Bug 4121

Summary:	conversion test fails on s390x
Product:	systemtap	Reporter:	David Wilder <dwilder>
Component:	testsuite	Assignee:	Frank Ch. Eigler <fche>
Status:	RESOLVED FIXED
Severity:	normal
Priority:	P2
Version:	unspecified
Target Milestone:	---
Host:		Target:
Build:		Last reconfirmed:

Description David Wilder 2007-03-01 19:21:34 UTC

The test testsuite/systemtap.stress/conversions.exp failes on s390x.  The test
expects kernel_*(0) to report an error. On s390x page 0 is a valid mapped page
and readable therefore the references to address 0 is valid and no error is
reported.

Comment 1 Frank Ch. Eigler 2007-03-02 02:54:26 UTC

Please check whether the patch just committed against
testsuite/.../conversions.exp fixes the problem.

BTW, if you could suggest another null-region pointer value that is known to be
invalid on the s390*, that'd be easy to add instead of 0.

Comment 2 David Wilder 2007-03-02 17:05:40 UTC

I think you wanted:

# PR 4121: address 0 is valid on s390x
if {[istarget s390x-*-*] && $value == 0 } { continue }

With out the test for $value==0 the test wont do anything :)

BTW: Address 0x7777777777 is the lowest address I could find that would fault.
I may have some lower pages unmapped, but I dont know what they are, if so they
may move around between versions.

Comment 3 Frank Ch. Eigler 2007-03-02 20:22:33 UTC

Er yeah, sorry; try it now.

Comment 4 David Wilder 2007-03-05 17:07:40 UTC

I have not tested the new patch yet.  But it looks good.  I will give it a go
when I do this weeks snapshot test run.  Thanks.

Comment 5 Frank Ch. Eigler 2007-08-21 19:07:37 UTC

closed due to inactivity, assumed working

Comment 6 Bruno Abinader 2008-03-03 18:09:45 UTC

Test testsuite/.../conversions.exp currently fails on arm target with this output:

FAIL: conversions.stp 0xffffffff (6)
FAIL: conversions.stp 0xffffffffffffffff (6)

This test depends on architecture-specific memory addressing conventions. There
is no special treatment for ARM in this case.

Comment 7 Frank Ch. Eigler 2008-03-03 18:24:12 UTC

Let's please open a new bug for ARM.  If you have access, please suggest some
ARM-specific known-bad address ranges for this test to use.