Skip to contentSkip to conversation

angstrom-2008.1: bump uclibc to 0.9.30.3 which unbreaks x264

PATCH In review Petr Štetiar 6 participants 17 messages Nov 26, 2010 – Jan 4, 2011

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 angstrom-2008.1: bump uclibc to 0.9.30.3 which unbreaks x264

Fixes the compile problem:

x264-r2245-r7/x264-snapshot-20100531-2245/encoder/set.c:236: undefined reference to log2f' libx264.a(encoder.o): In function x264_validate_parameters':

---
 conf/distro/angstrom-2008.1.conf |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
conf/distro/angstrom-2008.1.conf +2 -2
index 1fd5ab6..895deed 100644
@@ -78,8 +78,8 @@ PREFERRED_VERSION_linux-libc-headers-native = "${LINUX_LIBC_HEADERS_VERSION}"
 #Prefer glibc 2.6 and uclibc 0.9.30, these have had the most testing.
 PREFERRED_VERSION_glibc 			?= "2.9"
 PREFERRED_VERSION_glibc-initial 	?= "2.9"
-PREFERRED_VERSION_uclibc        	?= "0.9.30.2"
-PREFERRED_VERSION_uclibc-initial	?= "0.9.30.2"
+PREFERRED_VERSION_uclibc        	?= "0.9.30.3"
+PREFERRED_VERSION_uclibc-initial	?= "0.9.30.3"
 PREFERRED_VERSION_eglibc			?= "2.9"
 PREFERRED_VERSION_eglibc-initial	?= "2.9"
 
Signature · 1 line
1.7.0.4
#1 lore raw

conf/distro/angstrom-2008.1.conf PREFERRED_VERSION_linux-libc-headers-native = "${LINUX_LIBC_HEADERS_VERSION}" · line 82

Quoted 4 lines
conf/distro/angstrom-2008.1.conf +2 -2
-PREFERRED_VERSION_uclibc               ?= "0.9.30.2"
-PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.2"
+PREFERRED_VERSION_uclibc               ?= "0.9.30.3"
+PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.3"

Hi,

this would only fix x264/freempeg.

Unfortunately build will fail on e2fsprogs afterwards.
See http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2010-November/004257.html

So I suppose we should do a combined upgrade of uclibc and binutils. I'll test with current gcc-4.3.3 and let you know.

Andrea

#2 lore raw

Andrea Adami <[email protected]> [2010-11-26 11:53:50]:

Quoted 6 lines

conf/distro/angstrom-2008.1.conf PREFERRED_VERSION_linux-libc-headers-native = "${LINUX_LIBC_HEADERS_VERSION}" · line 82

Quoted 4 lines
conf/distro/angstrom-2008.1.conf +2 -2
-PREFERRED_VERSION_uclibc               ?= "0.9.30.2"
-PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.2"
+PREFERRED_VERSION_uclibc               ?= "0.9.30.3"
+PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.3"

Hi,

Hi,

Quoted 2 lines

Unfortunately build will fail on e2fsprogs afterwards.
See http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2010-November/004257.html

ok, so then I'll fix e2fsprogs :-)

Quoted 2 lines

So I suppose we should do a combined upgrade of uclibc and binutils. I'll test with current gcc-4.3.3 and let you know.

Ok, I don't mind how, just fix it please, it's all errors on release-2010.12 branch...

-- ynezz

#3 lore raw
In reply to Andrea Adami

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 7 lines · Andrea Adami

conf/distro/angstrom-2008.1.conf PREFERRED_VERSION_linux-libc-headers-native = "${LINUX_LIBC_HEADERS_VERSION}" · line 82

Quoted 4 lines
conf/distro/angstrom-2008.1.conf +2 -2
-PREFERRED_VERSION_uclibc               ?= "0.9.30.2"
-PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.2"
+PREFERRED_VERSION_uclibc               ?= "0.9.30.3"
+PREFERRED_VERSION_uclibc-initial       ?= "0.9.30.3"

Hi,

Quoted 7 lines

this would only fix x264/freempeg.

Unfortunately build will fail on e2fsprogs afterwards.
See http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2010-November/004257.html

So I suppose we should do a combined upgrade of uclibc and binutils. I'll test with current gcc-4.3.3 and let you know.

