Skip to contentSkip to conversation

gstreamer: fix unaligned detection for or1k

PATCH Superseded Peter Korsgaard 2 participants 8 messages Feb 18, 2017

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/3 gstreamer: fix unaligned detection for or1k

Similar to how we do it for the other niche/new architectures.

---
 package/gstreamer/gstreamer/gstreamer.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
package/gstreamer/gstreamer/gstreamer.mk +1 -1
index f2bde30d2..cde7f8f62 100644
@@ -15,7 +15,7 @@ GSTREAMER_LICENSE_FILES = COPYING
 # Checking if unaligned memory access works correctly cannot be done when cross
 # compiling. For the following architectures there is no information available
 # in the configure script.
-ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2),y)
+ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2)$(BR2_or1k),y)
 GSTREAMER_CONF_ENV = as_cv_unaligned_access=no
 endif
 ifeq ($(BR2_aarch64),y)
Signature · 1 line
2.11.0
#1 lore raw

PATCH 1/3 gstreamer: fix unaligned detection for or1k

Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Peter" == Peter Korsgaard <[email protected]> writes:

Similar to how we do it for the other niche/new architectures.

Committed, thanks.

Signature · 1 line
Bye, Peter Korsgaard
#2 lore raw

PATCH 2/3 gstreamer1: fix unaligned detection for or1k

In reply to Peter Korsgaard

Fixes:
http://autobuild.buildroot.net/results/717/717d78ce0935749f477bdf3133b6f20057a28c01/

Add patch to fix unaligned detection for or1k. Patch submitted upstream:

https://bugzilla.gnome.org/show_bug.cgi?id=778866

---
 ....in-Fix-unaligned-access-support-for-the-.patch | 33 ++++++++++++++++++++++
 package/gstreamer1/gstreamer1/gstreamer1.mk        |  2 +-
 2 files changed, 34 insertions(+), 1 deletion(-)
 create mode 100644 package/gstreamer1/gstreamer1/0001-gstconfig.h.in-Fix-unaligned-access-support-for-the-.patch
package/gstreamer1/gstreamer1/0001-gstconfig.h.in-Fix-unaligned-access-support-for-the-.patch new file +33 -0
new file mode 100644
index 000000000..88f337328
@@ -0,0 +1,33 @@
+From 437b3e38d9127ae021ca8510709c2e6be3e6b819 Mon Sep 17 00:00:00 2001
+From: Peter Korsgaard <[email protected]>
+Date: Sat, 18 Feb 2017 10:03:24 +0100
+Subject: [PATCH] gstconfig.h.in: Fix unaligned access support for the openrisc
+ architecture
+
+Teach gstconfig.h.in about the openrisc (or1k) architecture.  Fixes
+buildroot autobuild failure:
+
+http://autobuild.buildroot.net/results/717/717d78ce0935749f477bdf3133b6f20057a28c01/build-end.log
+
+Submitted-upstream: https://bugzilla.gnome.org/show_bug.cgi?id=778866
+Signed-off-by: Peter Korsgaard <[email protected]>
+---
+ gst/gstconfig.h.in | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/gst/gstconfig.h.in b/gst/gstconfig.h.in
+index cf95f61ab..1b2ca64ac 100644
+--- a/gst/gstconfig.h.in
++++ b/gst/gstconfig.h.in
+@@ -104,7 +104,7 @@
+  * http://docs.oracle.com/cd/E19205-01/820-4155/c++_faq.html#Vers6
+  * https://software.intel.com/en-us/node/583402
+  */
+-#if defined(__alpha__) || defined(__arc__) || defined(__arm__) || defined(__aarch64__) || defined(__bfin) || defined(__hppa__) || defined(__nios2__) || defined(__MICROBLAZE__) || defined(__mips__) || defined(__sh__) || defined(__SH4__) || defined(__sparc__) || defined(__sparc) || defined(__ia64__) || defined(_M_ALPHA) || defined(_M_ARM) || defined(_M_IA64) || defined(__xtensa__)
++#if defined(__alpha__) || defined(__arc__) || defined(__arm__) || defined(__aarch64__) || defined(__bfin) || defined(__hppa__) || defined(__nios2__) || defined(__MICROBLAZE__) || defined(__mips__) || defined(__or1k__) || defined(__sh__) || defined(__SH4__) || defined(__sparc__) || defined(__sparc) || defined(__ia64__) || defined(_M_ALPHA) || defined(_M_ARM) || defined(_M_IA64) || defined(__xtensa__)
+ #  define GST_HAVE_UNALIGNED_ACCESS 0
+ #elif defined(__i386__) || defined(__i386) || defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || defined(__ppc__) || defined(__ppc64__) || defined(__powerpc__) || defined(__powerpc64__) || defined(__m68k__) || defined(_M_IX86) || defined(_M_AMD64) || defined(_M_X64) || defined(__s390__) || defined(__s390x__) || defined(__zarch__)
+ #  define GST_HAVE_UNALIGNED_ACCESS 1
+-- 
+2.11.0
+
package/gstreamer1/gstreamer1/gstreamer1.mk +1 -1
index ac2c98870..7872ba13f 100644
@@ -14,7 +14,7 @@ GSTREAMER1_LICENSE = LGPLv2+, LGPLv2.1+
 # Checking if unaligned memory access works correctly cannot be done when cross
 # compiling. For the following architectures there is no information available
 # in the configure script.
-ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2),y)
+ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2)$(BR2_or1k),y)
 GSTREAMER1_CONF_ENV = as_cv_unaligned_access=no
 endif
 ifeq ($(BR2_aarch64),y)
Signature · 1 line
2.11.0
#3 lore raw

