Arch Linux User Repository

I've been getting this error since 5.62-1:

checking for dbus-1 >= 1.6... no
configure: error: D-Bus >= 1.6 is required
==> ERROR: A failure occurred in build().
    Aborting...
checking dependencies...

I checked my dbus version and it's the latest version, 1.12.20-1.

An ubuntu thread stated that libdbus-1-dev libudev-dev libical-dev libreadline-dev need to be installed, but I don't see an equivalent in the Arch repo or the AUR. I double checked the dependencies listed here and I don't see anything missing.

@Galaxy: Arch Linux is using systemd and the bluez systemd integration is needed.

If you need to disable it on Artix, you can add the flag locally on your machine (you can commit the change in your local repository; it will be kept when using "git pull". That's also working with most AUR helpers as they usually do a git pull)

Would you disable systemd so that it can be built in Artix Linux ?

@@ -38,6 +38,7 @@ build() {
           --enable-mesh \
           --enable-experimental \
           --enable-library \
+          --disable-systemd \
           --enable-deprecated # to enable deprectated tools
   make

hcitool lescan → Set scan parameters failed: Input/output error

When trying to use hcitool lescan, i can only get "Could not create connection: Input/output error"

What's going on here?

Note:

Package is updated and the below information is no longer relevant.

Building fails:

==> ERROR: Failure while downloading https://git.archlinux.org/svntogit/packages.git/plain/trunk/refresh_adv_manager_for_non-LE_devices.diff?h=packages/bluez
    Aborting...

The cause for this is that the diff file is not included in the bluez package anymore.

Workaround

In PKGBUILD, replace the URL for refresh_adv_manager_for_non-LE_devices.diff (the last value of source) with:

https://mirror.slackbuilds.org/slackware/slackware-current/source/n/bluez/refresh_adv_manager_for_non-LE_devices.diff.gz

Replace the last value of sha512sums with:

7e633031175298c4711f49843a6e3e2eb4ad93e49c200d5401aec159f36ae7a0f2172be7935f0f1dc88780b7aae22b8c08de61cfc26001c4e4579811aa8be451

makepkg will extract the diff file automatically, so do not worry about that.

Build and install the package as normal.

Suggestion to package maintainer

Include the diff file in the AUR package.

==> PGP keys need importing: -> E932D120BC2AEC444E558F0106CA9F5D1DCF2659, required by: bluez-utils-compat ==> Import? [Y/n] Y :: Importing keys with gpg... gpg: keyserver receive failed: General error problem importing keys

If there's an issue retrieving the key, you can always skip the pgp check by passing --skippgpcheck to makepkg when building.

I'm having the same issue.

bluez-5.54.tar ... FAILED (unknown public key 06CA9F5D1DCF2659) ==> ERROR: One or more PGP signatures could not be verified!

And then when I input: gpg --recv-keys 06CA9F5D1DCF2659

It throws: gpg: keyserver receive failed: General error

I got a "Cambridge Silicon Radio, Ltd Bluetooth Dongle (HCI mode)" that is not working, the only place I could make it to work was under Ubuntu thanks to this > https://github.com/jeremyb31/newbtfix-4.15. I tried as well here in Arch but it won't work. Then I came here just to try https://bugzilla.kernel.org/show_bug.cgi?id=60824, I couldn't install this package yet.

@Marc.2377 is correct. Run 'gpg --recv-keys 06CA9F5D1DCF2659' as the same user running yaourt.

@twa022 Do NOT use the sudo pacman-key solution suggested below. That will install it to the wrong (root's) gpg keyring.

@maid @CSYates98: see https://wiki.archlinux.org/index.php/Pacman/Package_signing#Adding_unofficial_keys

@Marc.2377 I am experiencing the same issue. When I first came accross the issue I tried running the command only to get the following as the result.

gpg: key 06CA9F5D1DCF2659: 18 signatures not checked due to missing keys

gpg: key 06CA9F5D1DCF2659: "Marcel Holtmann marcel@holtmann.org" not changed

gpg: Total number processed: 1

gpg: unchanged: 1

@maid: run 'gpg --recv-keys 06CA9F5D1DCF2659'.

key 06CA9F5D1DCF2659 not available!

@parkerlreed: I updated the 5.52 PKGBUILD to install the gatttool binary. Hope that helps.

please update to 5.52

Anybody found a way to reenable gatttool? Seems it was even removed from compat

Still Unable to find a built tarball for bluez-utils-compat

"Skip the test cases" doesn't sound like a good workaronud. Most likely there is a reason why they fail. Is there an upstream bug report about this?

@KoaTravis You were right. It worked, but wouldn't it be better to see what these tests are about?

