Bug 180477 - MEMORY2 doesn't test available RAM on 32-bit kernels
Summary: MEMORY2 doesn't test available RAM on 32-bit kernels
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Red Hat Ready Certification Tests
Classification: Retired
Component: memory
Version: 2
Hardware: All
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Will Woods
QA Contact: Rob Landry
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-02-08 14:10 UTC by Rob Landry
Modified: 2007-04-18 17:37 UTC (History)
2 users (show)

Fixed In Version: RHBA-2006-0278
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2006-05-08 13:39:28 UTC
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Red Hat Product Errata RHBA-2006:0278 0 normal SHIPPED_LIVE rhr2 bug fix update 2006-05-08 04:00:00 UTC
Red Hat Product Errata RHBA-2006:0284 0 normal SHIPPED_LIVE rhr2 bug fix update 2006-05-08 04:00:00 UTC

Description Rob Landry 2006-02-08 14:10:26 UTC
Description of problem:

$Summary as a result of memsize trying to malloc >3GB on 32-bit kernels

Steps to Reproduce:
1.find a 32-bit kernel with >~3GB of RAM
2.run memsize 3000
  
Actual results:

<snip>
$lmbench_dir/memsize $mem
++ /usr/lib/lmbench/bin/i686-pc-linux-gnu/memsize 3287
malloc: Cannot allocate memory
</snip>

the side effect causes this series of events; which nets us testing 500MB on a
4GB+ box:

<snip>
+ '[' -z '' ']'
+ echo 'WARNING: memsize failed! Using 80% of reported free RAM'
WARNING: memsize failed! Using 80% of reported free RAM
awk '/^MemFree:/ {print $2}' /proc/meminfo
++ awk '/^MemFree:/ {print $2}' /proc/meminfo
+ local freemem=1441
+ avail=1152
+ echo '1152MB available'
1152MB available
+ local MB=576
</snip>

Expected results:

Full testing of available ram; it would of been nice if memsize would fork to
properly flush the buffers cache's etc.; however the alternative is to make
MEMORY2 smarter and have it allocate no larger than 2.8 or so GB per process and
flush RAM out that way; at least on 32-bit kernels; another possibility might be
to add free + buffers + cache; minus some gap amount and try to test that; if we
can't flush the buffers and cache's at least mostly then something went wrong I
would guess.

Comment 1 Will Woods 2006-02-14 22:01:59 UTC
A patch is in CVS that makes MEMORY2 use multiple test processes on x86 machines
with large amounts of RAM.

Comment 4 Red Hat Bugzilla 2006-05-08 13:39:13 UTC
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2006-0284.html


Comment 5 Red Hat Bugzilla 2006-05-08 13:39:28 UTC
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2006-0278.html



Note You need to log in before you can comment on or make changes to this bug.