An upgrade of binutils is out of the question for angstrom 2008. -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFM75gYMkyGM64RGpERAo2GAJ9AQf8ixLASkpLnUwVhK0iRaYZgLACfQl/k hE6j+z4EssvPc0A+tiDN1OY=
=UYoM
-----END PGP SIGNATURE-----

#4 lore raw
Quoted 2 lines
Quoted 2 lines

So I suppose we should do a combined upgrade of uclibc and binutils. I'll test with current gcc-4.3.3 and let you know.

FWIW angstrom-2008.1 console, opie, x11, x11-gpe images build fine with binutils "2.20.1" and uclibc "git" like minimal distro but with gcc-4.3.3.

Quoted 1 line

An upgrade of binutils is out of the question for angstrom 2008.

I understand, too late now.
Shame on us for the lack of uclibc testing :/

Regards

Andrea

#7 lore raw
Quoted 4 lines · Andrea Adami

Ok, then I suggest to stay with 0.9.30.2 and add the two hacks documented here:

1) replacing log2f(i) with logf(i)/logf(2)
http://www.mail-archive.com/[email protected]/msg04949.html

Patch courtesy of Bernhard Guillon already done (bad that OE Patchwork is down...)
http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-November/027326.html

Quoted 3 lines

2) '--disable-tls' for e2fsprogs
http://patchwork.openembedded.org/patch/2069/

Here the patch is trivial...my concern is both should be Angstrom-only, isn't?

Andrea

#8 lore raw
Quoted 12 lines · Andrea Adami
Quoted 5 lines · Andrea Adami

Ok, then I suggest to stay with 0.9.30.2 and add the two hacks documented here:

1) replacing log2f(i) with logf(i)/logf(2)
http://www.mail-archive.com/[email protected]/msg04949.html

Patch courtesy of Bernhard Guillon already done (bad that OE Patchwork is down...)
http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-November/027326.html

Quoted 7 lines
Quoted 4 lines

2) '--disable-tls' for e2fsprogs
http://patchwork.openembedded.org/patch/2069/

Here the patch is trivial...my concern is both should be Angstrom-only, isn't?

This is true. We could add some python foo and check for angstrom? In my opinion the log2f patch is of no real harm. The tls support might be a problem.

Best regrads
Bernhard Guillon

#9 lore raw

Bernhard Guillon <[email protected]> [2010-12-01 12:41:31]:

Quoted 3 lines

This is true. We could add some python foo and check for angstrom? In my opinion the log2f patch is of no real harm. The tls support might be a problem.

Why to not move to some newer version of uClibc directly, without that patch at all?

-- ynezz

#10 lore raw
Quoted 8 lines · Petr Štetiar

Bernhard Guillon <[email protected]> [2010-12-01 12:41:31]:

Quoted 3 lines

This is true. We could add some python foo and check for angstrom? In my opinion the log2f patch is of no real harm. The tls support might be a problem.

Why to not move to some newer version of uClibc directly, without that patch at all?

because Angstrom 2008.1 is an already released distro. We don't like switching toolchains at last minute.

Graeme

#12 lore raw

2010/12/1 Andrea Adami <[email protected]>:

Quoted 8 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

This is true. We could add some python foo and check for angstrom? In my opinion the log2f patch is of no real harm. The tls support might be a problem.

About the tls:

http://bugs.gentoo.org/204102

Quoted 7 lines

... in e2fsprogs-1.40.4 the way uuids are generated has changed
(http://e2fsprogs.sourceforge.net/e2fsprogs-release.html#1.40.4).
e2fsprogs-1.40.3 works fine!

Perhaps Angstrom-2008.1 could pin 1.40.3

Andrea

You lost me here!

frans@frans-desktop:~/workspace/openembedded/recipes$ ls e2fs*
e2fsprogs_1.41.9.bb  e2fsprogs.inc

There is no 1.40.3 (git head master)

Frans

#13 lore raw
In reply to Petr Štetiar
Quoted 5 lines · Petr Štetiar

Bernhard Guillon <[email protected]> [2010-12-01 12:41:31]:

Quoted 3 lines

This is true. We could add some python foo and check for angstrom? In my opinion the log2f patch is of no real harm. The tls support might be a problem.

it seems so, khem hinted about git needing tls

Quoted 5 lines

Why to not move to some newer version of uClibc directly, without that patch at all?

