It looks like we're about to see a lot of changes in branch names in repos. If
we have the prune option here, those old names are lost, the changes propagate
to our source mirrors and our old releases break.
We have the force option so any replaced references should be replaced, its only
orphaned branches which will now be preserved.
I believe this behaviour will cause us fewer problems given the changes that
look likely to happen.
(Bitbake rev: 08aafc53ba911337d2d3558654b06f294ae634e2)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If a fetcher, e.g., git, is run when pseudo is active it will think it
is running as root. If it in turn uses ssh (as git does), ssh too will
think it is running as root. This will cause it to try to read root's
ssh configuration from /root/.ssh which will fail. If ssh then needs to
ask for credentials it will hang indefinitely as there is nowhere for it
to ask the user for them (and even if there was it would not access the
correct private keys).
The solution to the above is to temporarily disable pseudo while
executing any fetcher commands. There should be no reason for them to be
executed under pseudo anyway so this should not be a problem.
RP Ammendum:
We finally did get more information about how to reproduce this problem,
something needs to trigger bb.fetch2.get_srcrev() in a pseudo context,
for example when AUTOREV is in use or the recipe doesn't have a defined
SRCREV. That SRC_URI needs to be using protocol=ssh. This would trigger
an ls-remote of the remote repo and if that happens under pseudo, the
wrong ssh credentials may be attempted which can hang.
[YOCTO #12464]
(Bitbake rev: 5d42dce5e612060f4181c14e1cfdec2388bb7adf)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When running bitbake command with Python 3.6.5 always result in
import error causing by the change of distutils module.
This patch replaces the method to search executable in PATH by
"/usr/bin/env <command>".
(Bitbake rev: 86a4cc7248afd3e561dc70f3490f45549265470a)
Signed-off-by: Tzu Hsiang Lin <t9360341@ntut.org.tw>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since 2017-08-17 (git version 2.14.1.473.g3ec7d702a) using deprecated
git branch parameter "--set-upstream" causes a fetcher error. Replace
it by "--set-upstream-to".
https://git.kernel.org/pub/scm/git/git.git/commit/?id=52668846ea2d41ffbd87cda7cb8e492dea9f2c4d
says, it's deprecated since 2012-08-30 so hopefully all still supported
host distributions have new enough git to support "--set-upstream-to".
ERROR: PACKAGE do_unpack: Fetcher failure: ...;
git -c core.fsyncobjectfiles=0 branch --set-upstream master origin/master failed with exit code 128, output:
fatal: the '--set-upstream' option is no longer supported. Please use '--track' or '--set-upstream-to' instead.
ERROR: PACKAGE do_unpack: Function failed: base_do_unpack
(Bitbake rev: 698e36430761cd3cf81283d58ca5bb3426bd24dd)
Signed-off-by: Andre Rosa <andre.rosa@lge.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ensure that when an item fetched from a premirror has an invalid checksum the
fetcher falls back to the usual logic of trying the upstream and any configured
mirrors.
(Bitbake rev: cc52b9b12c60810142252b9cb5d4268e42371b8e)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This has long since been deprecated and is no longer used anywhere, FILESPATH
is the commonly used varaible which offers much more flexibility. Remove
the FILESDIR code and references from bitbake.
(Bitbake rev: 751c9dc51fd01fa64a1ff37ba2638110335f71af)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When the fetcher retrieves file:// URLs, there is no lock file being
used. This means that in case two separate tasks (typically from two
concurrent invocations of bitbake) want to download the same file://
URL at the same time, there is a very small chance that they also end
up wanting to create a symbolic link to the file at the same time.
This would previously lead to one of the tasks failing as the other
task would have created the link.
(Bitbake rev: 58a03531c8183b165bb7dcad86d8559c92bc150d)
Signed-off-by: Peter Kjellerstedt <pkj@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
URL decoding was improved in the core a while ago and this looks like
a leftover from those times which caused urls needing a user/password to
fail. Use the parameters from the core instead of the broken split
implementation.
[YOCTO #11262]
(Bitbake rev: 6a917ec99d659e684b15fa8af94c325172676062)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This commit looks to see if FETCHCMD_s3 is set and if not, sets
it.
This is needed because I've use cases where I don't use aws, but
s3cmd (due to license).
(Bitbake rev: fdeaed70a7d1ff8be1a1de937cb864130b0c2c86)
Signed-off-by: Elizabeth 'pidge' Flanagan <pidge@toganlabs.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Imagine you have an sstate mirror accessed over http and an SSTATE_MIRRORS
which maps file:// urls to http:// urls.
