Hi, men!
Ralf Brown wrote:
Does it mean that the interrupt will complete with ebx = 0 and CF set, or ebx can be non-zero? As I understand what is written, BIOS defines ebx != 0, and on the next iteration it sets the carry flag with ebx not defined. This interrupt completes with ebx = 0 on my system and I cannot test the described condition.

The question bothers me much, because error state can be interpreted incorrectly in my routine.

Or maybe it is enough to do 'CMP EAX, 534D4150h' to check if error occured? This simplifies the algo slightly...
Thanks.

Once the iteration starts (<=> CF=0),
then you don't have to consider CF.

http://www.delorie.com/djgpp/doc/rbinter/id/50/17.html


if CF=1, the system doesn't support int 15h with E820h.


if CF=0, EBX++ and ES:DI is filled with memory map data,


when all iterations are done, EBX becomes 0.

The maximum value of EBX is the number of system memory map.

Yup, fonolite, thanks for the link. I've read this for almost a hundred of times already Yes, I understand what you've written, and it's true for almost all BIOSes (and all that I met), but what confuses me is that was quoted. That BIOS IS PERMITTED TO SET CF AT THE END. And it seems that Ralph dealed with such BIOSes. If I don't consider CF, then the code won't work correctly under the described circumstances.