-- ynezz

uclibc-0.9.30.3 needs binutils 2.20.1

Quoted 2 lines

Koen:
An upgrade of binutils is out of the question for angstrom 2008.

so here we stay

Andrea

#14 lore raw

PATCH 1/1 x264-r2245: add a hack for uclibc - replace log2f(x) with logf(x)/logf(2)

In reply to Andrea Adami

From: Bernhard Guillon <[email protected]>

---
 recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch |   61 ++++++++++++++++++++
 recipes/vlc/x264_r2245.bb                          |    6 +-
 2 files changed, 64 insertions(+), 3 deletions(-)
 create mode 100644 recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch
recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch new file +61 -0
new file mode 100644
index 0000000..874a0ca
@@ -0,0 +1,61 @@
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/analyse.c x264-snapshot-20100531-2245/encoder/analyse.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/analyse.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/analyse.c	2010-08-20 02:52:11.505161943 +0200
+@@ -251,7 +251,7 @@
+     for( int i = 0; i <= 2*4*2048; i++ )
+     {
+         h->cost_mv[lambda][-i] =
+-        h->cost_mv[lambda][i]  = lambda * (log2f(i+1)*2 + 0.718f + !!i) + .5f;
++        h->cost_mv[lambda][i]  = lambda * ( (logf(i+1)/logf(2))*2 + 0.718f + !!i) + .5f;
+     }
+     x264_pthread_mutex_lock( &cost_ref_mutex );
+     for( int i = 0; i < 3; i++ )
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/encoder.c x264-snapshot-20100531-2245/encoder/encoder.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/encoder.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/encoder.c	2010-08-20 02:53:58.516661776 +0200
+@@ -479,8 +479,8 @@
+     if( h->param.rc.i_rc_method == X264_RC_CQP )
+     {
+         float qp_p = h->param.rc.i_qp_constant;
+-        float qp_i = qp_p - 6*log2f( h->param.rc.f_ip_factor );
+-        float qp_b = qp_p + 6*log2f( h->param.rc.f_pb_factor );
++        float qp_i = qp_p - 6*(logf(h->param.rc.f_ip_factor)/logf(2));
++        float qp_b = qp_p + 6*(logf(h->param.rc.f_pb_factor)/logf(2));
+         h->param.rc.i_qp_min = x264_clip3( (int)(X264_MIN3( qp_p, qp_i, qp_b )), 0, 51 );
+         h->param.rc.i_qp_max = x264_clip3( (int)(X264_MAX3( qp_p, qp_i, qp_b ) + .999), 0, 51 );
+         h->param.rc.i_aq_mode = 0;
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/ratecontrol.c x264-snapshot-20100531-2245/encoder/ratecontrol.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/ratecontrol.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/ratecontrol.c	2010-08-20 02:55:26.752661912 +0200
+@@ -549,8 +549,8 @@
+         rc->last_non_b_pict_type = SLICE_TYPE_I;
+     }
+ 
+-    rc->ip_offset = 6.0 * log2f( h->param.rc.f_ip_factor );
+-    rc->pb_offset = 6.0 * log2f( h->param.rc.f_pb_factor );
++    rc->ip_offset = 6.0 * (logf( h->param.rc.f_ip_factor )/logf(2));
++    rc->pb_offset = 6.0 * (logf( h->param.rc.f_pb_factor )/logf(2));
+     rc->qp_constant[SLICE_TYPE_P] = h->param.rc.i_qp_constant;
+     rc->qp_constant[SLICE_TYPE_I] = x264_clip3( h->param.rc.i_qp_constant - rc->ip_offset + 0.5, 0, 51 );
+     rc->qp_constant[SLICE_TYPE_B] = x264_clip3( h->param.rc.i_qp_constant + rc->pb_offset + 0.5, 0, 51 );
+@@ -1154,7 +1154,7 @@
+             if( zone->b_force_qp )
+                 q += zone->i_qp - rc->qp_constant[SLICE_TYPE_P];
+             else
+-                q -= 6*log2f( zone->f_bitrate_factor );
++                q -= 6*(logf( zone->f_bitrate_factor )/logf(2));
+         }
+     }
+ 
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/set.c x264-snapshot-20100531-2245/encoder/set.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/set.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/set.c	2010-08-20 02:58:19.400661820 +0200
+@@ -234,7 +234,7 @@
+         sps->vui.i_max_bytes_per_pic_denom = 0;
+         sps->vui.i_max_bits_per_mb_denom = 0;
+         sps->vui.i_log2_max_mv_length_horizontal =
+-        sps->vui.i_log2_max_mv_length_vertical = (int)log2f( X264_MAX( 1, param->analyse.i_mv_range*4-1 ) ) + 1;
++        sps->vui.i_log2_max_mv_length_vertical = (int)(logf( X264_MAX( 1, param->analyse.i_mv_range*4-1 ))/logf(2))  + 1;
+     }
+ }
+ 
recipes/vlc/x264_r2245.bb +3 -3
index f533bfc..f685c63 100644
@@ -1,10 +1,12 @@
 DESCRIPTION = "H.264 encoder"
 LICENSE = "GPL"
-PR = "r7"
+PR = "r8"
 
