Bug 1246155

This is quite odd; the error is from:

	stream = fdopen (fd, "r");
	if (stream == NULL) {
		g_set_error (error,
			     CD_ICC_ERROR,
			     CD_ICC_ERROR_FAILED_TO_OPEN,
			     "failed to open stream from fd %i",
			     fd);
		return FALSE;
	}

is there any way you can gdb the cd-test-daemon executable and find out the errno? That would help this along a huge amount.

For the other failure, I think this is a failure of the lcms2 library on PPC64:

	priv->lcms_profile = cmsOpenProfileFromMemTHR (priv->context_lcms,
						       data, data_len);
	if (priv->lcms_profile == NULL) {
		g_set_error_literal (error,
				     CD_ICC_ERROR,
				     CD_ICC_ERROR_FAILED_TO_PARSE,
				     "failed to load: not an ICC icc");
		return FALSE;
	}

i.e. i think there's some kind of endian issue when we're opening the file in lcms2.

(In reply to Richard Hughes from comment #1)
> is there any way you can gdb the cd-test-daemon executable and find out the
> errno? That would help this along a huge amount.

Sure, will do.

> i.e. i think there's some kind of endian issue when we're opening the file
> in lcms2.

Right, I think https://desktopqe-jenkins.rhev-ci-vms.eng.rdu2.redhat.com/job/desktop-lcms2-RHEL-7.2-2-runtest/26/artifact/testresults-FAKE/lcms2tests.log is not happy about endianess - I'll file a new bug on this and link it here

I can see this on x86_64 too

*** Bug 1224743 has been marked as a duplicate of this bug. ***

Let first fix the lcms2 issue (bug 1250914) and see if this helps

Probably still blocked by bug 1250914) which is not fixed yet.

So with the new lcms2 I got a bit different output:

=============================================
   colord 1.2.7: lib/colord/test-suite.log
=============================================

# TOTAL: 2
# PASS:  1
# SKIP:  0
# XFAIL: 0
# FAIL:  1
# XPASS: 0
# ERROR: 0

.. contents:: :depth: 2

FAIL: cd-test-daemon
====================

/colord/client: OK
/colord/device: OK
/colord/device{embedded}: OK
/colord/device{invalid-kind}: OK
/colord/device{duplicate}: OK
/colord/device{seat}: OK
/colord/device{enabled}: OK
/colord/device{invalid}: OK
/colord/device{qualifiers}: **
libcolord:ERROR:cd-test-daemon.c:165:colord_device_qualifiers_func: assertion failed (error == NULL): failed to load file: Error opening file: Permission denied (cd_client_error, 7)

Note that the test is executed as root, is it correct?

(In reply to Tomas Pelka from comment #8)
> So with the new lcms2 I got a bit different output:
> Note that the test is executed as root, is it correct?

No, you need to run the tests as a normal user. I don't think running the daemon tests (this actually interacts with a running colord process, rather than doing just self tests) in the rpm file is a good idea.

Can I close this as the daemon checks are nothing to do with ppc64le, and with the updated lcms2 package installed the self tests on the profile now pass?

No answer... I would say yes, lets close this.

*** Bug 1171718 has been marked as a duplicate of this bug. ***