Summary: | www-client/chromium-44.0.2383.0 doesn't compile with gcc-5.1 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Helmut Jarausch <jarausch> |
Component: | [OLD] GCC Porting | Assignee: | Chromium Project <chromium> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | alexxy, genzilla, hendrik, lucy, octoploid, rhill, robink, tdalman |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | 548412 | ||
Bug Blocks: | 536984 | ||
Attachments: |
build log xz-compressed
proposed work around patch from comment 4 of upstream bug 466760 chromium-42.0.2311.135-gcc-5.1.patch |
Description
Helmut Jarausch
2015-05-01 09:29:55 UTC
Will be fixed in 5.2: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65801 (In reply to octoploid from comment #1) > Will be fixed in 5.2: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65801 Will there be a 5.9999 ebuild (since it might take some time for 5.2 beeing released) ? (In reply to Helmut Jarausch from comment #2) > (In reply to octoploid from comment #1) > > Will be fixed in 5.2: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65801 > > Will there be a 5.9999 ebuild (since it might take some time for 5.2 beeing > released) ? I don't know. But it might be a good idea to add the patch to 5.1. Created attachment 402432 [details, diff] proposed work around patch from comment 4 of upstream bug 466760 I reproduced the problem with www-client/chromium-43.0.2357.18. There are some upstream bugs: https://code.google.com/p/cld2/issues/detail?id=26 https://code.google.com/p/chromium/issues/detail?id=466760 I have attached the "rather questionable patches" from comment 4: https://code.google.com/p/chromium/issues/detail?id=466760#c4 The patch is applying changes to generated code. The patch is for 44.0.2387.0 I am currently testing it with www-client/chromium-43.0.2357.18 with gcc 5.1.0. argus / # diff -ur /usr/portage/www-client/chromium/chromium-43.0.2357.18.ebuild /usr/local/portage/www-client/chromium/chromium-43.0.2357.18.ebuild --- /usr/portage/www-client/chromium/chromium-43.0.2357.18.ebuild 2015-04-20 12:31:19.000000000 +1000 +++ /usr/local/portage/www-client/chromium/chromium-43.0.2357.18.ebuild 2015-05-02 19:04:55.097170293 +1000 @@ -193,6 +193,7 @@ epatch "${FILESDIR}/${PN}-system-jinja-r7.patch" epatch "${FILESDIR}/${PN}-libsecret-r0.patch" epatch "${FILESDIR}/${PN}-system-libvpx-r0.patch" + epatch "${FILESDIR}/${PN}-44.0.2384.0-gcc-5.1.patch" if use widevine; then local WIDEVINE_VERSION="$(< "${ROOT}/usr/$(get_libdir)/chromium-browser/widevine.version")" argus / # In addtion, I have patched gcc-5.1.0, see https://bugs.gentoo.org/show_bug.cgi?id=548412 With this patch and the proposed work-around here, I could emerge chromium-44.0.2383.0 without any problems. The "fix" of casting to unsigned char is pretty ugly. I think we will just wait for the next gcc release unless chromium upstream applies some fix. The patch below seems to do the job. https://gcc.gnu.org/viewcvs/gcc/branches/gcc-5-branch/gcc/cp/typeck2.c?view=patch&r1=222636&r2=222635&pathrev=222636 (In reply to Mike Gilbert from comment #6) > The "fix" of casting to unsigned char is pretty ugly. > > I think we will just wait for the next gcc release unless chromium upstream > applies some fix. The patch below seems to do the job. > > https://gcc.gnu.org/viewcvs/gcc/branches/gcc-5-branch/gcc/cp/typeck2. > c?view=patch&r1=222636&r2=222635&pathrev=222636 Unfortunately this fix is buggy: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66007 (In reply to octoploid from comment #7) Thanks for the info. (In reply to octoploid from comment #7) > (In reply to Mike Gilbert from comment #6) > > The "fix" of casting to unsigned char is pretty ugly. > > > > I think we will just wait for the next gcc release unless chromium upstream > > applies some fix. The patch below seems to do the job. > > > > https://gcc.gnu.org/viewcvs/gcc/branches/gcc-5-branch/gcc/cp/typeck2. > > c?view=patch&r1=222636&r2=222635&pathrev=222636 > > Unfortunately this fix is buggy: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66007 But not the most recent one, see https://bugs.gentoo.org/show_bug.cgi?id=548412 Created attachment 403616 [details, diff] chromium-42.0.2311.135-gcc-5.1.patch For those experimenting with gcc 5.1: The proposed work around patch from comment 4 of upstream bug 466760 backported to apply to the stable www-client/chromium-42.0.2311.13, to apply, save the patch to this directory: mkdir -p /etc/portage/patches/www-client/chromium-42.0.2311.135 For later chromium versions, try the earlier patch (02432: proposed work around patch from comment 4 of upstream bug 466760),something like (depending on the chromium version): mkdir -p /etc/portage/patches/www-client/chromium-43.0.2357.18 cp chromium-44.0.2384.0-gcc-5.1.patch /etc/portage/patches/www-client/chromium-43.0.2357.18 (In reply to Mark Wright from comment #10) I would suggest patching gcc instead. gcc-5.2 will have a fix for this. This should really be fixed in Chromium. The code violates the stricter narrowing rules in the C++-11 standard. Our policy is to always prefer fixing the code over adding flags that relax standard conformance (eg. -fpermissive). By the time we unmask GCC 5 this will probably be have been fixed in both places so it doesn't really matter in this case. *** Bug 552550 has been marked as a duplicate of this bug. *** (In reply to Ryan Hill from comment #12) Sure. I a patch gets accepted by chromium upstream to fix the code, I certainly have no objection to it! I forgot we also have bug #550852, so we do need to fix something GCC-side. :D This should be fixed with gcc-5.2.0. |