January 2018 - Pacman-dev - lists.archlinux.org

[pacman-dev] [GIT] The official pacman repository branch, master, updated. v5.0.1-223-ga7dbe463
by allan＠archlinux.org 19 Jan '18

19 Jan '18

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "The official pacman repository". The branch, master has been updated via a7dbe4635b4af9b5bd927ec0e7a211d1f6c1b0f2 (commit) via 9058d7fe9aa2bb494d5a6572c167db75d6d02d30 (commit) via d911438352ebec5a887dac02ceb0bb99f54b34fc (commit) via 19dabcc19a476735c4273e7ba4a0a7437d83078f (commit) via b2e088e42cfb5be2ed7247040a50acf92916d1c2 (commit) via d9eda13fc69eb0bfbbab6d7488e99e5c7b033d3a (commit) via 170bb80a1f8a020595738c7febdb15c68574ce29 (commit) via fa745fb4677c3de37aad93b180fe9475179eb7e6 (commit) via 0bbf9ddf9dff45346fb95bc6b6f1db8320551b6a (commit) via e8462a4f88335b200ad0a1c5d96e05a44af7e67c (commit) via 653d2dc86d5340c6854818bdab3dae65bf62d474 (commit) via 4d80d87f9069fba17ebf4d91cdb097b433a57908 (commit) via 2bda849bf9b39b423175d1ee1d8796b856cc9988 (commit) via 59b6fdeee1722a58d1599bcbd1e2c0fc33debc99 (commit) via 24ab8717d459c07c51f47b7b66736de17ca23e9e (commit) from 3a9a510ca873df123a117133c7cb3ebaa711159c (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit a7dbe4635b4af9b5bd927ec0e7a211d1f6c1b0f2 Author: Jelle van der Waa <jelle(a)vdwaa.nl> Date: Sat Jan 13 16:43:12 2018 +0100 Add BUILDINFO man page Add a new man page which describes the structure of a BUILDINFO file included in a package produced by makepkg. Signed-off-by: Jelle van der Waa <jelle(a)vdwaa.nl> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 9058d7fe9aa2bb494d5a6572c167db75d6d02d30 Author: Eli Schwartz <eschwartz(a)archlinux.org> Date: Sat Jan 13 23:39:39 2018 -0500 pacman-key: do not set a default keyserver Recent versions of GnuPG are perfectly capable of using sane defaults, and the default SKS keyserver over hkps:// is better than hardcoding the same keyserver over hkp:// anyway. Fixes FS#55278 Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit d911438352ebec5a887dac02ceb0bb99f54b34fc Author: Eli Schwartz <eschwartz(a)archlinux.org> Date: Sun Jan 14 00:11:08 2018 -0500 vercmp: fail when the wrong number of arguments are provided Fixes FS#49093 Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 19dabcc19a476735c4273e7ba4a0a7437d83078f Author: Eli Schwartz <eschwartz(a)archlinux.org> Date: Sat Jan 13 23:52:38 2018 -0500 Fix usage of install_package In commit ab2be5794de8a57426ec3fc0f631cc766a0b8227 return codes were implemented, and the output of install_package was improperly assigned to a variable when the return code was wanted. All we need to do is restore the previous exit handling, but return $? instead of hardcoding "0". Reported-by: xftroxgpx <xftroxgpx(a)protonmail.com> Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit b2e088e42cfb5be2ed7247040a50acf92916d1c2 Author: Joel Teichroeb <joel(a)teichroeb.net> Date: Mon Jan 15 13:15:23 2018 -0800 use pacman-conf in bash completion Signed-off-by: Joel Teichroeb <joel(a)teichroeb.net> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit d9eda13fc69eb0bfbbab6d7488e99e5c7b033d3a Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sat Jan 13 11:50:00 2018 -0500 use pacman-conf in scripts Because parsing pacman.conf is so difficult that even we can't do it right. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 170bb80a1f8a020595738c7febdb15c68574ce29 Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sat Jan 13 11:49:59 2018 -0500 add pacman-conf utility Parsing pacman's configuration file is non-trivial and extremely difficult to do correctly from scripts; even our own do it incorrectly. pacman-conf is a dedicated tool specifically to allow scripts to parse config files, getting the same value that pacman itself would use. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit fa745fb4677c3de37aad93b180fe9475179eb7e6 Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sat Jan 13 11:49:58 2018 -0500 extract default settings to separate function Default values for configuration settings were being set during alpm setup and in some cases were never saved back to the original config struct. Refactoring all default settings into a separate function and saving them onto the original config struct will allow pacman-conf to resolve the defaults without having to setup alpm. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 0bbf9ddf9dff45346fb95bc6b6f1db8320551b6a Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sat Jan 13 11:49:57 2018 -0500 extract raw config file parser To allow pacman-conf to parse the configuration file without having to also setup alpm. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit e8462a4f88335b200ad0a1c5d96e05a44af7e67c Author: Allan McRae <allan(a)archlinux.org> Date: Wed Jan 10 18:18:34 2018 +1000 Support new OpenPGP format packets lengths RFC 4880 defines two packet formats for OpenPGP. Pacman aborted its key in keyring check with an error message if it encountered the new format. This was fine until some annoying Arch Trusted User generated a key using the new format! Implement the new format. This also required parsing the hashed sub packets. requiring the parsing code to moved to its own function. Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 653d2dc86d5340c6854818bdab3dae65bf62d474 Author: Allan McRae <allan(a)archlinux.org> Date: Tue Jan 9 12:50:37 2018 +1000 Fix moving source files into debug packages Commit 8bec63bf92d8dd028aa88dbd5109c314cdb9ebea attempted to switch to using -fdebug-prefix-map to set file locations in debug packages. It make a few mistakes... 1) Adding debug C{,XX}FLAGS only worked if DBGSRCDIR was defined in makepkg.conf. Fix this by falling back to the default value. 2) Using -fdebug-prefix-map altered a lot of assumptions about file locations when copying source files into debug packages. This resulted in lots of messages of failed cp in packaging output. Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 4d80d87f9069fba17ebf4d91cdb097b433a57908 Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sun Jan 7 19:30:41 2018 -0500 add missing newline to debug message Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 2bda849bf9b39b423175d1ee1d8796b856cc9988 Author: Andrew Gregory <andrew.gregory.8(a)gmail.com> Date: Sun Jan 7 19:30:42 2018 -0500 detect pkghash allocation failure If rehash ever failed with a full hash it would return the old hash that is already full. get_hash_position would then loop forever because it would never find an empty bucket. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 59b6fdeee1722a58d1599bcbd1e2c0fc33debc99 Author: Allan McRae <allan(a)archlinux.org> Date: Tue Jan 9 22:00:46 2018 +1000 Do not continuously try to open an invalid database If you manage to download a bad database (e.g. an html file when behind a proxy or with a badly configured webserver), pacman makes sure you know about it. Here is some example output: error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format error: could not open file /var/lib/pacman/sync/extra.db: Unrecognized archive format I don't know how many times that gets printed because it goes beyond my scrollback buffer. Flag a database that we can "open" and "fstat" but not read from as invalid to avoid this. Signed-off-by: Allan McRae <allan(a)archlinux.org> commit 24ab8717d459c07c51f47b7b66736de17ca23e9e Author: Allan McRae <allan(a)archlinux.org> Date: Tue Jan 9 20:48:35 2018 +1000 Handle empty string passed to query_owner Passing an empty string to pacman -Qo results in: error: No package owns <first directory in $PATH> Catch empty strings and report an error. Signed-off-by: Allan McRae <allan(a)archlinux.org> ----------------------------------------------------------------------- Summary of changes: doc/BUILDINFO.5.txt | 67 ++++++ doc/Makefile.am | 4 +- lib/libalpm/be_local.c | 5 +- lib/libalpm/be_sync.c | 9 +- lib/libalpm/db.c | 5 +- lib/libalpm/pkghash.c | 23 +- lib/libalpm/pkghash.h | 4 +- lib/libalpm/signing.c | 178 ++++++++------ scripts/completion/bash_completion.in | 6 +- scripts/completion/zsh_completion.in | 4 +- scripts/libmakepkg/tidy/strip.sh.in | 9 +- scripts/makepkg.sh.in | 14 +- scripts/pacman-db-upgrade.sh.in | 21 +- scripts/pacman-key.sh.in | 6 +- src/pacman/.gitignore | 2 + src/pacman/Makefile.am | 17 +- src/pacman/conf.c | 179 +++++++------- src/pacman/conf.h | 2 + src/pacman/pacman-conf.c | 437 ++++++++++++++++++++++++++++++++++ src/pacman/query.c | 5 + src/util/vercmp.c | 13 +- 21 files changed, 793 insertions(+), 217 deletions(-) create mode 100644 doc/BUILDINFO.5.txt create mode 100644 src/pacman/pacman-conf.c hooks/post-receive -- The official pacman repository

1 0

[pacman-dev] [PATCH 0/1] Add a BUILDINFO man page
by Jelle van der Waa 19 Jan '18

19 Jan '18

While working on the reprodcible build project I found it handy to have some resources about the contents of the BUILDINFO file. This patch adds a man page about the structure of the BUILDINFO file, I've based it on the PKGBUILD man page. If this patch is accepted, I'll be motivated to add one for PKGINFO as well. Jelle van der Waa (1): Add BUILDINFO man page doc/BUILDINFO.5.txt | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++ doc/Makefile.am | 4 +++- 2 files changed, 68 insertions(+), 1 deletion(-) create mode 100644 doc/BUILDINFO.5.txt -- 2.15.1

3 10

[pacman-dev] [PATCH] use pacman-conf in bash completion
by Joel Teichroeb 19 Jan '18

19 Jan '18

Signed-off-by: Joel Teichroeb <joel(a)teichroeb.net> --- scripts/completion/bash_completion.in | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/scripts/completion/bash_completion.in b/scripts/completion/bash_completion.in index 06963c42..9ef43238 100644 --- a/scripts/completion/bash_completion.in +++ b/scripts/completion/bash_completion.in @@ -96,6 +96,10 @@ _pacman_pkg() { )" } +_pacman_repo_list() { + _arch_compgen "$(pacman-conf --repo-list)" +} + _pacman() { local common core cur database files prev query remove sync upgrade o COMPREPLY=() @@ -136,7 +140,7 @@ _pacman() { _pacman_pkg Qq;; S) { _arch_incomp 'g groups' && _pacman_pkg Sg; } || - { _arch_incomp 'l list' && _pacman_pkg Sl sort; } || + { _arch_incomp 'l list' && _pacman_repo_list; } || _pacman_pkg Slq;; U) _pacman_file;; -- 2.15.1

2 1

[pacman-dev] [PATCH 0/4] add pacman-conf utility
by Andrew Gregory 18 Jan '18

18 Jan '18

Parsing pacman.conf is so hard that even our own scripts don't do it correctly. pacman-conf is a dedicated utility scripts can use to correctly parse values from pacman.conf. This is a conversion of an earlier version of pacconf from pacutils that has been relicensed and renamed to avoid a file conflict. I make no promises that I've correctly adjusted the automake configuration for this type of build, but it works for me. Andrew Gregory (4): extract raw config file parser extract default settings to separate function add pacman-conf utility use pacman-conf in scripts scripts/completion/zsh_completion.in | 4 +- scripts/pacman-db-upgrade.sh.in | 21 +- scripts/pacman-key.sh.in | 2 +- src/pacman/conf.c | 179 +++++++------- src/pacman/conf.h | 2 + src/util/.gitignore | 2 + src/util/Makefile.am | 21 +- src/util/pacman-conf.c | 437 +++++++++++++++++++++++++++++++++++ 8 files changed, 562 insertions(+), 106 deletions(-) create mode 100644 src/util/pacman-conf.c -- 2.15.1

