Re: [PATCH] ELF64 and changelog and 4Gb

Re: [PATCH] ELF64 and changelog and 4Gb

[Ruslan Nikolaev]

[-- Attachment #1: Type: text/plain, Size: 1682 bytes --]

> I agree. This is more readable.

I can change it of course... But it is not so bad:

1. if elf class == 64 => using ELF64 validate & loading code

2. else:

- validate header by grub_dl_check_header(). Function
grub_dl_check_header() also will check that elf class is 32-bit.

and so on...

I'm very busy now, but if you want I can update code... As for me I think
that is no need to check elf class twice (small optimization).


> Could you provide us detailed information? What information is passed
> precisely?

Ok... But first I have a question about grub2. Does it really provide
"mmap" information for OS as multiboot defined or memory size only?

As for 4Gb limit:

1. Memory size variable is a 32-bit. It measures in bytes => max 4Gb.
When it passes to OS it converts to KB measuring but it's still truncated
to 4Gb.

2. You can see the following code in file "init.c":

 /* Ignore >4GB.  */
 if (entry->addr <= 0xFFFFFFFF && entry->type == 1)
{
grub_addr_t addr;
grub_size_t len;

addr = (grub_addr_t) entry->addr;
len = ((addr + entry->len > 0xFFFFFFFF)
? 0xFFFFFFFF - addr
: (grub_size_t) entry->len);
add_mem_region (addr, len);
}

Memory regions addr and len also is a 32-bit when BIOS provides it as a
64-bit numbers.

Of course I can be wrong because I don't know grub code very well.

New ChangeLog:

2005-07-17 Ruslan Nikolaev <nruslan@mail.com>

* multiboot.c (grub_rescue_cmd_multiboot): using ELF64 loading code if
ELF CLASS is a 64-bit and CPU CLASS is X86_64

-- 
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm


[-- Attachment #2: Type: text/html, Size: 2244 bytes --]

Re: [PATCH] ELF64 and changelog and 4Gb

[Yoshinori K. Okuji]

On Wednesday 20 July 2005 01:01, Ruslan Nikolaev wrote:
> > I agree. This is more readable.
>
> I can change it of course... But it is not so bad:

Well, I think the problem is that there is no meaning to use a union here, and 
it makes the function too long.

> Ok... But first I have a question about grub2. Does it really provide
> "mmap" information for OS as multiboot defined or memory size only?

As this is written in the comment, this is not implemented yet.

> As for 4Gb limit:
>
> 1. Memory size variable is a 32-bit. It measures in bytes => max 4Gb.
> When it passes to OS it converts to KB measuring but it's still truncated
> to 4Gb.

But this itself is not a bug really. Because mem_upper is the size of a 
contiguous region starting from 1MB, and every BIOS has one or more memory 
holes before 4GB.

Well, there is a 4GB limit. We must implement memory map passing.

> 2005-07-17 Ruslan Nikolaev <nruslan@mail.com>
>
> * multiboot.c (grub_rescue_cmd_multiboot): using ELF64 loading code if
> ELF CLASS is a 64-bit and CPU CLASS is X86_64

Please look at the examples carefully. The rule is the same as C comments:

Also, please write complete sentences and capitalize the first word. If a 
lower-case identifier comes at the beginning of a sentence, don't capitalize 
it! Changing the spelling makes it a different identifier. If you don't like 
starting a sentence with a lower case letter, write the sentence differently 
(e.g., “The identifier lower-case is ...”).

Okuji

Re: [PATCH] ELF64 and changelog and 4Gb

[Marco Gerards]

"Ruslan Nikolaev" <nruslan@mail.com> writes:

>> I agree. This is more readable.
>
> I can change it of course... But it is not so bad:
>
> 1. if elf class == 64 => using ELF64 validate & loading code
>
> 2. else:
>
> - validate header by grub_dl_check_header(). Function
> grub_dl_check_header() also will check that elf class is 32-bit.

The important thing to remember is that GRUB support more machines
than the PC and other loaders than just multiboot.  So having a ELF32
and ELF64 loader like Hollis suggested would be the right thing to
do.  It will mean more shared code and a clean solution.

Thanks,
Marco