From: Li Z. <li...@cn...> - 2009-01-08 09:27:44
|
Subrata Modak wrote: > On Wed, 2009-01-07 at 09:53 +0800, Li Zefan wrote: >> Subrata Modak wrote: >>> Not sure whether everybody is back from vacation. But, this data is very >>> important for the project. Hope you will help me provide some inputs. >>> Some of you has already given me. Thanks for that. >>> >> A thought: >> >> We have many bug reports in LKML and the bugzilla, and some of them >> provided programs or steps to reproduce bugs. It is of benefit if >> we can make test programs and merge them into LTP based on those >> provided by bug reporters. Those test cases help to test corner >> cases. >> >> There is an excellent program to reliably reproduce a bug in direct-io, >> and the bug still remains to be resovled. Andrea Arcangeli <aar...@re...> >> sugguested it is probably worth adding to LTP, and so do I. :) >> >> The bug report and discussion can be found here: >> http://marc.info/?l=linux-mm&m=122668235304637&w=2 >> >> And the test program is attached. We have to ask Tim LaBerge <tim...@qu...> >> if we can use it in LTP. > > Yes, Definitely. I will write a patch to make it LTP specific and you > can review that before integrating. > You can write the test script based on the below one: #! /bin/sh # generate test files ./dma_thread # test different alignments: 512, 1024, ..., (4096-512) for ((i = 512; i < 4096; i += 512)) { ./dma_thread -a $i if [ $? -ne 0]; then echo "Test Failed !!" exit 1 fi } |