@KoaTravis, removing the check() part worked for me as well, but this should really be fixed to allow installation without manual intervention. "yay" users: use "yay -Syua --editmenu" to allow editing of PKGBUILD before installation attempts.

i try to compile on raspberry zero as a dependency for pi-bluetooth, this comes up: Unable to find a built tarball for bluez-utils-compat

@thisischrys I also believe there should be no manual intervention needed. It is just inconvenient.

@TheSaint I had the same problem. After removing the check function from the PKGBUILD I was able to successfully build and install it.

I can finish the installation, the unittest are failing ./test-driver: line 107: 14099 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-sdp PASS: unit/test-gobex-apparam ./test-driver: line 107: 14119 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-avrcp ./test-driver: line 107: 14098 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-avctp PASS: unit/test-lib PASS: unit/test-gattrib PASS: unit/test-midi PASS: unit/test-gobex ./test-driver: line 107: 14212 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-gatt ./test-driver: line 107: 14229 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-hog =============================================================================== Then it takes a long time and doesn't quit. I have to interrupt it manually.

It is not only the signal.h. The api in ell is also changed https://git.kernel.org/pub/scm/libs/ell/ell.git/commit/ell/signal.h?id=4a386a10d88ff52097a974167ab889b3fc2add70

also https://github.com/chrippa/ds4drv/issues/123

--nocheck is not a solution. Fix the PKGBUILD please.

Having problem to compile tools/btpclient.c: In function ‘signal_handler’: tools/btpclient.c:2834:7: error: ‘SIGINT’ undeclared (first use in this function) 2834 | case SIGINT: | ^~~~~~ tools/btpclient.c:2834:7: note: each undeclared identifier is reported only once for each function it appears in tools/btpclient.c:2835:7: error: ‘SIGTERM’ undeclared (first use in this function) 2835 | case SIGTERM: | ^~~~~~~ tools/btpclient.c: In function ‘main’: tools/btpclient.c:3197:2: warning: implicit declaration of function ‘sigemptyset’ [-Wimplicit-function-declaration] 3197 | sigemptyset(&mask); | ^~~~~~~~~~~ tools/btpclient.c:3198:2: warning: implicit declaration of function ‘sigaddset’ [-Wimplicit-function-declaration] 3198 | sigaddset(&mask, SIGINT); | ^~~~~~~~~ tools/btpclient.c:3198:19: error: ‘SIGINT’ undeclared (first use in this function) 3198 | sigaddset(&mask, SIGINT); | ^~~~~~ tools/btpclient.c:3199:19: error: ‘SIGTERM’ undeclared (first use in this function) 3199 | sigaddset(&mask, SIGTERM); | ^~~~~~~ tools/btpclient.c:3200:27: warning: passing argument 1 of ‘l_signal_create’ makes integer from pointer without a cast [-Wint-conversion] 3200 | signal = l_signal_create(&mask, signal_handler, NULL, NULL); | ^~~~~ | | | sigset_t * {aka struct <anonymous> } In file included from /usr/include/ell/ell.h:32, from tools/btpclient.c:33: /usr/include/ell/signal.h:37:18: note: expected ‘uint32_t’ {aka ‘unsigned int’} but argument is of type ‘sigset_t ’ {aka ‘struct <anonymous> ’} 37 | struct l_signal l_signal_create(uint32_t signo, l_signal_notify_cb_t callback, | ^~~~~~~~~~~~~~~ tools/btpclient.c:3200:34: warning: passing argument 2 of ‘l_signal_create’ from incompatible pointer type [-Wincompatible-pointer-types] 3200 | signal = l_signal_create(&mask, signal_handler, NULL, NULL); | ^~~~~~~~~~~~~~ | | | void ()(struct l_signal , uint32_t, void ) {aka void ()(struct l_signal , unsigned int, void )} In file included from /usr/include/ell/ell.h:32, from tools/btpclient.c:33: /usr/include/ell/signal.h:37:18: note: expected ‘l_signal_notify_cb_t’ {aka ‘void ()(void )’} but argument is of type ‘void ()(struct l_signal , uint32_t, void )’ {aka ‘void ()(struct l_signal , unsigned int, void )’} 37 | struct l_signal l_signal_create(uint32_t signo, l_signal_notify_cb_t callback, | ^~~~~~~~~~~~~~~ make[1]: [Makefile:8316: tools/tools_btpclient-btpclient.o] Error 1 make[1]: Attesa per i processi non terminati.... make: ** [Makefile:3278: all] Error 2

As bruceutut said, the following patch fixes it by including signal.h:

