CVS commit: pkgsrc/multimedia/libvpx

["Adam Ciarcinski" <adam%netbsd.org@localhost>]

Module Name:    pkgsrc
Committed By:   adam
Date:           Fri Oct  2 05:07:45 UTC 2026

Modified Files:
        pkgsrc/multimedia/libvpx: Makefile distinfo
        pkgsrc/multimedia/libvpx/patches: patch-build_make_configure.sh
Removed Files:
        pkgsrc/multimedia/libvpx/patches: patch-vp9_vp9__cx__iface.c

Log Message:
libvpx: updated to 1.17.0

1.17.0 "Yellowbilled Duck"
This release includes a new encoder control to validate high bitdepth input,
additional Neon and AVX2/AVX512 optimizations, support for the C11 standard,
and numerous bug fixes.

- Upgrading:
  This release is ABI compatible with the previous release.

  ISO C11 standard is now enforced for C files.

  SVC is now strictly disallowed for 2-pass encoding; enabling it will return
  VPX_CODEC_INVALID_PARAM.

  VP9 High Bitdepth (HBD) input validation is now enabled by default. If the
  input contains values outside the valid range, the encoder will return
  VPX_CODEC_INVALID_PARAM. Use the new VP9E_SET_VALIDATE_HBD_INPUT control
  (or vpxenc --validate-hbd-input=0) to disable this validation.

  VP9 now sets the returned image width (w) and height (h) to the actual
  width and height, rather than the stride and aligned height. This is a
  behavior change that may affect applications relying on w being equal to the
  stride or h being the aligned height.

  Strict validation has been added to some APIs:
  - Encoders (VP8/VP9) now reject input images with unequal U and V strides.
  - vpx_img_alloc() and vpx_img_wrap() now strictly validate the image format.
  - External rate control (RTC) API now validates configuration inputs.
  - VP9 SVC now validates spatial layer ID input.

- Enhancement:
  Added Arm Neon DotProd subpel variance paths.
  Optimized Neon HBD sadx4d and sad_skipx4d functions.
  Added AVX2/AVX512 implementations for various vpx_highbd predictor
  functions.
  Added support for MSVC v145 (Visual Studio 2026).
  Added configure option --disable-x86-asm.

- Bug fixes:
  Fix to heap buffer overflow / out-of-bounds write in vp9_postproc,
  vp8-multi-res-encoding, vpx_setup_noise, set_mb_ssim_rdmult_scaling,
  vp9_change_config (resolution change), vp9_dec_alloc_row_mt_mem (decode
  coeffs), and vp8_sixtap_predict (LoongArch).

  Fix to integer overflow in vp9-svc (layer resolution, duration),
  set_mb_ssim_rdmult_scaling, vp8_encode_frame, calc_pframe_target_size,
  y4minput buffer size calculations, and get_token_alloc.

  Fix to VP8 decoder crash on flush with no fragments.
  Fix to VP8 decoder potential hang/slowdown with corrupt partition 0 size.
  Fix to VP9 decoder crash with invalid reference frame index.
  Fix to VP9 highbitdepth encoder format mismatch.
  Fix to VP9 float-to-int overflow in validate_config (EOS count) and with
  zero bitrate.

  Fix to VP8 decoder multithreading corruption flag reset at start of
  frame.
  Fix to VP8 decoder accepting inter frames on resolution change before
  keyframe.
  Fix to VP8 encoder reference frame dimension validation.
  Fix to VP8 encoder temporal layer ID validation.
  Fix to VP8 encoder missing setjmp in VP8E_SET_SCALEMODE.

  Fix to VP9 SVC encoder crash with dynamic resize and cyclic refresh
  (aq_mode=3).
  Fix to VP9 1-pass rate control last_boosted_qindex clamping and
  VP9E_SET_QUANTIZER_ONE_PASS validation.

