- Pacman-dev - lists.archlinux.org

[pacman-dev] Missing sanity checks for path -> weird behaviors and segfaults
by Xavier 20 Aug '07

20 Aug '07

If Root isn't specified, paths like (null)var/ will be created. I think pacman should either default to / for Root, or just fail if it isn't set. Also, if no Cache Dirs is set, pacman will try to fallback to /tmp/ , but then it segfaults. (try pacman -Sw bar). Apparently it segfaults on this in libalpm/sync.c , sync_commit function, after downloading the file succesfully : 718 for(i = handle->dbs_sync; i; i = i->next) I found it weird that inside that loop, i was used a second time in an inner loop : 758 for(i = alpm_option_get_cachedirs(); i; i = alpm_list_next(i)) { I thought that was the problem, but it didn't fix the segfault. Second strange thing is that the files list seem to be freed in any cases : 795 FREELIST(files); but it's used at the end of that _alpm_sync_commit function : 1018 if(!validcache && !(trans->flags & PM_TRANS_FLAG_DOWNLOADONLY)) { 1019 /* delete packages */ 1020 for(i = files; i; i = i->next) { 1021 unlink(i->data); 1022 } 1023 } Finally, if the cache isn't valid (so it fallbacks to /tmp/) and DOWNLOADONLY is used, files won't be deleted, but will they be used next times? Because the fallback (lines 779-790) happen after the check of existence of the package in the cache (lines 740-744). I can't check what happens because of the segfault.

3 8

[pacman-dev] [patch] Conflict variables
by Nagy Gabor 18 Aug '07

18 Aug '07

Hi! After the professional cleanup of conflict.c (by Xavier, thx; anyway, do you want to add conflict checking to db_test?;-) we don't need to restrict conflicts to conflict-name only, we can use the depend syntax. So this patch doesn't change any important thing in the source code, mainly this is a testcase instead. Bye, ngaba

3 3

[pacman-dev] [patch] for resolvedeps + some notes
by Nagy Gabor 18 Aug '07

18 Aug '07

Hi! With this patch resolvedeps will search for other satisfiers, if we don't let it add a pkg from IgnorePkg. Note: this patch assumes my depcheck.diff patch applied. Bye PS: Resolvedeps is still dummy, because pulling a new package to the *list may break an earlier verified dependency of a pkg in the *list (because the new package will overwrite something in the localdb!). The best (but much slower) solution is to call depcheck with the *list parameter (plus syncpkg).

2 4

[pacman-dev] Pacman 3.0.5-3: strange behaviour
by Giovanni Scafora 18 Aug '07

18 Aug '07

Hi guys, I noticed that new pacman release doesn't ever show the message "is up to date" when I run pacman -Sy (or -Syu). Pacman thinks there are always new dbs. -- Giovanni Scafora Arch Linux Trusted User (voidnull) http://www.archlinux.org linuxmania(a)gmail.com

2 3

[pacman-dev] [GIT] pacman branch, master now at v3.0.0-349-gf81e619
by Dan McGee 17 Aug '07