--- bluez-5.50/tools/btpclient.c 2018-03-10 15:37:07.000000000 +0000

+++ bluez-5.50.patched/tools/btpclient.c 2019-06-28 15:28:05.821569524 +0000

@@ -29,6 +29,7 @@

#include <stdlib.h>

#include <assert.h>

#include <getopt.h>

+#include <signal.h>

#include <ell/ell.h>

what worked for me was dropping a line with --enable-btpclient and running makepkg with --nocheck

tools/btpclient.c: In function ‘signal_handler’:

tools/btpclient.c:2834:7: error: ‘SIGINT’ undeclared (first use in this function)

tools/btpclient.c:2835:7: error: ‘SIGTERM’ undeclared (first use in this function)

SIGTERM SIGINT cannot be found when building this package with gcc9.

Maybe need to include signal.h can solve this problem.

If you are having errors building because of the tests you can skip running the check() function in the PKGBUILD: makepkg -s --nocheck

This won't compile for me, failing at the same place of the last comment on may 24th.

PASS: unit/test-ecc ./test-driver: line 107: 21893 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-avctp ./test-driver: line 107: 21946 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-avrcp ./test-driver: line 107: 21948 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-hog CCLD unit/test-sdp ./test-driver: line 107: 22040 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-sdp PASS: unit/test-gobex-transfer CCLD unit/test-gatt ./test-driver: line 107: 22107 Aborted (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-gatt ^Cmake[3]: Deleting file 'unit/test-avdtp.log' make[3]: Deleting file 'unit/test-hfp.log' make[3]: [Makefile:8979: unit/test-avdtp.log] Error 130 make[3]: [Makefile:9000: unit/test-hfp.log] Error 130 make[2]: [Makefile:8877: check-TESTS] Interrupt make[1]: [Makefile:9261: check-am] Interrupt make: *** [Makefile:9263: check] Interrupt

Patroll: That's not an error, you need to install that key.

problem with verification signature with gpg: bluez-5.50.tar ... FAILED (unknown public key 06CA9F5D1DCF2659)

fail to start after readline 8.0
plz trigger a rebuild.

@yaron when it hangs, you can do a ps and grep for ./unit/test-gatt

Killing that process allows the build to continue. Oddly, it told me that the test passed.

Maybe this package should provide bluez-hcitool, as they both provide '/usr/bin/hcitool' and '/usr/share/man/man1/hcitool.1.gz'. Or both should provide hcitool.

Checksum is failing on refresh_adv_manager_for_non-LE_devices.diff, maybe it should come from https://git.archlinux.org/svntogit/packages.git/plain/trunk/refresh_adv_manager_for_non-LE_devices.diff?h=packages/bluez instead? (Note the "plain" instead of "tree".) It seems like the diff hasn't changed in that package since April, but it wouldn't surprise me if the HTML page representing it has.

@yaron you can just skip building the tests if you don't care about running them or they're not passing. Pass --nocheck to makepkg or if you globally want to disable building tests edit your makepkg.conf

Same issue yet again:

PASS: unit/test-lib

CC unit/test-gatt.o

CCLD unit/test-gatt

I'm out of ideas.

I have a major problem while running unit tests on this module.

The last message I see is: PASS: unit/test-lib CC unit/test-gatt.o CCLD unit/test-gatt

After that nothing happens.

can add armv6h to the architecture list as well. Package builds without error on the Raspberry Pi (armv6).

@nicolasvila: I have no way to trouble shoot the tests. Just pass --nocheck to makepkg when you build to skip the check function

Thanks for the update but the build fails during the test "unit/test-gatt" as you can see on the log below:

[...] PASS: unit/test-avdtp PASS: unit/test-avrcp ./test-driver: line 107: 1421 Segmentation fault (core dumped) "$@" > $log_file 2>&1 FAIL: unit/test-gatt make --no-print-directory all-am =================================== Testsuite summary for bluez 5.47 ===================================

TOTAL: 25

PASS: 24

SKIP: 0

XFAIL: 0

FAIL: 1

XPASS: 0

ERROR: 0

=================================== See ./test-suite.log =================================== make[3]: [Makefile:8597: test-suite.log] Error 1 make[2]: [Makefile:8705: check-TESTS] Error 2 make[1]: [Makefile:9089: check-am] Error 2 make: [Makefile:9091: check] Error 2 ==> ERROR: A failure occurred in check(). Aborting... ==> ERROR: Makepkg was unable to build bluez-utils-compat. ==> Restart building bluez-utils-compat ? [y/N] ==> ------------------------------------------- ==> ==> ERROR: unable to update

I'm unable to continue supporting this package. I have released it so someone can feel free to pick it up if it's still important enough.

Package conflicts with: bluez-hcitool