Re: [PATCH -v3 2/2] x86: more early console output from compressed/misc.c

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

On 08/02/2010 10:49 AM, Cyrill Gorcunov wrote:
> On Mon, Aug 02, 2010 at 02:18:39AM -0700, Yinghai Lu wrote:
>>
>> will get
>>
>> |Decompressing Linux... Parsing ELF... done.
>> |Booting the kernel.
>>
>> in serial console.
>>
>> reuse code from arch/x86/boot/
>> and we can use printf if needed
>>
>> -v2: define BOOT_BOOT_H to avoid include boot.h
>> -v3: early_serial_base need to be static in misc.c ?
>>
>> Signed-off-by: Yinghai Lu <yinghai@kernel.org>
>>
>> ---
>>  arch/x86/boot/compressed/misc.c |   65 +++++++++++++++++++++++++++++++++++++++-
>>  arch/x86/boot/main.c            |    6 +--
>>  2 files changed, 66 insertions(+), 5 deletions(-)
>>
> 
> It seems eventually we just add ~80 lines of code in compare with what we had before,
> just to print "Decompressing..." by serial line. Not sure Yinghai if it worth
> it. Though having printf that early might be quite useful for those who debugging
> or developing new compressors/decompressors. If it would be small non-intrusive
> patch I would both hands for it but now I simply don't know (to be fair mine
> proposal based on your initial patch not better either ;)
> 

Decompressing... is one thing, and if it was only that, I'd just suggest
killing off the console I/O in boot/compressed.  Getting an error
message out when there is a decompression failure is another thing.

Processing the command line a novo is somewhat painful, but it does deal
better with bootloaders that use the 32-bit entry point by necessity
(kexec, EFI) or due to sheer stupidity (Grub2).

As such, I think it's a worthwhile addition, as long as the source code
can be cleanly shared with the boot/ directory.

	-hpa