[Intel-gfx] [PATCH] Bump libdrm-intel dependency library to a newer version that support softpin.

[Emil Velikov]

On 14 July 2016 at 21:03, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> On Thu, Jul 14, 2016 at 07:54:59PM +0100, Emil Velikov wrote:
>> On 14 July 2016 at 17:49, Chris Wilson <chris at chris-wilson.co.uk> wrote:
>> > On Thu, Jul 14, 2016 at 05:29:55PM +0100, Emil Velikov wrote:
>> >> Hi Marius,
>> >>
>> >> Just double-checking - this is an i-g-t patch, isn't it ?
>> >>
>> >> On 14 July 2016 at 11:39, Marius Vlad <marius.c.vlad at intel.com> wrote:
>> >> > Required by commit 2603b98ca (aubdump: Support softpin bos).
>> >> >
>> >> > Signed-off-by: Marius Vlad <marius.c.vlad at intel.com>
>> >> > CC: Kristian Høgsberg Kristensen <kristian.h.kristensen at intel.com>
>> >> > ---
>> >> >  configure.ac | 2 +-
>> >> >  1 file changed, 1 insertion(+), 1 deletion(-)
>> >> >
>> >> > diff --git a/configure.ac b/configure.ac
>> >> > index f05bcb0..ade9756 100644
>> >> > --- a/configure.ac
>> >> > +++ b/configure.ac
>> >> > @@ -119,7 +119,7 @@ if test "x$GCC" = "xyes"; then
>> >> >  fi
>> >> >  AC_SUBST(ASSEMBLER_WARN_CFLAGS)
>> >> >
>> >> > -PKG_CHECK_MODULES(DRM, [libdrm_intel >= 2.4.64 libdrm])
>> >> > +PKG_CHECK_MODULES(DRM, [libdrm_intel >= 2.4.68 libdrm])
>> >> Yes please. As you do that one can nuke most/all the "define LOCAL_"
>> >> and "struct local_" (in lib/ioctl_wrappers.h)
>> >> and replace with with the official symbols. A very nice plus imho ;-)
>> >
>> > Please don't. It makes running on older installations even more
>> > cumbersome.
>> Slightly confused here: are you against the libdrm_intel bump, or the
>> removal of the local symbols ?
>
> Local symbols. They save a lot of time if you can just get the test you
> need compiling and not worry about dependencies. One of the basic
> tenents is that we drop a new kernel into an old userspace and expect to
> have not broken anything. Being lazy, for smoke testing I just build
> in situ.
>
Fully agree.

>> Admittedly sometimes distros don't bother/refuse to update libdrm
>> which could be an issue in the former case. Although if the package
>> (with all the definitions) is compulsory, how would that cause an
>> issue ?
>
> The package may not even exist when testing on v.old distro images.
> It is mostly a major of convenience, but since the work is already done
> to be independent, removing causes more work.
Seems that things are not completely independent, as illustrated by
the version bump.

Thus was wondering what's the benefit of (or why object against
nuking) the local copies. Since even if they were needed before, they
won't be after the bump.That said, it seems that it's a picky/sore
topic so I won't dig into it.

If anything I'll just send patches to be acked/nacked.
-Emil