3 10

[pacman-dev] [PATCH 1/3] vercmp: fail when the wrong number of arguments are provided
by Eli Schwartz 14 Jan '18

14 Jan '18

Fixes FS#49093 Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> --- src/util/vercmp.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/src/util/vercmp.c b/src/util/vercmp.c index 3521a96a..7b34b30a 100644 --- a/src/util/vercmp.c +++ b/src/util/vercmp.c @@ -40,8 +40,6 @@ static void usage(void) int main(int argc, char *argv[]) { - const char *s1 = ""; - const char *s2 = ""; int ret; if(argc == 1) { @@ -54,14 +52,12 @@ int main(int argc, char *argv[]) usage(); return 0; } - if(argc > 2) { - s2 = argv[2]; - } - if(argc > 1) { - s1 = argv[1]; + if(argc != 3) { + printf("error: %d arguments specified but vercmp needs 2\n", argc-1); + return EXIT_FAILURE; } - ret = alpm_pkg_vercmp(s1, s2); + ret = alpm_pkg_vercmp(argv[1], argv[2]); printf("%d\n", ret); return EXIT_SUCCESS; } -- 2.15.1

3 7

[pacman-dev] [PATCH] Fix usage of install_package
by Eli Schwartz 14 Jan '18

14 Jan '18

In commit ab2be5794de8a57426ec3fc0f631cc766a0b8227 return codes were implemented, and the output of install_package was improperly assigned to a variable when the return code was wanted. All we need to do is restore the previous exit handling, but return $? instead of hardcoding "0". Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> --- scripts/makepkg.sh.in | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index bbf7fc59..e9e04d99 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -1071,8 +1071,8 @@ check_build_status() { && ! (( FORCE || SOURCEONLY || NOBUILD || NOARCHIVE)); then if (( INSTALL )); then warning "$(gettext "A package has already been built, installing existing package...")" - status=$(install_package) - exit $status + install_package + exit $? else error "$(gettext "A package has already been built. (use %s to overwrite)")" "-f" exit $E_ALREADY_BUILT @@ -1094,8 +1094,8 @@ check_build_status() { if (( allpkgbuilt )); then if (( INSTALL )); then warning "$(gettext "The package group has already been built, installing existing packages...")" - status=$(install_package) - exit $status + install_package + exit $? else error "$(gettext "The package group has already been built. (use %s to overwrite)")" "-f" exit $E_ALREADY_BUILT -- 2.15.1

1 1

[pacman-dev] [PATCH] pacman-key: do not set a default keyserver
by Eli Schwartz 14 Jan '18

14 Jan '18

Recent versions of GnuPG are perfectly capable of using sane defaults, and the default SKS keyserver over hkps:// is better than hardcoding the same keyserver over hkp:// anyway. Fixes FS#55278 Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> --- scripts/pacman-key.sh.in | 4 ---- 1 file changed, 4 deletions(-) diff --git a/scripts/pacman-key.sh.in b/scripts/pacman-key.sh.in index 7b158da7..8996e127 100644 --- a/scripts/pacman-key.sh.in +++ b/scripts/pacman-key.sh.in @@ -51,8 +51,6 @@ UPDATEDB=0 USE_COLOR='y' VERIFY=0 -DEFAULT_KEYSERVER='hkp://pool.sks-keyservers.net' - m4_include(library/output_format.sh) usage() { @@ -210,8 +208,6 @@ initialize() { add_gpg_conf_option "$conffile" 'no-greeting' add_gpg_conf_option "$conffile" 'no-permission-warning' add_gpg_conf_option "$conffile" 'lock-never' - keyserv=${KEYSERVER:-$DEFAULT_KEYSERVER} - add_gpg_conf_option "$conffile" 'keyserver' "$keyserv" add_gpg_conf_option "$conffile" 'keyserver-options' 'timeout=10' # set up a private signing key (if none available) -- 2.15.1

1 0

[pacman-dev] Road to 5.1
by Allan McRae 11 Jan '18

11 Jan '18

Hi all, We have not had a release in a long, long time. Lets fix that! There is a substantial number of bugs tagged for a 5.1 release [1]. These were added when we were all younger and more optimistic about the world... I have added a 5.2 target, so we can start bumping some to a future release. Please reply here with anything you think is essential to the next release - even better if you provide a patch! Cheers, Allan [1] https://bugs.archlinux.org/roadmap/proj3

4 4

[pacman-dev] [PATCH] Support new OpenPGP format packets lengths
by Allan McRae 10 Jan '18

10 Jan '18

RFC 4880 defines two packet formats for OpenPGP. Pacman aborted its key in keyring check with an error message if it encountered the new format. This was fine until some annoying Arch Trusted User generated a key using the new format! Implement the new format. This also required parsing the hashed sub packets. requiring the parsing code to moved to its own function. Signed-off-by: Allan McRae <allan(a)archlinux.org> --- This patch is less daunting than it appears. One large pice of code got and increased indent and one large part got moved to its own function. I adjusted some of the length_checks to always explicitly check != 0. I also had to adjust one that was giving false positives - we don't care if the potential read is going beyond the packet, only if it is going beyond the end of the string. Can someone check those? lib/libalpm/signing.c | 178 ++++++++++++++++++++++++++++++-------------------- 1 file changed, 107 insertions(+), 71 deletions(-) diff --git a/lib/libalpm/signing.c b/lib/libalpm/signing.c index 51b11df6..f88b7a02 100644 --- a/lib/libalpm/signing.c +++ b/lib/libalpm/signing.c @@ -999,6 +999,51 @@ static size_t length_check(size_t length, size_t position, size_t a, } } +static int parse_subpacket(alpm_handle_t *handle, const char *identifier, + const unsigned char *sig, const size_t len, const size_t pos, + const size_t plen, alpm_list_t **keys) +{ + size_t slen; + size_t spos = pos; + + while(spos < pos + plen) { + if(sig[spos] < 192) { + slen = sig[spos]; + spos = spos + 1; + } else if(sig[spos] < 255) { + if(length_check(len, spos, 2, handle, identifier) != 0){ + return -1; + } + slen = (sig[spos] << 8) | sig[spos + 1]; + spos = spos + 2; + } else { + if(length_check(len, spos, 5, handle, identifier) != 0) { + return -1; + } + slen = (sig[spos + 1] << 24) | (sig[spos + 2] << 16) | (sig[spos + 3] << 8) | sig[spos + 4]; + spos = spos + 5; + } + if(sig[spos] == 16) { + /* issuer key ID */ + char key[17]; + size_t i; + if(length_check(len, spos, 8, handle, identifier) != 0) { + return -1; + } + for (i = 0; i < 8; i++) { + sprintf(&key[i * 2], "%02X", sig[spos + i + 1]); + } + *keys = alpm_list_add(*keys, strdup(key)); + break; + } + if(length_check(len, spos, slen, handle, identifier) != 0) { + return -1; + } + spos = spos + slen; + } + return 0; +} + /** * Extract the Issuer Key ID from a signature * @param sig PGP signature @@ -1009,7 +1054,7 @@ static size_t length_check(size_t length, size_t position, size_t a, int SYMEXPORT alpm_extract_keyid(alpm_handle_t *handle, const char *identifier, const unsigned char *sig, const size_t len, alpm_list_t **keys) { - size_t pos, spos, blen, hlen, ulen, slen; + size_t pos, blen, hlen, ulen; pos = 0; while(pos < len) { @@ -1020,49 +1065,69 @@ int SYMEXPORT alpm_extract_keyid(alpm_handle_t *handle, const char *identifier, } if(sig[pos] & 0x40) { - /* "new" packet format is not supported */ - _alpm_log(handle, ALPM_LOG_ERROR, - _("%s: unsupported signature format\n"), identifier); - return -1; - } - - if(((sig[pos] & 0x3f) >> 2) != 2) { - /* signature is not a "Signature Packet" */ - _alpm_log(handle, ALPM_LOG_ERROR, - _("%s: signature format error\n"), identifier); - return -1; - } + /* new packet format */ + if(length_check(len, pos, 