has align any dependencies from org directive ?
In my code I use directive org 7c00h+50h

When calling fillsector the sector fill is missing exact 50h bytes.
So sector is filled up to 7af not up to 7ff.

I dont understand why align references to org (if my suspicion is true).
Align 800h should be 800h in all cases and should not fail when org directive is 100h for example.

Maybe somebody can bring some light in my confusion.
Could live with workaround in fillsector2 but want to understand align in general.

By using org 7c50h you are telling the assembler that the code will be loaded at that address in memory. So later when you use align 800h you are telling the assembler to align the next byte at that next 2kB boundary. The next 2kB boundary is 0x8000h.

fasm's ALIGN is by definition memory-wise, thus it is affected by ORG and family. Perhaps it's not stated clearly enough in the manual, I may correct it.

Well sometime it helps to ask a question in the forum to find the answer myself. So in the night after writing my post it got clear to me. Building a general question helps finding the answer sometimes.

Sure ORG has to affect align, what else ???

I thought of a possible error but then I found that i missed exact 50h bytes. This corresponds to my offset ORG xxxx+50h.

I wrote first a bootloader and because it's getting to big for putting on a hard disk including partition table I wrote a preloader to load the rest. And loaded it +50h bytes and then made a relocation of code. That caused the problem because the second code has ORG +50h but was loaded in a new sector at address "0". So I missed exact 50h bytes. I correct this with a manual fill "db 50h dup 0".

Sorry for disturbing community with this bad question.

there are no stupid (bad) questions - there are lost (unclear) points.