I'm experimenting with pe.inc. I've tried to rename the PE address space and encountered a strange error, and I can't understand what's wrong.

How to reproduce (or you can get premodified pe.inc):
1. Open pe.inc
2. Replace PE:: to XXX::
3. Replace PE: to XXX: everywhere (but don't replace where it is not a whole word, like here: load TYPE:word).
4. Replace "define CheckSumBlocks XXX,0,HeadersEnd" to "define CheckSumBlocks PE,0,HeadersEnd".
5. Try to compile win32.asm example. Error:

Quote:

win32.asm [25]: section '.idata' import data readable writeable macro section [3] macro section [14]: store SECTION_SIZE at XXX:SectionTable.VirtualSize+SECTION_INDEX*SectionTable.ENTRY_LENGTH Processed: store SECTION_SIZE at XXX:SectionTable.VirtualSize+SECTION_INDEX*SectionTable.ENTRY_LENGTH Error: address out of range.

Trying to debug it...
1. Replace:

Code:

if DEFINED_SECTION | SECTION_SIZE > 0                         SECTION_INDEX = SECTION_INDEX + 1                 end if

to

Code:

if DEFINED_SECTION | SECTION_SIZE > 0                         SECTION_INDEX = SECTION_INDEX + 1                         repeat 1, n:SECTION_INDEX                                 display 'section: ', `n, 0x0A                         end repeat                 end if

2. Replace:

Code:

NUMBER_OF_SECTIONS := SECTION_INDEX

to

Code:

NUMBER_OF_SECTIONS := SECTION_INDEX                 repeat 1, n:SECTION_INDEX                         display 'postpone: ', `n, 0x0A                 end repeat

3. The result:

Code:

section: 1 section: 2 section: 3 section: 1 postpone: 1

What's wrong? Why SECTION_INDEX resets to the 0 before the last section?

VEG wrote:

What's wrong? Why SECTION_INDEX resets to the 0 before the last section?

These macros use "namespace PE" construction in many places and it is only safe when "PE" is defined as a global symbol, otherwise "namespace PE" is going to switch to the namespace of a local "PE".
The "PE::" label has an additional function there of defining a global "PE" symbol which serves a role of a base for global namespace (this still can be fooled if you define a local "PE" symbol somewhere and there are better ways to make macros that are foolproof against that, but that is another story). So to make them work correctly when you replace "PE::" with "XXX::" you need to ensure that "PE" is a global symbol in some other way:

Code:

XXX:: PE := 1

Oh, thank you for the explanation.

Actually, I was trying to move the PE address space into the PE namespace, and to rename it to avoid ambiguity (for a reader of the source, not for the fasmg) of the same name for two different things. IMHO, something like "PE.Header::" or even "PE.Section0::" would be more clear for a programmer who will read the pe.inc. The latter variant opens a door for the idea when all following sections will be accessible using names like "PE.Section1::", "PE.Section2::", etc. It even conforms with the PE documentation which says that sections start numbering from 1:

pecoff.docx wrote:

The number of entries in the section table is given by the NumberOfSections field in the file header. Entries in the section table are numbered starting from one (1).

But it's just some thoughts. It doesn't mean that it is a good idea

This not a bad idea. The PE.INC was one the first sets of formatting macros I made for fasmg and at the time I wanted to write something simple and compact to include in the examples. But then the x86 examples became much more complex anyway. I think that I could in the future fork PE.INC into a basic version included in the examples and a more "advanced" version in the Windows headers package.