File urls set donestampneeded = False, http urls don't. This can result in
races in the try_mirror_url() code since it will trigger new downloads after
aquiring the lockfile as verify_donestamp() doesn't look at origud and there
is no donestamp.
verify_donestamp() already has code to look at origud, we're just missing
some code at the start of the function to do this. Fix it to avoid
these races.
(Bitbake rev: b8b14d975a254444461ba857fc6fb8c725de8874)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We call git ls-remote to get the latest revision from a git repository,
however by calling runfetchcmd() we can end up recursively running
git ls-remote a number of times with OE e.g. if ${SRCPV} is in PV, ${PV}
is in WORKDIR, and ${WORKDIR} is in PATH (as a result of recipe-specific
sysroots), our call to runfetchcmd() exports PATH so _lsremote() will
get called again - with the end result that we run git ls-remote 30
times in quick succession (!). Prevent that from happening by using a
guard variable and returning a dummy value if it's called recursively.
Fixes [YOCTO #11185].
(Bitbake rev: ff1ccd1db5d70b3fc9ad0d3e8f3d7b804c22bf36)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
FetchError isn't defined, use bb.fetch2.FetchError in this context.
(Bitbake rev: 945fa980e027753df2c21d84eb63dcaddb2caaee)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The touch of .done explicitly specifies the path, so there's no need for
workdir=, and "os.path.join('.')" is identical to just '.'.
(Bitbake rev: 955cbfdaa2400d15ec428b65848e6835c9f44860)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
list.index() isn't a particularly efficient operation, so keep track of our
position via enumerate() instead, which is more pythonic as well.
(Bitbake rev: dec6e90a4d27ee335e9c78aeebd277098fec94d1)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Treat tar files compressed with 7-Zip in the same way as tar files
compressed with other compression formats.
(Bitbake rev: 363a0f54dc7d9930537f0df25173fa31ca1f98ac)
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Class for fetching files from Amazon S3 using the AWS Command Line
Interface. The aws tool must be correctly installed and configured
prior to use.
The class supports both download() and checkstatus(), which therefore
allows S3 mirrors to be used for SSTATE_MIRRORS.
(Bitbake rev: 6fe07ed25457dd7952b60f4b2153d56b15d5eea6)
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Forcing the use of "\n" in mirror variables is pointless, we can just require that
there are pairs of values.
(Bitbake rev: 044fb04dbe69313ee6908bf4d3cee7f797d0c41c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Cleanup some more usage of bb.data APIs in the fetchers.
(Bitbake rev: 9752fd1c10b8fcc819822fa6eabc2c1050fcc03b)
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Most angular2 packages have names of the form @angular/xxx.
The / obviously can't be used in a file name, replace it with -.
(Bitbake rev: d3bd41d0ec9621307c362b394872b18b8b7ed8d6)
Signed-off-by: Anders Darander <anders@chargestorm.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A number of npm packages use @ as a leading chararacter.
Examples are most of the angular2 packages.
(Bitbake rev: 628c4bf6c89b3d62c9b864380b5c8e131a899bff)
Signed-off-by: Anders Darander <anders@chargestorm.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The functionality around the 'rsh' parameter seemed to be broken. The
'rsh' parameter was only used when 'protocol' was set to 'svn+ssh' which
is confusing. The 'rsh' parameter was used for setting the value of
'svn_RSH' environment variable, which however, is not supported by svn
(not at least according to SVN documentation).
This patch removes the 'rsh' parameter and replaces it with 'ssh'. This
new (optional) parameter is used when svn+ssh protocol is used and it
can be used to specify the ssh program used by svn. This is achieved by
setting the SVN_SSH environment variable which is mentioned in SVN
documentation.
(Bitbake rev: 5b364b02270b0d7c2b7ca8d67fa2731bf93720ee)
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
- In some cases the file descriptor
is held by nfs client and none of os.path.* is catching
that, it could mean that error is not doled out because
client has cached the stat info. In this case we are
out of luck. Needed to catch IOError, which would be
causing the Stale error.
- In download method, update_stamp is invoked
md5sum validation which is found to be throwing
Stale errors.
- Added error handling to fix the stale errors.
(Bitbake rev: 5a53e7d7b017769a6eb0f0a6335735a1fe51a5ec)
Signed-off-by: Balaji Punnuru <balaji_punnuru@cable.comcast.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For spelling's sake, rename Python routine "setup_revisons" to
"setup_revisions."
