Re: [Qemu-devel] [PATCH v3 22/22] qidl: unit tests and build infrastructure

[Michael Roth <mdroth@linux.vnet.ibm.com>]

On Mon, Oct 15, 2012 at 10:52:37AM +0200, Kevin Wolf wrote:
> Am 12.10.2012 23:39, schrieb Michael Roth:
> > On Fri, Oct 05, 2012 at 10:24:30AM +0200, Paolo Bonzini wrote:
> >> Il 04/10/2012 19:33, Michael Roth ha scritto:
> >>> +
> >>> +%.qidl.c: %.c $(SRC_PATH)/qidl.h $(addprefix $(SRC_PATH)/scripts/,lexer.py qidl.py qidl_parser.py qapi.py qapi_visit.py)
> >>> +	$(call rm -f $(*D)/qidl-generated/$(*F).qidl.c)
> >>> +	$(if $(strip $(shell grep "QIDL_ENABLE()" $< 1>/dev/null && echo "true")), \
> >>> +	  $(call quiet-command, \
> >>> +	    $(CC) $(QEMU_INCLUDES) $(QEMU_CFLAGS) $(CFLAGS) -E -c -DQIDL_GEN $< | \
> >>> +	    $(PYTHON) $(SRC_PATH)/scripts/qidl.py \
> >>> +	    --output-filepath=$(*D)/qidl-generated/$(*F).qidl.c || [ "$$?" -eq 2 ], \
> >>> +	    "qidl PP $(*D)/$(*F).c"),)
> >>> +%.o: %.c %.qidl.c
> >>> +	$(if $(strip $(shell test -f $(*D)/qidl-generated/$(*F).qidl.c && echo "true")), \
> >>> +	  $(call quiet-command, \
> >>> +	    $(CC) $(QEMU_INCLUDES) $(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) -c \
> >>> +		-DQIDL_ENABLED -include $< -o $@ $(*D)/qidl-generated/$(*F).qidl.c, \
> >>> +		"qidl CC $@"), \
> >>> +	  $(call quiet-command, \
> >>> +	    $(CC) $(QEMU_INCLUDES) $(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) -c \
> >>> +	      -o $@ $<,"  CC    $@"))
> >>
> >>
> >> Because the .qidl.c files are not created for files without a QIDL_ENABLE()
> >> directive, all those files will be grepped on every invocation of the makefile.
> >>
> >> It is better to define the list of QIDL_ENABLEd files in an auxiliary makefile
> >> like this (untested):
> > 
> > Thanks for the suggestion / example.
> > 
> > FYI, I tried to get this working but didn't manage to get it in for v4. Everything
> > seems to be falling back to the non-qidl %o: target, and I haven't had time to
> > debug it.
> > 
> > FWIW, it doesn't seem to be a major factor performance-wise. Current build
> > times on my laptop are (for all-target builds):
> > 
> > QIDL v4:
> >  real    8m35.383s
> >  user    31m1.844s
> >  sys     1m33.998s
> > 
> > upstream:
> >  real    8m28.181s
> >  user    30m44.983s
> >  sys     1m29.926s
> 
> Isn't the more interesting case that I only touched block/qcow2.c and
> run make to update my binaries? I haven't really looked at the code, but
> Paolo's comment suggests that even in this case a lot of files would be
> grepped, and I think that losing a few seconds in such cases would
> really hurt.

True, that's the more interesting case and I missed that point in
my response (had actually thought Paolo was driving at making
qemu-idl-files.mak a top-level target to avoid re-grepping on each
--target-list entry's build)

But Paolo posted some clarification in the v4 comments for this patch
on why this isn't the case for individual compilation units. The re-grep
would only occur for cases where the .c file changes, or something else
that might affect the output of the code generators changed
(scripts/(qidl|qapi)*, qidl.h, etc.). So in your example only
block/qcow2.c would be re-grepped.

> 
> Kevin
>