Re: [PATCH] Further update of the i386 boot documentation

["H. Peter Anvin" <hpa@zytor.com>]

Jeremy Fitzhardinge wrote:
> H. Peter Anvin wrote:
>> +Field name:	boot_flag
>> +Type:		read
>> +Offset/size:	0x1fe/2
>> +Protocol:	ALL
>> +
>> +  Contains 0xAA55.  This is the closest thing old Linux kernels have
>> +  to a magic number.
>>   
> 
> Endianess?  I guess a blanket statement saying that all constants are
> stored little-endian enough (which is obvious, but its always good to be
> explicit).

Next time I make a revision I'll put that in; since we're talking about
x86 here it's, as you say, obvious that it's littleendian.  As such, I
think it can wait.

>> +
>> +Field name:	header
>> +Type:		read
>> +Offset/size:	0x202/4
>> +Protocol:	2.00+
>> +
>> +  Contains the magic number "HdrS" (0x53726448).
>>   
> 
> This is a bit confusing from an endian perspective.  Does the "HdrS"
> notation mean that it is a byte array containing 'H', 'd', 'r', 'S', as
> the string syntax suggests?  Or is that the ascii interpretation of each
> byte of a 4-byte value read as a little-endian encoding from that location?

Given that we have already established littleendian byte order, it's the
same thing.

>> +
>> +Field name:	version
>> +Type:		read
>> +Offset/size:	0x206/2
>> +Protocol:	2.00+
>> +
>> +  Contains the boot protocol version, e.g. 0x0204 for version 2.04.
>>   
> 
> So the version is in BCD?

Valid objection.  It probably should be considered as (major, minor)
bytes, but we haven't had any releases where it hasn't also been valid
BCD.  I would prefer separate bytes myself, so 2.10 = 0x20a instead of
2.10 = 0x210.

>> +Field name:	readmode_swtch
>> +Type:		modify (optional)
>> +Offset/size:	0x208/4
>> +Protocol:	2.00+
>> +
>> +  Boot loader hook (see separate chapter.)
>>   
> 
> Chapter?  Is there a more specific reference you could make?

Fair enough...

>> +Field name:	kernel_version
>> +Type:		read
>> +Offset/size:	0x20e/2
>> +Protocol:	2.00+
>> +
>> +  If set to a nonzero value, contains a pointer to a null-terminated
>>   
> 
> "nil-terminated"? "\0-terminated"?

Uh?  That seems more than a little silly.  Yes, I guess formally
speaking we're talking about "NUL-terminated", but the term
"null-terminated" has over 800,000 hits on Google -- 10 times as many as
"NUL-terminated" -- and is hardly an ambiguous term ("NUL-terminated" is
ugly, and "zero-terminated" is ambiguous.)

>> +  human-readable kernel version number string, less 0x200.  This can
>> +  be used to display the kernel version to the user.  This value
>> +  should be less than (0x200*setup_sects).  For example, if this value
>> +  is set to 0x1c00, the kernel version number string can be found at
>> +  offset 0x1e00 in the kernel file.  This is a valid value if and only
>> +  if the "setup_sects" field contains the value 14 or higher.
>>   
> How about something like:
> 
>     This example is only valid if "setup_sects" is greater than
> 
>         ((0x1e00 - 0x200) / 0x200) = 14.

Probably makes sense.

	-hpa