(Bitbake rev: 4df59b027c02ef39d72476251ccd3fd62fc20bf6)
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(Bitbake rev: 05f5421b2e44cd58c5912848de43d5884d070150)
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When I originally added this check I didn't quite understand how the
values in this field should be expressed - it seems from reading the
documentation if there is an entry starting with '!' then the list is
a blacklist and we shouldn't expect "linux" to be in the list, or we'll
end up skipping important dependencies.
This fixes fetching the "statsd" npm package.
Fixes [YOCTO #10760].
(Bitbake rev: 7aa6d1586417e0e7d9925917a82caee5884957db)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
An npm package.json file has two dependency fields: dependencies and
optionalDependencies. An item in optionalDependencies *may* also be
listed in dependencies, but this is not required (and not necessary
since if it's in optionalDependencies it will be optional, adding it to
dependencies won't do anything). The code here was assuming that an
optional dependency would always be in both, that's probably because
that was true of the examples I was looking at at the time. To fix it,
just add the optional ones to the list we're iterating over.
(Bitbake rev: c0c50d43266150a80be31ae2c6fcaf37f5ba231d)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If PATH contains WORKDIR which contains PV which contains SRCPV we can end
up in circular recursion within the fetcher. This code change allows for the recursion
to be broken by giving PV a temporary dummy value in a data store copy.
(Bitbake rev: ce1e70b8018340b54dba3a81d7d379182cb77514)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When the gitsm fetcher is used with a repo that includes a .gitattributes
file that makes git modify files on cloning (e.g. line break characters),
the subsequent checkout performed in the update_submodules function fails.
This is fixed by adding the force flag (-f) to the checkout command.
(Bitbake rev: c05e1396625b14e66d795408ea2ae4cd2afc3209)
Signed-off-by: Ola Redell <ola.redell@retotech.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Commit 873e33d0479e977520106b65d149ff1799195bf6 [fetch2/wget:
add Basic Auth from netrc to checkstatus()] causes "Fetcher failure
for URL: 'https://www.example.com/'. URL https://www.example.com/
doesn't work." on new builds when a user has a .netrc file but there
is no default and no matching host. The call to netrc.authenticators()
will return None in these cases and the attempted assignment to the
3-tuple will raise a TypeError exception. Add the TypeError to the
exceptions caught to get around this issue.
(Bitbake rev: c0c0af40ebddaf9dc99353c580a65d4c04295613)
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
fetch2/wget uses urllib to check the status of the mirrors, wget will
use netrc to pass login and password information however checkstatus
will skip that.
This adds netrc login and password to checkstatus so both will work the
same.
(Bitbake rev: 873e33d0479e977520106b65d149ff1799195bf6)
Signed-off-by: Matthew McClintock <msm-oss@mcclintock.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
getVarFlag() now defaults to expanding by default, thus remove the
True option from getVarFlag() calls with a regex search and
replace.
Search made with the following regex:
getVarFlag ?\(( ?[^,()]*, ?[^,()]*), True\)
(Bitbake rev: c19baa8c19ea8ab9b9b64fd30298d8764c6fd2cd)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
getVar() now defaults to expanding by default, thus remove the True
option from getVar() calls with a regex search and replace.
Search made with the following regex: getVar ?\(( ?[^,()]*), True\)
(Bitbake rev: 3b45c479de8640f92dd1d9f147b02e1eecfaadc8)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The old style bb.data.getVar/setVar API is obsolete. Most of bitbake
doesn't use it but there were some pieces that escaped conversion. This
patch fixes the remaining users mostly in the fetchers.
(Bitbake rev: ff7892fa808116acc1ac50effa023a4cb031a5fc)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If ud.ignore_checksums is set (which we currently use to suppress the
warnings for missing SRC_URI checksums when fetching files from
scripts), then if we're fetching an npm package we should similarly
suppress the warnings when NPM_LOCKDOWN and NPM_SHRINKWRAP aren't set.
At the same time, make any errors reading either of these files actual
errors since if the file is specified and could not be found, that
should be an error - not the exact same warning.
Fixes [YOCTO #10464].
(Bitbake rev: cefb8c93c8299e68352cf7ec5ad9ca50c0d499ed)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently if you use the subdir parameter in a SRC_URI and pass an absolute path
then it gets appended to the unpack directory instead of being used directly.
This is inconvenient as it may be useful to use ${S} when you want to unpack a
file into the source tree.
