[bug#42678] [PATCH, WIP] gnu: libimagequant: Update to 2.12.6

[bug#42678] [PATCH, WIP] gnu: libimagequant: Update to 2.12.6

[Hendur Saga via Guix-patches via]

[-- Attachment #1.1: Type: text/plain, Size: 2998 bytes --]

Hello,

Just a simple update, but sadly, I am unable to build using QEMU for other platforms. Same as with previous version, it does NOT build correctly on armhf.


starting phase `configure'
source directory: "/tmp/guix-build-libimagequant-2.12.6.drv-0/source" (relative from build: ".")
build directory: "/tmp/guix-build-libimagequant-2.12.6.drv-0/source"
configure flags: ("CONFIG_SHELL=/gnu/store/7zp9ifpgm3zj481nk6jg1im13g4mza2g-bash-minimal-5.0.16/bin/bash" "SHELL=/gnu/store/7zp9ifpgm3zj481nk6jg1im13g4mza2g-bash-minimal-5.0.16/bin/bash" "--prefix=/gnu/store/m65h6a45pz3wpraiid99hqq0rcbk5s9w-libimagequant-2.12.6" "--enable-fast-install" "--build=arm-unknown-linux-gnueabihf")
warning: unknown switch 'CONFIG_SHELL' (see ./configure --help for the list)
warning: unknown switch 'SHELL' (see ./configure --help for the list)
warning: unknown switch '--enable-fast-install' (see ./configure --help for the list)
warning: unknown switch '--build' (see ./configure --help for the list)

  Compiler: gcc
     Debug: no
       SSE: yes
    OpenMP: no

phase `configure' succeeded after 3.3 seconds
starting phase `patch-generated-file-shebangs'
phase `patch-generated-file-shebangs' succeeded after 0.0 seconds
starting phase `build'
gcc -fno-math-errno -funroll-loops -fomit-frame-pointer -Wall -std=c99 -I. -O3 -DNDEBUG -DUSE_SSE=1 -msse -Wno-unknown-pragmas -fexcess-precision=fast   -c -o pam.o pam.c
gcc -fno-math-errno -funroll-loops -fomit-frame-pointer -Wall -std=c99 -I. -O3 -DNDEBUG -DUSE_SSE=1 -msse -Wno-unknown-pragmas -fexcess-precision=fast   -c -o mediancut.o mediancut.c
gcc -fno-math-errno -funroll-loops -fomit-frame-pointer -Wall -std=c99 -I. -O3 -DNDEBUG -DUSE_SSE=1 -msse -Wno-unknown-pragmas -fexcess-precision=fast   -c -o blur.o blur.c
gcc -fno-math-errno -funroll-loops -fomit-frame-pointer -Wall -std=c99 -I. -O3 -DNDEBUG -DUSE_SSE=1 -msse -Wno-unknown-pragmas -fexcess-precision=fast   -c -o mempool.o mempool.c
gcc: error: unrecognized command line option ‘-msse’; did you mean ‘-fdse’?
make: *** [<builtin>: pam.o] Error 1
make: *** Waiting for unfinished jobs....
gcc: error: unrecognized command line option ‘-msse’; did you mean ‘-fdse’?
gcc: error: unrecognized command line option ‘-msse’; did you mean ‘-fdse’?
make: *** [<builtin>: mediancut.o] Error 1
make: *** [<builtin>: blur.o] Error 1
gcc: error: unrecognized command line option ‘-msse’; did you mean ‘-fdse’?
make: *** [<builtin>: mempool.o] Error 1
command "make" "-j" "4" failed with status 2


It appears to be an false assumption of SSE2 support where there is none? In addition, the same error happens when building this new version on aarch64, but not the previous one. The package definition for gemmlowp might have a clue as to what to do here.

Since QEMU is a machine emulator/virtualizer, I'm not sure how relevant this is, but my machine's CPU does not have SSE2 itself, AFAIK.

- Hendursaga

[-- Attachment #1.2: Type: text/html, Size: 3409 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-gnu-libimagequant-Update-to-2.12.6.patch --]
[-- Type: text/x-patch, Size: 1208 bytes --]

From 13d5468c407ef569b343ff1e7cbdb6be2bfc5f3c Mon Sep 17 00:00:00 2001
From: Hendursaga <hendursaga@yahoo.com>
Date: Sun, 2 Aug 2020 15:59:55 +0000
Subject: [PATCH] gnu: libimagequant: Update to 2.12.6

* gnu/packages/image.scm (libimagequant): Update to 2.12.6
---
 gnu/packages/image.scm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gnu/packages/image.scm b/gnu/packages/image.scm
index ac523172de..159e75150f 100644
--- a/gnu/packages/image.scm
+++ b/gnu/packages/image.scm
@@ -304,7 +304,7 @@ Currently all documentation resides in @file{pnglite.h}.")
 (define-public libimagequant
   (package
     (name "libimagequant")
-    (version "2.12.5")
+    (version "2.12.6")
     (source
      (origin
        (method git-fetch)
@@ -313,7 +313,7 @@ Currently all documentation resides in @file{pnglite.h}.")
              (commit version)))
        (file-name (git-file-name name version))
        (sha256
-        (base32 "0cp68w04ja5pv77ssfafsn958w9hh9zb8crrlb5j3gsrcmdc032k"))))
+        (base32 "00w7fny3xf14cfyhbdnmqyh9ddqdh1irvgzxd35a2z65kp7vnvj0"))))
     (build-system gnu-build-system)
     (arguments
      '(#:tests? #f))                    ; no check target
-- 
2.28.0

[bug#42678] [PATCH, WIP] gnu: libimagequant: Update to 2.12.6

[Mathieu Othacehe]

Hello,

> Just a simple update, but sadly, I am unable to build using QEMU for other platforms. Same as with previous version, it does NOT build correctly on armhf.

You mean using "-s armhf-linux"?

I tried to build it using this command:

--8<---------------cut here---------------start------------->8---
guix build libimagequant --target=arm-linux-gnueabihf
--8<---------------cut here---------------end--------------->8---

and I have a build trace similar to yours. The issue seems to be that
the configure script does not detect our intent to cross-compile,
because it uses "gcc" instead of "arm-linux-gnueabihf-gcc".

Mathieu

[bug#42678] [PATCH, WIP] gnu: libimagequant: Update to 2.12.6

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 1176 bytes --]

On Mon, Aug 03, 2020 at 09:26:41AM +0200, Mathieu Othacehe wrote:
> 
> Hello,
> 
> > Just a simple update, but sadly, I am unable to build using QEMU for other platforms. Same as with previous version, it does NOT build correctly on armhf.
> 
> You mean using "-s armhf-linux"?
> 
> I tried to build it using this command:
> 
> --8<---------------cut here---------------start------------->8---
> guix build libimagequant --target=arm-linux-gnueabihf
> --8<---------------cut here---------------end--------------->8---
> 
> and I have a build trace similar to yours. The issue seems to be that
> the configure script does not detect our intent to cross-compile,
> because it uses "gcc" instead of "arm-linux-gnueabihf-gcc".
> 
> Mathieu
> 

Also armhf and aarch64 don't have the SSE instruction code as part of
the architecture. lensfun in gnu/packages/photo.scm has an example of
disabling SSE on all platforms except for x64_64 and i686.

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#42678: [PATCH, WIP] gnu: libimagequant: Update to 2.12.6

[Jelle Licht]

Superseded by d6f1e8135676a4d38fda9927434999312b60ba21 on master. Closing.