New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
media-libs/x265: Fix build on x32 by disabling assem… #173
Conversation
39fe282
to
8b8f2dc
Compare
8b8f2dc
to
1aae5e0
Compare
Commit message is too long. Put very short explanation on summary line, then explain the whole thing in detail in body, i.e.:
|
@gentoo/video |
1aae5e0
to
daee057
Compare
Pull requested updated as requested, upstream report made in videolan/x265#1 |
Errr, sorry for being picky, but I think pasting the build log in commit message is not a good idea ;-). Word the minimal description someone may need to understand the change. In particular, the description you gave on upstream commit is better than the one here ;-). |
Upstream videolan/x265#1 is not the official repo, report was made on https://bitbucket.org/multicoreware/x265/pull-requests/21/build-disable-march-selection-from/diff |
daee057
to
2c03843
Compare
Can you tell me if this is better now ? |
Yes, looks good. Just to be clear, what happens when you enable assembly on x32? Does it fail? I suggest you file a bug upstream about that too, they may want to update their asm, or implicitly disable it when unsupported ABI is used. |
Build fails, as reported upstream: https://bitbucket.org/multicoreware/x265/issues/198/assembly-code-fails-on-x32 |
2c03843
to
b8b5f1d
Compare
PR updaded for x265 1.8 |
b8b5f1d
to
182f180
Compare
lgtm, thanks! just one nitpick: it'd be good to have the bug number in a comment close to where you disable asm, as in my original suggestion: Unless asm optimizations are never intended to work on x32, I imagine a few years from now, people will wonder why this is disabled without warning :) |
@@ -54,6 +58,8 @@ multilib_src_configure() { | |||
if [ "${ABI}" = x86 ] ; then | |||
use 10bit && ewarn "Disabling 10bit support on x86 as it does not build (or requires to disable assembly optimizations)" | |||
mycmakeargs+=( -DHIGH_BIT_DEPTH=OFF ) | |||
elif [ "${ABI}" = x32 ] ; then |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If I may add my own nitpick, please use bash [[ ... ]]
. Then quoting becomes unnecessary:
elif [[ ${ABI} = x32 ]] ; then
If that wouldn't be a problem, please also change the other conditional.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
PR updated as per your comment
x32 arch as defined on https://sites.google.com/site/x32abi is neither X86 nor X64, then forcing -march=i686 leads to build failure as wrong -march is used. Forcing -march, -mfloat-abi and -mfpu for ARM is also wrong As a global sanity sake, disable all forced -march in CMakeLists Upstream report: https://bitbucket.org/multicoreware/x265/pull-requests/21/build-disable-march-selection-from/diff Package-Manager: portage-2.2.20.1
182f180
to
5658f1d
Compare
PR updated as per your comment |
if you wish to update per above nitpicks, feel free, otherwise it's good to go and I now see why you were using MULTILIB_ABI_FLAG: it is used in src_test... |
@bjacquin, your PGP key still seems to be missing on sks keyservers ;-). Please upload it. ( |
Hi Michał, I'm using a EdDSA GnuPG key and therefore keys.gnupg.net does not Cheers, On 14/10/2015 22:10, Michał Górny wrote:
Bertrand |
…bly code, patch to disable automagic march setup, bug #510890
Package-Manager: portage-2.2.20.1