17 Aug '07

Hello, This is an automated email from the git hooks/update script, it was generated because a ref change was pushed to the repository. Updating branch, master, via f81e619974ca47f6dba8c3002577447a71e0f1ec (commit) via fc65a9bcb1722ae9768518aa921c5f02cf282b96 (commit) via 16c91bedf844f5f7764aa78b801ca023810611c3 (commit) via 1cb6d80631359306d70bbe5fbcbbca1474594ada (commit) via ba67fdae63122b40d8fb4abf9c1c7695999f143d (commit) via ee977019e39fe33a508a60abe4a3ad7eff819b74 (commit) via 47cada81a0a8fdd4c77645233ed8406b829b552c (commit) via 8038190c7c4786e1c49494eea1b40cdddcbd5136 (commit) via 0f539832f45fa2c183cfe52c1881b3acaa54e3e6 (commit) via 064de149b7e4f6ea2891ae522102c4ae958f55e5 (commit) via f4dbb204f1ad99179c145558ababf1190f067083 (commit) via a65ad4efc1102449a087b464d3b8d4670a614b99 (commit) via 49c29e16b31ce35e81be210bc74a3834530bb9e0 (commit) from 47ee89ada9d13935510ef92541649901789c23c8 (commit) - Log ----------------------------------------------------------------- commit f81e619974ca47f6dba8c3002577447a71e0f1ec Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Fri Aug 17 14:07:56 2007 +0200 libalpm/sync.c : don't go back on STATE_PREPARED when committing fails. In my opinion, a commiting transaction that failed isn't equivalent to a prepared transaction. Some things could have been done in the meantime. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> commit fc65a9bcb1722ae9768518aa921c5f02cf282b96 Author: Dan McGee <dan(a)archlinux.org> Date: Thu Aug 16 16:19:06 2007 -0400 Fix some errors spit out by -Wextra Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 16c91bedf844f5f7764aa78b801ca023810611c3 Author: Dan McGee <dan(a)archlinux.org> Date: Thu Aug 16 15:03:00 2007 -0400 configure.ac: Clean up and document version number increments Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 1cb6d80631359306d70bbe5fbcbbca1474594ada Author: Andrew Fyfe <andrew(a)neptune-one.net> Date: Wed Jul 25 22:49:32 2007 +0100 Clean up MD5 code. * 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(a)neptune-one.net> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit ba67fdae63122b40d8fb4abf9c1c7695999f143d Author: Andrew Fyfe <andrew(a)neptune-one.net> Date: Wed Jul 25 22:35:29 2007 +0100 Remove support for SHA1 from pacman. 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(a)neptune-one.net> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit ee977019e39fe33a508a60abe4a3ad7eff819b74 Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Mon Jul 9 17:02:29 2007 +0200 libalpm/db.c : add alpm_db_unregister_all. This basically moves the code from alpm_release, which was mostly about unregistering all databases, to a safer alpm_db_unregister_all. This allows to avoid modifying the dbs_sync list while iterating over it, and and also prevent alpm_release from looping infinitely when a database can't be unregistered. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> commit 47cada81a0a8fdd4c77645233ed8406b829b552c Author: Dan McGee <dan(a)archlinux.org> Date: Thu Aug 16 11:04:58 2007 -0400 pacman.c: Alphabetize listing of query options Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 8038190c7c4786e1c49494eea1b40cdddcbd5136 Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Sat Jul 7 21:24:30 2007 +0200 libalpm/db.c : allow unregistering db for interrupted transaction. This prevents alpm_release to loop infinitely in case of an interrupted transaction, where the database wasn't unregistered. alpm_release should probably also be fixed, as it can still loop if db_unregister fails for another reason. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> commit 0f539832f45fa2c183cfe52c1881b3acaa54e3e6 Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Sun Jul 15 00:44:06 2007 +0200 new query options : explicit (-Qe) and deps (-Qd). The t shortcut for --test was removed, the orphan option (previously -Qe) was renamed to -Qt, -Qe lists all packages installed explictly, and -Qd lists all packages installed as dependencies. Besides, t can be combined with either e or d. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> commit 064de149b7e4f6ea2891ae522102c4ae958f55e5 Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Sat Jul 14 20:45:30 2007 +0200 libalpm/query.c : makes orphans and foreign options as filters. The --foreign and --orphans functions now behave as a filter for the other options. This cleans the code a bit, and will make easier the adding of new filter options, like explicit (show only explictly installed packages) or depends (show only packages installed as dependencies). Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> commit f4dbb204f1ad99179c145558ababf1190f067083 Author: Dan McGee <dan(a)archlinux.org> Date: Wed Aug 15 20:16:46 2007 -0400 src/util: Clean up headers and a few fixes Remove some unnecessary headers in the two utilities as well as fix a possible non-null termination issue in vercmp. Signed-off-by: Dan McGee <dan(a)archlinux.org> commit a65ad4efc1102449a087b464d3b8d4670a614b99 Author: Dan McGee <dan(a)archlinux.org> Date: Tue Aug 14 10:14:35 2007 -0400 package.h: implement origin_data union in pmpkg_t struct 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(a)archlinux.org> commit 49c29e16b31ce35e81be210bc74a3834530bb9e0 Author: Dan McGee <dan(a)archlinux.org> Date: Tue Aug 14 09:33:55 2007 -0400 Code cleanup Remove the commented desc_localized stuff, we can find it later in version control. Also remove some unnecessary includes of the stat header and use -fstack-protector-all which is a bit more broad. Signed-off-by: Dan McGee <dan(a)archlinux.org> ----------------------------------------------------------------------- Diffstat: configure.ac | 72 +++++- contrib/vimproject | 2 - lib/libalpm/Makefile.am | 2 - lib/libalpm/add.c | 53 ++--- lib/libalpm/alpm.c | 16 +-- lib/libalpm/alpm.h | 4 +- lib/libalpm/backup.c | 6 +- lib/libalpm/be_files.c | 49 +---- lib/libalpm/cache.c | 3 +- lib/libalpm/conflict.c | 3 - lib/libalpm/db.c | 64 +++++- lib/libalpm/db.h | 1 + lib/libalpm/md5.c | 561 ++++++++++++++++++++++++---------------------- lib/libalpm/md5.h | 122 +++++++---- lib/libalpm/md5driver.c | 93 -------- lib/libalpm/package.c | 153 ++++---------- lib/libalpm/package.h | 12 +- lib/libalpm/remove.c | 6 +- lib/libalpm/sha1.c | 431 ------------------------------------ lib/libalpm/sha1.h | 72 ------ lib/libalpm/sync.c | 25 +-- lib/libalpm/trans.c | 2 +- lib/libalpm/trans.h | 2 +- lib/libalpm/util.c | 45 ++++- src/pacman/callback.c | 3 +- src/pacman/conf.h | 2 + src/pacman/package.c | 21 +-- src/pacman/pacman.c | 36 ++-- src/pacman/query.c | 215 +++++++----------- src/pacman/sync.c | 1 - src/pacman/util.c | 2 +- src/util/Makefile.am | 2 +- src/util/testpkg.c | 8 +- src/util/vercmp.c | 23 +- 34 files changed, 742 insertions(+), 1370 deletions(-) hooks/update --- Git Source Code Management System hooks/update refs/heads/master \ 47ee89ada9d13935510ef92541649901789c23c8 \ f81e619974ca47f6dba8c3002577447a71e0f1ec