PATCH 3/3 gstreamer1: drop unused configure time unaligned access handling logic

In reply to Peter Korsgaard

gstreamer has moved to build time unaligned access support detection using ifdefs in gstconfig.h since 1.9.2:

commit 6ef601367e0f5adb7a8b02cec94dec04fccf701a

Author: Nirbheek Chauhan <[email protected]>
Date:   Tue Jun 21 18:59:49 2016 +0530

gstconfig.h: Detect unaligned access support at compile-time

This makes gstconfig.h completely arch-independent. Should cover all compilers that gstreamer is known to build on, and all architectures that I could find information on. People are encouraged to file bugs if their platform/arch is missing.

So ac_cv_unaligned_access is no longer used and our logic can be removed.

---
 package/gstreamer1/gstreamer1/gstreamer1.mk | 10 ----------
 1 file changed, 10 deletions(-)
package/gstreamer1/gstreamer1/gstreamer1.mk +0 -10
index 7872ba13f..e110dee7d 100644
@@ -11,16 +11,6 @@ GSTREAMER1_INSTALL_STAGING = YES
 GSTREAMER1_LICENSE_FILES = COPYING
 GSTREAMER1_LICENSE = LGPLv2+, LGPLv2.1+
 
-# Checking if unaligned memory access works correctly cannot be done when cross
-# compiling. For the following architectures there is no information available
-# in the configure script.
-ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2)$(BR2_or1k),y)
-GSTREAMER1_CONF_ENV = as_cv_unaligned_access=no
-endif
-ifeq ($(BR2_aarch64),y)
-GSTREAMER1_CONF_ENV = as_cv_unaligned_access=yes
-endif
-
 GSTREAMER1_CONF_OPTS = \
 	--disable-examples \
 	--disable-tests \
Signature · 1 line
2.11.0
#4 lore raw

PATCH 3/3 gstreamer1: drop unused configure time unaligned access handling logic

Peter, All,

On 2017-02-18 10:25 +0100, Peter Korsgaard spake thusly:

Quoted 7 lines

gstreamer has moved to build time unaligned access support detection using ifdefs in gstconfig.h since 1.9.2:

commit 6ef601367e0f5adb7a8b02cec94dec04fccf701a

Author: Nirbheek Chauhan <[email protected]>
Date:   Tue Jun 21 18:59:49 2016 +0530
Quoted 8 lines

gstconfig.h: Detect unaligned access support at compile-time

This makes gstconfig.h completely arch-independent. Should cover all compilers that gstreamer is known to build on, and all architectures that I could find information on. People are encouraged to file bugs if their platform/arch is missing.

So ac_cv_unaligned_access is no longer used and our logic can be removed.

In the previous patch, you add the or1k conditions to the if-block, but it is unused and you remove it here.

So I think this patch should come before you fix or1k, i.e. swap patches 2 and 3, no?

Regards,
Yann E. MORIN.

Quoted 33 lines
---
 package/gstreamer1/gstreamer1/gstreamer1.mk | 10 ----------
 1 file changed, 10 deletions(-)
package/gstreamer1/gstreamer1/gstreamer1.mk +0 -10
index 7872ba13f..e110dee7d 100644
@@ -11,16 +11,6 @@ GSTREAMER1_INSTALL_STAGING = YES
 GSTREAMER1_LICENSE_FILES = COPYING
 GSTREAMER1_LICENSE = LGPLv2+, LGPLv2.1+
 
-# Checking if unaligned memory access works correctly cannot be done when cross
-# compiling. For the following architectures there is no information available
-# in the configure script.
-ifeq ($(BR2_arc)$(BR2_xtensa)$(BR2_microblaze)$(BR2_nios2)$(BR2_or1k),y)
-GSTREAMER1_CONF_ENV = as_cv_unaligned_access=no
-endif
-ifeq ($(BR2_aarch64),y)
-GSTREAMER1_CONF_ENV = as_cv_unaligned_access=yes
-endif
-
 GSTREAMER1_CONF_OPTS = \
 	--disable-examples \
 	--disable-tests \
Signature · 6 lines
2.11.0

