portage.git
13 years agodepgraph: queue ignored deps for later traversal
Zac Medico [Mon, 2 May 2011 03:09:33 +0000 (20:09 -0700)]
depgraph: queue ignored deps for later traversal

This will fix an issue with some build-time deps being ignored by
--complete-graph in cases when --with-bdeps is not enabled. It
should help --rebuild to find more rebuilds in this case.

13 years agodepgraph: account for new-style virtual/libc
Zac Medico [Mon, 2 May 2011 01:42:32 +0000 (18:42 -0700)]
depgraph: account for new-style virtual/libc

This will fix bug #364681.

13 years agotrigger_rebuilds: handle circular deps
Zac Medico [Sun, 1 May 2011 23:27:48 +0000 (16:27 -0700)]
trigger_rebuilds: handle circular deps

13 years agodepgraph: traverse bdeps in complete mode
Zac Medico [Sun, 1 May 2011 21:17:06 +0000 (14:17 -0700)]
depgraph: traverse bdeps in complete mode

This makes the graph as complete as possible, and can help --rebuild to
find more rebuilds.

13 years agodepgraph: ignore unwanted virtuals
Zac Medico [Sun, 1 May 2011 20:23:13 +0000 (13:23 -0700)]
depgraph: ignore unwanted virtuals

This fixes an unwanted side-effect from commit
470871eeafa89a05486d4eb6f3f7626c1f813e4d.

13 years agoemerge: add --rebuild and --norebuild-atoms opts
David James [Sun, 1 May 2011 17:21:45 +0000 (10:21 -0700)]
emerge: add --rebuild and --norebuild-atoms opts

Rebuild when build-time/run-time deps are upgraded.

If pkgA has been updated, and pkgB depends on pkgA at both
build-time and run-time, pkgB needs to be rebuilt. This
feature ensures that all packages are consistent when
dependencies that are used at both runtime and build time
are changed.

This feature only rebuilds packages one layer deep. That
means that if you upgrade libcros, for example, packages
that depend directly on libcros will be rebuilt and
reinstalled, but indirect dependencies will not be rebuilt.

BUG=chromium-os:14296

TEST=Test whether packages rebuilding a bunch of packages.

Change-Id: Idbc0532b4b1de28fd9e5a0abe3b7dbe1a3abd2c8

Review URL: http://codereview.chromium.org/6905107

13 years agorepoman: remove most old-style virtual checks
Zac Medico [Sun, 1 May 2011 16:30:36 +0000 (09:30 -0700)]
repoman: remove most old-style virtual checks

PROVIDE virtuals have been banned from Gentoo's tree (see bug #365505),
so these checks aren't needed anymore. The PROVIDE.syntax check is kept
in the Package class since PROVIDE is still supported for backward
compatibility, and therefore invalid syntax can still have negative
effects. This check can be removed if/when backward compatibilty is
completely dropped.

13 years agorepoman: Check for PROVIDE.
Ulrich Mueller [Sun, 1 May 2011 15:57:33 +0000 (08:57 -0700)]
repoman: Check for PROVIDE.

This will fix bug #365505.

13 years agoSupport multiple arguments in set_unless_changed() and unset_unless_changed().
Arfrever Frehtes Taifersar Arahesis [Sun, 1 May 2011 14:50:30 +0000 (16:50 +0200)]
Support multiple arguments in set_unless_changed() and unset_unless_changed().
Use VARIABLE=VALUE syntax for arguments of set_unless_changed().

13 years agoruntests.sh: propagate arguments to runTests
Zac Medico [Sun, 1 May 2011 14:18:42 +0000 (07:18 -0700)]
runtests.sh: propagate arguments to runTests

13 years agoBug #365439: Add set_unless_changed() and unset_unless_changed().
Arfrever Frehtes Taifersar Arahesis [Sun, 1 May 2011 01:21:57 +0000 (03:21 +0200)]
Bug #365439: Add set_unless_changed() and unset_unless_changed().

