[lkml]   [2002]   [Apr]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: [PATCH] x86 boot enhancements, Clean up the 32bit entry points 6/11
     Seems that previous message did not go through, rewrite.

    I am sorry I did not check enough your patch.
    You are speaking of: arch/i386/boot/compressed/head.S
    I am speaking of: arch/i386/kernel/head.S

    Gujin skip completely arch/i386/boot/compressed/* and really
    boots the file '$$tmppiggy.gz' line 44 of file:

    So you can do whatever you want with the "first" 32 bits entry point,
    I am just concerned by the "second" kernel 32 bits entry point, in

    I still have a problem to detect the size of your decompressor, and that
    is my use of the "lss" instruction.
    This "lss SYMBOL_NAME(stack_start),%esp" gives an access to the symbol
    'stack_start', so it is quite easy to find back the GZIP signature
    of the initial '$$tmppiggy.gz' in what I call my "compatibility" mode,
    i.e. booting the legacy vmlinuz files - and skipping all of the real mode
    code and the decompressor code.

    This "lss" line has not always been at the same offset, but is around
    since maybe even the 0.01 kernel, it is quite easy to find it from its
    hexadecimal form. (function vmlinuz_header_treat() in vmlinuz.c of

    The loaded high/loaded low stuff is just to know if I have to remove
    0x100000 or 0x1000 from this symbol to have the number of bytes
    to skip on the file.
    By the way, the bit in the kernel header is set by the bootloader to say
    where it has loaded the kernel, not by the compiler/linker chain.

    So is it possible to write somewhere how much code to skip or the offset
    of the kernel GZIP signature?
    Something like:
    jmp next
    lss SYMBOL_NAME(stack_start),%esp
    Would make me really happy, but is dirty.
    Changing the 'tmppiggy.lnk' in the Makefile can be done, but the value
    (to know the length of the decompressor code) has to be _before_ the code
    itself in the raw file.
    Else whatever signature at whatever fixed address with the code+rodata
    size following would make me happy.

    Sorry again for the confusion,

    Do You Yahoo!? -- Une adresse gratuite et en français !
    Yahoo! Mail :
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2005-03-22 13:25    [W:0.023 / U:13.804 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site