Patchwork libdrm: Remove Cairo dependency

login
register
mail settings
Submitter Ross Burton
Date Oct. 1, 2012, 11:56 a.m.
Message ID <1349092569-26895-1-git-send-email-ross.burton@intel.com>
Download mbox | patch
Permalink /patch/37517/
State Accepted
Commit a6d305261dc925210185d8b70fb1a923e012153b
Headers show

Comments

Ross Burton - Oct. 1, 2012, 11:56 a.m.
From: Daniel Stone <daniel@fooishbar.org>

This causes a build loop, when DRM depends on Cairo depends on Mesa
depends on DRM.  We can safely remove it as it's only one libdrm example
program which uses Cairo, which we won't be needing.  At least it's not
worth the build loop.

Signed-off-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
---
 meta/recipes-graphics/drm/libdrm.inc |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Martin Jansa - Oct. 1, 2012, 12:02 p.m.
On Mon, Oct 01, 2012 at 12:56:09PM +0100, Ross Burton wrote:
> From: Daniel Stone <daniel@fooishbar.org>
> 
> This causes a build loop, when DRM depends on Cairo depends on Mesa
> depends on DRM.  We can safely remove it as it's only one libdrm example
> program which uses Cairo, which we won't be needing.  At least it's not
> worth the build loop.
> 
> Signed-off-by: Daniel Stone <daniel@fooishbar.org>
> Signed-off-by: Ross Burton <ross.burton@intel.com>
> ---
>  meta/recipes-graphics/drm/libdrm.inc |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/meta/recipes-graphics/drm/libdrm.inc b/meta/recipes-graphics/drm/libdrm.inc
> index 5ddab85..cc09791 100644
> --- a/meta/recipes-graphics/drm/libdrm.inc
> +++ b/meta/recipes-graphics/drm/libdrm.inc
> @@ -9,9 +9,9 @@ LICENSE = "MIT"
>  LIC_FILES_CHKSUM = "file://xf86drm.c;beginline=9;endline=32;md5=c8a3b961af7667c530816761e949dc71"
>  SRC_URI = "http://dri.freedesktop.org/libdrm/libdrm-${PV}.tar.bz2"
>  PROVIDES = "drm"
> -DEPENDS = "libpthread-stubs udev cairo"
> +DEPENDS = "libpthread-stubs udev"
>  
> -INC_PR = "r2"
> +INC_PR = "r3"
>  
>  #libpciaccess is required starting from libdrm 2.4.26
>  DEPENDS += " libpciaccess"

shouldn't we disable that example explicitly, so that the output is not
different when cairo gets built before libdrm?

Cheers,
Ross Burton - Oct. 1, 2012, 12:04 p.m.
On 1 October 2012 13:02, Martin Jansa <martin.jansa@gmail.com> wrote:
> shouldn't we disable that example explicitly, so that the output is not
> different when cairo gets built before libdrm?

The test (tests/modetest/) is noinst, so the output doesn't actually change.

Ross
Richard Purdie - Oct. 2, 2012, 10:56 a.m.
On Mon, 2012-10-01 at 13:04 +0100, Burton, Ross wrote:
> On 1 October 2012 13:02, Martin Jansa <martin.jansa@gmail.com> wrote:
> > shouldn't we disable that example explicitly, so that the output is not
> > different when cairo gets built before libdrm?
> 
> The test (tests/modetest/) is noinst, so the output doesn't actually change.

Shouldn't we just disable the things to gain a small bit of build speed
if they don't matter? :)

Cheers,

Richard
Ross Burton - Oct. 2, 2012, 10:58 a.m.
On 2 October 2012 11:56, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> Shouldn't we just disable the things to gain a small bit of build speed
> if they don't matter? :)

That means carrying a patch and the maintenance burden when upgrading.
 I'm not entirely sure you'll notice all ten lines of Cairo code being
compiled if cairo happens to be installed when libdrm2 builds.

Ross
Richard Purdie - Oct. 2, 2012, 11:06 a.m.
On Tue, 2012-10-02 at 11:58 +0100, Burton, Ross wrote:
> On 2 October 2012 11:56, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
> > Shouldn't we just disable the things to gain a small bit of build speed
> > if they don't matter? :)
> 
> That means carrying a patch and the maintenance burden when upgrading.
>  I'm not entirely sure you'll notice all ten lines of Cairo code being
> compiled if cairo happens to be installed when libdrm2 builds.

I don't feel strongly about it but it sounds like the kind of thing it
would be useful to switch off and upstream would probably take such a
patch?

Cheers,

Richard
Saul Wold - Oct. 3, 2012, 1:42 a.m.
On 10/01/2012 04:56 AM, Ross Burton wrote:
> From: Daniel Stone <daniel@fooishbar.org>
>
> This causes a build loop, when DRM depends on Cairo depends on Mesa
> depends on DRM.  We can safely remove it as it's only one libdrm example
> program which uses Cairo, which we won't be needing.  At least it's not
> worth the build loop.
>
> Signed-off-by: Daniel Stone <daniel@fooishbar.org>
> Signed-off-by: Ross Burton <ross.burton@intel.com>
> ---
>   meta/recipes-graphics/drm/libdrm.inc |    4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/meta/recipes-graphics/drm/libdrm.inc b/meta/recipes-graphics/drm/libdrm.inc
> index 5ddab85..cc09791 100644
> --- a/meta/recipes-graphics/drm/libdrm.inc
> +++ b/meta/recipes-graphics/drm/libdrm.inc
> @@ -9,9 +9,9 @@ LICENSE = "MIT"
>   LIC_FILES_CHKSUM = "file://xf86drm.c;beginline=9;endline=32;md5=c8a3b961af7667c530816761e949dc71"
>   SRC_URI = "http://dri.freedesktop.org/libdrm/libdrm-${PV}.tar.bz2"
>   PROVIDES = "drm"
> -DEPENDS = "libpthread-stubs udev cairo"
> +DEPENDS = "libpthread-stubs udev"
>
> -INC_PR = "r2"
> +INC_PR = "r3"
>
>   #libpciaccess is required starting from libdrm 2.4.26
>   DEPENDS += " libpciaccess"
>
Merged into OE-Core

Thanks
	Sau!

Patch

diff --git a/meta/recipes-graphics/drm/libdrm.inc b/meta/recipes-graphics/drm/libdrm.inc
index 5ddab85..cc09791 100644
--- a/meta/recipes-graphics/drm/libdrm.inc
+++ b/meta/recipes-graphics/drm/libdrm.inc
@@ -9,9 +9,9 @@  LICENSE = "MIT"
 LIC_FILES_CHKSUM = "file://xf86drm.c;beginline=9;endline=32;md5=c8a3b961af7667c530816761e949dc71"
 SRC_URI = "http://dri.freedesktop.org/libdrm/libdrm-${PV}.tar.bz2"
 PROVIDES = "drm"
-DEPENDS = "libpthread-stubs udev cairo"
+DEPENDS = "libpthread-stubs udev"
 
-INC_PR = "r2"
+INC_PR = "r3"
 
 #libpciaccess is required starting from libdrm 2.4.26
 DEPENDS += " libpciaccess"