Re: call for testing: xen 4.1 packages

["Toby Karyadi" <toby.karyadi%gmail.com@localhost>]

On 3/31/2011, "Christoph Egger" <Christoph_Egger%gmx.de@localhost> wrote:

>>
>> I found it:
>>
>> In xen/arch/x86/boot/build32.mk there are these lines:
>>
>> # NB. awk invocation is a portable alternative to 'head -n -1'
>> %.S: %.bin
>>         (od -v -t x $< | awk 'NR > 1 {print s} {s=$$0}' | \
>>         sed 's/ /,0x/g' | sed 's/^[0-9]*,/ .long /') >$@
>>
>> These generate different output on NetBSD than on Linux.
>> That is why xen boots when compiled on Linux host.
>>
>> For comparison have a look at tools/firmware/hvmloader/32bit/mkhex
>> where this generates the same output on both NetBSD and Linux:
>>
>> echo "unsigned $1[] = {"
>> od -v -t x $2 | sed 's/^[0-9]*  */0x/' | sed 's/  */, 0x/g' | sed
>> 's/$/,/' | sed 's/0x,//' | sed 's/^[0-9]*,//'
>> echo "};"
>
>I think, I fixed the boot problem now. The generated reloc.S in
>xen/arch/x86/boot/  was just garbage.
>
>Christoph

Yup you did! I can now boot successfully using my 5.99.45 dom0 kernel
(haven't been able to start any domain though, but that's a different
story). Doesn't it make you wonder how many more of these problems
exist due to bsd vs gnu tools differences? Looking at your patch-aa
seems to me that od produces slightly different output, yes?

Anyway, thanks for the quick fix!

Toby