Change this behaviour so that absolute paths are used directly instead of being
appended to the root directory. To ensure that recipes cannot write files to an
arbitrary location enforce that the subdir starts with the unpack root.
(Bitbake rev: c3873346c6fa1021a1d63bddd9b898a77c618432)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The implementation of SRCREV_FORMAT has at least two issues:
1. Given two names "foo" and "foobar" and SRCREV_FORMAT = "foo_foobar",
"foo" might currently get substituted twice, and "foobar" not at
all.
2. If the revision substitued for some name happens to contain another
name as a substring, then that substring might incorrectly get
replaced.
Fix both issues by sorting the names with the longest ones first and
replacing all names at once with a regular expression. This was inspired
by
http://stackoverflow.com/questions/6116978/python-replace-multiple-strings.
(Bitbake rev: 8e6a893cb7f13ea14051fc40c6c9baf41aa47fee)
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If we've already fetched a particular URL then we do not need to do so
again within in the same operation. Maintain an internal list of fetched
URLs to avoid doing that.
(Bitbake rev: b4705c80add1f618c11a9223cdd9578d763b50ec)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The correct name of the parameter is "version" not "ver" so ensure we
aren't misleading the user by giving the latter in an example.
(Bitbake rev: 14c045c6a20993d389b91ae2459d811a1430a7b2)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Allow using a top-level shrinkwrap file with one or more npm://
dependencies, i.e. if the module isn't found at the top level then look
one level down.
Part of the fix for [YOCTO #9537].
(Bitbake rev: f7de3f8b5f628dee043fe783148812914ab20813)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
"npmpkg" can be a default, but it should respect the subdir parameter as
with other FetchMethods. This allows us to have more than one npm://
entry in SRC_URI without nasty hacks.
Fix required in order to support [YOCTO #9537].
(Bitbake rev: e6a94d2091ec5d42f25102334a8492a731b8dec3)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
You cannot set a URL-specific value in an object-level variable on
the FetchMethod in urldata_init() or the result is the value specific to
the last URL will be the one that gets set. This prevented fetching more
than one npm:// URL correctly - the other tarballs would not download to
the correct location and do_unpack failed to find them as a result.
Fix required in order to support [YOCTO #9537].
(Bitbake rev: 1435b49ea7d0f9d4cc4a665fb2aa83d1eea7900f)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We were downloading into the current directory here, which is fine if
that current directory can be expected to be the right place - but
that's not true when called from recipetool within OE. We should
explicitly specify the directory to run the command in and then there
won't be a problem.
(Bitbake rev: 0ddaf725e5a0675b252b7f80b1706370e478175b)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The ud.pkgdir argument was being passed as the 'quiet' argument to
runfetchcmd, not the 'workdir' argument, resulting in fetching the svn module
into the root of DL_DIR, not where it belongs.
Cc: Matt Madison <matt@madison.systems>
(Bitbake rev: dc756510a95f88b192352be6fcd1d5d77852c348)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If http basic auth creds were added to sstate mirrors like so:
https://foo.com/sstate/PATH;user=foo:bar;downloadfilename=PATH
The sstate mirror check would silently fail with 401 unauthorized.
This patch allows both the check, and the wget download to succeed by
checking for user credentials and if present adding the correct
headers, or wget params as needed.
[ YOCTO #9815 ]
(Bitbake rev: cea8113d14da9e12db80a5b6b5811a47a7dfdeef)
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We need a separate fetcher cache per multiconfig as the revisions and other
SRC_URI data can potentially be different. For now, this is the simplest way
to achieve that and avoids linux-yocto kernel build failures when targeting
multiple machines for example.
(Bitbake rev: d98cc31d6668bc1d6372664593126b5e5132ef2c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Now that the fetchers all preserve the current working
directory, the cwd changes in the try_mirror_url,
download, and checkstatus methods are no longer needed.
(Bitbake rev: 0ed8975c42718342a104a9764a58816f964ec4ea)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie
Peter Kjellerstedt
Tzu Hsiang Lin
Andre Rosa
Joshua Lock
Elizabeth 'pidge' Flanagan
Paul Eggleton
Christopher Larson
Andre McCurdy
Alexander Kanavin
Anders Darander
Markus Lehtonen
Khem Raj
Robert P. J. Day
Ola Redell
Mark Asselstine
Matthew McClintock
Ross Burton
Ulf Magnusson
Stephano Cetola
Matt Madison