The attached patch adds a build slave for MIPS, and enables a builder for Qt WebKit MIPS32R2 little-endian release build. The builder uses Qt 5. Who should I send the password for the MIPS slave?
Created attachment 164016 [details] Add MIPS slave to buildbot configuration
Comment on attachment 164016 [details] Add MIPS slave to buildbot configuration View in context: https://bugs.webkit.org/attachment.cgi?id=164016&action=review > Tools/BuildSlaveSupport/build.webkit.org-config/config.json:233 > + "name": "Qt Linux MIPS32R2 little-endian Release", "type": "Build", "builddir": "qt-linux-mipsel-mips32r2-release", This is a really long name. Can we simply call it "Qt Linux MIPS Release" instead?
(In reply to comment #2) > (From update of attachment 164016 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=164016&action=review > > > Tools/BuildSlaveSupport/build.webkit.org-config/config.json:233 > > + "name": "Qt Linux MIPS32R2 little-endian Release", "type": "Build", "builddir": "qt-linux-mipsel-mips32r2-release", > > This is a really long name. Can we simply call it "Qt Linux MIPS Release" instead? The reason for the name is that we plan to add different configurations: - MIPS32R1 little-endian and possibly big-endian (R1 is used by e.g. Broadcom chipsets) - MIPS32R2 litte-endian and big-endian - we might want to support MIPS64 as well down the road If you want, we could make the name shorter by using LE and BE instead of little-endian and big-endian. Would this solution be acceptable for you?
Comment on attachment 164016 [details] Add MIPS slave to buildbot configuration View in context: https://bugs.webkit.org/attachment.cgi?id=164016&action=review >>> Tools/BuildSlaveSupport/build.webkit.org-config/config.json:233 >>> + "name": "Qt Linux MIPS32R2 little-endian Release", "type": "Build", "builddir": "qt-linux-mipsel-mips32r2-release", >> >> This is a really long name. Can we simply call it "Qt Linux MIPS Release" instead? > > The reason for the name is that we plan to add different configurations: > - MIPS32R1 little-endian and possibly big-endian (R1 is used by e.g. Broadcom chipsets) > - MIPS32R2 litte-endian and big-endian > - we might want to support MIPS64 as well down the road > > If you want, we could make the name shorter by using LE and BE instead of little-endian and big-endian. Would this solution be acceptable for you? I think "Qt Linux MIPS32R2 LE Release" would be a good choice. Of course, you can add detailed information to your slave info file, see this example http://build.webkit.org/buildslaves/szeged-linux-1. (admin and host file in the info directory, close to the tac file on the slave)
(In reply to comment #0) > Who should I send the password for the MIPS slave? Lucas or Bill can generate you a random password. (I cc-ed them.)
Created attachment 164107 [details] Add MIPS slave to buildbot configuration with shorter builder name
(In reply to comment #4) > I think "Qt Linux MIPS32R2 LE Release" would be a good choice. Uploaded the modified patch. > > Of course, you can add detailed information to your slave info file, > see this example http://build.webkit.org/buildslaves/szeged-linux-1. > (admin and host file in the info directory, close to the tac file on the slave) Yes, that is already done, thank you.
Do you still need anything from us to enable the MIPS buildbot? For example solving bug 96916 would benefit from having a working MIPS buildbot. Thank you, Gergely
Comment on attachment 164107 [details] Add MIPS slave to buildbot configuration with shorter builder name r=me. Bill or Lucas, could you generate a password for the new mips-1 user, send it to Gergely, and restart the master of course, please?
password sent via email.
password has been updated. We need to land this before the bot can start.
Created attachment 165123 [details] Updated patch to include a ChangeLog
Comment on attachment 165123 [details] Updated patch to include a ChangeLog r=me, but cq- now, it should be landed in IDLE time not to break too many slave's work.
Comment on attachment 165123 [details] Updated patch to include a ChangeLog Clearing flags on attachment: 165123 Committed r129229: <http://trac.webkit.org/changeset/129229>
All reviewed patches have been landed. Closing bug.
(In reply to comment #14) > (From update of attachment 165123 [details]) > Clearing flags on attachment: 165123 > > Committed r129229: <http://trac.webkit.org/changeset/129229> It seems master's automatic restart still doesn't work. Bill or Lucas, could you restart it in an IDLE time, please?