 X264PV = "snapshot-20100531-2245"
 
 SRC_URI = "http://download.videolan.org/pub/videolan/x264/snapshots/x264-${X264PV}.tar.bz2"
+SRC_URI_append_libc-uclibc = "file://uclibc_log2f_fix.HACK.patch" 
+
 SRC_URI[md5sum] = "d1ccb8122bd418291a9576a2bffdf662"
 SRC_URI[sha256sum] = "929e946947701a0b3a336a4b9cfe65daf4c52480f45d4363335ae2a3d5596fa9"
 
@@ -24,5 +26,3 @@ X264_DISABLE_ASM_x86 = ""
 DEPENDS_x86 = "yasm-native"
 
 EXTRA_OECONF = '--enable-shared ${X264_DISABLE_ASM} --extra-cflags="${X264_ECFLAGS}"'
-
-
Signature · 1 line
1.7.1
#15 lore raw

Re: [PATCH] x264-r2245: add a hack for uclibc - replace log2f(x) with logf(x)/logf(2)

Thanks for this patch, have applied modified to apply for avr32 only as that platform cannot move toolchains.

Graeme

Quoted 101 lines · On 26/11/2010 23:22, [email protected] wrote:

From: Bernhard Guillon <[email protected]>

---
 recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch |   61 ++++++++++++++++++++
 recipes/vlc/x264_r2245.bb                          |    6 +-
 2 files changed, 64 insertions(+), 3 deletions(-)
 create mode 100644 recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch
recipes/vlc/x264-r2245/uclibc_log2f_fix.HACK.patch new file +61 -0
new file mode 100644
index 0000000..874a0ca
@@ -0,0 +1,61 @@
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/analyse.c x264-snapshot-20100531-2245/encoder/analyse.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/analyse.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/analyse.c	2010-08-20 02:52:11.505161943 +0200
+@@ -251,7 +251,7 @@
+     for( int i = 0; i <= 2*4*2048; i++ )
+     {
+         h->cost_mv[lambda][-i] =
+-        h->cost_mv[lambda][i]  = lambda * (log2f(i+1)*2 + 0.718f + !!i) + .5f;
++        h->cost_mv[lambda][i]  = lambda * ( (logf(i+1)/logf(2))*2 + 0.718f + !!i) + .5f;
+     }
+     x264_pthread_mutex_lock( &cost_ref_mutex );
+     for( int i = 0; i < 3; i++ )
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/encoder.c x264-snapshot-20100531-2245/encoder/encoder.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/encoder.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/encoder.c	2010-08-20 02:53:58.516661776 +0200
+@@ -479,8 +479,8 @@
+     if( h->param.rc.i_rc_method == X264_RC_CQP )
+     {
+         float qp_p = h->param.rc.i_qp_constant;
+-        float qp_i = qp_p - 6*log2f( h->param.rc.f_ip_factor );
+-        float qp_b = qp_p + 6*log2f( h->param.rc.f_pb_factor );
++        float qp_i = qp_p - 6*(logf(h->param.rc.f_ip_factor)/logf(2));
++        float qp_b = qp_p + 6*(logf(h->param.rc.f_pb_factor)/logf(2));
+         h->param.rc.i_qp_min = x264_clip3( (int)(X264_MIN3( qp_p, qp_i, qp_b )), 0, 51 );
+         h->param.rc.i_qp_max = x264_clip3( (int)(X264_MAX3( qp_p, qp_i, qp_b ) + .999), 0, 51 );
+         h->param.rc.i_aq_mode = 0;