13 years agodepgraph: migrate implicit libc deps from Sched
Zac Medico [Sat, 30 Apr 2011 20:38:00 +0000 (13:38 -0700)]
depgraph: migrate implicit libc deps from Sched

Now all the virtual/libc handling is done by the depgraph, which
will be helpful when fixing bug #364681.

13 years agoinstall_qa_check(): Simplify code and make 'l' a local variable.
Arfrever Frehtes Taifersar Arahesis [Sat, 30 Apr 2011 16:30:29 +0000 (18:30 +0200)]
install_qa_check(): Simplify code and make 'l' a local variable.

13 years agoCheck for references to ${ROOT} in install_qa_checks.
David James [Sat, 30 Apr 2011 00:21:58 +0000 (17:21 -0700)]
Check for references to ${ROOT} in install_qa_checks.

When ROOT != /, binaries that reference ROOT will load their
dependencies from ROOT first rather than from the system-configured
path. This is a problem because the ROOT will be / on the target
system.

Besides the above, this patch also fixes incorrect parsing of scanelf
output, where we would treat the RPATHs returned by scanelf as the
names of binaries.

TEST=When "stricter" FEATURE is enabled, verify that emerge
fails when an ebuild references broken rpaths referencing
ROOT. When "stricter" FEATURE is not enabled, verify that such
references are automatically fixed. Also verify that ebuilds
with non-broken RPATHs (e.g. RPATHs referencing $ORIGIN/../lib)
are not touched by the change.

BUG=chromium-os:14271

Change-Id: I4f29cc4ea9195a1255f080284da1f676e4a2c26b

Review URL: http://codereview.chromium.org/6903153

13 years agoemergelog: only set permission on creation
Zac Medico [Fri, 29 Apr 2011 16:07:25 +0000 (09:07 -0700)]
emergelog: only set permission on creation

This will fix bug #365319.

13 years agoaction_info: eliminate duplicate info_pkgs match
Zac Medico [Fri, 29 Apr 2011 15:04:13 +0000 (08:04 -0700)]
action_info: eliminate duplicate info_pkgs match

Currently, sys-kernel/linux-headers is matched by both a plain
sys-kernel/linux-headers atom and by the virtual/os-headers new-style
virtual. For backward compatibility, we're going to have duplicates
like this for at least a few months (see bug #364673, comment #5).
Therefore, automatically eliminate duplicates in the display. Entries
that include virtual provider info are preferred over those that do
not.

13 years ago_postinst_bsdflags: apply to $ROOT, not $D
Zac Medico [Thu, 28 Apr 2011 21:31:03 +0000 (14:31 -0700)]
_postinst_bsdflags: apply to $ROOT, not $D

13 years agoMove preinst/postinst_bsdflags from bash to py
David James [Thu, 28 Apr 2011 21:22:15 +0000 (14:22 -0700)]
Move preinst/postinst_bsdflags from bash to py

Moving these commands from shell to Python helps avoid an unnecessary
call to misc-functions.sh for the postinst_bsdflags. This improves
performance.

BUG=none
TEST=Run emerge-x86-generic -eg --jobs=16 libchrome

Change-Id: I0c2af50b4d2f7644cabac84fde7fe4d682010c69

Review URL: http://codereview.chromium.org/6676107

13 years agodisplay_preserved_libs: load plib_registry
Zac Medico [Thu, 28 Apr 2011 21:07:08 +0000 (14:07 -0700)]
display_preserved_libs: load plib_registry

Since commit 7535cabdf2fab76fc55df83643157613dfd66be9, the
plib_registry is modified in a subprocess, and the parent
process' copy can become stale because of this. Therefore,
explicitly load it.

13 years agoemerge: add --misspell-suggestions=n option
Zac Medico [Thu, 28 Apr 2011 16:10:02 +0000 (09:10 -0700)]
emerge: add --misspell-suggestions=n option

This will fix bug #363137.