  Fix to NV12 x_chroma_shift calculation in vpx_image and image2yuvconfig.
  Fix to AVX-512 CPU feature detection on older x86 CPUs.
  Fix to x86_abi_support.asm placing read-only data tables in .rdata for
  Windows.
  Fix to configure disabling _FORTIFY_SOURCE on Linux.
  Fix to file leaks on error paths in vpxdec and video reader/writer
  utilities.

  Fix to crash/undefined behavior in vpx_img_flip with formats without alpha.


To generate a diff of this commit:
cvs rdiff -u -r1.111 -r1.112 pkgsrc/multimedia/libvpx/Makefile
cvs rdiff -u -r1.57 -r1.58 pkgsrc/multimedia/libvpx/distinfo
cvs rdiff -u -r1.4 -r1.5 \
    pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh
cvs rdiff -u -r1.1 -r0 \
    pkgsrc/multimedia/libvpx/patches/patch-vp9_vp9__cx__iface.c

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.

Modified files:

Index: pkgsrc/multimedia/libvpx/Makefile
diff -u pkgsrc/multimedia/libvpx/Makefile:1.111 pkgsrc/multimedia/libvpx/Makefile:1.112
--- pkgsrc/multimedia/libvpx/Makefile:1.111     Fri Mar 27 21:35:19 2026
+++ pkgsrc/multimedia/libvpx/Makefile   Fri Oct  2 05:07:45 2026
@@ -1,7 +1,6 @@
-# $NetBSD: Makefile,v 1.111 2026/03/27 21:35:19 tnn Exp $
+# $NetBSD: Makefile,v 1.112 2026/10/02 05:07:45 adam Exp $
 
-DISTNAME=      libvpx-1.16.0
-PKGREVISION=   1
+DISTNAME=      libvpx-1.17.0
 CATEGORIES=    multimedia
 MASTER_SITES=  ${MASTER_SITE_GITHUB:=webmproject/}
 GITHUB_TAG=    v${PKGVERSION_NOREV}

Index: pkgsrc/multimedia/libvpx/distinfo
diff -u pkgsrc/multimedia/libvpx/distinfo:1.57 pkgsrc/multimedia/libvpx/distinfo:1.58
--- pkgsrc/multimedia/libvpx/distinfo:1.57      Fri Sep 18 17:53:11 2026
+++ pkgsrc/multimedia/libvpx/distinfo   Fri Oct  2 05:07:45 2026
@@ -1,11 +1,10 @@
-$NetBSD: distinfo,v 1.57 2026/09/18 17:53:11 agc Exp $
+$NetBSD: distinfo,v 1.58 2026/10/02 05:07:45 adam Exp $
 
-BLAKE2s (libvpx-1.16.0.tar.gz) = 17341f5c9ce829528b4df6b3287470492041fbea5de712c19459102dfe35cb41
-SHA512 (libvpx-1.16.0.tar.gz) = 07f5e352411d6c0be331706d1835ac89bafbeddcbbac5542b473323766e9e974f4f68b33590f2aa50a7d8d69468a642b508cbb0a7c49a82c9933b07820f9c9d9
-Size (libvpx-1.16.0.tar.gz) = 5635379 bytes
+BLAKE2s (libvpx-1.17.0.tar.gz) = cce90acd3e7bc91a12d6d4ad67c92a25f2e4b4c07955a1d8e8dbf6afc4b98e8e
+SHA512 (libvpx-1.17.0.tar.gz) = be98d1b913575e538fefefe657c9944df85118fe2d0558361495a72c75ac6a2251048cb645fca49e60d6941c823a923f31d4ebc4de9b778008c267c42805b339
+Size (libvpx-1.17.0.tar.gz) = 5670546 bytes
 SHA1 (patch-build_make_Makefile) = d72c391b66a89c883664450b13cecdf61c64b52e
-SHA1 (patch-build_make_configure.sh) = ef4247ed3712ed81654f465f813160685dc09e8b
+SHA1 (patch-build_make_configure.sh) = e54b3b6f218a3859c155b94046fed5428f1bfbcb
 SHA1 (patch-configure) = aeb5bfd9d58b06b4f2fdbdb8c73b03339de313e7
 SHA1 (patch-examples.mk) = 17410f43ff9952d616be3211ca697f37c107610a
 SHA1 (patch-libs.mk) = 4fe233a421ee6f998b2cd0328b66b1d759706a5f
-SHA1 (patch-vp9_vp9__cx__iface.c) = 9a3e4e2c68f2a6aede22c502b07450a7f5d43e48

