Re: [PATCH 0/5] Boot protocol changes

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

Jeremy Fitzhardinge wrote:
> H. Peter Anvin wrote:
>> I'm proposing that the existing bzImage format be retained, but that
>> the payload of the decompressor (already a gzip file) simply be
>> vmlinux.gz -- i.e. a gzip compressed ELF file, notes and all.  A
>> pointer in the header will point to the offset of the payload (this is
>> new, obviously.)
>>
>> The decompression stub is adjusted to expect an ELF image, instead of
>> a raw binary.
> 
> It could, or just treat it as a raw binary at 1M+offset to skip the headers.

It would be cleaner to actually parse the ELF; it's only a handful of 
lines of code (we don't have to support arbitrary placement of sections, 
obviously, which makes the problem simpler.)

>> Existing bootloaders (16- or 32-bit) simply load the bzImage the way
>> they do now; new bootloaders have the option of accessing the
>> vmlinux.gz directly if they either want to load it themselves or want
>> to examine the notes. 
> 
> OK, but that has the same problem as making the payload an ELF file:
> 32-bit bootloaders which simply jump to 1M will be jumping into data
> rather than code - and I got the impression from taking to Eric at KS
> that there are such bootloaders.  

Uhm, no it doesn't.  Those bootloaders jump to the decompressor, not to 
the payload.  The decompressor interface hasn't changed.

> If that's not an issue, then I still think the payload should be a plain
> ELF file (possibly self-decompressing, or just a plain uncompressed
> vmlinux, if that's what's desired).  Still think making a protected-mode
> bootloader do the decompression is the wrong way to go about this; ELF
> is enough.

It doesn't have to if it doesn't want to; it only needs to do so if it 
wants to access the kernel as an ELF.  Again, it has the advantage that 
the ELF is the real vmlinux, no funnies.

	-hpa