13 years agoemerge: fix indent for --binpkg-respect-use help
Zac Medico [Thu, 28 Apr 2011 12:58:35 +0000 (05:58 -0700)]
emerge: fix indent for --binpkg-respect-use help

13 years agoemerge: add --help for 3 new options
Zac Medico [Wed, 27 Apr 2011 22:58:16 +0000 (15:58 -0700)]
emerge: add --help for 3 new options

13 years agoemerge: add 3 new options similar to --exclude
David James [Wed, 27 Apr 2011 22:07:12 +0000 (15:07 -0700)]
emerge: add 3 new options similar to --exclude

Add --nousepkg-atoms, --useoldpkg-atoms, and --reinstall-atoms flag
to Portage

reinstall-atoms accepts a space separated list of package names or
slot atoms. Emerge will treat matching packages as if they are not
installed, and reinstall them if necessary.

useoldpkg-atoms accepts a space separated list of package names or
slot atoms. Emerge will prefer matching binary packages over newer
unbuilt packages. This is useful in case you want to request that a
particular package won't be rebuilt from source.

nousepkg-atoms accepts a space separated list of package names or
slot atoms. Emerge will ignore matching binary packages.

Change-Id: I0d73039c6a4cd63695b28ffc80215628e0e05c95

BUG=chromium-os:12507 TEST=Try out the flag

Review URL: http://codereview.chromium.org/6577024

13 years ago--autounmask: Don't change masked/forced flags
Sebastian Luther [Wed, 27 Apr 2011 20:36:15 +0000 (13:36 -0700)]
--autounmask: Don't change masked/forced flags

This will fix bug #364701.

13 years agoexpand_new_virt: don't traverse blockers
Zac Medico [Wed, 27 Apr 2011 16:13:51 +0000 (09:13 -0700)]
expand_new_virt: don't traverse blockers

13 years agoexpand_new_virt: validate EAPI, IUSE and USE
Zac Medico [Wed, 27 Apr 2011 14:35:57 +0000 (07:35 -0700)]
expand_new_virt: validate EAPI, IUSE and USE

13 years agoexpand_new_virt: eliminate cpv_getkey
Zac Medico [Wed, 27 Apr 2011 14:05:24 +0000 (07:05 -0700)]
expand_new_virt: eliminate cpv_getkey

13 years agoexpand_new_virt: use stack for recursion
Zac Medico [Tue, 26 Apr 2011 23:57:20 +0000 (16:57 -0700)]
expand_new_virt: use stack for recursion

13 years agoBug #364467 - handle gcc-4.6-style QA warnings
Ryan Hill [Tue, 26 Apr 2011 21:16:20 +0000 (14:16 -0700)]
Bug #364467 - handle gcc-4.6-style QA warnings

13 years agoFix a typo.
Arfrever Frehtes Taifersar Arahesis [Tue, 26 Apr 2011 20:50:03 +0000 (22:50 +0200)]
Fix a typo.

13 years agoemerge --info: support new-style virtuals
Zac Medico [Tue, 26 Apr 2011 20:19:34 +0000 (13:19 -0700)]
emerge --info: support new-style virtuals

This will fix bug #364673.

13 years agoebuild(1): fix env FEATURES overrides bug #364035
Zac Medico [Sat, 23 Apr 2011 15:41:07 +0000 (08:41 -0700)]
ebuild(1): fix env FEATURES overrides bug #364035

13 years agorepoman: deprecate more eclasses
Tomas Chvatal [Wed, 20 Apr 2011 23:56:06 +0000 (16:56 -0700)]
repoman: deprecate more eclasses

13 years agorepoman: cleanup gpgsign v2.2.0_alpha30
Zac Medico [Thu, 14 Apr 2011 01:54:30 +0000 (18:54 -0700)]
repoman: cleanup gpgsign

13 years agoUpdate timestamps in headers of modified files.
Zac Medico [Wed, 13 Apr 2011 07:59:38 +0000 (00:59 -0700)]
Update timestamps in headers of modified files.

13 years agodoins: tweak $DISTDIR symlink handling
Zac Medico [Tue, 12 Apr 2011 01:09:20 +0000 (18:09 -0700)]
doins: tweak $DISTDIR symlink handling

Only match stuff inside $PORTAGE_ACTUAL_DISTDIR/ so that we don't
stray from PMS.