_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'
#5 lore raw

PATCH 3/3 gstreamer1: drop unused configure time unaligned access handling logic

Quoted 18 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Yann" == Yann E MORIN <[email protected]> writes:

Peter, All,
On 2017-02-18 10:25 +0100, Peter Korsgaard spake thusly:

Quoted 14 lines

gstreamer has moved to build time unaligned access support detection using ifdefs in gstconfig.h since 1.9.2:

commit 6ef601367e0f5adb7a8b02cec94dec04fccf701a

Author: Nirbheek Chauhan <[email protected]>
Date:   Tue Jun 21 18:59:49 2016 +0530

gstconfig.h: Detect unaligned access support at compile-time

This makes gstconfig.h completely arch-independent. Should cover all compilers that gstreamer is known to build on, and all architectures that I could find information on. People are encouraged to file bugs if their platform/arch is missing.

Quoted 7 lines
Quoted 1 line

So ac_cv_unaligned_access is no longer used and our logic can be removed.

In the previous patch, you add the or1k conditions to the if-block, but it is unused and you remove it here.

So I think this patch should come before you fix or1k, i.e. swap patches 2 and 3, no?

No, I think you are mixing up gstreamer and gstreamer1.

Patch 1 fixes gstreamer by adding or1k to the configure handling Patch 2 fixes gstreamer1 by adding a patch fixing gst/gstconfig.h Patch 3 cleans up gstreamer1 by removing the unused configure handling

Signature · 1 line
Bye, Peter Korsgaard
#6 lore raw

PATCH 3/3 gstreamer1: drop unused configure time unaligned access handling logic

Peter, All,

On 2017-02-18 11:56 +0100, Peter Korsgaard spake thusly:

Quoted 20 lines
Quoted 19 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Yann" == Yann E MORIN <[email protected]> writes:

Peter, All,
On 2017-02-18 10:25 +0100, Peter Korsgaard spake thusly:

Quoted 15 lines

gstreamer has moved to build time unaligned access support detection using ifdefs in gstconfig.h since 1.9.2:

commit 6ef601367e0f5adb7a8b02cec94dec04fccf701a

Author: Nirbheek Chauhan <[email protected]>
Date:   Tue Jun 21 18:59:49 2016 +0530

gstconfig.h: Detect unaligned access support at compile-time

This makes gstconfig.h completely arch-independent. Should cover all compilers that gstreamer is known to build on, and all architectures that I could find information on. People are encouraged to file bugs if their platform/arch is missing.

So ac_cv_unaligned_access is no longer used and our logic can be removed.

Quoted 7 lines
Quoted 5 lines

In the previous patch, you add the or1k conditions to the if-block, but it is unused and you remove it here.

So I think this patch should come before you fix or1k, i.e. swap patches 2 and 3, no?

No, I think you are mixing up gstreamer and gstreamer1.

No, I don't think so...

Quoted 2 lines

Patch 1 fixes gstreamer by adding or1k to the configure handling Patch 2 fixes gstreamer1 by adding a patch fixing gst/gstconfig.h

Here you patch package/gstreamer1/gstreamer1/gstreamer1.mk to also apply the as_cv_unaligned_access=no in the or1k case, but...

Quoted 1 line

Patch 3 cleans up gstreamer1 by removing the unused configure handling

... you remove it right here, in patch 3, since that passing as_cv_unaligned_access=no no longer works anyway.

So I question the fact that patch 2 is wrong (at least partly useless) in passing as_cv_unaligned_access=no when it is not used.

So, I would argue that we should do:

Patch 1 fixes gstreamer by adding or1k to the configure handling Patch 3 cleans up gstreamer1 by removing the unused configure handling Patch 2 fixes gstreamer1 by adding a patch fixing gst/gstconfig.h

Regards,
Yann E. MORIN.

Signature · 6 lines
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'
#7 lore raw

PATCH 3/3 gstreamer1: drop unused configure time unaligned access handling logic

Quoted 7 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line

"Yann" == Yann E MORIN <[email protected]> writes:

So I think this patch should come before you fix or1k, i.e. swap patches 2 and 3, no?

No, I think you are mixing up gstreamer and gstreamer1.

Quoted 7 lines

No, I don't think so...

Quoted 2 lines

Patch 1 fixes gstreamer by adding or1k to the configure handling Patch 2 fixes gstreamer1 by adding a patch fixing gst/gstconfig.h

Here you patch package/gstreamer1/gstreamer1/gstreamer1.mk to also apply the as_cv_unaligned_access=no in the or1k case, but...

Gaah, you are right - I apparently ended up doing a commit -a. I'll fix that.

Signature · 1 line
Bye, Peter Korsgaard