Some make installcheck results 2.6.18-92.1.22.el5/0.9/0.140 on i686 commit b41a544e

[Mark Wielaard <mjw at redhat dot com>]

Another summary, now for another 2.6.18 kernel, but on i686, the
previous one was x86_64. This one also uses elfutils 0.140 (from
source), the other 0.137 as packaged.

Host: Linux localhost.localdomain 2.6.18-92.1.22.el5 #1 SMP Tue Dec 16 12:03:43 EST 2008 i686 i686 i386 GNU/Linux
Snapshot: version 0.9/0.140 commit b41a544e
Distro: CentOS release 5.2 (Final)

The results are very similar as with the x86_64 one.
With some things sticking out as i686 specific.

# of expected passes            735
# of unexpected failures        13
# of unexpected successes       8
# of expected failures          202
# of unknown successes          4
# of known failures             10
# of untested testcases         24

FAIL: backtrace (1 0)
FAIL: backtrace-unwindsyms (1 0)

Always fails.
Tracked as PR6961.

FAIL: gtod (265)

So, I am seeing this on i686, but not on x86_64 independent of kernel it
seems.

FAIL: itrace1 shutdown (eof)
FAIL: itrace2 shutdown (eof)

Only work on newer (fedora) kernels.
Tracked as RHBZ 463333.

FAIL: sdt additional_flags=-O2
FAIL: sdt c++ additional_flags=-O2
FAIL: sdt additional_flags="-O3"
FAIL: sdt c++ additional_flags="-O3"

Fails with the older gcc 4.2.1, works with newer gcc versions.
Tracked as PR9956.

FAIL: systemtap.base/stmt_rel.stp

Hmmm, weird. The log file says:
line 3: expected "PASS wildcard"
Got "FAIL wildcard 3"

But running it by hand seems to run fine:
PASS bio_init
PASS line number
PASS wildcard

FAIL: systemtap.base/uprobes_lib.stp

Double calling of uprobe probes. New test.
Tracked as PR9940

FAIL: buildok/seventeen.stp
FAIL: buildok/seventeen.stp with server

"integer constant is too large"
Tracked as PR10000