13 years agodoins: dereference abs symlinks to files
Zac Medico [Mon, 11 Apr 2011 23:23:54 +0000 (16:23 -0700)]
doins: dereference abs symlinks to files

Our fake $DISTDIR contains symlinks that should not be reproduced
inside $D. In order to ensure that things like dodoc
"$DISTDIR"/foo.pdf work as expected, we dereference symlinked files
that are referenced by absolute paths. Thanks to James Cloos
<cloos@jhcloos.com> for reporting this issue which he observed with
the sci-mathematics/minisat ebuild.

13 years agoextract_affecting_use: test bug #363073
Zac Medico [Mon, 11 Apr 2011 23:00:02 +0000 (16:00 -0700)]
extract_affecting_use: test bug #363073

13 years agoextract_affecting_use: allow parens in atoms
Zac Medico [Mon, 11 Apr 2011 22:30:13 +0000 (15:30 -0700)]
extract_affecting_use: allow parens in atoms

This fixes bug #363073 in which an InvalidDependString exception is
erroneously triggered by atoms containing EAPI 4 USE dependency
defaults. This case is very similar to bug #354003 which was fixed in
commit 8735222b77e66850213e2aa6a7ea48e744ba0d4f.

13 years agoRemove unnecessary sync in Scheduler.py.
David James [Mon, 11 Apr 2011 19:00:39 +0000 (12:00 -0700)]
Remove unnecessary sync in Scheduler.py.

If we are passed a fakevartree object, it should already be sync'd, so
there is no need to sync it again here. This sync is only needed when
we create a new FakeVartree, so I moved the sync to the right place to
fix this.

BUG=chromium-os:14035
TEST=Build a bunch of packages with this patch.

Change-Id: I89d79cf946f4c0c27ad585ad7c88a41985260342

Review URL: http://codereview.chromium.org/6813084

13 years agoFix a typo in a comment.
Zac Medico [Sun, 10 Apr 2011 20:05:45 +0000 (13:05 -0700)]
Fix a typo in a comment.

13 years agoScheduler: add more info in docstring comments
Zac Medico [Sun, 10 Apr 2011 20:03:11 +0000 (13:03 -0700)]
Scheduler: add more info in docstring comments

This clarifies the purposes of _system_merge_started and _choose_pkg.

13 years agoAdd force-mirror FEATURE.
David James [Thu, 7 Apr 2011 23:49:58 +0000 (16:49 -0700)]
Add force-mirror FEATURE.

force-mirror: Only fetch files from configured mirrors, ignoring
SRC_URI, except when mirror is in the ebuild(5) RESTRICT variable.

BUG=chromium-os:13221
TEST=Test that ebuilds now fail with FEATURES=force-mirror if files are
not in the configured mirror.

Change-Id: I8484d5af8dff97b431398030b33c024ff1295ba0

Review URL: http://codereview.chromium.org/6677171

13 years agofilter PORTAGE_GPG_SIGNING_COMMAND for ebuild env
Zac Medico [Mon, 4 Apr 2011 22:03:01 +0000 (15:03 -0700)]
filter PORTAGE_GPG_SIGNING_COMMAND for ebuild env

13 years agoMerge branch 'master' of git+ssh://git.overlays.gentoo.org/proj/portage
Arfrever Frehtes Taifersar Arahesis [Sun, 3 Apr 2011 17:30:33 +0000 (19:30 +0200)]
Merge branch 'master' of git+ssh://git.overlays.gentoo.org/proj/portage

13 years agoSupport PORTAGE_GPG_SIGNING_COMMAND variable.
Arfrever Frehtes Taifersar Arahesis [Sun, 3 Apr 2011 17:28:58 +0000 (19:28 +0200)]
Support PORTAGE_GPG_SIGNING_COMMAND variable.

13 years agorepoman: fix profiles.desc overlay path
Zac Medico [Sun, 3 Apr 2011 15:54:10 +0000 (08:54 -0700)]
repoman: fix profiles.desc overlay path

Thanks to Alexey Shvetsov <alexxy@gentoo.org> for reporting this issue
and testing the fix.

13 years agoDisplay: fix fetch restrict msg for RO_DISTDIRS
Zac Medico [Thu, 31 Mar 2011 15:41:52 +0000 (08:41 -0700)]
Display: fix fetch restrict msg for RO_DISTDIRS

This will fix bug #361463.

13 years agoMinor improvements in gpgsign().
Arfrever Frehtes Taifersar Arahesis [Wed, 30 Mar 2011 22:19:54 +0000 (00:19 +0200)]
Minor improvements in gpgsign().
repoman_settings["PORTAGE_GPG_DIR"] is always set.

13 years agoSimplify calculation of repoman_settings["PORTAGE_GPG_DIR"] in gpgsign().
Arfrever Frehtes Taifersar Arahesis [Wed, 30 Mar 2011 20:32:45 +0000 (22:32 +0200)]
Simplify calculation of repoman_settings["PORTAGE_GPG_DIR"] in gpgsign().
os.path.expanduser() works even when HOME environment variable is not set.

13 years agochk_updated_info_files: handle --quiet
Zac Medico [Wed, 30 Mar 2011 18:36:35 +0000 (11:36 -0700)]
chk_updated_info_files: handle --quiet

This will fix bug #361257.

13 years agoUse os.path.expanduser() to simplify code.
Arfrever Frehtes Taifersar Arahesis [Wed, 30 Mar 2011 15:03:22 +0000 (17:03 +0200)]
Use os.path.expanduser() to simplify code.

13 years agodisplay_merge: tweak background/logging logic v2.2.0_alpha29
Zac Medico [Sun, 27 Mar 2011 22:37:47 +0000 (15:37 -0700)]
display_merge: tweak background/logging logic

13 years agodblink: rename 'buffer' var in _elog_process
Zac Medico [Sun, 27 Mar 2011 21:26:08 +0000 (14:26 -0700)]
dblink: rename 'buffer' var in _elog_process

Avoid name collision with built-in 'buffer' function.

13 years agounmerge: fix lock/UnsupportedAPIException logic
Zac Medico [Sun, 27 Mar 2011 20:57:35 +0000 (13:57 -0700)]
unmerge: fix lock/UnsupportedAPIException logic

13 years agoUnsupportedAPIException: handle unicode in EAPI
Zac Medico [Sun, 27 Mar 2011 20:55:11 +0000 (13:55 -0700)]
UnsupportedAPIException: handle unicode in EAPI

Normally EAPI doesn't contain unicode, but as in bug #359675, it can
contain practically anything if files in /var/db/pkg are corrupt.

13 years agoemerge-webrsync: drop lzma support
Zac Medico [Sun, 27 Mar 2011 20:01:10 +0000 (13:01 -0700)]
emerge-webrsync: drop lzma support

The lzma snapshots have been replaced by xz snapshots, as discussed in
the "RFC: Remove .lzma in favor of .xz portage snapshots" thread on the
gentoo-dev ml:

http://archives.gentoo.org/gentoo-dev/msg_7de1b57ce531fdf33d33b5b070006436.xml

13 years agoMergeProcess: add fallback setcpv call
Zac Medico [Sun, 27 Mar 2011 06:09:27 +0000 (23:09 -0700)]
MergeProcess: add fallback setcpv call

It's important that this metadata access happens in the parent process,
since closing of file descriptors in the subprocess can prevent access
to open database connections such as that used by the sqlite metadata
cache module.

13 years agoMergeProcess: query blockers in the main process
Zac Medico [Sat, 26 Mar 2011 23:25:15 +0000 (16:25 -0700)]
MergeProcess: query blockers in the main process

Metadata cache queries may not work for some databases from within a
subprocess. For example, sqlite is known to misbehave.

13 years agodisplay_merge: pass through noiselevel args
Zac Medico [Sat, 26 Mar 2011 18:00:44 +0000 (11:00 -0700)]
display_merge: pass through noiselevel args

13 years agoScheduler: pkg_to_replace beware old-style virt
Zac Medico [Sat, 26 Mar 2011 17:10:24 +0000 (10:10 -0700)]
Scheduler: pkg_to_replace beware old-style virt

13 years agoBlockerDB: add SLOT match in discardBlocker
Zac Medico [Sat, 26 Mar 2011 16:12:38 +0000 (09:12 -0700)]
BlockerDB: add SLOT match in discardBlocker

13 years agoMergeListItem: inherit _poll and _wait
Zac Medico [Sat, 26 Mar 2011 08:52:49 +0000 (01:52 -0700)]
MergeListItem: inherit _poll and _wait

13 years agoMergeProcess: relocate portage reinstall code
Zac Medico [Sat, 26 Mar 2011 08:18:12 +0000 (01:18 -0700)]
MergeProcess: relocate portage reinstall code

This code goes inside _start since it needs to execute in the parent
process.

13 years agounmerge: fix PORTAGE_BACKGROUND logic
Zac Medico [Sat, 26 Mar 2011 07:38:23 +0000 (00:38 -0700)]
unmerge: fix PORTAGE_BACKGROUND logic

13 years agoReoranize PackageMerge/MergeListItem interaction.
Zac Medico [Sat, 26 Mar 2011 07:06:01 +0000 (00:06 -0700)]
Reoranize PackageMerge/MergeListItem interaction.

13 years agoPackageMerge: call wait() for conformity
Zac Medico [Sat, 26 Mar 2011 03:38:11 +0000 (20:38 -0700)]
PackageMerge: call wait() for conformity

13 years agoUpdate header timestamps for modified files.
Zac Medico [Sat, 26 Mar 2011 03:36:47 +0000 (20:36 -0700)]
Update header timestamps for modified files.

13 years agoMergeProcess: call elog_process for replaced pkgs
Zac Medico [Sat, 26 Mar 2011 03:24:10 +0000 (20:24 -0700)]
MergeProcess: call elog_process for replaced pkgs

13 years agodblink: use self.mycpv in _elog
Zac Medico [Sat, 26 Mar 2011 02:43:12 +0000 (19:43 -0700)]
dblink: use self.mycpv in _elog

13 years agoMergeProcess: toggle vardbapi._pkgs_changed
Zac Medico [Sat, 26 Mar 2011 02:00:16 +0000 (19:00 -0700)]
MergeProcess: toggle vardbapi._pkgs_changed

13 years agoconfig: blacklist PORTAGE_BACKGROUND* vars
Zac Medico [Sat, 26 Mar 2011 01:44:46 +0000 (18:44 -0700)]
config: blacklist PORTAGE_BACKGROUND* vars

13 years agoMergeProcess: handle unicode in elog pipe
Zac Medico [Fri, 25 Mar 2011 20:52:16 +0000 (13:52 -0700)]
MergeProcess: handle unicode in elog pipe

13 years agoenv_update: remove sleep_for_mtime_granularity
David James [Fri, 25 Mar 2011 17:10:54 +0000 (10:10 -0700)]
env_update: remove sleep_for_mtime_granularity

Instead of sleeping, use os.utime to avoid mtime ambiguity.

Change-Id: I359ececca19b6be9f5997ac5c8fe6fea6ea49dd5

Review URL: http://codereview.chromium.org/6676150

13 years agoPollScheduler: tweak _unregister event cleanup
Zac Medico [Fri, 25 Mar 2011 16:52:40 +0000 (09:52 -0700)]
PollScheduler: tweak _unregister event cleanup

13 years agoPollScheduler: handle missing log directory
Zac Medico [Fri, 25 Mar 2011 15:57:10 +0000 (08:57 -0700)]
PollScheduler: handle missing log directory

This can be triggered by AbstractPollTask._log_poll_exception(), as
reported by Michael Haubenwallner <haubi@gentoo.org> for AIX.

13 years agodblink: fix elog bugs for unmerge phases
Zac Medico [Fri, 25 Mar 2011 10:39:15 +0000 (03:39 -0700)]
dblink: fix elog bugs for unmerge phases

13 years agovartree: remove broken scheduler reference
Zac Medico [Fri, 25 Mar 2011 10:12:55 +0000 (03:12 -0700)]
vartree: remove broken scheduler reference

13 years agovartree: remove unused scheduler references
Zac Medico [Fri, 25 Mar 2011 09:47:17 +0000 (02:47 -0700)]
vartree: remove unused scheduler references

Since all the merge code runs inside MergeProcess now, there's no
reason to yield to the scheduler.

13 years agoMergeProcess: separate unmerge output from merge
Zac Medico [Fri, 25 Mar 2011 09:29:06 +0000 (02:29 -0700)]
MergeProcess: separate unmerge output from merge

The unmerge output has been mixed together with the merge output since
commit 7535cabdf2fab76fc55df83643157613dfd66be9 because
dblink._scheduler was set to None. Now it's fixed to produce separate
logs like it used to.

13 years agoebuild(1): fix AttributeError for merge phase
Zac Medico [Fri, 25 Mar 2011 08:23:51 +0000 (01:23 -0700)]
ebuild(1): fix AttributeError for merge phase

  File "pym/portage/dbapi/vartree.py", line 4043, in merge
    merge_task.start()
  File "pym/_emerge/AsynchronousTask.py", line 23, in start
    self._start()
  File "pym/_emerge/SpawnProcess.py", line 118, in _start
    self._reg_id = self.scheduler.register(files.process.fileno(),
AttributeError: 'NoneType' object has no attribute 'register'

13 years agoMergeProcess: close elog_writer_fd leak
Zac Medico [Fri, 25 Mar 2011 08:11:45 +0000 (01:11 -0700)]
MergeProcess: close elog_writer_fd leak

13 years agoMergeProcess: Fix PORTAGE_BACKGROUND/LOG_FILE use
Zac Medico [Fri, 25 Mar 2011 04:32:38 +0000 (21:32 -0700)]
MergeProcess: Fix PORTAGE_BACKGROUND/LOG_FILE use

In this subprocess we don't want PORTAGE_BACKGROUND to suppress
stdout/stderr output since they are pipes. We also don't want to open
PORTAGE_LOG_FILE, since it will already be opened by the parent
process, so we set the PORTAGE_BACKGROUND="subprocess" value for use
in conditional logging code involving PORTAGE_LOG_FILE.

13 years agoPreserve elog message continuity during updates.
Zac Medico [Fri, 25 Mar 2011 03:03:28 +0000 (20:03 -0700)]
Preserve elog message continuity during updates.

This integrates the fix from commit
8209aeab647b1ab80a64d5931069b3533776ef75 with the asynchronous merge
changes from commit 7535cabdf2fab76fc55df83643157613dfd66be9.

13 years agoMerge packages asynchronously in Portage.
David James [Fri, 25 Mar 2011 02:36:33 +0000 (19:36 -0700)]
Merge packages asynchronously in Portage.

This allows for the scheduler to continue to run while packages are
being merged and installed, allowing for additional parallelism and
making better use of the CPUs.

Review URL: http://codereview.chromium.org/6713043

13 years agoUpdate Portage to sync BlockerDB at init.
David James [Fri, 25 Mar 2011 01:14:45 +0000 (18:14 -0700)]
Update Portage to sync BlockerDB at init.

Currently, Portage syncs the BlockerDB before every merge. This slows
down merges considerably because it requires reading the entire vardb
from disk. Since Portage doesn't merge conflicting packages anyway, we
can optimize this by just reading the vardb at initialization and
caching that.

Change-Id: I6701926f022ef3aa2da10482fc8b09573ae24610

Review URL: http://codereview.chromium.org/6688037

13 years agoemerge: support [ y | n ] for --ask
Zac Medico [Thu, 24 Mar 2011 18:15:47 +0000 (11:15 -0700)]
emerge: support [ y | n ] for --ask

This allows --ask to be enabled via EMERGE_DEFAULT_OPTS, and then
temporarily disabled via the command-line. This may be useful for
programs that call emerge non-interactively, as in bug #360233.

13 years agoOptimize ebuild.sh to cache preprocessing output.
David James [Thu, 24 Mar 2011 05:05:15 +0000 (22:05 -0700)]
Optimize ebuild.sh to cache preprocessing output.

Right now ebuild.sh runs preprocessing steps multiple times during the
same build. It processes the same environment and produces the same
output every time. This slows down the build unnecessarily.

This optimization cuts build time for 50 binary packages with 16 jobs
from 3m22 to 2m33 (33%).

Change-Id: I72ac84cce2eb60d17736cc5f5343d4a858ec86f4

Review URL: http://codereview.chromium.org/6733008

13 years agoget_dep_chain: handle InvalidDependString
Zac Medico [Thu, 24 Mar 2011 04:08:35 +0000 (21:08 -0700)]
get_dep_chain: handle InvalidDependString

13 years agochk_updated_cfg_files: fix whitespace in message
Zac Medico [Mon, 21 Mar 2011 00:11:14 +0000 (17:11 -0700)]
chk_updated_cfg_files: fix whitespace in message

Thanks to Dennis Schridde in bug #359681.

13 years agoportageq owners: list orphans via stderr
Zac Medico [Sun, 20 Mar 2011 17:41:24 +0000 (10:41 -0700)]
portageq owners: list orphans via stderr

13 years agoemerge_main: add "args" parameter (sys.argv[1:])
Zac Medico [Sun, 20 Mar 2011 15:57:42 +0000 (08:57 -0700)]
emerge_main: add "args" parameter (sys.argv[1:])

Also, fix local portage.debug import so that "global portage" is
not necessary.

13 years agoTaskSequence: also call CompositeTask._cancel()
Zac Medico [Fri, 18 Mar 2011 23:45:47 +0000 (16:45 -0700)]
TaskSequence: also call CompositeTask._cancel()

This is a regression from commit
363a7478a100495d2e8597bc6c6d2fdac94339e0 but it's usually not
noticeable with Ctrl-C since SIGINT is propagated to nested processes.

13 years agoUpdate timestamps in headers of modified files. v2.2.0_alpha28
Zac Medico [Fri, 18 Mar 2011 21:44:05 +0000 (14:44 -0700)]
Update timestamps in headers of modified files.

13 years agoPackageVirtualDbapi: fix get() for reposyntax
Zac Medico [Fri, 18 Mar 2011 21:38:38 +0000 (14:38 -0700)]
PackageVirtualDbapi: fix get() for reposyntax

13 years agoCompositeTask: fix _wait for TASK_QUEUED
Zac Medico [Fri, 18 Mar 2011 21:05:07 +0000 (14:05 -0700)]
CompositeTask: fix _wait for TASK_QUEUED

Though this case might never have been triggered, the logic was broken
for cases in which self.cancelled was false and a task was queued. In
this case we need to call back into the scheduler until the queued task
is started or we are cancelled, whichever comes first.

13 years agoebuild-ipc: import errno
Zac Medico [Thu, 17 Mar 2011 21:52:41 +0000 (14:52 -0700)]
ebuild-ipc: import errno

13 years agoEbuildIpcDaemon: use non-blocking write
Zac Medico [Thu, 17 Mar 2011 21:39:53 +0000 (14:39 -0700)]
EbuildIpcDaemon: use non-blocking write

This prevents the Scheduler from hanging if the client is killed before
we can send the reply, as reported by David James:

  http://codereview.chromium.org/6713003

This commit depends on ebuild-ipc (the client) opening the other side
of this fifo before it sends its request, which has already been added
in commit eff879ff0ce7dcc1ce68d5f16de1ec73051f8c18.

13 years agoebuild-ipc: use non-blocking read
Zac Medico [Thu, 17 Mar 2011 21:12:13 +0000 (14:12 -0700)]
ebuild-ipc: use non-blocking read

This makes it possible for the daemon to send a reply without blocking,
thereby improving performance and also making it possible for the
daemon to do a non-blocking write without a race condition.

This reverts part of commit 81fc303212b8379219cf5d463c8717359b972dba,
which probably didn't help portability anyway. Now, ebuild-ipc is using
non-blocking read with os.read and EAGAIN handling, just like
EbuildIpcDaemon since commit 7e5b81da12dd7bd59f6620840dc0d824e3f4d69a
(known compatible with FreeBSD).

13 years agoAsynchronousLock: implement _cancel
Zac Medico [Thu, 17 Mar 2011 20:08:35 +0000 (13:08 -0700)]
AsynchronousLock: implement _cancel