2 1

[pacman-dev] CVS update of pacman-lib/lib/libalpm (add.c)
by Dan McGee 17 Aug '07

17 Aug '07

Date: Friday, August 17, 2007 @ 16:07:05 Author: dan Path: /home/cvs-pacman/pacman-lib/lib/libalpm Modified: add.c (1.133 -> 1.134) Fix symlink overwrite issue in CVS -------+ add.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) Index: pacman-lib/lib/libalpm/add.c diff -u pacman-lib/lib/libalpm/add.c:1.133 pacman-lib/lib/libalpm/add.c:1.134 --- pacman-lib/lib/libalpm/add.c:1.133 Wed Jun 20 17:52:24 2007 +++ pacman-lib/lib/libalpm/add.c Fri Aug 17 16:07:05 2007 @@ -700,7 +700,9 @@ archive_entry_set_pathname(entry, filename); - if(archive_read_extract(archive, entry, ARCHIVE_EXTRACT_FLAGS) != ARCHIVE_OK) { + int ret = archive_read_extract(archive, entry, + ARCHIVE_EXTRACT_FLAGS | ARCHIVE_EXTRACT_NO_OVERWRITE); + if(ret != ARCHIVE_OK && ret != ARCHIVE_WARN) { _alpm_log(PM_LOG_ERROR, _("could not extract %s (%s)"), filename, strerror(errno)); alpm_logaction(_("error: could not extract %s (%s)"), filename, strerror(errno)); errors++;

1 0

[pacman-dev] Better handling of broken databases
by Xavier 16 Aug '07

16 Aug '07

An user had a broken database with duplicated entries, and pacman behaved quite weird with it : http://bbs.archlinux.org/viewtopic.php?id=36117 It's maybe possible to handle better such cases, but that's a low priority todo.

1 0

[pacman-dev] _alpm_checkconflicts may add duplicated entries to baddeps
by Nagy Gabor 14 Aug '07

14 Aug '07

Hi! As I see, if there is a conflict in package list, then we may add this twice to the conflict list (because find that conflict twice). Anyway, I don't like this assymmetric "conflict" storing. Bye, ngaba ---------------------------------------------------- SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu This mail sent through IMP: http://horde.org/imp/

3 4

[pacman-dev] resolvedeps<->resolveconflict discussion
by Nagy Gabor 14 Aug '07

14 Aug '07

Hi! Now I'd like to discuss some things I dislike in the current resolvedeps<->resolveconflict relation (they are not bugs just annoying things): 1. They are too separated from each other: I mean, we first resolve deps, but this may cause conflict, we resolve it by removing some (target) packages [we stop here now], but this may cause some broken deps, we should resolve them (by adding new targets), but this may cause a conflict again... Possible (semi)solution: resolvedeps should choose a non-conflicting package (yes, monster functions are the more effective again ;-) 2. Resolveconflicts may cause orphan install: If we remove a package from the sync target list, we don't touch its induced dependencies, so we may install some unneeded dependencies (which we pulled earlier to the target list during dep resolution). Possible solution: pmsyncpkg_t-like induced dependency-storing or we could use something similar to canremove. (On the other hand if it is needed to remove a dependency (from the target list) to resolve a conflict, we could do a cascade remove; but it's easier to say that an unresolveable conflict detected.) Bye, ngaba ---------------------------------------------------- SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu This mail sent through IMP: http://horde.org/imp/

1 0

[pacman-dev] [patch] resolveconflict clean-up
by Nagy Gabor 13 Aug '07

13 Aug '07

Hi! If you want to apply this patch, please follow my patch flow ;-) I think my alpm_sync_prepare is much cleaner than the old one. As you see, sync040.py and sync041.py now fails, because patched pacman autoresolves target vs target conflict now (I marked this as TODO: user should be asked to 'remove pkg1' or 'remove pkg2' or don't resolve this conflict.) Bye, ngaba Ps: Both the old and patched version of sync_prepare have a problem: If conflict resolution induces a replace, we can loose the PM_SYNC_TYPE_DEPEND information!!

3 4