Index: pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh
diff -u pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh:1.4 pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh:1.5
--- pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh:1.4  Mon Jan 22 13:09:26 2024
+++ pkgsrc/multimedia/libvpx/patches/patch-build_make_configure.sh      Fri Oct  2 05:07:45 2026
@@ -1,4 +1,4 @@
-$NetBSD: patch-build_make_configure.sh,v 1.4 2024/01/22 13:09:26 ryoon Exp $
+$NetBSD: patch-build_make_configure.sh,v 1.5 2026/10/02 05:07:45 adam Exp $
 
 *BSD and qnx are identified as linux.
 Add another SDK path on Mac OS X.
@@ -6,9 +6,9 @@ All sparc cpus can not do unaligned acce
 Detect NetBSD ARMv7 hardfloat toolchain.
 Recognize powerpc as a target ISA, so we don't end up with generic-gnu
 
---- build/make/configure.sh.orig       2024-01-09 21:12:22.000000000 +0000
+--- build/make/configure.sh.orig       2026-08-20 19:06:35.000000000 +0000
 +++ build/make/configure.sh
-@@ -752,7 +752,7 @@ process_common_toolchain() {
+@@ -792,7 +792,7 @@ process_common_toolchain() {
        aarch64*)
          tgt_isa=arm64
          ;;
@@ -17,7 +17,7 @@ Recognize powerpc as a target ISA, so we
          tgt_isa=armv7
          float_abi=hard
          ;;
-@@ -784,6 +784,9 @@ process_common_toolchain() {
+@@ -824,6 +824,9 @@ process_common_toolchain() {
        loongarch64*)
          tgt_isa=loongarch64
          ;;
@@ -27,7 +27,7 @@ Recognize powerpc as a target ISA, so we
      esac
  
      # detect tgt_os
-@@ -806,7 +809,7 @@ process_common_toolchain() {
+@@ -846,7 +849,7 @@ process_common_toolchain() {
          [ -z "$tgt_isa" ] && tgt_isa=x86
          tgt_os=win32
          ;;
@@ -36,7 +36,7 @@ Recognize powerpc as a target ISA, so we
          tgt_os=linux
          ;;
        *solaris2.10)
-@@ -861,6 +864,9 @@ process_common_toolchain() {
+@@ -901,6 +904,9 @@ process_common_toolchain() {
        soft_enable lasx
        enable_feature loongarch
        ;;
@@ -46,7 +46,7 @@ Recognize powerpc as a target ISA, so we
    esac
  
    # Position independent code (PIC) is probably what we want when building
-@@ -1533,7 +1539,7 @@ EOF
+@@ -1610,7 +1616,7 @@ EOF
    check_cc <<EOF
  unsigned int e = 'O'<<24 | '2'<<16 | 'B'<<8 | 'E';
  EOF
@@ -55,8 +55,8 @@ Recognize powerpc as a target ISA, so we
          grep '4f *32 *42 *45' >/dev/null 2>&1 && enable_feature big_endian
  
      # Try to find which inline keywords are supported
-@@ -1550,7 +1556,7 @@ EOF
-         # bionic includes basic pthread functionality, obviating -lpthread.
+@@ -1628,7 +1634,7 @@ EOF
+         soft_enable pthread_setname_np
          ;;
        *)
 -        check_header pthread.h && check_lib -lpthread <<EOF && add_extralibs -lpthread || disable_feature pthread_h