linux/poky.git/bitbake, branch 1.2_M1.rc2

siggen.py: If both sigs have a variable in it's whitelist then don't say it's changed

2011-12-05T16:23:52+00:00

Some BB_HASHBASE_WHITELIST variables are in the lists of variable dependencies for signatures. Ignore those differences in lists since this difference does not matter (Bitbake rev: 71b53a3f0766ca464560a1f6a449f9424fbdf7ae) Signed-off-by: Matthew McClintock Signed-off-by: Richard Purdie

Remove the async_cmds and sync_cmds from command.py

2011-12-05T16:16:10+00:00

In bitbake/lib/bb/command.py::Command::__init__, we have the following lines: for attr in CommandsSync.__dict__: command = attr[:].lower() method = getattr(CommandsSync, attr) sync_cmds[command] = (method) for attr in CommandsAsync.__dict__: command = attr[:].lower() method = getattr(CommandsAsync, attr) async_cmds[command] = (method) The sync_cmds and async_cmds are defined as global dictionaries, but it seems that we've never used them (I did a "grep -r async_cmds bitbake/", , there is no result except the ones that I have removed), and I can't find the history of it from "git log -p", I guess that they have been replaced by the self.cmds_sync and self.cmds_async. [YOCTO #1791] (Bitbake rev: 24e99460800856035bb54a84c7aa33b3517436e9) Signed-off-by: Robert Yang Signed-off-by: Richard Purdie

fetch2: fix Exception parameters when BB_STRICT_CHECKSUM enabled

2011-12-05T16:16:10+00:00

BB_STRICT_CHECKSUM caused a backtrace as the FetchError parameters are incorrectly specified such that FetchError is being passed 8 params when it's expecting 3. This fixes the parameters so we're passing a formatted string and the url. (Bitbake rev: b8b2f4287c9125542b18a294c0a94ed89a7e73a8) Signed-off-by: Joshua Lock Signed-off-by: Richard Purdie

Remove the duplicated assignments of self.configuration.cmd

2011-12-05T16:15:46+00:00

The assignments of self.configuration.cmd in BBCooker seems duplicated, have the followings in both BBCooker::__init__ and BBCooker::loadConfigurationData: if not self.configuration.cmd: self.configuration.cmd = self.configuration.data.getVar("BB_DEFAULT_TASK", True) or "build" The __init__ invokes the loadConfigurationData, and it would make sure that self.configuration.cmd has been assigned a proper value, so we can remove the one in __init__. [YOCTO #1791] Signed-off-by: Robert Yang Signed-off-by: Richard Purdie

cooker.py: Allow the -e option to work with virtual classes and -b

2011-12-01T23:14:05+00:00

Using bitbake -e -b virtual:xxxx:/path/to/the.bb would result in zero matches since the virtual:xxxx piece wasn't being processed. This adds in the necessary functionality to handle it correctly. [YOCTO #1793] (Bitbake rev: bd5a727c8447bcb747c1d2463b7de2ab6d21a7de) Signed-off-by: Richard Purdie

build.py: Be determistic about a function's cwd

2011-12-01T12:06:27+00:00

There is a subtle but nasty problem that a function's cwd can vary depending on whether ${B} (often ${S}) exists before the funciton is called or not. Most functions in the system can cope with this but its bad practise and I've just witnessed build failures resulting from this during image generation from bootimg.bbclass. I also suspect this could explain some odd fetcher behaviour witnessed in the past. This change ensures we always call funcitons with a specific build directory making things deterministic. (Bitbake rev: ef0888f83fa4408eb768257d7e03700202faad18) Signed-off-by: Richard Purdie

bitbake: Update users of getVar/setVar to use the data store functions directly

2011-11-27T10:35:30+00:00

Signed-off-by: Richard Purdie

bitbake/siggen.py: Don't backtrace if the taskhash data isn't present

2011-11-25T11:25:59+00:00

This allows the code to safely fall back to dumping the basehash data if the taskhash data isn't present for some reason. We could effecitvely obsolete the runtime option and use this approach instead exclusively. (Bitbake rev: 5ace320ccc01f4e326f90b7ba060dcbff3380dca) Signed-off-by: Richard Purdie

bitbake: Update and fix bitbake-runtask

2011-11-25T11:25:53+00:00

Since bitbake switched back to the fork instead of the exec model, it no longer used bitbake-runtask and the code has suffered some bitrot. bitbake-runtask is a useful tool for excuting the task without the scheduler of bitbake, so that the external tool can invoke it easily. It also provides a useful example of how to invoke exec_task() with low overhead without a lot of the bitbake threading/UI overhead. Significant changes: * This patch changes the argument order so that the commonly used and mandatory arguments come first. * The taskhash file and dryrun options are now optional * It now uses the bitbake logging mechanisms to provide processed logging output to the console. * The process handling to do with stdout/stderr redirection are removed since they're no longer required. [YOCTO #1229] RP: Logging updates to the patch based on Roberts original patch Signed-off-by: Robert Yang Signed-off-by: Richard Purdie

Nothing uses USERNAME, remove it - can cause sstate-cache conflicts

2011-11-25T00:30:30+00:00

USER is the correct variable to use, also this can affect sstate cache as well. (Bitbake rev: d7f9edda65dae2e046871afa275c5a51dff48fc4) Signed-off-by: Matthew McClintock Signed-off-by: Richard Purdie