Re: [Qemu-devel] [Xen-devel] Hvmloader: Modify ACPI to only supply _EJ0 methods for PCIslots that support hotplug by runtime patching

[Markus Armbruster <armbru@redhat.com>]

Ian Campbell <Ian.Campbell@citrix.com> writes:

> On Wed, 2013-10-16 at 12:49 +0200, Markus Armbruster wrote:
>> Ian Campbell <Ian.Campbell@citrix.com> writes:
>> 
>> > On Wed, 2013-10-16 at 10:54 +0100, Jan Beulich wrote:
>> >> >>> On 16.10.13 at 08:30, "Gonglei (Arei)" <arei.gonglei@huawei.com> wrote:
>> >> > --- hvmloader/acpi/Makefile	2013-10-16 11:51:53.000000000 +0800
>> >> > +++ hvmloader_new//acpi/Makefile	2013-10-16 11:51:58.000000000 +0800
>> >> > @@ -36,18 +36,34 @@
>> >> >  
>> >> >  dsdt_anycpu_qemu_xen.asl: dsdt.asl mk_dsdt
>> >> >  	awk 'NR > 1 {print s} {s=$$0}' $< > $@
>> >> > -	./mk_dsdt --dm-version qemu-xen >> $@
>> >> > +	sed -i 's/AmlCode/dsdt_anycpu_qemu_xen/g' $@
>> >> 
>> >> This must never be done - if someone hits Ctrl-C in the middle of
>> >> this, you'll have a modified but incomplete generated file. You
>> >> either need to use properly chained rules, or do all output to a
>> >> temporary file which you rename as the last step.
>> >> 
>> >> I realize that the problem existed before your change, but you
>> >> making it worse requires doing it properly now.
>> >
>> > The correct way to do this is to generate to a temporary file and then
>> > use $(call move-if-changed,$@.tmp,$@). There are a bunch of examples in
>> > tools/libxl/Makefile and elsewhere.
>> 
>> Quoting the GNU make manual[*]:
>> 
>>     So generally the right thing to do is to delete the target file if
>>     the recipe fails after beginning to change the file.  make will do
>>     this if .DELETE_ON_ERROR appears as a target.  This is almost always
>>     what you want make to do, but it is not historical practice; so for
>>     compatibility, you must explicitly request it.
>> 
>> In my opinion, every Makefile should request it.
>> 
>> [*] https://www.gnu.org/software/make/manual/html_node/Errors.html
>
> I think this is somewhat orthogonal to the use of move-if-changed which
> is there to ensure that the timestamp doesn't change gratuitously and
> cause a load of knock on rebuilds even though the regenerated content is
> identical.

Yes.  I was latching onto the "if someone hits Ctrl-C in the middle of
this, you'll have a modified but incomplete generated file" phrase.