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/ratecontrol.c x264-snapshot-20100531-2245/encoder/ratecontrol.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/ratecontrol.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/ratecontrol.c	2010-08-20 02:55:26.752661912 +0200
+@@ -549,8 +549,8 @@
+         rc->last_non_b_pict_type = SLICE_TYPE_I;
+     }
+ 
+-    rc->ip_offset = 6.0 * log2f( h->param.rc.f_ip_factor );
+-    rc->pb_offset = 6.0 * log2f( h->param.rc.f_pb_factor );
++    rc->ip_offset = 6.0 * (logf( h->param.rc.f_ip_factor )/logf(2));
++    rc->pb_offset = 6.0 * (logf( h->param.rc.f_pb_factor )/logf(2));
+     rc->qp_constant[SLICE_TYPE_P] = h->param.rc.i_qp_constant;
+     rc->qp_constant[SLICE_TYPE_I] = x264_clip3( h->param.rc.i_qp_constant - rc->ip_offset + 0.5, 0, 51 );
+     rc->qp_constant[SLICE_TYPE_B] = x264_clip3( h->param.rc.i_qp_constant + rc->pb_offset + 0.5, 0, 51 );
+@@ -1154,7 +1154,7 @@
+             if( zone->b_force_qp )
+                 q += zone->i_qp - rc->qp_constant[SLICE_TYPE_P];
+             else
+-                q -= 6*log2f( zone->f_bitrate_factor );
++                q -= 6*(logf( zone->f_bitrate_factor )/logf(2));
+         }
+     }
+ 
+diff -Nur x264-snapshot-20100531-2245.vanilla/encoder/set.c x264-snapshot-20100531-2245/encoder/set.c
+--- x264-snapshot-20100531-2245.vanilla/encoder/set.c	2010-05-31 22:45:11.000000000 +0200
++++ x264-snapshot-20100531-2245/encoder/set.c	2010-08-20 02:58:19.400661820 +0200
+@@ -234,7 +234,7 @@
+         sps->vui.i_max_bytes_per_pic_denom = 0;
+         sps->vui.i_max_bits_per_mb_denom = 0;
+         sps->vui.i_log2_max_mv_length_horizontal =
+-        sps->vui.i_log2_max_mv_length_vertical = (int)log2f( X264_MAX( 1, param->analyse.i_mv_range*4-1 ) ) + 1;
++        sps->vui.i_log2_max_mv_length_vertical = (int)(logf( X264_MAX( 1, param->analyse.i_mv_range*4-1 ))/logf(2))  + 1;
+     }
+ }
+ 
recipes/vlc/x264_r2245.bb +3 -3
index f533bfc..f685c63 100644
@@ -1,10 +1,12 @@
 DESCRIPTION = "H.264 encoder"
 LICENSE = "GPL"
-PR = "r7"
+PR = "r8"
 
 X264PV = "snapshot-20100531-2245"
 
 SRC_URI = "http://download.videolan.org/pub/videolan/x264/snapshots/x264-${X264PV}.tar.bz2"
+SRC_URI_append_libc-uclibc = "file://uclibc_log2f_fix.HACK.patch" 
+
 SRC_URI[md5sum] = "d1ccb8122bd418291a9576a2bffdf662"
 SRC_URI[sha256sum] = "929e946947701a0b3a336a4b9cfe65daf4c52480f45d4363335ae2a3d5596fa9"
 
@@ -24,5 +26,3 @@ X264_DISABLE_ASM_x86 = ""
 DEPENDS_x86 = "yasm-native"
 
 EXTRA_OECONF = '--enable-shared ${X264_DISABLE_ASM} --extra-cflags="${X264_ECFLAGS}"'
-
-
#16 lore raw
In reply to Andrea Adami
Quoted 2 lines · Andrea Adami

this would only fix x264/freempeg.

Hi, I just sent a patch to the list which fixes x264 with a hack that simply replaces all occurrences of log2f(x) with logf(x)/logf(2).

best regards
Bernhard Guillon