Bug 116992

Summary: fastest checksum algorithm not chosen for RAID 5
Product: [Fedora] Fedora Reporter: Ellen Shull <ellenshull>
Component: kernelAssignee: Arjan van de Ven <arjanv>
Status: CLOSED NOTABUG QA Contact: Brian Brock <bbrock>
Severity: medium Docs Contact:
Priority: medium    
Version: rawhide   
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2004-02-27 14:07:33 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
full dmesg none

Description Ellen Shull 2004-02-27 04:59:02 UTC 
Description of problem: 
 
from dmesg: 
 
raid5: measuring checksumming speed 
   8regs     :  2212.000 MB/sec 
   8regs_prefetch:  2020.000 MB/sec 
   32regs    :  1488.000 MB/sec 
   32regs_prefetch:  1388.000 MB/sec 
   pIII_sse  :  1600.000 MB/sec 
   pII_mmx   :  3964.000 MB/sec 
   p5_mmx    :  5272.000 MB/sec 
raid5: using function: pIII_sse (1600.000 MB/sec) 
 
Version-Release number of selected component (if applicable): 
kernel-2.6.1-1.65.i686 
kernel-2.6.3-1.109.i686 
(probably all kernels between them too, didn't install those) 
 
How reproducible: 
always on my system (haven't tried on others) 
 
Steps to Reproduce: 
1. Boot with an affected kernel on a system with a Linux software 
RAID 5 volume (duh!) 
   
Actual results: 
says it's using the pIII_sse function 
 
Expected results: 
It should use the p5_mmx function, which is listed as fastest 
 
Etc: 
Of course, I'm trusting the kernel isn't lying.  Also, could be 
other factors that are leading it to choose the slower function...  
if so, would be nice if it would say so. 
 
For all I know, this could have affected older (FC1, RH9, ...) 
release kernels too...  my logs don't go back far enough for me to 
check.
Comment 1 Ellen Shull 2004-02-27 05:00:24 UTC 
Created attachment 98095 [details]
full dmesg
Comment 2 Dave Jones 2004-02-27 14:07:33 UTC 
This isn't a bug. It's intended to work this way.
It chooses the SSE routine over the others if it can use it, because
it avoids trashing whats in the cache.