1, handle, identifier) != 0) { + return -1; + } + pos = pos + 1; - switch(sig[pos] & 0x03) { - case 0: - if(length_check(len, pos, 2, handle, identifier) != 0) { + if(sig[pos] < 192) { + if(length_check(len, pos, 1, handle, identifier) != 0) { return -1; } - blen = sig[pos + 1]; - pos = pos + 2; - break; - - case 1: - if(length_check(len, pos, 3, handle, identifier)) { + blen = sig[pos]; + pos = pos + 1; + } else if (sig[pos] < 224) { + if(length_check(len, pos, 2, handle, identifier) != 0) { return -1; } - blen = (sig[pos + 1] << 8) | sig[pos + 2]; - pos = pos + 3; - break; - - case 2: + blen = ((sig[pos] - 192) << 8) + sig[pos + 1] + 192; + pos = pos + 2; + } else if (sig[pos] == 255) { if(length_check(len, pos, 5, handle, identifier)) { return -1; } blen = (sig[pos + 1] << 24) | (sig[pos + 2] << 16) | (sig[pos + 3] << 8) | sig[pos + 4]; pos = pos + 5; - break; - - case 3: + } else { /* partial body length not supported */ _alpm_log(handle, ALPM_LOG_ERROR, _("%s: unsupported signature format\n"), identifier); return -1; + } + } else { + /* old package format */ + switch(sig[pos] & 0x03) { + case 0: + if(length_check(len, pos, 2, handle, identifier) != 0) { + return -1; + } + blen = sig[pos + 1]; + pos = pos + 2; + break; + + case 1: + if(length_check(len, pos, 3, handle, identifier) != 0) { + return -1; + } + blen = (sig[pos + 1] << 8) | sig[pos + 2]; + pos = pos + 3; + break; + + case 2: + if(length_check(len, pos, 5, handle, identifier) != 0) { + return -1; + } + blen = (sig[pos + 1] << 24) | (sig[pos + 2] << 16) | (sig[pos + 3] << 8) | sig[pos + 4]; + pos = pos + 5; + break; + + case 3: + /* partial body length not supported */ + _alpm_log(handle, ALPM_LOG_ERROR, + _("%s: unsupported signature format\n"), identifier); + return -1; + } } if(sig[pos] != 4) { @@ -1071,14 +1136,12 @@ int SYMEXPORT alpm_extract_keyid(alpm_handle_t *handle, const char *identifier, _("%s: unsupported signature format\n"), identifier); return -1; } - if(sig[pos + 1] != 0x00) { /* not a signature of a binary document */ _alpm_log(handle, ALPM_LOG_ERROR, _("%s: signature format error\n"), identifier); return -1; } - pos = pos + 4; /* pos got changed above, so an explicit check is necessary @@ -1087,55 +1150,28 @@ int SYMEXPORT alpm_extract_keyid(alpm_handle_t *handle, const char *identifier, return -1; } hlen = (sig[pos] << 8) | sig[pos + 1]; + if(length_check(len, pos, hlen + 2, handle, identifier) != 0) { + return -1; + } + pos = pos + 2; - if(length_check(len, pos, hlen + 2, handle, identifier)) { + if(parse_subpacket(handle, identifier, sig, len, pos, hlen, keys) == -1) { return -1; } - pos = pos + hlen + 2; + pos = pos + hlen; ulen = (sig[pos] << 8) | sig[pos + 1]; + if(length_check(len, pos, hlen + 2, handle, identifier) != 0) { + return -1; + } pos = pos + 2; - spos = pos; - - while(spos < pos + ulen) { - if(sig[spos] < 192) { - slen = sig[spos]; - spos = spos + 1; - } else if(sig[spos] < 255) { - if(length_check(pos + ulen, spos, 2, handle, identifier)){ - return -1; - } - slen = (sig[spos] << 8) | sig[spos + 1]; - spos = spos + 2; - } else { - /* check for pos and spos, as spos is still pos */ - if(length_check(len, pos, 5, handle, identifier)) { - return -1; - } - slen = (sig[spos + 1] << 24) | (sig[spos + 2] << 16) | (sig[spos + 3] << 8) | sig[spos + 4]; - spos = spos + 5; - } - if(sig[spos] == 16) { - /* issuer key ID */ - char key[17]; - size_t i; - if(length_check(pos + ulen, spos, 8, handle, identifier)) { - return -1; - } - for (i = 0; i < 8; i++) { - sprintf(&key[i * 2], "%02X", sig[spos + i + 1]); - } - *keys = alpm_list_add(*keys, strdup(key)); - break; - } - if(length_check(pos + ulen + 1, spos, slen, handle, identifier)) { - return -1; - } - spos = spos + slen; + if(parse_subpacket(handle, identifier, sig, len, pos, hlen, keys) == -1) { + return -1; } pos = pos + (blen - hlen - 8); } + return 0; } -- 2.15.1

