[Beignet] [Intel-gfx] [PATCH i-g-t 2/2] configure: Bump required libdrm version to 2.4.60

[Daniel Vetter]

On Tue, Mar 10, 2015 at 01:06:44PM -0700, Jeff McGee wrote:
> On Tue, Mar 10, 2015 at 07:47:03PM +0100, Daniel Vetter wrote:
> > On Tue, Mar 10, 2015 at 01:58:52PM -0400, Rob Clark wrote:
> > > On Tue, Mar 10, 2015 at 12:59 PM, Jeff McGee <jeff.mcgee at intel.com> wrote:
> > > > On Tue, Mar 10, 2015 at 08:37:30AM +0100, Daniel Vetter wrote:
> > > >> On Mon, Mar 09, 2015 at 04:41:02PM -0700, jeff.mcgee at intel.com wrote:
> > > >> > From: Jeff McGee <jeff.mcgee at intel.com>
> > > >> >
> > > >> > tests/core_getparams needs the new libdrm interfaces for
> > > >> > querying subslice and EU counts.
> > > >> >
> > > >> > For: VIZ-4636
> > > >> > Signed-off-by: Jeff McGee <jeff.mcgee at intel.com>
> > > >> > ---
> > > >> >  configure.ac | 2 +-
> > > >> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > >> >
> > > >> > diff --git a/configure.ac b/configure.ac
> > > >> > index 16d6a2e..88a1c3d 100644
> > > >> > --- a/configure.ac
> > > >> > +++ b/configure.ac
> > > >> > @@ -82,7 +82,7 @@ if test "x$GCC" = "xyes"; then
> > > >> >  fi
> > > >> >  AC_SUBST(ASSEMBLER_WARN_CFLAGS)
> > > >> >
> > > >> > -PKG_CHECK_MODULES(DRM, [libdrm_intel >= 2.4.52 libdrm])
> > > >> > +PKG_CHECK_MODULES(DRM, [libdrm_intel >= 2.4.60 libdrm])
> > > >>
> > > >> Please don't and instead copypaste the new structs/defines with a local_
> > > >> prefix like we do it for all the other new igt testcases. Forcing libdrm
> > > >> to get updated for igt all the time can get annoying fast.
> > > >> -Daniel
> > > >>
> > > > In this case I'm trying to exercise new API functions in libdrm which
> > > > wrap the GETPARAM ioctl. Would you rather me bypass the wrapper to
> > > > avoid requiring updated libdrm? I can do that, but it fails to test the
> > > > complete path that client would use.
> > > 
> > > 
> > > Am I missing something, or does 2.4.60 not exist yet?
> > > 
> > > That said dependency bumps for igt seem like less of an issue than
> > > dependency bumps for mesa..  I mean if you are using igt you are
> > > probably on the latest anyways..  I'm not sure why Daniel is so
> > > concerned about that..
> > > 
> > > (but dependency bumps to something that doesn't exist yet should
> > > perhaps be avoided)
> > 
> > I'd like to avoid massive depency loops for igt tests so that I can merge
> > the testcase right when the patches land in -nightly. Otherwise there's
> > always a small delay involved where regression can creep in. Also if I
> > have to update libdrm every time I update igt that's annoying since
> > without that I don't have to install/update anything at all - I run igt
> > in-place. And we've used the LOCAL_ prefixes for pretty much every abi
> > addition in igt tests thus far.
> > -Daniel
> 
> I understand that and it certainly makes sense when libdrm is only
> providing defines or structs. But as I said, in this case there is
> code in libdrm (the wrapper) that we could test as part of the
> complete path. Are you recommending that I implement duplicate
> wrapper functions in igt with the local prefix?

Sorry I totally didn't realize that. Generally we don't have a lot of igt
testcase for libdrm really, imo it's usually simpler to just add the
interface to each part. Since this is such a simple one there's no need to
have a low-level test and the libdrm test on top, but that's what I'd do
if there's something worth testing in libdrm. Because for complex
functionality I really want to get the bare-metal tests in together with
the kernel part. Stalling for libdrm release could take longer.

And yes, personally I'd just have open-coded the getparam call here in
igt, but that's just a bikeshed.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch