There were a few issues with this code:
1. We already had an open fd to a file, but never used it to our benefit.
Use the libarchive convienence method to write the current file contents
straight to a file descriptor.
2. The real problem cropped up on Windows where the locking semantics caused
the old way of extraction to fail because we had an open file descriptor.
By using the file descriptor and closing it ASAP, we prevent these
failures.
Signed-off-by: Dan McGee <dan@archlinux.org>
After the libarchive upgrade from 2.4.12 to 2.4.14, our usage of
archive_entry_pathname became dangerous. We were using the result of that
function even after calls to archive_entry_set_pathname.
With 2.4.14, the entryname becomes wrong after these calls, and so all the
future use of entryname are bogus. entryname is used quite a lot for
logging, so that's not so bad. But it's also used for the backup handling,
so that's not very cool. For example, reinstalling a package with backup
entries will erase all the md5 entries from the DB, because they won't be
found back.
entryname is now a static string so that we can easily keep the result of
archive_entry_pathname.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
[Dan: fixed version numbers in commit message]
Signed-off-by: Dan McGee <dan@archlinux.org>
The _alpm_backup_split function always alloced memory for the fname, and we
let it disappear in a specific case (upgrade026.py). Fix the issue.
Signed-off-by: Dan McGee <dan@archlinux.org>
Its implementation was quite broken:
* add_loadtarget() might have silently filtered out some targets when
replacing an older version.
* This was used in sync.c to determine whether a target is implicit or not,
which is incorrect behavior. Before this patch we silently removed user
confirmed replacements; now we always warn on a replacement.
* remove001.py behavior was quite odd in adding same target 5 times to the
target list, we can change this behavior to be a failure.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
[Xav: changed remove001 pactest accordingly]
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
[Dan: rewrote commit message]
Signed-off-by: Dan McGee <dan@archlinux.org>
Fixes FS#9235.
We already had the following case in extract_single_file :
/* cases 1,2,3: couldn't stat an existing file, skip all backup checks */
But we actually only did a lstat here. And if lstat worked, we did a stat
without checking.
When lstat works and stat fails, it means we have a broken symlink, like in
FS#9235. We can actually treat this case like a non-existing file.
The broken symlink will then be simply overwritten.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
This is the symmetric of --asdeps, install packages explicitly.
Documentation and completion files were updated accordingly.
Added sync301.py and upgrade032.py pactest files to test this.
I also made a little modification in ALLDEPS handling too.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
This also affects all structures with static strings, such as depmiss,
conflict, etc. This should help a lot with memory usage, and hopefully make
things a bit more "idiot proof".
Currently our pactest pass/fail rate is identical before and after this
patch. This is not to say it is a perfect patch- I have yet to pull valgrind
out. However, this should be quite safe to use in all situations from here
on out, and we can start plugging the memleaks.
Original-work-by: Aaron Griffin <aaronmgriffin@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This was the case of the bash packaging error where a file was removed from
the package but not the backup array.
I just added a sanity check so that only the files from the backup array
that are also in the filelist are used.
I had to edit upgrade026 pactest slightly : it required the file to be
copied to .pacsave instead of moved. But just moving it should be enough, as
we agreed on the ML :
http://www.archlinux.org/pipermail/pacman-dev/2007-December/010440.html
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This is the bash case when the /etc/profile file was removed by error from
the package, but stayed in the backup array.
Ref:
http://www.archlinux.org/pipermail/arch-dev-public/2007-December/003556.html
Also fixed a little typo in add.c, but it's disabled code.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
As I mentioned earlier on the ML :
http://www.archlinux.org/pipermail/pacman-dev/2007-December/010416.html
the first part of commit 843d368ef6 had no effect because of a bug.
So I fixed the bug, but since this would change backup handling behavior,
and possibly require other bigger changes to work right, I decided to just
disable that part temporarily, and left a TODO in the code.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Update the GPL boilerplate to direct people to the GNU website for a copy of
the license, as well as bump all of Judd's copyrights to 2007.
Signed-off-by: Dan McGee <dan@archlinux.org>
alpm_list_find and alpm_list_find_ptr will now return a void *, and
alpm_list_find_str will return a char *, instead of an int.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
I started playing around with gcov today and it showed a few places in the
code that we don't test at all. This is the start of ensuring that we
execute most of the code in our codebase.
Signed-off-by: Dan McGee <dan@archlinux.org>
_alpm_innerconflicts: check for target<->target conflicts
_alpm_outerconflicts: check for target<->localpkg conflicts
This will be useful in sync.c clean-up and in testdb.c
As an application the patch also fixes a misleading message (and a memleak)
in add.c
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
checkdeps and resolvedeps now take both a remove list and an install list as
arguments, allowing dependencies to be calculated correctly.
This broke the sync990 pactest, but this pactest used dependencies and
provides in an unusual way, so it has been changed.
Dan: the sync990 pactest was just plain wrong. It didn't satisfy the
dependencies correctly, so should never have succeeded.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
[Dan: some variable renaming, clarification in commit message]
Signed-off-by: Dan McGee <dan@archlinux.org>
This operation made sense in the days before sync DBs existed, but it no
longer has the same usefulness it once did.
Signed-off-by: Dan McGee <dan@archlinux.org>
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
pmdepmissing_t was used for two totally different things :
missing dependencies, and dependency conflicts.
So this patch simply adds a type for dep conflicts,
and convert the code to use it.
This fix the TODO in conflict.c :
/* TODO WTF is a 'depmissing' doing indicating a conflict? */
Additionally, the code in conflict.c now eliminates the duplicated conflicts.
If pkg1 conflicts with pkg2, and pkg2 conflicts with pkg1, only one of them will be stored.
However the conflict handling in sync_prepare (sync.c) is still very asymetrical, and very ugly too.
This should be improved in the future (there is already a pending patch from Nagy that cleans it a lot).
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
The names related to conflicts are misleading :
For dependencies conflicts, the type is pmdepmissing,
and the function names contain just "conflict".
For file conflicts, the type is pmconflict,
and some functions contained just "conflict", some others "fileconflict".
So this is the first step for improving the situation.
Original idea/patch from Nagy, but the patch already didn't apply anymore,
so I did it again.
The main difference is that I kept the conflictype, with the following renaming :
pmconflicttype_t -> pmfileconflicttype_t
PM_CONFLICT_TYPE_TARGET -> PM_FILECONFLICT_TARGET
PM_CONFLICT_TYPE_FILE -> PM_FILECONFLICT_FILESYSTEM
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
This file only contained one private function : _alpm_db_whatprovides .
And the public alpm_db_whatprovides was in db.c , so I moved everything there.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
[Dan: updated POTFILES.in as well]
Signed-off-by: Dan McGee <dan@archlinux.org>
The old alpm_list_find was renamed to alpm_list_find_ptr, and a new
alpm_list_find was introduced, which uses the fn comparison-function
parameter in its decision.
Now both alpm_list_find_ptr (a new ptrcmp helper function was also
added) and alpm_list_find_str are just an alpm_list_find call.
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
[Dan: made ptrcmp a static function]
Signed-off-by: Dan McGee <dan@archlinux.org>
Instead of using the often-busted REQUIREDBY entries in the pacman database,
compute them each time they are required. This should help many things:
1. Simplify the codebase
2. Prevent future database corruption
3. Ensure when we do use requiredby, it is always correct
4. Shrink the pmpkg_t memory overhead
Signed-off-by: Dan McGee <dan@archlinux.org>
Linux lstat follows POSIX standards and dereferences a symlink pointing
to a directory if there is a trailing slash. For purposes of libalpm, we
don't want this so make a lstat wrapper that suppresses this behavior.
Signed-off-by: Dan McGee <dan@archlinux.org>
I broke scriptlet logging with ad691001e2.
Readd more or less what was there before, although it still needs a lot of
work including hopefully rewriting it to a new event subsystem and having
it log to a seperate file.
Signed-off-by: Dan McGee <dan@archlinux.org>
Packages and DBs now support using the UNIX epoch (seconds since Jan 1, 1970)
for use in builddate and installdate. This will only affect newly built
packages. Old existing packages with the text format are still supported, but
this is deprecated.
In the case of removal of text time support, this code will fail gracefully,
returning the start of the epoch for broken packages.
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
In most cases, we want to fully scan a package when we load it, which serves
as a integrity verification check. However, there are times when it is only
desired to read the metadata and nothing else, so allow the caller of pkg_load
to choose the behavior they need.
This pays big dividends in speeding up pacman cache cleaning functionality.
Old (729 packages):
real 1m43.717s
user 1m20.785s
sys 0m2.993s
New (729 packages):
real 0m25.607s
user 0m19.389s
sys 0m0.543s
Signed-off-by: Dan McGee <dan@archlinux.org>
Previously, package names must match a specified scheme or they will cause
pacman add operations to fail. This is not a very intelligent or necessary
way to act, so remove the dependency on the name of the package to be
installed and read all relevant information from the metadata instead.
This does have one causality to be addressed later- pacman cache cleaning
functionality, which has never been phenomenal, just lost most capability.
Signed-off-by: Dan McGee <dan@archlinux.org>
As seen with the recent upgrade of pacman and the removal of the
pacman.d/current mirrorlist, files that were formerly in the backup array
get deleted upon their removal, which could be dangerous. Instead, we should
use the combined backup array of the old and new package. This fix should
address this issue in a relatively straightforward way.
In addition, old files should be moved to pacsave locations as expected.
Signed-off-by: Dan McGee <dan@archlinux.org>
The mistake fixed in commit 26441cf65c
was actually done at two places.
This fix the second one.
Also remove one unnecessary newline introduced by
commit d34b2c4ed8
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
My two previous hacks related to this part
(8038190c7c and
b15a5194d1) were caused by the lack of
understanding of a feature introduced a while ago:
Better control over CTRL-C interruptions -- do not leave the DB in an
inconsistent state (54008798ef).
Now I have been looking at this commit, and the added feature is indeed
interesting. The main problem I had with it is that it does a rather
unusual use of alpm_trans_release, which caused a few problems that I tried
to fix in a weird way. I think these problems were caused by the fact that
there weren't any difference between "interrupt transaction" and "release a
transaction which failed" actions from the alpm_trans_release POV. So I
decided to add a new function instead, alpm_trans_interrupt, which is
called on Ctrl+C, and which only sets trans->state to STATE_INTERRUPTED so
that remove_commit and add_commit can exit cleanly at a safe moment. This
allowed me to revert my two previous hacks as well.
Also ensure we handle SIGINT correctly in all cases- if a transaction is
not ongoing, then we can free the transaction and exit quickly.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This allows us to remove the hack in the frontend where we added a newline
to everything coming out of the pm_printf functions, and instead let the
developer put newlines where they want them. This should be the last hangover
of that auto-newline stuff.
Signed-off-by: Dan McGee <dan@archlinux.org>
This patch cleans up and fix the _alpm_trans_update_depends function
and ensure that all requiredby fields are filled in case of multiple satisfiers
(previously, the handling of mutliple satisfiers in that function was inconsistent).
This makes a special case handling of requiredby in commit_single_pkg() obsolete,
and so allows cleaning that code as well.
Also fixed upgrade056 pactest because :
1) the requiredby fields were wrong, and this wouldn't happen with the fixed _alpm_trans_update_depends().
2) this is a very unusual case anyway (and handling all corner cases combined to a broken database
seems nearly impossible to achieve).
References :
http://www.archlinux.org/pipermail/pacman-dev/2007-July/008919.htmlhttp://www.archlinux.org/pipermail/pacman-dev/2007-July/008920.html
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
* Move alpm md5 functions to lib/libalpm/util.c
* Remove unneeded includes for md5.h
* Replace md5 implementation with one from http://www.xyssl.org
Dan: clean up XySSL code by removing parts we don't use, and add a note
saying what changed.
Dan: fix alpm_get_md5sum, off by one error on the malloc call and other
small things.
Signed-off-by: Andrew Fyfe <andrew@neptune-one.net>
Signed-off-by: Dan McGee <dan@archlinux.org>
There's no need for a second hashing algorithm. MD5 serves the purpose
of verifying that a package file hasn't been corrupted during download.
Signed-off-by: Andrew Fyfe <andrew@neptune-one.net>
Signed-off-by: Dan McGee <dan@archlinux.org>
We were using a void *data element in pmpkg_t before, which is unsafe by its
nature of being untyped. Reimplement data as origin_data being a union that
can hold either a path to a package file or a pointer to a cache database,
and make the other necesary updates in the code to reflect this. See package.h
for details.
Signed-off-by: Dan McGee <dan@archlinux.org>
This is a full refactoring of _alpm_add_commit into multiple functions and a
better layout for later. Some highlights:
* No regressions in pactest performance.
* Much cleaner layout. add_commit is split into four functions:
upgrade_remove, extract_single_file, commit_single_pkg, _alpm_add_commit
* No more super-functions, the heaviest (extract_single_file) is finally
under 400 lines, with much improved commenting as well.
* Incorporation of Andrew Fyfe's ideas for fixing FS #7484. Try to handle
all possible cases of extraction combinations with files, symlinks, dirs,
and non-existant entries on the filesystem. This should make our dealings
with libarchive fall more into our control.
Signed-off-by: Dan McGee <dan@archlinux.org>
This replaces the former -D operation that was undocumented and rather
hacky. It can be used with add, upgrade, or sync transactions and will affect
all packages installed. Should close FS #7193.
Also tell makepkg to use this new flag.
Signed-off-by: Dan McGee <dan@archlinux.org>
This fixes the failure of the requiredby004 pactest in a not so pretty way,
but it gets the job done. I purposely used the extremely long name of
PM_TRANS_TYPE_REMOVEUPGRADE to be both clear and in the hope that someone
else will figure out a better solution.
Original idea from Nagy Gabor, patch updated and cleaned for current code.
Signed-off-by: Dan McGee <dan@archlinux.org>
Remove versioncmp.c by moving all functions to locations that make sense.
Move replacement functions (for building without glibc) into util.c where
they belong, and do proper checks for them instead of using __sun__, etc.
Signed-off-by: Dan McGee <dan@archlinux.org>
There is no real reason to burden our translators with these messages, as
anyone helping to debug these will probably want them in English.
Signed-off-by: Dan McGee <dan@archlinux.org>
We shouldn't translate log messages to pacman.log so it is consistant and
can be parsed by other tools. Remove all gettext _() around these strings.
Signed-off-by: Dan McGee <dan@archlinux.org>
I ran flawfinder and sparse over the pacman source code and found a few
things that were worth fixing (and were quick fixes).
Signed-off-by: Dan McGee <dan@archlinux.org>
Adds the ARCHIVE_EXTRACT_NO_OVERWRITE libarchive flags
for extracting files and directories.
This will prevent symlinks for being overwritten by directories.
All other files that need to be extracted should already have
been deleted previously by pacman.
This flag is not used for extracting files in backup array
to /tmp/alpm_XXXX, because this file is created by mkstemp first,
and so needs to be overwritten by the file from the archive.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
With libarchive 2.2.3 (previously 1.3.1), archive_read_extract now returns ARCHIVE_WARN
when a package is extracted as user, because for example, UID=0 or SUID bit can't be set.
This patch makes pacman not treating these warnings as errors anymore,
but simply ignoring them.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com>
For future possibilities, don't extract any files that start with '.'.
This will allow us to add features such as the ChangeLog viewing without
having to wait to include these files in packages, because older versions
of pacman will be forward compatable with 'hidden' files at the root level
of the package.
Signed-off-by: Dan McGee <dan@archlinux.org>
This was a bad way to fix a problem upstream. Doing this resulted in
symlink permissions being applied to files installed on the system, leaving
some binaries with 77 permission, etc.
This reverts commit 4e6b7c1cde.
Add some 'const' keywords all over the code to make it a bit more strict on
what you can and can't do with data. This is especially important when we
return pointers to the pacman frontend- ideally this would always be
untouchable data.
Signed-off-by: Dan McGee <dan@archlinux.org>
This code depends on /etc/mtab existance, which is not very reliable in all
cases, especially in a chroot or non-Linux environment. Dump it for now
until we can find a better way.
Signed-off-by: Dan McGee <dan@archlinux.org>
Hackish fix to ensure libarchive extracts files and dirs with the right
permissions. For some reason extracting /tmp on install wasn't handled properly
by librachive, so an explicit chmod will fix this.
Signed-off-by: Aaron Griffin <aaronmgriffin@gmail.com>
* Remove some unnecessary conditional compiling in util.h- move the
functions tha required it to trans.c (along with a bunch of new header
includes).
* Clean up util.h a bit- remove some header includes, remove universal
libarchive include and only put it in the files that need it.
Signed-off-by: Dan McGee <dan@archlinux.org>
Remove inclusion of libintl.h from all files, because we can do it once
in util.c where the _() macro is defined.
Signed-off-by: Dan McGee <dan@archlinux.org>
the alpm strcmp operation which takes void* references.
* We had this great visibility patch, but never actually took advantage of
it. Added the right compile flag to make it work and added some more
SYMEXPORTs where necessary to have a successful compile.
reordering and adding ones that were forgotten (noticed when trying to
compile after reordering).
* Updated the HACKING file to include information on #include usage.
* print -> vprint in "making dir" function in pactest.
specified
* Use db->path when appropriate
* Commented out the FAKEROOT checks in libalpm. This should never ever be done.
TODO test this quite a bit, as this will never cause the transactions to fail
if RW operations are requested... right now it is totally up to the front end
to decide when to fail
* Use realpath() to canonicalize the root path when specified, so
_alpm_makepath() doesn't freak out
* Fixed some output/indent of MDFile and SHAFile algorithms
* More efficient sprintf() usage in MDFile/SHAFile
* Added real error output to _alpm_makepath
* Moved entirely to alpm_pkg_get_* accessors, to read data on demand
* Mostly removed the INFRQ_ parameters from outside the be_files backend (making
the backend more extensible in the long run)
* packages created from _alpm_db_scan now have the db and origin set (making
accessors actually work for these packages)
* removed _alpm_db_ensure_pkgcache
* totally revamped the _alpm_checkconflicts function, making it cleaner and
easier to read (and thus fix in the long run) - maintainable code ftw
NOTE: feel free to rename the functions... I couldn't think of anything better
* removed an extra loop in sync.c:find_replacements - no sense in looping over
an entire DB while strcmp'ing the name, when we have get_pkgfromcache
Other:
* package struct "license" -> "licenses"
* Created _alpm_sync_find (duplicate code in some places, find_pkginsync
* Minor const correctness changes along the way
* fixed a couple extra '/' pathing issues (non-issues really)
* removed a duplicate pkg_cmp function
* Added some provision switching hackery. This could probably use some
refactoring,.. it solves the following case:
pkg1 and pkg2 provide 'foo' and are both installed
pkg3 depends on 'foo' and so lists 'pkg1' in the REQUIREDBY db section
pkg1 is upgraded and no longer provides 'foo'
** This code ensures that the REQUIREDBY of pkg3 is updated to require pkg2
now instead of pkg1
* Unified some functions names "package" -> "pkg" for consistency
* Removed the goofy 'faketarget' stuff used for dep testing
* Renamed alpm_pkg_isin -> alpm_pkg_find
* Renamed alpm_db_readpkg -> alpm_db_get_pkg
* Fixed a segfault in the conflict checking code
* Added an automatic failure in the case of -A/-U for a replacement of an
existing package. This requires a large amount of work and is postponed for
now. Example:
If ncmpc is installed, pacman -U /path/to/ncmpc-svn.pkg.tar.gz will fail
with and appropriate error message
from one package to another seemlessly (knock on wood). This is implemented
through the use of two skip lists in the trans struct- skip_add and
skip_remove, which replace the former trans->skiplist.
* Removed an unnecessary function parameter, added a necessary one.
* If a package has no backup files, print '(none)' under the heading so it is
more obvious.
* Updated my TODO list.
- the code should be clearer, more organized, commented, and have worthwhile
variable names now
- proactive backup=()s now work. That is, adding a file to a backup array
does what it should on the upgrade to that package, no longer forcing you to
wait a full upgrade cycle for it to take effect
* ldconfig was being run twice on an upgrade operation - fixed
* fixed another pm_fprintf/printf output corruption with the progress bars
* refactored some duplicate code for adjusting 'requiredby' lists
* Added config.rpath to .cvsignore
from the PKGINFO, and size is the stat() size of the archive
* Removed the useless 'output' param from package.c:parse_descfile
* Installation progress
- Call progress callback once at 0% for initialization
- 'needdisp' was useless
- alpm_list_count is called an excessive amount in these nested loops. Now we
only call it once per iteration
- Use the compressed sizes for PROGRESS calcs as uncompressed (isize) is not
exact (it is missing metadata sizes), and thus produces > 100% numbers
* The --debug params were goofy. New setup allows --debug without params,
--debug=<level> where level 1=debug output, 2=debug and download output,
3=debug, download, and function tracing output. This seems more sane to me.
* Removed PM_LOG_FLOW1 and PM_LOG_FLOW2. They were just confusing. When adding
new functions, it is near impossible to determin if your output should be
"flow1" or "flow2" without tracking all the way up the call chain. Rarely
would one ever say "ok, lets just show "flow2" output. These have both been
replaced with PM_LOG_DEBUG
* Removed the need for the root parameter on alpm_initialize. it is now
defaulted to PM_ROOT just like dbpath and cachedir. This allows alpm to be
initialized BEFORE option parsing in the front end, saving us some duplicate
variables in the frontend.
* Cleaned up front end variables due to early alpm_initialize call.
* Fix up add.c a bit better than it was in regards to FS #3492.
* Optimized the sqrt call in dependency cycle checking to a single call.
* Removal of an outdated comment.
* alpm_list_is_in --> alpm_list_find
* alpm_list_is_strin --> alpm_list_find_str
* Flip parameters of both functions to be inline with rest of alpm_list.
First commit, woohoo.
optimizations in order to make the code much more readable and
type-checkable. Every enum in the library now has it's own type that
should be used instead of the generic 'unsigned char'. In addition,
several #define statements dealing with constants were converted to
enums.
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
