How to detect flags

[ecm-pushbx]

The most flags of any interrupt.lst entries is the MSW debugging INT 41 interface:

TheList/source/Interrupt List/INT 41 SYSTEM DATA HARD DISK 0 PARAMETER TABLE CPU MS Windows/INT 410000 MS Windows debugging kernel OUTPUT CHARACTER FOR USER.txt

Line 7 in 166c02b

INT 41 CPU - MS Windows debugging kernel - OUTPUT CHARACTER FOR USER

This has "CPU" which according to https://github.com/LoopZ/TheList/blob/166c02bba343294e8318630cddcef03636b4f8c8/source/Interrupt%20List/_Flags.txt means Callout, Protected mode, and Undocumented.

Do all interrupt entries contain the dash after the slot for the flags, even if no flags are present? That would simplify scanning for the "name" needed for some hyperlinks.

[shidel]

First, thanks for pointing me to this one. I either forgot about multiple flags, or there was a bug I overlooked, when I wrote the program to break it down into pieces. The header for that one should really look like this:

--------------------------------------------------------------------------------
 Unique ID: 410000
  Category: G 	debuggers/debugging tools
      Flag: C 	callout or callback (usually hooked rather than called)
      Flag: P 	available only in protected mode
      Flag: U 	undocumented function
--------------------------------------------------------------------------------

So today, I will whip up a quick one-off to go through all the entries and correct that oversight.

To answer your question. Yes. Sort-of...

This is also true for the other LST files like FARCALL, CMOS, PORTS, MEMORY, MSR, I2C. Like entry: https://github.com/LoopZ/TheList/blob/166c02bba343294e8318630cddcef03636b4f8c8/source/Far%20Call%20Interface/CALL%20C000h%200068h%20ATI%20Mach32%20VIDEO%20BIOS.txt

But, not the entries for SMM, BIBLIO, OVERVIEW, TABLES or LINKS.

When there are flags present, they are always the field prior to the first dash (-). I am going to double check this today while I create the Flag fix-up routine, but I think there are only ever two or three space separated fields that ever exist before that first (-). If the third field exists, it is Flag(s).

The sort-of comes into play, when there is no description. While I do not recall if there are any entries without one at this time, it is possible to just have an entry like INT 21 U or INT 21. But, if there is any discription, that - is present.

[ecm-pushbx]

First, thanks for pointing me to this one. I either forgot about multiple flags, or there was a bug I overlooked, when I wrote the program to break it down into pieces. The header for that one should really look like this:

--------------------------------------------------------------------------------
 Unique ID: 410000
  Category: G 	debuggers/debugging tools
      Flag: C 	callout or callback (usually hooked rather than called)
      Flag: P 	available only in protected mode
      Flag: U 	undocumented function
--------------------------------------------------------------------------------

So today, I will whip up a quick one-off to go through all the entries and correct that oversight.

Oh yeah, figures it should list all of them.

To answer your question. Yes. Sort-of...

This is also true for the other LST files like FARCALL, CMOS, PORTS, MEMORY, MSR, I2C. Like entry: https://github.com/LoopZ/TheList/blob/166c02bba343294e8318630cddcef03636b4f8c8/source/Far%20Call%20Interface/CALL%20C000h%200068h%20ATI%20Mach32%20VIDEO%20BIOS.txt

But, not the entries for SMM, BIBLIO, OVERVIEW, TABLES or LINKS.

When there are flags present, they are always the field prior to the first dash (-). I am going to double check this today while I create the Flag fix-up routine, but I think there are only ever two or three space separated fields that ever exist before that first (-). If the third field exists, it is Flag(s).

The sort-of comes into play, when there is no description. While I do not recall if there are any entries without one at this time, it is possible to just have an entry like INT 21 U or INT 21. But, if there is any discription, that - is present.

Okay, that helps me on how to handle this. Thank you!

[shidel]

Flags are now fixed. Yay.

INT 310700 DPMI 09 MARK PAGES AS PAGING CANDIDATES

[shidel]

During the creation of the program to fix the flags, I noticed that occasionally parsing could get a little confused when handling the title. This was because sometimes TABS were instead of spaces. While I let that slide during the fix-flags, I whipped up a quick fix-titles to normalize the spacing in the Titles.

Now that has been done, you can reliably just check the 3rd space delimited field of a title (first non-blank line after the second bar of 80 dashes). If it exists and is not a dash, it is the flags.

Now on to promoting Uniqueness. :-)

[andrewbird]

This was because sometimes TABS were instead of spaces.

Will you add that to the validation tool, it would be good to catch those before merging updates?

[shidel]

@andrewbird, that is very likely.

After I fix the ID Uniqueness, I plan on going back into makelist and adding support to not use the extended IDs when run in --legacy mode. At that time, I plan on adding a couple more checks for --cicd mode regarding the file header. Also, testing the Title for extra whitespace or tab usage would be a good QC add at that time.