1 0

[pacman-dev] [PATCH] detect pkghash allocation failure
by Andrew Gregory 10 Jan '18

10 Jan '18

If rehash ever failed with a full hash it would return the old hash that is already full. get_hash_position would then loop forever because it would never find an empty bucket. Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> --- We still don't actually handle pkghash creation errors anywhere, but at least it won't loop forever anymore. lib/libalpm/be_local.c | 5 ++++- lib/libalpm/be_sync.c | 5 ++++- lib/libalpm/db.c | 5 ++++- lib/libalpm/pkghash.c | 23 ++++++++++++++--------- lib/libalpm/pkghash.h | 4 ++-- 5 files changed, 28 insertions(+), 14 deletions(-) diff --git a/lib/libalpm/be_local.c b/lib/libalpm/be_local.c index 97a49688..1ec6a20f 100644 --- a/lib/libalpm/be_local.c +++ b/lib/libalpm/be_local.c @@ -601,7 +601,10 @@ static int local_db_populate(alpm_db_t *db) /* add to the collection */ _alpm_log(db->handle, ALPM_LOG_FUNCTION, "adding '%s' to package cache for db '%s'\n", pkg->name, db->treename); - db->pkgcache = _alpm_pkghash_add(db->pkgcache, pkg); + if(_alpm_pkghash_add(&db->pkgcache, pkg) == NULL) { + _alpm_pkg_free(pkg); + RET_ERR(db->handle, ALPM_ERR_MEMORY, -1); + } count++; } diff --git a/lib/libalpm/be_sync.c b/lib/libalpm/be_sync.c index 84a00c0f..35cd8a4d 100644 --- a/lib/libalpm/be_sync.c +++ b/lib/libalpm/be_sync.c @@ -413,7 +413,10 @@ static alpm_pkg_t *load_pkg_for_entry(alpm_db_t *db, const char *entryname, /* add to the collection */ _alpm_log(db->handle, ALPM_LOG_FUNCTION, "adding '%s' to package cache for db '%s'\n", pkg->name, db->treename); - db->pkgcache = _alpm_pkghash_add(db->pkgcache, pkg); + if(_alpm_pkghash_add(&db->pkgcache, pkg) == NULL) { + _alpm_pkg_free(pkg); + RET_ERR(db->handle, ALPM_ERR_MEMORY, NULL); + } } else { free(pkgname); free(pkgver); diff --git a/lib/libalpm/db.c b/lib/libalpm/db.c index 789478e8..308fb9c7 100644 --- a/lib/libalpm/db.c +++ b/lib/libalpm/db.c @@ -589,7 +589,10 @@ int _alpm_db_add_pkgincache(alpm_db_t *db, alpm_pkg_t *pkg) ? ALPM_PKG_FROM_LOCALDB : ALPM_PKG_FROM_SYNCDB; newpkg->origin_data.db = db; - db->pkgcache = _alpm_pkghash_add_sorted(db->pkgcache, newpkg); + if(_alpm_pkghash_add_sorted(&db->pkgcache, newpkg) == NULL) { + _alpm_pkg_free(newpkg); + RET_ERR(db->handle, ALPM_ERR_MEMORY, -1); + } free_groupcache(db); diff --git a/lib/libalpm/pkghash.c b/lib/libalpm/pkghash.c index 45e63082..a0e81f28 100644 --- a/lib/libalpm/pkghash.c +++ b/lib/libalpm/pkghash.c @@ -132,8 +132,7 @@ static alpm_pkghash_t *rehash(alpm_pkghash_t *oldhash) newhash = _alpm_pkghash_create(newsize); if(newhash == NULL) { - /* creation of newhash failed, stick with old one... */ - return oldhash; + return NULL; } newhash->list = oldhash->list; @@ -156,23 +155,29 @@ static alpm_pkghash_t *rehash(alpm_pkghash_t *oldhash) return newhash; } -static alpm_pkghash_t *pkghash_add_pkg(alpm_pkghash_t *hash, alpm_pkg_t *pkg, +static alpm_pkghash_t *pkghash_add_pkg(alpm_pkghash_t **hashref, alpm_pkg_t *pkg, int sorted) { alpm_list_t *ptr; unsigned int position; + alpm_pkghash_t *hash; - if(pkg == NULL || hash == NULL) { - return hash; + if(pkg == NULL || hashref == NULL || *hashref == NULL) { + return NULL; } + hash = *hashref; if(hash->entries >= hash->limit) { - hash = rehash(hash); + if((hash = rehash(hash)) == NULL) { + /* resizing failed and there are no more open buckets */ + return NULL; + } + *hashref = hash; } position = get_hash_position(pkg->name_hash, hash); - MALLOC(ptr, sizeof(alpm_list_t), return hash); + MALLOC(ptr, sizeof(alpm_list_t), return NULL); ptr->data = pkg; ptr->prev = ptr; @@ -189,12 +194,12 @@ static alpm_pkghash_t *pkghash_add_pkg(alpm_pkghash_t *hash, alpm_pkg_t *pkg, return hash; } -alpm_pkghash_t *_alpm_pkghash_add(alpm_pkghash_t *hash, alpm_pkg_t *pkg) +alpm_pkghash_t *_alpm_pkghash_add(alpm_pkghash_t **hash, alpm_pkg_t *pkg) { return pkghash_add_pkg(hash, pkg, 0); } -alpm_pkghash_t *_alpm_pkghash_add_sorted(alpm_pkghash_t *hash, alpm_pkg_t *pkg) +alpm_pkghash_t *_alpm_pkghash_add_sorted(alpm_pkghash_t **hash, alpm_pkg_t *pkg) { return pkghash_add_pkg(hash, pkg, 1); } diff --git a/lib/libalpm/pkghash.h b/lib/libalpm/pkghash.h index dd163346..a5c02c19 100644 --- a/lib/libalpm/pkghash.h +++ b/lib/libalpm/pkghash.h @@ -49,8 +49,8 @@ typedef struct __alpm_pkghash_t alpm_pkghash_t; alpm_pkghash_t *_alpm_pkghash_create(unsigned int size); -alpm_pkghash_t *_alpm_pkghash_add(alpm_pkghash_t *hash, alpm_pkg_t *pkg); -alpm_pkghash_t *_alpm_pkghash_add_sorted(alpm_pkghash_t *hash, alpm_pkg_t *pkg); +alpm_pkghash_t *_alpm_pkghash_add(alpm_pkghash_t **hash, alpm_pkg_t *pkg); +alpm_pkghash_t *_alpm_pkghash_add_sorted(alpm_pkghash_t **hash, alpm_pkg_t *pkg); alpm_pkghash_t *_alpm_pkghash_remove(alpm_pkghash_t *hash, alpm_pkg_t *pkg, alpm_pkg_t **data); void _alpm_pkghash_free(alpm_pkghash_t *hash); -- 2.15.1

2 1